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 void 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;
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;
2390                 } else if (line[0] == '+') {
2391                         diff_words_append(line, len,
2392                                           &ecbdata->diff_words->plus);
2393                         return;
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;
2403                 }
2404                 diff_words_flush(ecbdata);
2405                 emit_diff_symbol(o, s, line, len, 0);
2406                 return;
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 }
2431
2432 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2433 {
2434         const char *old_name = a;
2435         const char *new_name = b;
2436         int pfx_length, sfx_length;
2437         int pfx_adjust_for_slash;
2438         int len_a = strlen(a);
2439         int len_b = strlen(b);
2440         int a_midlen, b_midlen;
2441         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2442         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2443
2444         if (qlen_a || qlen_b) {
2445                 quote_c_style(a, name, NULL, 0);
2446                 strbuf_addstr(name, " => ");
2447                 quote_c_style(b, name, NULL, 0);
2448                 return;
2449         }
2450
2451         /* Find common prefix */
2452         pfx_length = 0;
2453         while (*old_name && *new_name && *old_name == *new_name) {
2454                 if (*old_name == '/')
2455                         pfx_length = old_name - a + 1;
2456                 old_name++;
2457                 new_name++;
2458         }
2459
2460         /* Find common suffix */
2461         old_name = a + len_a;
2462         new_name = b + len_b;
2463         sfx_length = 0;
2464         /*
2465          * If there is a common prefix, it must end in a slash.  In
2466          * that case we let this loop run 1 into the prefix to see the
2467          * same slash.
2468          *
2469          * If there is no common prefix, we cannot do this as it would
2470          * underrun the input strings.
2471          */
2472         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2473         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2474                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2475                *old_name == *new_name) {
2476                 if (*old_name == '/')
2477                         sfx_length = len_a - (old_name - a);
2478                 old_name--;
2479                 new_name--;
2480         }
2481
2482         /*
2483          * pfx{mid-a => mid-b}sfx
2484          * {pfx-a => pfx-b}sfx
2485          * pfx{sfx-a => sfx-b}
2486          * name-a => name-b
2487          */
2488         a_midlen = len_a - pfx_length - sfx_length;
2489         b_midlen = len_b - pfx_length - sfx_length;
2490         if (a_midlen < 0)
2491                 a_midlen = 0;
2492         if (b_midlen < 0)
2493                 b_midlen = 0;
2494
2495         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2496         if (pfx_length + sfx_length) {
2497                 strbuf_add(name, a, pfx_length);
2498                 strbuf_addch(name, '{');
2499         }
2500         strbuf_add(name, a + pfx_length, a_midlen);
2501         strbuf_addstr(name, " => ");
2502         strbuf_add(name, b + pfx_length, b_midlen);
2503         if (pfx_length + sfx_length) {
2504                 strbuf_addch(name, '}');
2505                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2506         }
2507 }
2508
2509 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2510                                           const char *name_a,
2511                                           const char *name_b)
2512 {
2513         struct diffstat_file *x;
2514         CALLOC_ARRAY(x, 1);
2515         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2516         diffstat->files[diffstat->nr++] = x;
2517         if (name_b) {
2518                 x->from_name = xstrdup(name_a);
2519                 x->name = xstrdup(name_b);
2520                 x->is_renamed = 1;
2521         }
2522         else {
2523                 x->from_name = NULL;
2524                 x->name = xstrdup(name_a);
2525         }
2526         return x;
2527 }
2528
2529 static void diffstat_consume(void *priv, char *line, unsigned long len)
2530 {
2531         struct diffstat_t *diffstat = priv;
2532         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2533
2534         if (line[0] == '+')
2535                 x->added++;
2536         else if (line[0] == '-')
2537                 x->deleted++;
2538 }
2539
2540 const char mime_boundary_leader[] = "------------";
2541
2542 static int scale_linear(int it, int width, int max_change)
2543 {
2544         if (!it)
2545                 return 0;
2546         /*
2547          * make sure that at least one '-' or '+' is printed if
2548          * there is any change to this path. The easiest way is to
2549          * scale linearly as if the allotted width is one column shorter
2550          * than it is, and then add 1 to the result.
2551          */
2552         return 1 + (it * (width - 1) / max_change);
2553 }
2554
2555 static void show_graph(struct strbuf *out, char ch, int cnt,
2556                        const char *set, const char *reset)
2557 {
2558         if (cnt <= 0)
2559                 return;
2560         strbuf_addstr(out, set);
2561         strbuf_addchars(out, ch, cnt);
2562         strbuf_addstr(out, reset);
2563 }
2564
2565 static void fill_print_name(struct diffstat_file *file)
2566 {
2567         struct strbuf pname = STRBUF_INIT;
2568
2569         if (file->print_name)
2570                 return;
2571
2572         if (file->is_renamed)
2573                 pprint_rename(&pname, file->from_name, file->name);
2574         else
2575                 quote_c_style(file->name, &pname, NULL, 0);
2576
2577         if (file->comments)
2578                 strbuf_addf(&pname, " (%s)", file->comments);
2579
2580         file->print_name = strbuf_detach(&pname, NULL);
2581 }
2582
2583 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2584                 int files, int insertions, int deletions)
2585 {
2586         struct strbuf sb = STRBUF_INIT;
2587
2588         if (!files) {
2589                 assert(insertions == 0 && deletions == 0);
2590                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2591                                  NULL, 0, 0);
2592                 return;
2593         }
2594
2595         strbuf_addf(&sb,
2596                     (files == 1) ? " %d file changed" : " %d files changed",
2597                     files);
2598
2599         /*
2600          * For binary diff, the caller may want to print "x files
2601          * changed" with insertions == 0 && deletions == 0.
2602          *
2603          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2604          * is probably less confusing (i.e skip over "2 files changed
2605          * but nothing about added/removed lines? Is this a bug in Git?").
2606          */
2607         if (insertions || deletions == 0) {
2608                 strbuf_addf(&sb,
2609                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2610                             insertions);
2611         }
2612
2613         if (deletions || insertions == 0) {
2614                 strbuf_addf(&sb,
2615                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2616                             deletions);
2617         }
2618         strbuf_addch(&sb, '\n');
2619         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2620                          sb.buf, sb.len, 0);
2621         strbuf_release(&sb);
2622 }
2623
2624 void print_stat_summary(FILE *fp, int files,
2625                         int insertions, int deletions)
2626 {
2627         struct diff_options o;
2628         memset(&o, 0, sizeof(o));
2629         o.file = fp;
2630
2631         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2632 }
2633
2634 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2635 {
2636         int i, len, add, del, adds = 0, dels = 0;
2637         uintmax_t max_change = 0, max_len = 0;
2638         int total_files = data->nr, count;
2639         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2640         const char *reset, *add_c, *del_c;
2641         int extra_shown = 0;
2642         const char *line_prefix = diff_line_prefix(options);
2643         struct strbuf out = STRBUF_INIT;
2644
2645         if (data->nr == 0)
2646                 return;
2647
2648         count = options->stat_count ? options->stat_count : data->nr;
2649
2650         reset = diff_get_color_opt(options, DIFF_RESET);
2651         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2652         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2653
2654         /*
2655          * Find the longest filename and max number of changes
2656          */
2657         for (i = 0; (i < count) && (i < data->nr); i++) {
2658                 struct diffstat_file *file = data->files[i];
2659                 uintmax_t change = file->added + file->deleted;
2660
2661                 if (!file->is_interesting && (change == 0)) {
2662                         count++; /* not shown == room for one more */
2663                         continue;
2664                 }
2665                 fill_print_name(file);
2666                 len = strlen(file->print_name);
2667                 if (max_len < len)
2668                         max_len = len;
2669
2670                 if (file->is_unmerged) {
2671                         /* "Unmerged" is 8 characters */
2672                         bin_width = bin_width < 8 ? 8 : bin_width;
2673                         continue;
2674                 }
2675                 if (file->is_binary) {
2676                         /* "Bin XXX -> YYY bytes" */
2677                         int w = 14 + decimal_width(file->added)
2678                                 + decimal_width(file->deleted);
2679                         bin_width = bin_width < w ? w : bin_width;
2680                         /* Display change counts aligned with "Bin" */
2681                         number_width = 3;
2682                         continue;
2683                 }
2684
2685                 if (max_change < change)
2686                         max_change = change;
2687         }
2688         count = i; /* where we can stop scanning in data->files[] */
2689
2690         /*
2691          * We have width = stat_width or term_columns() columns total.
2692          * We want a maximum of min(max_len, stat_name_width) for the name part.
2693          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2694          * We also need 1 for " " and 4 + decimal_width(max_change)
2695          * for " | NNNN " and one the empty column at the end, altogether
2696          * 6 + decimal_width(max_change).
2697          *
2698          * If there's not enough space, we will use the smaller of
2699          * stat_name_width (if set) and 5/8*width for the filename,
2700          * and the rest for constant elements + graph part, but no more
2701          * than stat_graph_width for the graph part.
2702          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2703          * for the standard terminal size).
2704          *
2705          * In other words: stat_width limits the maximum width, and
2706          * stat_name_width fixes the maximum width of the filename,
2707          * and is also used to divide available columns if there
2708          * aren't enough.
2709          *
2710          * Binary files are displayed with "Bin XXX -> YYY bytes"
2711          * instead of the change count and graph. This part is treated
2712          * similarly to the graph part, except that it is not
2713          * "scaled". If total width is too small to accommodate the
2714          * guaranteed minimum width of the filename part and the
2715          * separators and this message, this message will "overflow"
2716          * making the line longer than the maximum width.
2717          */
2718
2719         if (options->stat_width == -1)
2720                 width = term_columns() - strlen(line_prefix);
2721         else
2722                 width = options->stat_width ? options->stat_width : 80;
2723         number_width = decimal_width(max_change) > number_width ?
2724                 decimal_width(max_change) : number_width;
2725
2726         if (options->stat_graph_width == -1)
2727                 options->stat_graph_width = diff_stat_graph_width;
2728
2729         /*
2730          * Guarantee 3/8*16==6 for the graph part
2731          * and 5/8*16==10 for the filename part
2732          */
2733         if (width < 16 + 6 + number_width)
2734                 width = 16 + 6 + number_width;
2735
2736         /*
2737          * First assign sizes that are wanted, ignoring available width.
2738          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2739          * starting from "XXX" should fit in graph_width.
2740          */
2741         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2742         if (options->stat_graph_width &&
2743             options->stat_graph_width < graph_width)
2744                 graph_width = options->stat_graph_width;
2745
2746         name_width = (options->stat_name_width > 0 &&
2747                       options->stat_name_width < max_len) ?
2748                 options->stat_name_width : max_len;
2749
2750         /*
2751          * Adjust adjustable widths not to exceed maximum width
2752          */
2753         if (name_width + number_width + 6 + graph_width > width) {
2754                 if (graph_width > width * 3/8 - number_width - 6) {
2755                         graph_width = width * 3/8 - number_width - 6;
2756                         if (graph_width < 6)
2757                                 graph_width = 6;
2758                 }
2759
2760                 if (options->stat_graph_width &&
2761                     graph_width > options->stat_graph_width)
2762                         graph_width = options->stat_graph_width;
2763                 if (name_width > width - number_width - 6 - graph_width)
2764                         name_width = width - number_width - 6 - graph_width;
2765                 else
2766                         graph_width = width - number_width - 6 - name_width;
2767         }
2768
2769         /*
2770          * From here name_width is the width of the name area,
2771          * and graph_width is the width of the graph area.
2772          * max_change is used to scale graph properly.
2773          */
2774         for (i = 0; i < count; i++) {
2775                 const char *prefix = "";
2776                 struct diffstat_file *file = data->files[i];
2777                 char *name = file->print_name;
2778                 uintmax_t added = file->added;
2779                 uintmax_t deleted = file->deleted;
2780                 int name_len;
2781
2782                 if (!file->is_interesting && (added + deleted == 0))
2783                         continue;
2784
2785                 /*
2786                  * "scale" the filename
2787                  */
2788                 len = name_width;
2789                 name_len = strlen(name);
2790                 if (name_width < name_len) {
2791                         char *slash;
2792                         prefix = "...";
2793                         len -= 3;
2794                         name += name_len - len;
2795                         slash = strchr(name, '/');
2796                         if (slash)
2797                                 name = slash;
2798                 }
2799
2800                 if (file->is_binary) {
2801                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2802                         strbuf_addf(&out, " %*s", number_width, "Bin");
2803                         if (!added && !deleted) {
2804                                 strbuf_addch(&out, '\n');
2805                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2806                                                  out.buf, out.len, 0);
2807                                 strbuf_reset(&out);
2808                                 continue;
2809                         }
2810                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2811                                 del_c, deleted, reset);
2812                         strbuf_addstr(&out, " -> ");
2813                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2814                                 add_c, added, reset);
2815                         strbuf_addstr(&out, " bytes\n");
2816                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2817                                          out.buf, out.len, 0);
2818                         strbuf_reset(&out);
2819                         continue;
2820                 }
2821                 else if (file->is_unmerged) {
2822                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2823                         strbuf_addstr(&out, " Unmerged\n");
2824                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2825                                          out.buf, out.len, 0);
2826                         strbuf_reset(&out);
2827                         continue;
2828                 }
2829
2830                 /*
2831                  * scale the add/delete
2832                  */
2833                 add = added;
2834                 del = deleted;
2835
2836                 if (graph_width <= max_change) {
2837                         int total = scale_linear(add + del, graph_width, max_change);
2838                         if (total < 2 && add && del)
2839                                 /* width >= 2 due to the sanity check */
2840                                 total = 2;
2841                         if (add < del) {
2842                                 add = scale_linear(add, graph_width, max_change);
2843                                 del = total - add;
2844                         } else {
2845                                 del = scale_linear(del, graph_width, max_change);
2846                                 add = total - del;
2847                         }
2848                 }
2849                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2850                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2851                         number_width, added + deleted,
2852                         added + deleted ? " " : "");
2853                 show_graph(&out, '+', add, add_c, reset);
2854                 show_graph(&out, '-', del, del_c, reset);
2855                 strbuf_addch(&out, '\n');
2856                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2857                                  out.buf, out.len, 0);
2858                 strbuf_reset(&out);
2859         }
2860
2861         for (i = 0; i < data->nr; i++) {
2862                 struct diffstat_file *file = data->files[i];
2863                 uintmax_t added = file->added;
2864                 uintmax_t deleted = file->deleted;
2865
2866                 if (file->is_unmerged ||
2867                     (!file->is_interesting && (added + deleted == 0))) {
2868                         total_files--;
2869                         continue;
2870                 }
2871
2872                 if (!file->is_binary) {
2873                         adds += added;
2874                         dels += deleted;
2875                 }
2876                 if (i < count)
2877                         continue;
2878                 if (!extra_shown)
2879                         emit_diff_symbol(options,
2880                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2881                                          NULL, 0, 0);
2882                 extra_shown = 1;
2883         }
2884
2885         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2886         strbuf_release(&out);
2887 }
2888
2889 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2890 {
2891         int i, adds = 0, dels = 0, total_files = data->nr;
2892
2893         if (data->nr == 0)
2894                 return;
2895
2896         for (i = 0; i < data->nr; i++) {
2897                 int added = data->files[i]->added;
2898                 int deleted = data->files[i]->deleted;
2899
2900                 if (data->files[i]->is_unmerged ||
2901                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2902                         total_files--;
2903                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2904                         adds += added;
2905                         dels += deleted;
2906                 }
2907         }
2908         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2909 }
2910
2911 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2912 {
2913         int i;
2914
2915         if (data->nr == 0)
2916                 return;
2917
2918         for (i = 0; i < data->nr; i++) {
2919                 struct diffstat_file *file = data->files[i];
2920
2921                 fprintf(options->file, "%s", diff_line_prefix(options));
2922
2923                 if (file->is_binary)
2924                         fprintf(options->file, "-\t-\t");
2925                 else
2926                         fprintf(options->file,
2927                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2928                                 file->added, file->deleted);
2929                 if (options->line_termination) {
2930                         fill_print_name(file);
2931                         if (!file->is_renamed)
2932                                 write_name_quoted(file->name, options->file,
2933                                                   options->line_termination);
2934                         else {
2935                                 fputs(file->print_name, options->file);
2936                                 putc(options->line_termination, options->file);
2937                         }
2938                 } else {
2939                         if (file->is_renamed) {
2940                                 putc('\0', options->file);
2941                                 write_name_quoted(file->from_name, options->file, '\0');
2942                         }
2943                         write_name_quoted(file->name, options->file, '\0');
2944                 }
2945         }
2946 }
2947
2948 struct dirstat_file {
2949         const char *name;
2950         unsigned long changed;
2951 };
2952
2953 struct dirstat_dir {
2954         struct dirstat_file *files;
2955         int alloc, nr, permille, cumulative;
2956 };
2957
2958 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2959                 unsigned long changed, const char *base, int baselen)
2960 {
2961         unsigned long sum_changes = 0;
2962         unsigned int sources = 0;
2963         const char *line_prefix = diff_line_prefix(opt);
2964
2965         while (dir->nr) {
2966                 struct dirstat_file *f = dir->files;
2967                 int namelen = strlen(f->name);
2968                 unsigned long changes;
2969                 char *slash;
2970
2971                 if (namelen < baselen)
2972                         break;
2973                 if (memcmp(f->name, base, baselen))
2974                         break;
2975                 slash = strchr(f->name + baselen, '/');
2976                 if (slash) {
2977                         int newbaselen = slash + 1 - f->name;
2978                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2979                         sources++;
2980                 } else {
2981                         changes = f->changed;
2982                         dir->files++;
2983                         dir->nr--;
2984                         sources += 2;
2985                 }
2986                 sum_changes += changes;
2987         }
2988
2989         /*
2990          * We don't report dirstat's for
2991          *  - the top level
2992          *  - or cases where everything came from a single directory
2993          *    under this directory (sources == 1).
2994          */
2995         if (baselen && sources != 1) {
2996                 if (sum_changes) {
2997                         int permille = sum_changes * 1000 / changed;
2998                         if (permille >= dir->permille) {
2999                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
3000                                         permille / 10, permille % 10, baselen, base);
3001                                 if (!dir->cumulative)
3002                                         return 0;
3003                         }
3004                 }
3005         }
3006         return sum_changes;
3007 }
3008
3009 static int dirstat_compare(const void *_a, const void *_b)
3010 {
3011         const struct dirstat_file *a = _a;
3012         const struct dirstat_file *b = _b;
3013         return strcmp(a->name, b->name);
3014 }
3015
3016 static void show_dirstat(struct diff_options *options)
3017 {
3018         int i;
3019         unsigned long changed;
3020         struct dirstat_dir dir;
3021         struct diff_queue_struct *q = &diff_queued_diff;
3022
3023         dir.files = NULL;
3024         dir.alloc = 0;
3025         dir.nr = 0;
3026         dir.permille = options->dirstat_permille;
3027         dir.cumulative = options->flags.dirstat_cumulative;
3028
3029         changed = 0;
3030         for (i = 0; i < q->nr; i++) {
3031                 struct diff_filepair *p = q->queue[i];
3032                 const char *name;
3033                 unsigned long copied, added, damage;
3034                 struct diff_populate_filespec_options dpf_options = {
3035                         .check_size_only = 1,
3036                 };
3037
3038                 name = p->two->path ? p->two->path : p->one->path;
3039
3040                 if (p->one->oid_valid && p->two->oid_valid &&
3041                     oideq(&p->one->oid, &p->two->oid)) {
3042                         /*
3043                          * The SHA1 has not changed, so pre-/post-content is
3044                          * identical. We can therefore skip looking at the
3045                          * file contents altogether.
3046                          */
3047                         damage = 0;
3048                         goto found_damage;
3049                 }
3050
3051                 if (options->flags.dirstat_by_file) {
3052                         /*
3053                          * In --dirstat-by-file mode, we don't really need to
3054                          * look at the actual file contents at all.
3055                          * The fact that the SHA1 changed is enough for us to
3056                          * add this file to the list of results
3057                          * (with each file contributing equal damage).
3058                          */
3059                         damage = 1;
3060                         goto found_damage;
3061                 }
3062
3063                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
3064                         diff_populate_filespec(options->repo, p->one, NULL);
3065                         diff_populate_filespec(options->repo, p->two, NULL);
3066                         diffcore_count_changes(options->repo,
3067                                                p->one, p->two, NULL, NULL,
3068                                                &copied, &added);
3069                         diff_free_filespec_data(p->one);
3070                         diff_free_filespec_data(p->two);
3071                 } else if (DIFF_FILE_VALID(p->one)) {
3072                         diff_populate_filespec(options->repo, p->one, &dpf_options);
3073                         copied = added = 0;
3074                         diff_free_filespec_data(p->one);
3075                 } else if (DIFF_FILE_VALID(p->two)) {
3076                         diff_populate_filespec(options->repo, p->two, &dpf_options);
3077                         copied = 0;
3078                         added = p->two->size;
3079                         diff_free_filespec_data(p->two);
3080                 } else
3081                         continue;
3082
3083                 /*
3084                  * Original minus copied is the removed material,
3085                  * added is the new material.  They are both damages
3086                  * made to the preimage.
3087                  * If the resulting damage is zero, we know that
3088                  * diffcore_count_changes() considers the two entries to
3089                  * be identical, but since the oid changed, we
3090                  * know that there must have been _some_ kind of change,
3091                  * so we force all entries to have damage > 0.
3092                  */
3093                 damage = (p->one->size - copied) + added;
3094                 if (!damage)
3095                         damage = 1;
3096
3097 found_damage:
3098                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3099                 dir.files[dir.nr].name = name;
3100                 dir.files[dir.nr].changed = damage;
3101                 changed += damage;
3102                 dir.nr++;
3103         }
3104
3105         /* This can happen even with many files, if everything was renames */
3106         if (!changed)
3107                 return;
3108
3109         /* Show all directories with more than x% of the changes */
3110         QSORT(dir.files, dir.nr, dirstat_compare);
3111         gather_dirstat(options, &dir, changed, "", 0);
3112 }
3113
3114 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3115 {
3116         int i;
3117         unsigned long changed;
3118         struct dirstat_dir dir;
3119
3120         if (data->nr == 0)
3121                 return;
3122
3123         dir.files = NULL;
3124         dir.alloc = 0;
3125         dir.nr = 0;
3126         dir.permille = options->dirstat_permille;
3127         dir.cumulative = options->flags.dirstat_cumulative;
3128
3129         changed = 0;
3130         for (i = 0; i < data->nr; i++) {
3131                 struct diffstat_file *file = data->files[i];
3132                 unsigned long damage = file->added + file->deleted;
3133                 if (file->is_binary)
3134                         /*
3135                          * binary files counts bytes, not lines. Must find some
3136                          * way to normalize binary bytes vs. textual lines.
3137                          * The following heuristic assumes that there are 64
3138                          * bytes per "line".
3139                          * This is stupid and ugly, but very cheap...
3140                          */
3141                         damage = DIV_ROUND_UP(damage, 64);
3142                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3143                 dir.files[dir.nr].name = file->name;
3144                 dir.files[dir.nr].changed = damage;
3145                 changed += damage;
3146                 dir.nr++;
3147         }
3148
3149         /* This can happen even with many files, if everything was renames */
3150         if (!changed)
3151                 return;
3152
3153         /* Show all directories with more than x% of the changes */
3154         QSORT(dir.files, dir.nr, dirstat_compare);
3155         gather_dirstat(options, &dir, changed, "", 0);
3156 }
3157
3158 static void free_diffstat_file(struct diffstat_file *f)
3159 {
3160         free(f->print_name);
3161         free(f->name);
3162         free(f->from_name);
3163         free(f);
3164 }
3165
3166 void free_diffstat_info(struct diffstat_t *diffstat)
3167 {
3168         int i;
3169         for (i = 0; i < diffstat->nr; i++)
3170                 free_diffstat_file(diffstat->files[i]);
3171         free(diffstat->files);
3172 }
3173
3174 struct checkdiff_t {
3175         const char *filename;
3176         int lineno;
3177         int conflict_marker_size;
3178         struct diff_options *o;
3179         unsigned ws_rule;
3180         unsigned status;
3181 };
3182
3183 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3184 {
3185         char firstchar;
3186         int cnt;
3187
3188         if (len < marker_size + 1)
3189                 return 0;
3190         firstchar = line[0];
3191         switch (firstchar) {
3192         case '=': case '>': case '<': case '|':
3193                 break;
3194         default:
3195                 return 0;
3196         }
3197         for (cnt = 1; cnt < marker_size; cnt++)
3198                 if (line[cnt] != firstchar)
3199                         return 0;
3200         /* line[1] through line[marker_size-1] are same as firstchar */
3201         if (len < marker_size + 1 || !isspace(line[marker_size]))
3202                 return 0;
3203         return 1;
3204 }
3205
3206 static void checkdiff_consume_hunk(void *priv,
3207                                    long ob, long on, long nb, long nn,
3208                                    const char *func, long funclen)
3209
3210 {
3211         struct checkdiff_t *data = priv;
3212         data->lineno = nb - 1;
3213 }
3214
3215 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3216 {
3217         struct checkdiff_t *data = priv;
3218         int marker_size = data->conflict_marker_size;
3219         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3220         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3221         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3222         char *err;
3223         const char *line_prefix;
3224
3225         assert(data->o);
3226         line_prefix = diff_line_prefix(data->o);
3227
3228         if (line[0] == '+') {
3229                 unsigned bad;
3230                 data->lineno++;
3231                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3232                         data->status |= 1;
3233                         fprintf(data->o->file,
3234                                 "%s%s:%d: leftover conflict marker\n",
3235                                 line_prefix, data->filename, data->lineno);
3236                 }
3237                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3238                 if (!bad)
3239                         return;
3240                 data->status |= bad;
3241                 err = whitespace_error_string(bad);
3242                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3243                         line_prefix, data->filename, data->lineno, err);
3244                 free(err);
3245                 emit_line(data->o, set, reset, line, 1);
3246                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3247                               data->o->file, set, reset, ws);
3248         } else if (line[0] == ' ') {
3249                 data->lineno++;
3250         }
3251 }
3252
3253 static unsigned char *deflate_it(char *data,
3254                                  unsigned long size,
3255                                  unsigned long *result_size)
3256 {
3257         int bound;
3258         unsigned char *deflated;
3259         git_zstream stream;
3260
3261         git_deflate_init(&stream, zlib_compression_level);
3262         bound = git_deflate_bound(&stream, size);
3263         deflated = xmalloc(bound);
3264         stream.next_out = deflated;
3265         stream.avail_out = bound;
3266
3267         stream.next_in = (unsigned char *)data;
3268         stream.avail_in = size;
3269         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3270                 ; /* nothing */
3271         git_deflate_end(&stream);
3272         *result_size = stream.total_out;
3273         return deflated;
3274 }
3275
3276 static void emit_binary_diff_body(struct diff_options *o,
3277                                   mmfile_t *one, mmfile_t *two)
3278 {
3279         void *cp;
3280         void *delta;
3281         void *deflated;
3282         void *data;
3283         unsigned long orig_size;
3284         unsigned long delta_size;
3285         unsigned long deflate_size;
3286         unsigned long data_size;
3287
3288         /* We could do deflated delta, or we could do just deflated two,
3289          * whichever is smaller.
3290          */
3291         delta = NULL;
3292         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3293         if (one->size && two->size) {
3294                 delta = diff_delta(one->ptr, one->size,
3295                                    two->ptr, two->size,
3296                                    &delta_size, deflate_size);
3297                 if (delta) {
3298                         void *to_free = delta;
3299                         orig_size = delta_size;
3300                         delta = deflate_it(delta, delta_size, &delta_size);
3301                         free(to_free);
3302                 }
3303         }
3304
3305         if (delta && delta_size < deflate_size) {
3306                 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
3307                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3308                                  s, strlen(s), 0);
3309                 free(s);
3310                 free(deflated);
3311                 data = delta;
3312                 data_size = delta_size;
3313         } else {
3314                 char *s = xstrfmt("%lu", two->size);
3315                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3316                                  s, strlen(s), 0);
3317                 free(s);
3318                 free(delta);
3319                 data = deflated;
3320                 data_size = deflate_size;
3321         }
3322
3323         /* emit data encoded in base85 */
3324         cp = data;
3325         while (data_size) {
3326                 int len;
3327                 int bytes = (52 < data_size) ? 52 : data_size;
3328                 char line[71];
3329                 data_size -= bytes;
3330                 if (bytes <= 26)
3331                         line[0] = bytes + 'A' - 1;
3332                 else
3333                         line[0] = bytes - 26 + 'a' - 1;
3334                 encode_85(line + 1, cp, bytes);
3335                 cp = (char *) cp + bytes;
3336
3337                 len = strlen(line);
3338                 line[len++] = '\n';
3339                 line[len] = '\0';
3340
3341                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3342                                  line, len, 0);
3343         }
3344         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3345         free(data);
3346 }
3347
3348 static void emit_binary_diff(struct diff_options *o,
3349                              mmfile_t *one, mmfile_t *two)
3350 {
3351         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3352         emit_binary_diff_body(o, one, two);
3353         emit_binary_diff_body(o, two, one);
3354 }
3355
3356 int diff_filespec_is_binary(struct repository *r,
3357                             struct diff_filespec *one)
3358 {
3359         struct diff_populate_filespec_options dpf_options = {
3360                 .check_binary = 1,
3361         };
3362
3363         if (one->is_binary == -1) {
3364                 diff_filespec_load_driver(one, r->index);
3365                 if (one->driver->binary != -1)
3366                         one->is_binary = one->driver->binary;
3367                 else {
3368                         if (!one->data && DIFF_FILE_VALID(one))
3369                                 diff_populate_filespec(r, one, &dpf_options);
3370                         if (one->is_binary == -1 && one->data)
3371                                 one->is_binary = buffer_is_binary(one->data,
3372                                                 one->size);
3373                         if (one->is_binary == -1)
3374                                 one->is_binary = 0;
3375                 }
3376         }
3377         return one->is_binary;
3378 }
3379
3380 static const struct userdiff_funcname *
3381 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3382 {
3383         diff_filespec_load_driver(one, o->repo->index);
3384         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3385 }
3386
3387 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3388 {
3389         if (!options->a_prefix)
3390                 options->a_prefix = a;
3391         if (!options->b_prefix)
3392                 options->b_prefix = b;
3393 }
3394
3395 struct userdiff_driver *get_textconv(struct repository *r,
3396                                      struct diff_filespec *one)
3397 {
3398         if (!DIFF_FILE_VALID(one))
3399                 return NULL;
3400
3401         diff_filespec_load_driver(one, r->index);
3402         return userdiff_get_textconv(r, one->driver);
3403 }
3404
3405 static void builtin_diff(const char *name_a,
3406                          const char *name_b,
3407                          struct diff_filespec *one,
3408                          struct diff_filespec *two,
3409                          const char *xfrm_msg,
3410                          int must_show_header,
3411                          struct diff_options *o,
3412                          int complete_rewrite)
3413 {
3414         mmfile_t mf1, mf2;
3415         const char *lbl[2];
3416         char *a_one, *b_two;
3417         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3418         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3419         const char *a_prefix, *b_prefix;
3420         struct userdiff_driver *textconv_one = NULL;
3421         struct userdiff_driver *textconv_two = NULL;
3422         struct strbuf header = STRBUF_INIT;
3423         const char *line_prefix = diff_line_prefix(o);
3424
3425         diff_set_mnemonic_prefix(o, "a/", "b/");
3426         if (o->flags.reverse_diff) {
3427                 a_prefix = o->b_prefix;
3428                 b_prefix = o->a_prefix;
3429         } else {
3430                 a_prefix = o->a_prefix;
3431                 b_prefix = o->b_prefix;
3432         }
3433
3434         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3435             (!one->mode || S_ISGITLINK(one->mode)) &&
3436             (!two->mode || S_ISGITLINK(two->mode))) {
3437                 show_submodule_diff_summary(o, one->path ? one->path : two->path,
3438                                 &one->oid, &two->oid,
3439                                 two->dirty_submodule);
3440                 return;
3441         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3442                    (!one->mode || S_ISGITLINK(one->mode)) &&
3443                    (!two->mode || S_ISGITLINK(two->mode))) {
3444                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3445                                 &one->oid, &two->oid,
3446                                 two->dirty_submodule);
3447                 return;
3448         }
3449
3450         if (o->flags.allow_textconv) {
3451                 textconv_one = get_textconv(o->repo, one);
3452                 textconv_two = get_textconv(o->repo, two);
3453         }
3454
3455         /* Never use a non-valid filename anywhere if at all possible */
3456         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3457         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3458
3459         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3460         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3461         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3462         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3463         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3464         if (lbl[0][0] == '/') {
3465                 /* /dev/null */
3466                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3467                 if (xfrm_msg)
3468                         strbuf_addstr(&header, xfrm_msg);
3469                 must_show_header = 1;
3470         }
3471         else if (lbl[1][0] == '/') {
3472                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3473                 if (xfrm_msg)
3474                         strbuf_addstr(&header, xfrm_msg);
3475                 must_show_header = 1;
3476         }
3477         else {
3478                 if (one->mode != two->mode) {
3479                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3480                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3481                         must_show_header = 1;
3482                 }
3483                 if (xfrm_msg)
3484                         strbuf_addstr(&header, xfrm_msg);
3485
3486                 /*
3487                  * we do not run diff between different kind
3488                  * of objects.
3489                  */
3490                 if ((one->mode ^ two->mode) & S_IFMT)
3491                         goto free_ab_and_return;
3492                 if (complete_rewrite &&
3493                     (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3494                     (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3495                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3496                                          header.buf, header.len, 0);
3497                         strbuf_reset(&header);
3498                         emit_rewrite_diff(name_a, name_b, one, two,
3499                                           textconv_one, textconv_two, o);
3500                         o->found_changes = 1;
3501                         goto free_ab_and_return;
3502                 }
3503         }
3504
3505         if (o->irreversible_delete && lbl[1][0] == '/') {
3506                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3507                                  header.len, 0);
3508                 strbuf_reset(&header);
3509                 goto free_ab_and_return;
3510         } else if (!o->flags.text &&
3511                    ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3512                      (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3513                 struct strbuf sb = STRBUF_INIT;
3514                 if (!one->data && !two->data &&
3515                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3516                     !o->flags.binary) {
3517                         if (oideq(&one->oid, &two->oid)) {
3518                                 if (must_show_header)
3519                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3520                                                          header.buf, header.len,
3521                                                          0);
3522                                 goto free_ab_and_return;
3523                         }
3524                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3525                                          header.buf, header.len, 0);
3526                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3527                                     diff_line_prefix(o), lbl[0], lbl[1]);
3528                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3529                                          sb.buf, sb.len, 0);
3530                         strbuf_release(&sb);
3531                         goto free_ab_and_return;
3532                 }
3533                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3534                     fill_mmfile(o->repo, &mf2, two) < 0)
3535                         die("unable to read files to diff");
3536                 /* Quite common confusing case */
3537                 if (mf1.size == mf2.size &&
3538                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3539                         if (must_show_header)
3540                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3541                                                  header.buf, header.len, 0);
3542                         goto free_ab_and_return;
3543                 }
3544                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3545                 strbuf_reset(&header);
3546                 if (o->flags.binary)
3547                         emit_binary_diff(o, &mf1, &mf2);
3548                 else {
3549                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3550                                     diff_line_prefix(o), lbl[0], lbl[1]);
3551                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3552                                          sb.buf, sb.len, 0);
3553                         strbuf_release(&sb);
3554                 }
3555                 o->found_changes = 1;
3556         } else {
3557                 /* Crazy xdl interfaces.. */
3558                 const char *diffopts;
3559                 const char *v;
3560                 xpparam_t xpp;
3561                 xdemitconf_t xecfg;
3562                 struct emit_callback ecbdata;
3563                 const struct userdiff_funcname *pe;
3564
3565                 if (must_show_header) {
3566                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3567                                          header.buf, header.len, 0);
3568                         strbuf_reset(&header);
3569                 }
3570
3571                 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3572                 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3573
3574                 pe = diff_funcname_pattern(o, one);
3575                 if (!pe)
3576                         pe = diff_funcname_pattern(o, two);
3577
3578                 memset(&xpp, 0, sizeof(xpp));
3579                 memset(&xecfg, 0, sizeof(xecfg));
3580                 memset(&ecbdata, 0, sizeof(ecbdata));
3581                 if (o->flags.suppress_diff_headers)
3582                         lbl[0] = NULL;
3583                 ecbdata.label_path = lbl;
3584                 ecbdata.color_diff = want_color(o->use_color);
3585                 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3586                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3587                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3588                 ecbdata.opt = o;
3589                 if (header.len && !o->flags.suppress_diff_headers)
3590                         ecbdata.header = &header;
3591                 xpp.flags = o->xdl_opts;
3592                 xpp.ignore_regex = o->ignore_regex;
3593                 xpp.ignore_regex_nr = o->ignore_regex_nr;
3594                 xpp.anchors = o->anchors;
3595                 xpp.anchors_nr = o->anchors_nr;
3596                 xecfg.ctxlen = o->context;
3597                 xecfg.interhunkctxlen = o->interhunkcontext;
3598                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3599                 if (o->flags.funccontext)
3600                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3601                 if (pe)
3602                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3603
3604                 diffopts = getenv("GIT_DIFF_OPTS");
3605                 if (!diffopts)
3606                         ;
3607                 else if (skip_prefix(diffopts, "--unified=", &v))
3608                         xecfg.ctxlen = strtoul(v, NULL, 10);
3609                 else if (skip_prefix(diffopts, "-u", &v))
3610                         xecfg.ctxlen = strtoul(v, NULL, 10);
3611
3612                 if (o->word_diff)
3613                         init_diff_words_data(&ecbdata, o, one, two);
3614                 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3615                                   &ecbdata, &xpp, &xecfg))
3616                         die("unable to generate diff for %s", one->path);
3617                 if (o->word_diff)
3618                         free_diff_words_data(&ecbdata);
3619                 if (textconv_one)
3620                         free(mf1.ptr);
3621                 if (textconv_two)
3622                         free(mf2.ptr);
3623                 xdiff_clear_find_func(&xecfg);
3624         }
3625
3626  free_ab_and_return:
3627         strbuf_release(&header);
3628         diff_free_filespec_data(one);
3629         diff_free_filespec_data(two);
3630         free(a_one);
3631         free(b_two);
3632         return;
3633 }
3634
3635 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3636 {
3637         if (!is_renamed) {
3638                 if (p->status == DIFF_STATUS_ADDED) {
3639                         if (S_ISLNK(p->two->mode))
3640                                 return "new +l";
3641                         else if ((p->two->mode & 0777) == 0755)
3642                                 return "new +x";
3643                         else
3644                                 return "new";
3645                 } else if (p->status == DIFF_STATUS_DELETED)
3646                         return "gone";
3647         }
3648         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3649                 return "mode -l";
3650         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3651                 return "mode +l";
3652         else if ((p->one->mode & 0777) == 0644 &&
3653                  (p->two->mode & 0777) == 0755)
3654                 return "mode +x";
3655         else if ((p->one->mode & 0777) == 0755 &&
3656                  (p->two->mode & 0777) == 0644)
3657                 return "mode -x";
3658         return NULL;
3659 }
3660
3661 static void builtin_diffstat(const char *name_a, const char *name_b,
3662                              struct diff_filespec *one,
3663                              struct diff_filespec *two,
3664                              struct diffstat_t *diffstat,
3665                              struct diff_options *o,
3666                              struct diff_filepair *p)
3667 {
3668         mmfile_t mf1, mf2;
3669         struct diffstat_file *data;
3670         int may_differ;
3671         int complete_rewrite = 0;
3672
3673         if (!DIFF_PAIR_UNMERGED(p)) {
3674                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3675                         complete_rewrite = 1;
3676         }
3677
3678         data = diffstat_add(diffstat, name_a, name_b);
3679         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3680         if (o->flags.stat_with_summary)
3681                 data->comments = get_compact_summary(p, data->is_renamed);
3682
3683         if (!one || !two) {
3684                 data->is_unmerged = 1;
3685                 return;
3686         }
3687
3688         /* saves some reads if true, not a guarantee of diff outcome */
3689         may_differ = !(one->oid_valid && two->oid_valid &&
3690                         oideq(&one->oid, &two->oid));
3691
3692         if (diff_filespec_is_binary(o->repo, one) ||
3693             diff_filespec_is_binary(o->repo, two)) {
3694                 data->is_binary = 1;
3695                 if (!may_differ) {
3696                         data->added = 0;
3697                         data->deleted = 0;
3698                 } else {
3699                         data->added = diff_filespec_size(o->repo, two);
3700                         data->deleted = diff_filespec_size(o->repo, one);
3701                 }
3702         }
3703
3704         else if (complete_rewrite) {
3705                 diff_populate_filespec(o->repo, one, NULL);
3706                 diff_populate_filespec(o->repo, two, NULL);
3707                 data->deleted = count_lines(one->data, one->size);
3708                 data->added = count_lines(two->data, two->size);
3709         }
3710
3711         else if (may_differ) {
3712                 /* Crazy xdl interfaces.. */
3713                 xpparam_t xpp;
3714                 xdemitconf_t xecfg;
3715
3716                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3717                     fill_mmfile(o->repo, &mf2, two) < 0)
3718                         die("unable to read files to diff");
3719
3720                 memset(&xpp, 0, sizeof(xpp));
3721                 memset(&xecfg, 0, sizeof(xecfg));
3722                 xpp.flags = o->xdl_opts;
3723                 xpp.ignore_regex = o->ignore_regex;
3724                 xpp.ignore_regex_nr = o->ignore_regex_nr;
3725                 xpp.anchors = o->anchors;
3726                 xpp.anchors_nr = o->anchors_nr;
3727                 xecfg.ctxlen = o->context;
3728                 xecfg.interhunkctxlen = o->interhunkcontext;
3729                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
3730                                   diffstat_consume, diffstat, &xpp, &xecfg))
3731                         die("unable to generate diffstat for %s", one->path);
3732
3733                 if (DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two)) {
3734                         struct diffstat_file *file =
3735                                 diffstat->files[diffstat->nr - 1];
3736                         /*
3737                          * Omit diffstats of modified files where nothing changed.
3738                          * Even if may_differ, this might be the case due to
3739                          * ignoring whitespace changes, etc.
3740                          *
3741                          * But note that we special-case additions, deletions,
3742                          * renames, and mode changes as adding an empty file,
3743                          * for example is still of interest.
3744                          */
3745                         if ((p->status == DIFF_STATUS_MODIFIED)
3746                                 && !file->added
3747                                 && !file->deleted
3748                                 && one->mode == two->mode) {
3749                                 free_diffstat_file(file);
3750                                 diffstat->nr--;
3751                         }
3752                 }
3753         }
3754
3755         diff_free_filespec_data(one);
3756         diff_free_filespec_data(two);
3757 }
3758
3759 static void builtin_checkdiff(const char *name_a, const char *name_b,
3760                               const char *attr_path,
3761                               struct diff_filespec *one,
3762                               struct diff_filespec *two,
3763                               struct diff_options *o)
3764 {
3765         mmfile_t mf1, mf2;
3766         struct checkdiff_t data;
3767
3768         if (!two)
3769                 return;
3770
3771         memset(&data, 0, sizeof(data));
3772         data.filename = name_b ? name_b : name_a;
3773         data.lineno = 0;
3774         data.o = o;
3775         data.ws_rule = whitespace_rule(o->repo->index, attr_path);
3776         data.conflict_marker_size = ll_merge_marker_size(o->repo->index, attr_path);
3777
3778         if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3779             fill_mmfile(o->repo, &mf2, two) < 0)
3780                 die("unable to read files to diff");
3781
3782         /*
3783          * All the other codepaths check both sides, but not checking
3784          * the "old" side here is deliberate.  We are checking the newly
3785          * introduced changes, and as long as the "new" side is text, we
3786          * can and should check what it introduces.
3787          */
3788         if (diff_filespec_is_binary(o->repo, two))
3789                 goto free_and_return;
3790         else {
3791                 /* Crazy xdl interfaces.. */
3792                 xpparam_t xpp;
3793                 xdemitconf_t xecfg;
3794
3795                 memset(&xpp, 0, sizeof(xpp));
3796                 memset(&xecfg, 0, sizeof(xecfg));
3797                 xecfg.ctxlen = 1; /* at least one context line */
3798                 xpp.flags = 0;
3799                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume_hunk,
3800                                   checkdiff_consume, &data,
3801                                   &xpp, &xecfg))
3802                         die("unable to generate checkdiff for %s", one->path);
3803
3804                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3805                         struct emit_callback ecbdata;
3806                         int blank_at_eof;
3807
3808                         ecbdata.ws_rule = data.ws_rule;
3809                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3810                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3811
3812                         if (blank_at_eof) {
3813                                 static char *err;
3814                                 if (!err)
3815                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3816                                 fprintf(o->file, "%s:%d: %s.\n",
3817                                         data.filename, blank_at_eof, err);
3818                                 data.status = 1; /* report errors */
3819                         }
3820                 }
3821         }
3822  free_and_return:
3823         diff_free_filespec_data(one);
3824         diff_free_filespec_data(two);
3825         if (data.status)
3826                 o->flags.check_failed = 1;
3827 }
3828
3829 struct diff_filespec *alloc_filespec(const char *path)
3830 {
3831         struct diff_filespec *spec;
3832
3833         FLEXPTR_ALLOC_STR(spec, path, path);
3834         spec->count = 1;
3835         spec->is_binary = -1;
3836         return spec;
3837 }
3838
3839 void free_filespec(struct diff_filespec *spec)
3840 {
3841         if (!--spec->count) {
3842                 diff_free_filespec_data(spec);
3843                 free(spec);
3844         }
3845 }
3846
3847 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3848                    int oid_valid, unsigned short mode)
3849 {
3850         if (mode) {
3851                 spec->mode = canon_mode(mode);
3852                 oidcpy(&spec->oid, oid);
3853                 spec->oid_valid = oid_valid;
3854         }
3855 }
3856
3857 /*
3858  * Given a name and sha1 pair, if the index tells us the file in
3859  * the work tree has that object contents, return true, so that
3860  * prepare_temp_file() does not have to inflate and extract.
3861  */
3862 static int reuse_worktree_file(struct index_state *istate,
3863                                const char *name,
3864                                const struct object_id *oid,
3865                                int want_file)
3866 {
3867         const struct cache_entry *ce;
3868         struct stat st;
3869         int pos, len;
3870
3871         /*
3872          * We do not read the cache ourselves here, because the
3873          * benchmark with my previous version that always reads cache
3874          * shows that it makes things worse for diff-tree comparing
3875          * two linux-2.6 kernel trees in an already checked out work
3876          * tree.  This is because most diff-tree comparisons deal with
3877          * only a small number of files, while reading the cache is
3878          * expensive for a large project, and its cost outweighs the
3879          * savings we get by not inflating the object to a temporary
3880          * file.  Practically, this code only helps when we are used
3881          * by diff-cache --cached, which does read the cache before
3882          * calling us.
3883          */
3884         if (!istate->cache)
3885                 return 0;
3886
3887         /* We want to avoid the working directory if our caller
3888          * doesn't need the data in a normal file, this system
3889          * is rather slow with its stat/open/mmap/close syscalls,
3890          * and the object is contained in a pack file.  The pack
3891          * is probably already open and will be faster to obtain
3892          * the data through than the working directory.  Loose
3893          * objects however would tend to be slower as they need
3894          * to be individually opened and inflated.
3895          */
3896         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3897                 return 0;
3898
3899         /*
3900          * Similarly, if we'd have to convert the file contents anyway, that
3901          * makes the optimization not worthwhile.
3902          */
3903         if (!want_file && would_convert_to_git(istate, name))
3904                 return 0;
3905
3906         len = strlen(name);
3907         pos = index_name_pos(istate, name, len);
3908         if (pos < 0)
3909                 return 0;
3910         ce = istate->cache[pos];
3911
3912         /*
3913          * This is not the sha1 we are looking for, or
3914          * unreusable because it is not a regular file.
3915          */
3916         if (!oideq(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3917                 return 0;
3918
3919         /*
3920          * If ce is marked as "assume unchanged", there is no
3921          * guarantee that work tree matches what we are looking for.
3922          */
3923         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3924                 return 0;
3925
3926         /*
3927          * If ce matches the file in the work tree, we can reuse it.
3928          */
3929         if (ce_uptodate(ce) ||
3930             (!lstat(name, &st) && !ie_match_stat(istate, ce, &st, 0)))
3931                 return 1;
3932
3933         return 0;
3934 }
3935
3936 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3937 {
3938         struct strbuf buf = STRBUF_INIT;
3939         char *dirty = "";
3940
3941         /* Are we looking at the work tree? */
3942         if (s->dirty_submodule)
3943                 dirty = "-dirty";
3944
3945         strbuf_addf(&buf, "Subproject commit %s%s\n",
3946                     oid_to_hex(&s->oid), dirty);
3947         s->size = buf.len;
3948         if (size_only) {
3949                 s->data = NULL;
3950                 strbuf_release(&buf);
3951         } else {
3952                 s->data = strbuf_detach(&buf, NULL);
3953                 s->should_free = 1;
3954         }
3955         return 0;
3956 }
3957
3958 /*
3959  * While doing rename detection and pickaxe operation, we may need to
3960  * grab the data for the blob (or file) for our own in-core comparison.
3961  * diff_filespec has data and size fields for this purpose.
3962  */
3963 int diff_populate_filespec(struct repository *r,
3964                            struct diff_filespec *s,
3965                            const struct diff_populate_filespec_options *options)
3966 {
3967         int size_only = options ? options->check_size_only : 0;
3968         int check_binary = options ? options->check_binary : 0;
3969         int err = 0;
3970         int conv_flags = global_conv_flags_eol;
3971         /*
3972          * demote FAIL to WARN to allow inspecting the situation
3973          * instead of refusing.
3974          */
3975         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3976                 conv_flags = CONV_EOL_RNDTRP_WARN;
3977
3978         if (!DIFF_FILE_VALID(s))
3979                 die("internal error: asking to populate invalid file.");
3980         if (S_ISDIR(s->mode))
3981                 return -1;
3982
3983         if (s->data)
3984                 return 0;
3985
3986         if (size_only && 0 < s->size)
3987                 return 0;
3988
3989         if (S_ISGITLINK(s->mode))
3990                 return diff_populate_gitlink(s, size_only);
3991
3992         if (!s->oid_valid ||
3993             reuse_worktree_file(r->index, s->path, &s->oid, 0)) {
3994                 struct strbuf buf = STRBUF_INIT;
3995                 struct stat st;
3996                 int fd;
3997
3998                 if (lstat(s->path, &st) < 0) {
3999                 err_empty:
4000                         err = -1;
4001                 empty:
4002                         s->data = (char *)"";
4003                         s->size = 0;
4004                         return err;
4005                 }
4006                 s->size = xsize_t(st.st_size);
4007                 if (!s->size)
4008                         goto empty;
4009                 if (S_ISLNK(st.st_mode)) {
4010                         struct strbuf sb = STRBUF_INIT;
4011
4012                         if (strbuf_readlink(&sb, s->path, s->size))
4013                                 goto err_empty;
4014                         s->size = sb.len;
4015                         s->data = strbuf_detach(&sb, NULL);
4016                         s->should_free = 1;
4017                         return 0;
4018                 }
4019
4020                 /*
4021                  * Even if the caller would be happy with getting
4022                  * only the size, we cannot return early at this
4023                  * point if the path requires us to run the content
4024                  * conversion.
4025                  */
4026                 if (size_only && !would_convert_to_git(r->index, s->path))
4027                         return 0;
4028
4029                 /*
4030                  * Note: this check uses xsize_t(st.st_size) that may
4031                  * not be the true size of the blob after it goes
4032                  * through convert_to_git().  This may not strictly be
4033                  * correct, but the whole point of big_file_threshold
4034                  * and is_binary check being that we want to avoid
4035                  * opening the file and inspecting the contents, this
4036                  * is probably fine.
4037                  */
4038                 if (check_binary &&
4039                     s->size > big_file_threshold && s->is_binary == -1) {
4040                         s->is_binary = 1;
4041                         return 0;
4042                 }
4043                 fd = open(s->path, O_RDONLY);
4044                 if (fd < 0)
4045                         goto err_empty;
4046                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
4047                 close(fd);
4048                 s->should_munmap = 1;
4049
4050                 /*
4051                  * Convert from working tree format to canonical git format
4052                  */
4053                 if (convert_to_git(r->index, s->path, s->data, s->size, &buf, conv_flags)) {
4054                         size_t size = 0;
4055                         munmap(s->data, s->size);
4056                         s->should_munmap = 0;
4057                         s->data = strbuf_detach(&buf, &size);
4058                         s->size = size;
4059                         s->should_free = 1;
4060                 }
4061         }
4062         else {
4063                 struct object_info info = {
4064                         .sizep = &s->size
4065                 };
4066
4067                 if (!(size_only || check_binary))
4068                         /*
4069                          * Set contentp, since there is no chance that merely
4070                          * the size is sufficient.
4071                          */
4072                         info.contentp = &s->data;
4073
4074                 if (options && options->missing_object_cb) {
4075                         if (!oid_object_info_extended(r, &s->oid, &info,
4076                                                       OBJECT_INFO_LOOKUP_REPLACE |
4077                                                       OBJECT_INFO_SKIP_FETCH_OBJECT))
4078                                 goto object_read;
4079                         options->missing_object_cb(options->missing_object_data);
4080                 }
4081                 if (oid_object_info_extended(r, &s->oid, &info,
4082                                              OBJECT_INFO_LOOKUP_REPLACE))
4083                         die("unable to read %s", oid_to_hex(&s->oid));
4084
4085 object_read:
4086                 if (size_only || check_binary) {
4087                         if (size_only)
4088                                 return 0;
4089                         if (s->size > big_file_threshold && s->is_binary == -1) {
4090                                 s->is_binary = 1;
4091                                 return 0;
4092                         }
4093                 }
4094                 if (!info.contentp) {
4095                         info.contentp = &s->data;
4096                         if (oid_object_info_extended(r, &s->oid, &info,
4097                                                      OBJECT_INFO_LOOKUP_REPLACE))
4098                                 die("unable to read %s", oid_to_hex(&s->oid));
4099                 }
4100                 s->should_free = 1;
4101         }
4102         return 0;
4103 }
4104
4105 void diff_free_filespec_blob(struct diff_filespec *s)
4106 {
4107         if (s->should_free)
4108                 free(s->data);
4109         else if (s->should_munmap)
4110                 munmap(s->data, s->size);
4111
4112         if (s->should_free || s->should_munmap) {
4113                 s->should_free = s->should_munmap = 0;
4114                 s->data = NULL;
4115         }
4116 }
4117
4118 void diff_free_filespec_data(struct diff_filespec *s)
4119 {
4120         if (!s)
4121                 return;
4122
4123         diff_free_filespec_blob(s);
4124         FREE_AND_NULL(s->cnt_data);
4125 }
4126
4127 static void prep_temp_blob(struct index_state *istate,
4128                            const char *path, struct diff_tempfile *temp,
4129                            void *blob,
4130                            unsigned long size,
4131                            const struct object_id *oid,
4132                            int mode)
4133 {
4134         struct strbuf buf = STRBUF_INIT;
4135         struct strbuf tempfile = STRBUF_INIT;
4136         char *path_dup = xstrdup(path);
4137         const char *base = basename(path_dup);
4138         struct checkout_metadata meta;
4139
4140         init_checkout_metadata(&meta, NULL, NULL, oid);
4141
4142         /* Generate "XXXXXX_basename.ext" */
4143         strbuf_addstr(&tempfile, "XXXXXX_");
4144         strbuf_addstr(&tempfile, base);
4145
4146         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
4147         if (!temp->tempfile)
4148                 die_errno("unable to create temp-file");
4149         if (convert_to_working_tree(istate, path,
4150                         (const char *)blob, (size_t)size, &buf, &meta)) {
4151                 blob = buf.buf;
4152                 size = buf.len;
4153         }
4154         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
4155             close_tempfile_gently(temp->tempfile))
4156                 die_errno("unable to write temp-file");
4157         temp->name = get_tempfile_path(temp->tempfile);
4158         oid_to_hex_r(temp->hex, oid);
4159         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
4160         strbuf_release(&buf);
4161         strbuf_release(&tempfile);
4162         free(path_dup);
4163 }
4164
4165 static struct diff_tempfile *prepare_temp_file(struct repository *r,
4166                                                const char *name,
4167                                                struct diff_filespec *one)
4168 {
4169         struct diff_tempfile *temp = claim_diff_tempfile();
4170
4171         if (!DIFF_FILE_VALID(one)) {
4172         not_a_valid_file:
4173                 /* A '-' entry produces this for file-2, and
4174                  * a '+' entry produces this for file-1.
4175                  */
4176                 temp->name = "/dev/null";
4177                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
4178                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
4179                 return temp;
4180         }
4181
4182         if (!S_ISGITLINK(one->mode) &&
4183             (!one->oid_valid ||
4184              reuse_worktree_file(r->index, name, &one->oid, 1))) {
4185                 struct stat st;
4186                 if (lstat(name, &st) < 0) {
4187                         if (errno == ENOENT)
4188                                 goto not_a_valid_file;
4189                         die_errno("stat(%s)", name);
4190                 }
4191                 if (S_ISLNK(st.st_mode)) {
4192                         struct strbuf sb = STRBUF_INIT;
4193                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4194                                 die_errno("readlink(%s)", name);
4195                         prep_temp_blob(r->index, name, temp, sb.buf, sb.len,
4196                                        (one->oid_valid ?
4197                                         &one->oid : null_oid()),
4198                                        (one->oid_valid ?
4199                                         one->mode : S_IFLNK));
4200                         strbuf_release(&sb);
4201                 }
4202                 else {
4203                         /* we can borrow from the file in the work tree */
4204                         temp->name = name;
4205                         if (!one->oid_valid)
4206                                 oid_to_hex_r(temp->hex, null_oid());
4207                         else
4208                                 oid_to_hex_r(temp->hex, &one->oid);
4209                         /* Even though we may sometimes borrow the
4210                          * contents from the work tree, we always want
4211                          * one->mode.  mode is trustworthy even when
4212                          * !(one->oid_valid), as long as
4213                          * DIFF_FILE_VALID(one).
4214                          */
4215                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4216                 }
4217                 return temp;
4218         }
4219         else {
4220                 if (diff_populate_filespec(r, one, NULL))
4221                         die("cannot read data blob for %s", one->path);
4222                 prep_temp_blob(r->index, name, temp,
4223                                one->data, one->size,
4224                                &one->oid, one->mode);
4225         }
4226         return temp;
4227 }
4228
4229 static void add_external_diff_name(struct repository *r,
4230                                    struct strvec *argv,
4231                                    const char *name,
4232                                    struct diff_filespec *df)
4233 {
4234         struct diff_tempfile *temp = prepare_temp_file(r, name, df);
4235         strvec_push(argv, temp->name);
4236         strvec_push(argv, temp->hex);
4237         strvec_push(argv, temp->mode);
4238 }
4239
4240 /* An external diff command takes:
4241  *
4242  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4243  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4244  *
4245  */
4246 static void run_external_diff(const char *pgm,
4247                               const char *name,
4248                               const char *other,
4249                               struct diff_filespec *one,
4250                               struct diff_filespec *two,
4251                               const char *xfrm_msg,
4252                               struct diff_options *o)
4253 {
4254         struct strvec argv = STRVEC_INIT;
4255         struct strvec env = STRVEC_INIT;
4256         struct diff_queue_struct *q = &diff_queued_diff;
4257
4258         strvec_push(&argv, pgm);
4259         strvec_push(&argv, name);
4260
4261         if (one && two) {
4262                 add_external_diff_name(o->repo, &argv, name, one);
4263                 if (!other)
4264                         add_external_diff_name(o->repo, &argv, name, two);
4265                 else {
4266                         add_external_diff_name(o->repo, &argv, other, two);
4267                         strvec_push(&argv, other);
4268                         strvec_push(&argv, xfrm_msg);
4269                 }
4270         }
4271
4272         strvec_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4273         strvec_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4274
4275         diff_free_filespec_data(one);
4276         diff_free_filespec_data(two);
4277         if (run_command_v_opt_cd_env(argv.v, RUN_USING_SHELL, NULL, env.v))
4278                 die(_("external diff died, stopping at %s"), name);
4279
4280         remove_tempfile();
4281         strvec_clear(&argv);
4282         strvec_clear(&env);
4283 }
4284
4285 static int similarity_index(struct diff_filepair *p)
4286 {
4287         return p->score * 100 / MAX_SCORE;
4288 }
4289
4290 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4291 {
4292         if (startup_info->have_repository)
4293                 return find_unique_abbrev(oid, abbrev);
4294         else {
4295                 char *hex = oid_to_hex(oid);
4296                 if (abbrev < 0)
4297                         abbrev = FALLBACK_DEFAULT_ABBREV;
4298                 if (abbrev > the_hash_algo->hexsz)
4299                         BUG("oid abbreviation out of range: %d", abbrev);
4300                 if (abbrev)
4301                         hex[abbrev] = '\0';
4302                 return hex;
4303         }
4304 }
4305
4306 static void fill_metainfo(struct strbuf *msg,
4307                           const char *name,
4308                           const char *other,
4309                           struct diff_filespec *one,
4310                           struct diff_filespec *two,
4311                           struct diff_options *o,
4312                           struct diff_filepair *p,
4313                           int *must_show_header,
4314                           int use_color)
4315 {
4316         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4317         const char *reset = diff_get_color(use_color, DIFF_RESET);
4318         const char *line_prefix = diff_line_prefix(o);
4319
4320         *must_show_header = 1;
4321         strbuf_init(msg, PATH_MAX * 2 + 300);
4322         switch (p->status) {
4323         case DIFF_STATUS_COPIED:
4324                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4325                             line_prefix, set, similarity_index(p));
4326                 strbuf_addf(msg, "%s\n%s%scopy from ",
4327                             reset,  line_prefix, set);
4328                 quote_c_style(name, msg, NULL, 0);
4329                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4330                 quote_c_style(other, msg, NULL, 0);
4331                 strbuf_addf(msg, "%s\n", reset);
4332                 break;
4333         case DIFF_STATUS_RENAMED:
4334                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4335                             line_prefix, set, similarity_index(p));
4336                 strbuf_addf(msg, "%s\n%s%srename from ",
4337                             reset, line_prefix, set);
4338                 quote_c_style(name, msg, NULL, 0);
4339                 strbuf_addf(msg, "%s\n%s%srename to ",
4340                             reset, line_prefix, set);
4341                 quote_c_style(other, msg, NULL, 0);
4342                 strbuf_addf(msg, "%s\n", reset);
4343                 break;
4344         case DIFF_STATUS_MODIFIED:
4345                 if (p->score) {
4346                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4347                                     line_prefix,
4348                                     set, similarity_index(p), reset);
4349                         break;
4350                 }
4351                 /* fallthru */
4352         default:
4353                 *must_show_header = 0;
4354         }
4355         if (one && two && !oideq(&one->oid, &two->oid)) {
4356                 const unsigned hexsz = the_hash_algo->hexsz;
4357                 int abbrev = o->abbrev ? o->abbrev : DEFAULT_ABBREV;
4358
4359                 if (o->flags.full_index)
4360                         abbrev = hexsz;
4361
4362                 if (o->flags.binary) {
4363                         mmfile_t mf;
4364                         if ((!fill_mmfile(o->repo, &mf, one) &&
4365                              diff_filespec_is_binary(o->repo, one)) ||
4366                             (!fill_mmfile(o->repo, &mf, two) &&
4367                              diff_filespec_is_binary(o->repo, two)))
4368                                 abbrev = hexsz;
4369                 }
4370                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4371                             diff_abbrev_oid(&one->oid, abbrev),
4372                             diff_abbrev_oid(&two->oid, abbrev));
4373                 if (one->mode == two->mode)
4374                         strbuf_addf(msg, " %06o", one->mode);
4375                 strbuf_addf(msg, "%s\n", reset);
4376         }
4377 }
4378
4379 static void run_diff_cmd(const char *pgm,
4380                          const char *name,
4381                          const char *other,
4382                          const char *attr_path,
4383                          struct diff_filespec *one,
4384                          struct diff_filespec *two,
4385                          struct strbuf *msg,
4386                          struct diff_options *o,
4387                          struct diff_filepair *p)
4388 {
4389         const char *xfrm_msg = NULL;
4390         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4391         int must_show_header = 0;
4392
4393
4394         if (o->flags.allow_external) {
4395                 struct userdiff_driver *drv;
4396
4397                 drv = userdiff_find_by_path(o->repo->index, attr_path);
4398                 if (drv && drv->external)
4399                         pgm = drv->external;
4400         }
4401
4402         if (msg) {
4403                 /*
4404                  * don't use colors when the header is intended for an
4405                  * external diff driver
4406                  */
4407                 fill_metainfo(msg, name, other, one, two, o, p,
4408                               &must_show_header,
4409                               want_color(o->use_color) && !pgm);
4410                 xfrm_msg = msg->len ? msg->buf : NULL;
4411         }
4412
4413         if (pgm) {
4414                 run_external_diff(pgm, name, other, one, two, xfrm_msg, o);
4415                 return;
4416         }
4417         if (one && two)
4418                 builtin_diff(name, other ? other : name,
4419                              one, two, xfrm_msg, must_show_header,
4420                              o, complete_rewrite);
4421         else
4422                 fprintf(o->file, "* Unmerged path %s\n", name);
4423 }
4424
4425 static void diff_fill_oid_info(struct diff_filespec *one, struct index_state *istate)
4426 {
4427         if (DIFF_FILE_VALID(one)) {
4428                 if (!one->oid_valid) {
4429                         struct stat st;
4430                         if (one->is_stdin) {
4431                                 oidclr(&one->oid);
4432                                 return;
4433                         }
4434                         if (lstat(one->path, &st) < 0)
4435                                 die_errno("stat '%s'", one->path);
4436                         if (index_path(istate, &one->oid, one->path, &st, 0))
4437                                 die("cannot hash %s", one->path);
4438                 }
4439         }
4440         else
4441                 oidclr(&one->oid);
4442 }
4443
4444 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4445 {
4446         /* Strip the prefix but do not molest /dev/null and absolute paths */
4447         if (*namep && !is_absolute_path(*namep)) {
4448                 *namep += prefix_length;
4449                 if (**namep == '/')
4450                         ++*namep;
4451         }
4452         if (*otherp && !is_absolute_path(*otherp)) {
4453                 *otherp += prefix_length;
4454                 if (**otherp == '/')
4455                         ++*otherp;
4456         }
4457 }
4458
4459 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4460 {
4461         const char *pgm = external_diff();
4462         struct strbuf msg;
4463         struct diff_filespec *one = p->one;
4464         struct diff_filespec *two = p->two;
4465         const char *name;
4466         const char *other;
4467         const char *attr_path;
4468
4469         name  = one->path;
4470         other = (strcmp(name, two->path) ? two->path : NULL);
4471         attr_path = name;
4472         if (o->prefix_length)
4473                 strip_prefix(o->prefix_length, &name, &other);
4474
4475         if (!o->flags.allow_external)
4476                 pgm = NULL;
4477
4478         if (DIFF_PAIR_UNMERGED(p)) {
4479                 run_diff_cmd(pgm, name, NULL, attr_path,
4480                              NULL, NULL, NULL, o, p);
4481                 return;
4482         }
4483
4484         diff_fill_oid_info(one, o->repo->index);
4485         diff_fill_oid_info(two, o->repo->index);
4486
4487         if (!pgm &&
4488             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4489             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4490                 /*
4491                  * a filepair that changes between file and symlink
4492                  * needs to be split into deletion and creation.
4493                  */
4494                 struct diff_filespec *null = alloc_filespec(two->path);
4495                 run_diff_cmd(NULL, name, other, attr_path,
4496                              one, null, &msg,
4497                              o, p);
4498                 free(null);
4499                 strbuf_release(&msg);
4500
4501                 null = alloc_filespec(one->path);
4502                 run_diff_cmd(NULL, name, other, attr_path,
4503                              null, two, &msg, o, p);
4504                 free(null);
4505         }
4506         else
4507                 run_diff_cmd(pgm, name, other, attr_path,
4508                              one, two, &msg, o, p);
4509
4510         strbuf_release(&msg);
4511 }
4512
4513 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4514                          struct diffstat_t *diffstat)
4515 {
4516         const char *name;
4517         const char *other;
4518
4519         if (DIFF_PAIR_UNMERGED(p)) {
4520                 /* unmerged */
4521                 builtin_diffstat(p->one->path, NULL, NULL, NULL,
4522                                  diffstat, o, p);
4523                 return;
4524         }
4525
4526         name = p->one->path;
4527         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4528
4529         if (o->prefix_length)
4530                 strip_prefix(o->prefix_length, &name, &other);
4531
4532         diff_fill_oid_info(p->one, o->repo->index);
4533         diff_fill_oid_info(p->two, o->repo->index);
4534
4535         builtin_diffstat(name, other, p->one, p->two,
4536                          diffstat, o, p);
4537 }
4538
4539 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4540 {
4541         const char *name;
4542         const char *other;
4543         const char *attr_path;
4544
4545         if (DIFF_PAIR_UNMERGED(p)) {
4546                 /* unmerged */
4547                 return;
4548         }
4549
4550         name = p->one->path;
4551         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4552         attr_path = other ? other : name;
4553
4554         if (o->prefix_length)
4555                 strip_prefix(o->prefix_length, &name, &other);
4556
4557         diff_fill_oid_info(p->one, o->repo->index);
4558         diff_fill_oid_info(p->two, o->repo->index);
4559
4560         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4561 }
4562
4563 static void prep_parse_options(struct diff_options *options);
4564
4565 void repo_diff_setup(struct repository *r, struct diff_options *options)
4566 {
4567         memcpy(options, &default_diff_options, sizeof(*options));
4568
4569         options->file = stdout;
4570         options->repo = r;
4571
4572         options->output_indicators[OUTPUT_INDICATOR_NEW] = '+';
4573         options->output_indicators[OUTPUT_INDICATOR_OLD] = '-';
4574         options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = ' ';
4575         options->abbrev = DEFAULT_ABBREV;
4576         options->line_termination = '\n';
4577         options->break_opt = -1;
4578         options->rename_limit = -1;
4579         options->dirstat_permille = diff_dirstat_permille_default;
4580         options->context = diff_context_default;
4581         options->interhunkcontext = diff_interhunk_context_default;
4582         options->ws_error_highlight = ws_error_highlight_default;
4583         options->flags.rename_empty = 1;
4584         options->flags.relative_name = diff_relative;
4585         options->objfind = NULL;
4586
4587         /* pathchange left =NULL by default */
4588         options->change = diff_change;
4589         options->add_remove = diff_addremove;
4590         options->use_color = diff_use_color_default;
4591         options->detect_rename = diff_detect_rename_default;
4592         options->xdl_opts |= diff_algorithm;
4593         if (diff_indent_heuristic)
4594                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4595
4596         options->orderfile = diff_order_file_cfg;
4597
4598         if (!options->flags.ignore_submodule_set)
4599                 options->flags.ignore_untracked_in_submodules = 1;
4600
4601         if (diff_no_prefix) {
4602                 options->a_prefix = options->b_prefix = "";
4603         } else if (!diff_mnemonic_prefix) {
4604                 options->a_prefix = "a/";
4605                 options->b_prefix = "b/";
4606         }
4607
4608         options->color_moved = diff_color_moved_default;
4609         options->color_moved_ws_handling = diff_color_moved_ws_default;
4610
4611         prep_parse_options(options);
4612 }
4613
4614 void diff_setup_done(struct diff_options *options)
4615 {
4616         unsigned check_mask = DIFF_FORMAT_NAME |
4617                               DIFF_FORMAT_NAME_STATUS |
4618                               DIFF_FORMAT_CHECKDIFF |
4619                               DIFF_FORMAT_NO_OUTPUT;
4620         /*
4621          * This must be signed because we're comparing against a potentially
4622          * negative value.
4623          */
4624         const int hexsz = the_hash_algo->hexsz;
4625
4626         if (options->set_default)
4627                 options->set_default(options);
4628
4629         if (HAS_MULTI_BITS(options->output_format & check_mask))
4630                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4631
4632         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4633                 die(_("-G, -S and --find-object are mutually exclusive"));
4634
4635         /*
4636          * Most of the time we can say "there are changes"
4637          * only by checking if there are changed paths, but
4638          * --ignore-whitespace* options force us to look
4639          * inside contents.
4640          */
4641
4642         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS) ||
4643             options->ignore_regex_nr)
4644                 options->flags.diff_from_contents = 1;
4645         else
4646                 options->flags.diff_from_contents = 0;
4647
4648         if (options->flags.find_copies_harder)
4649                 options->detect_rename = DIFF_DETECT_COPY;
4650
4651         if (!options->flags.relative_name)
4652                 options->prefix = NULL;
4653         if (options->prefix)
4654                 options->prefix_length = strlen(options->prefix);
4655         else
4656                 options->prefix_length = 0;
4657
4658         if (options->output_format & (DIFF_FORMAT_NAME |
4659                                       DIFF_FORMAT_NAME_STATUS |
4660                                       DIFF_FORMAT_CHECKDIFF |
4661                                       DIFF_FORMAT_NO_OUTPUT))
4662                 options->output_format &= ~(DIFF_FORMAT_RAW |
4663                                             DIFF_FORMAT_NUMSTAT |
4664                                             DIFF_FORMAT_DIFFSTAT |
4665                                             DIFF_FORMAT_SHORTSTAT |
4666                                             DIFF_FORMAT_DIRSTAT |
4667                                             DIFF_FORMAT_SUMMARY |
4668                                             DIFF_FORMAT_PATCH);
4669
4670         /*
4671          * These cases always need recursive; we do not drop caller-supplied
4672          * recursive bits for other formats here.
4673          */
4674         if (options->output_format & (DIFF_FORMAT_PATCH |
4675                                       DIFF_FORMAT_NUMSTAT |
4676                                       DIFF_FORMAT_DIFFSTAT |
4677                                       DIFF_FORMAT_SHORTSTAT |
4678                                       DIFF_FORMAT_DIRSTAT |
4679                                       DIFF_FORMAT_SUMMARY |
4680                                       DIFF_FORMAT_CHECKDIFF))
4681                 options->flags.recursive = 1;
4682         /*
4683          * Also pickaxe would not work very well if you do not say recursive
4684          */
4685         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4686                 options->flags.recursive = 1;
4687         /*
4688          * When patches are generated, submodules diffed against the work tree
4689          * must be checked for dirtiness too so it can be shown in the output
4690          */
4691         if (options->output_format & DIFF_FORMAT_PATCH)
4692                 options->flags.dirty_submodules = 1;
4693
4694         if (options->detect_rename && options->rename_limit < 0)
4695                 options->rename_limit = diff_rename_limit_default;
4696         if (hexsz < options->abbrev)
4697                 options->abbrev = hexsz; /* full */
4698
4699         /*
4700          * It does not make sense to show the first hit we happened
4701          * to have found.  It does not make sense not to return with
4702          * exit code in such a case either.
4703          */
4704         if (options->flags.quick) {
4705                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4706                 options->flags.exit_with_status = 1;
4707         }
4708
4709         options->diff_path_counter = 0;
4710
4711         if (options->flags.follow_renames && options->pathspec.nr != 1)
4712                 die(_("--follow requires exactly one pathspec"));
4713
4714         if (!options->use_color || external_diff())
4715                 options->color_moved = 0;
4716
4717         FREE_AND_NULL(options->parseopts);
4718 }
4719
4720 int parse_long_opt(const char *opt, const char **argv,
4721                    const char **optarg)
4722 {
4723         const char *arg = argv[0];
4724         if (!skip_prefix(arg, "--", &arg))
4725                 return 0;
4726         if (!skip_prefix(arg, opt, &arg))
4727                 return 0;
4728         if (*arg == '=') { /* stuck form: --option=value */
4729                 *optarg = arg + 1;
4730                 return 1;
4731         }
4732         if (*arg != '\0')
4733                 return 0;
4734         /* separate form: --option value */
4735         if (!argv[1])
4736                 die("Option '--%s' requires a value", opt);
4737         *optarg = argv[1];
4738         return 2;
4739 }
4740
4741 static int diff_opt_stat(const struct option *opt, const char *value, int unset)
4742 {
4743         struct diff_options *options = opt->value;
4744         int width = options->stat_width;
4745         int name_width = options->stat_name_width;
4746         int graph_width = options->stat_graph_width;
4747         int count = options->stat_count;
4748         char *end;
4749
4750         BUG_ON_OPT_NEG(unset);
4751
4752         if (!strcmp(opt->long_name, "stat")) {
4753                 if (value) {
4754                         width = strtoul(value, &end, 10);
4755                         if (*end == ',')
4756                                 name_width = strtoul(end+1, &end, 10);
4757                         if (*end == ',')
4758                                 count = strtoul(end+1, &end, 10);
4759                         if (*end)
4760                                 return error(_("invalid --stat value: %s"), value);
4761                 }
4762         } else if (!strcmp(opt->long_name, "stat-width")) {
4763                 width = strtoul(value, &end, 10);
4764                 if (*end)
4765                         return error(_("%s expects a numerical value"),
4766                                      opt->long_name);
4767         } else if (!strcmp(opt->long_name, "stat-name-width")) {
4768                 name_width = strtoul(value, &end, 10);
4769                 if (*end)
4770                         return error(_("%s expects a numerical value"),
4771                                      opt->long_name);
4772         } else if (!strcmp(opt->long_name, "stat-graph-width")) {
4773                 graph_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-count")) {
4778                 count = strtoul(value, &end, 10);
4779                 if (*end)
4780                         return error(_("%s expects a numerical value"),
4781                                      opt->long_name);
4782         } else
4783                 BUG("%s should not get here", opt->long_name);
4784
4785         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4786         options->stat_name_width = name_width;
4787         options->stat_graph_width = graph_width;
4788         options->stat_width = width;
4789         options->stat_count = count;
4790         return 0;
4791 }
4792
4793 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4794 {
4795         struct strbuf errmsg = STRBUF_INIT;
4796         if (parse_dirstat_params(options, params, &errmsg))
4797                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4798                     errmsg.buf);
4799         strbuf_release(&errmsg);
4800         /*
4801          * The caller knows a dirstat-related option is given from the command
4802          * line; allow it to say "return this_function();"
4803          */
4804         options->output_format |= DIFF_FORMAT_DIRSTAT;
4805         return 1;
4806 }
4807
4808 static const char diff_status_letters[] = {
4809         DIFF_STATUS_ADDED,
4810         DIFF_STATUS_COPIED,
4811         DIFF_STATUS_DELETED,
4812         DIFF_STATUS_MODIFIED,
4813         DIFF_STATUS_RENAMED,
4814         DIFF_STATUS_TYPE_CHANGED,
4815         DIFF_STATUS_UNKNOWN,
4816         DIFF_STATUS_UNMERGED,
4817         DIFF_STATUS_FILTER_AON,
4818         DIFF_STATUS_FILTER_BROKEN,
4819         '\0',
4820 };
4821
4822 static unsigned int filter_bit['Z' + 1];
4823
4824 static void prepare_filter_bits(void)
4825 {
4826         int i;
4827
4828         if (!filter_bit[DIFF_STATUS_ADDED]) {
4829                 for (i = 0; diff_status_letters[i]; i++)
4830                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4831         }
4832 }
4833
4834 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4835 {
4836         return opt->filter & filter_bit[(int) status];
4837 }
4838
4839 unsigned diff_filter_bit(char status)
4840 {
4841         prepare_filter_bits();
4842         return filter_bit[(int) status];
4843 }
4844
4845 static int diff_opt_diff_filter(const struct option *option,
4846                                 const char *optarg, int unset)
4847 {
4848         struct diff_options *opt = option->value;
4849         int i, optch;
4850
4851         BUG_ON_OPT_NEG(unset);
4852         prepare_filter_bits();
4853
4854         /*
4855          * If there is a negation e.g. 'd' in the input, and we haven't
4856          * initialized the filter field with another --diff-filter, start
4857          * from full set of bits, except for AON.
4858          */
4859         if (!opt->filter) {
4860                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4861                         if (optch < 'a' || 'z' < optch)
4862                                 continue;
4863                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4864                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4865                         break;
4866                 }
4867         }
4868
4869         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4870                 unsigned int bit;
4871                 int negate;
4872
4873                 if ('a' <= optch && optch <= 'z') {
4874                         negate = 1;
4875                         optch = toupper(optch);
4876                 } else {
4877                         negate = 0;
4878                 }
4879
4880                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4881                 if (!bit)
4882                         return error(_("unknown change class '%c' in --diff-filter=%s"),
4883                                      optarg[i], optarg);
4884                 if (negate)
4885                         opt->filter &= ~bit;
4886                 else
4887                         opt->filter |= bit;
4888         }
4889         return 0;
4890 }
4891
4892 static void enable_patch_output(int *fmt)
4893 {
4894         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4895         *fmt |= DIFF_FORMAT_PATCH;
4896 }
4897
4898 static int diff_opt_ws_error_highlight(const struct option *option,
4899                                        const char *arg, int unset)
4900 {
4901         struct diff_options *opt = option->value;
4902         int val = parse_ws_error_highlight(arg);
4903
4904         BUG_ON_OPT_NEG(unset);
4905         if (val < 0)
4906                 return error(_("unknown value after ws-error-highlight=%.*s"),
4907                              -1 - val, arg);
4908         opt->ws_error_highlight = val;
4909         return 0;
4910 }
4911
4912 static int diff_opt_find_object(const struct option *option,
4913                                 const char *arg, int unset)
4914 {
4915         struct diff_options *opt = option->value;
4916         struct object_id oid;
4917
4918         BUG_ON_OPT_NEG(unset);
4919         if (get_oid(arg, &oid))
4920                 return error(_("unable to resolve '%s'"), arg);
4921
4922         if (!opt->objfind)
4923                 CALLOC_ARRAY(opt->objfind, 1);
4924
4925         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4926         opt->flags.recursive = 1;
4927         opt->flags.tree_in_recursive = 1;
4928         oidset_insert(opt->objfind, &oid);
4929         return 0;
4930 }
4931
4932 static int diff_opt_anchored(const struct option *opt,
4933                              const char *arg, int unset)
4934 {
4935         struct diff_options *options = opt->value;
4936
4937         BUG_ON_OPT_NEG(unset);
4938         options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4939         ALLOC_GROW(options->anchors, options->anchors_nr + 1,
4940                    options->anchors_alloc);
4941         options->anchors[options->anchors_nr++] = xstrdup(arg);
4942         return 0;
4943 }
4944
4945 static int diff_opt_binary(const struct option *opt,
4946                            const char *arg, int unset)
4947 {
4948         struct diff_options *options = opt->value;
4949
4950         BUG_ON_OPT_NEG(unset);
4951         BUG_ON_OPT_ARG(arg);
4952         enable_patch_output(&options->output_format);
4953         options->flags.binary = 1;
4954         return 0;
4955 }
4956
4957 static int diff_opt_break_rewrites(const struct option *opt,
4958                                    const char *arg, int unset)
4959 {
4960         int *break_opt = opt->value;
4961         int opt1, opt2;
4962
4963         BUG_ON_OPT_NEG(unset);
4964         if (!arg)
4965                 arg = "";
4966         opt1 = parse_rename_score(&arg);
4967         if (*arg == 0)
4968                 opt2 = 0;
4969         else if (*arg != '/')
4970                 return error(_("%s expects <n>/<m> form"), opt->long_name);
4971         else {
4972                 arg++;
4973                 opt2 = parse_rename_score(&arg);
4974         }
4975         if (*arg != 0)
4976                 return error(_("%s expects <n>/<m> form"), opt->long_name);
4977         *break_opt = opt1 | (opt2 << 16);
4978         return 0;
4979 }
4980
4981 static int diff_opt_char(const struct option *opt,
4982                          const char *arg, int unset)
4983 {
4984         char *value = opt->value;
4985
4986         BUG_ON_OPT_NEG(unset);
4987         if (arg[1])
4988                 return error(_("%s expects a character, got '%s'"),
4989                              opt->long_name, arg);
4990         *value = arg[0];
4991         return 0;
4992 }
4993
4994 static int diff_opt_color_moved(const struct option *opt,
4995                                 const char *arg, int unset)
4996 {
4997         struct diff_options *options = opt->value;
4998
4999         if (unset) {
5000                 options->color_moved = COLOR_MOVED_NO;
5001         } else if (!arg) {
5002                 if (diff_color_moved_default)
5003                         options->color_moved = diff_color_moved_default;
5004                 if (options->color_moved == COLOR_MOVED_NO)
5005                         options->color_moved = COLOR_MOVED_DEFAULT;
5006         } else {
5007                 int cm = parse_color_moved(arg);
5008                 if (cm < 0)
5009                         return error(_("bad --color-moved argument: %s"), arg);
5010                 options->color_moved = cm;
5011         }
5012         return 0;
5013 }
5014
5015 static int diff_opt_color_moved_ws(const struct option *opt,
5016                                    const char *arg, int unset)
5017 {
5018         struct diff_options *options = opt->value;
5019         unsigned cm;
5020
5021         if (unset) {
5022                 options->color_moved_ws_handling = 0;
5023                 return 0;
5024         }
5025
5026         cm = parse_color_moved_ws(arg);
5027         if (cm & COLOR_MOVED_WS_ERROR)
5028                 return error(_("invalid mode '%s' in --color-moved-ws"), arg);
5029         options->color_moved_ws_handling = cm;
5030         return 0;
5031 }
5032
5033 static int diff_opt_color_words(const struct option *opt,
5034                                 const char *arg, int unset)
5035 {
5036         struct diff_options *options = opt->value;
5037
5038         BUG_ON_OPT_NEG(unset);
5039         options->use_color = 1;
5040         options->word_diff = DIFF_WORDS_COLOR;
5041         options->word_regex = arg;
5042         return 0;
5043 }
5044
5045 static int diff_opt_compact_summary(const struct option *opt,
5046                                     const char *arg, int unset)
5047 {
5048         struct diff_options *options = opt->value;
5049
5050         BUG_ON_OPT_ARG(arg);
5051         if (unset) {
5052                 options->flags.stat_with_summary = 0;
5053         } else {
5054                 options->flags.stat_with_summary = 1;
5055                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
5056         }
5057         return 0;
5058 }
5059
5060 static int diff_opt_diff_algorithm(const struct option *opt,
5061                                    const char *arg, int unset)
5062 {
5063         struct diff_options *options = opt->value;
5064         long value = parse_algorithm_value(arg);
5065
5066         BUG_ON_OPT_NEG(unset);
5067         if (value < 0)
5068                 return error(_("option diff-algorithm accepts \"myers\", "
5069                                "\"minimal\", \"patience\" and \"histogram\""));
5070
5071         /* clear out previous settings */
5072         DIFF_XDL_CLR(options, NEED_MINIMAL);
5073         options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
5074         options->xdl_opts |= value;
5075         return 0;
5076 }
5077
5078 static int diff_opt_dirstat(const struct option *opt,
5079                             const char *arg, int unset)
5080 {
5081         struct diff_options *options = opt->value;
5082
5083         BUG_ON_OPT_NEG(unset);
5084         if (!strcmp(opt->long_name, "cumulative")) {
5085                 if (arg)
5086                         BUG("how come --cumulative take a value?");
5087                 arg = "cumulative";
5088         } else if (!strcmp(opt->long_name, "dirstat-by-file"))
5089                 parse_dirstat_opt(options, "files");
5090         parse_dirstat_opt(options, arg ? arg : "");
5091         return 0;
5092 }
5093
5094 static int diff_opt_find_copies(const struct option *opt,
5095                                 const char *arg, int unset)
5096 {
5097         struct diff_options *options = opt->value;
5098
5099         BUG_ON_OPT_NEG(unset);
5100         if (!arg)
5101                 arg = "";
5102         options->rename_score = parse_rename_score(&arg);
5103         if (*arg != 0)
5104                 return error(_("invalid argument to %s"), opt->long_name);
5105
5106         if (options->detect_rename == DIFF_DETECT_COPY)
5107                 options->flags.find_copies_harder = 1;
5108         else
5109                 options->detect_rename = DIFF_DETECT_COPY;
5110
5111         return 0;
5112 }
5113
5114 static int diff_opt_find_renames(const struct option *opt,
5115                                  const char *arg, int unset)
5116 {
5117         struct diff_options *options = opt->value;
5118
5119         BUG_ON_OPT_NEG(unset);
5120         if (!arg)
5121                 arg = "";
5122         options->rename_score = parse_rename_score(&arg);
5123         if (*arg != 0)
5124                 return error(_("invalid argument to %s"), opt->long_name);
5125
5126         options->detect_rename = DIFF_DETECT_RENAME;
5127         return 0;
5128 }
5129
5130 static int diff_opt_follow(const struct option *opt,
5131                            const char *arg, int unset)
5132 {
5133         struct diff_options *options = opt->value;
5134
5135         BUG_ON_OPT_ARG(arg);
5136         if (unset) {
5137                 options->flags.follow_renames = 0;
5138                 options->flags.default_follow_renames = 0;
5139         } else {
5140                 options->flags.follow_renames = 1;
5141         }
5142         return 0;
5143 }
5144
5145 static int diff_opt_ignore_submodules(const struct option *opt,
5146                                       const char *arg, int unset)
5147 {
5148         struct diff_options *options = opt->value;
5149
5150         BUG_ON_OPT_NEG(unset);
5151         if (!arg)
5152                 arg = "all";
5153         options->flags.override_submodule_config = 1;
5154         handle_ignore_submodules_arg(options, arg);
5155         return 0;
5156 }
5157
5158 static int diff_opt_line_prefix(const struct option *opt,
5159                                 const char *optarg, int unset)
5160 {
5161         struct diff_options *options = opt->value;
5162
5163         BUG_ON_OPT_NEG(unset);
5164         options->line_prefix = optarg;
5165         options->line_prefix_length = strlen(options->line_prefix);
5166         graph_setup_line_prefix(options);
5167         return 0;
5168 }
5169
5170 static int diff_opt_no_prefix(const struct option *opt,
5171                               const char *optarg, int unset)
5172 {
5173         struct diff_options *options = opt->value;
5174
5175         BUG_ON_OPT_NEG(unset);
5176         BUG_ON_OPT_ARG(optarg);
5177         options->a_prefix = "";
5178         options->b_prefix = "";
5179         return 0;
5180 }
5181
5182 static enum parse_opt_result diff_opt_output(struct parse_opt_ctx_t *ctx,
5183                                              const struct option *opt,
5184                                              const char *arg, int unset)
5185 {
5186         struct diff_options *options = opt->value;
5187         char *path;
5188
5189         BUG_ON_OPT_NEG(unset);
5190         path = prefix_filename(ctx->prefix, arg);
5191         options->file = xfopen(path, "w");
5192         options->close_file = 1;
5193         if (options->use_color != GIT_COLOR_ALWAYS)
5194                 options->use_color = GIT_COLOR_NEVER;
5195         free(path);
5196         return 0;
5197 }
5198
5199 static int diff_opt_patience(const struct option *opt,
5200                              const char *arg, int unset)
5201 {
5202         struct diff_options *options = opt->value;
5203         int i;
5204
5205         BUG_ON_OPT_NEG(unset);
5206         BUG_ON_OPT_ARG(arg);
5207         options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5208         /*
5209          * Both --patience and --anchored use PATIENCE_DIFF
5210          * internally, so remove any anchors previously
5211          * specified.
5212          */
5213         for (i = 0; i < options->anchors_nr; i++)
5214                 free(options->anchors[i]);
5215         options->anchors_nr = 0;
5216         return 0;
5217 }
5218
5219 static int diff_opt_ignore_regex(const struct option *opt,
5220                                  const char *arg, int unset)
5221 {
5222         struct diff_options *options = opt->value;
5223         regex_t *regex;
5224
5225         BUG_ON_OPT_NEG(unset);
5226         regex = xmalloc(sizeof(*regex));
5227         if (regcomp(regex, arg, REG_EXTENDED | REG_NEWLINE))
5228                 return error(_("invalid regex given to -I: '%s'"), arg);
5229         ALLOC_GROW(options->ignore_regex, options->ignore_regex_nr + 1,
5230                    options->ignore_regex_alloc);
5231         options->ignore_regex[options->ignore_regex_nr++] = regex;
5232         return 0;
5233 }
5234
5235 static int diff_opt_pickaxe_regex(const struct option *opt,
5236                                   const char *arg, int unset)
5237 {
5238         struct diff_options *options = opt->value;
5239
5240         BUG_ON_OPT_NEG(unset);
5241         options->pickaxe = arg;
5242         options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5243         return 0;
5244 }
5245
5246 static int diff_opt_pickaxe_string(const struct option *opt,
5247                                    const char *arg, int unset)
5248 {
5249         struct diff_options *options = opt->value;
5250
5251         BUG_ON_OPT_NEG(unset);
5252         options->pickaxe = arg;
5253         options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5254         return 0;
5255 }
5256
5257 static int diff_opt_relative(const struct option *opt,
5258                              const char *arg, int unset)
5259 {
5260         struct diff_options *options = opt->value;
5261
5262         options->flags.relative_name = !unset;
5263         if (arg)
5264                 options->prefix = arg;
5265         return 0;
5266 }
5267
5268 static int diff_opt_submodule(const struct option *opt,
5269                               const char *arg, int unset)
5270 {
5271         struct diff_options *options = opt->value;
5272
5273         BUG_ON_OPT_NEG(unset);
5274         if (!arg)
5275                 arg = "log";
5276         if (parse_submodule_params(options, arg))
5277                 return error(_("failed to parse --submodule option parameter: '%s'"),
5278                              arg);
5279         return 0;
5280 }
5281
5282 static int diff_opt_textconv(const struct option *opt,
5283                              const char *arg, int unset)
5284 {
5285         struct diff_options *options = opt->value;
5286
5287         BUG_ON_OPT_ARG(arg);
5288         if (unset) {
5289                 options->flags.allow_textconv = 0;
5290         } else {
5291                 options->flags.allow_textconv = 1;
5292                 options->flags.textconv_set_via_cmdline = 1;
5293         }
5294         return 0;
5295 }
5296
5297 static int diff_opt_unified(const struct option *opt,
5298                             const char *arg, int unset)
5299 {
5300         struct diff_options *options = opt->value;
5301         char *s;
5302
5303         BUG_ON_OPT_NEG(unset);
5304
5305         if (arg) {
5306                 options->context = strtol(arg, &s, 10);
5307                 if (*s)
5308                         return error(_("%s expects a numerical value"), "--unified");
5309         }
5310         enable_patch_output(&options->output_format);
5311
5312         return 0;
5313 }
5314
5315 static int diff_opt_word_diff(const struct option *opt,
5316                               const char *arg, int unset)
5317 {
5318         struct diff_options *options = opt->value;
5319
5320         BUG_ON_OPT_NEG(unset);
5321         if (arg) {
5322                 if (!strcmp(arg, "plain"))
5323                         options->word_diff = DIFF_WORDS_PLAIN;
5324                 else if (!strcmp(arg, "color")) {
5325                         options->use_color = 1;
5326                         options->word_diff = DIFF_WORDS_COLOR;
5327                 }
5328                 else if (!strcmp(arg, "porcelain"))
5329                         options->word_diff = DIFF_WORDS_PORCELAIN;
5330                 else if (!strcmp(arg, "none"))
5331                         options->word_diff = DIFF_WORDS_NONE;
5332                 else
5333                         return error(_("bad --word-diff argument: %s"), arg);
5334         } else {
5335                 if (options->word_diff == DIFF_WORDS_NONE)
5336                         options->word_diff = DIFF_WORDS_PLAIN;
5337         }
5338         return 0;
5339 }
5340
5341 static int diff_opt_word_diff_regex(const struct option *opt,
5342                                     const char *arg, int unset)
5343 {
5344         struct diff_options *options = opt->value;
5345
5346         BUG_ON_OPT_NEG(unset);
5347         if (options->word_diff == DIFF_WORDS_NONE)
5348                 options->word_diff = DIFF_WORDS_PLAIN;
5349         options->word_regex = arg;
5350         return 0;
5351 }
5352
5353 static int diff_opt_rotate_to(const struct option *opt, const char *arg, int unset)
5354 {
5355         struct diff_options *options = opt->value;
5356
5357         BUG_ON_OPT_NEG(unset);
5358         if (!strcmp(opt->long_name, "skip-to"))
5359                 options->skip_instead_of_rotate = 1;
5360         else
5361                 options->skip_instead_of_rotate = 0;
5362         options->rotate_to = arg;
5363         return 0;
5364 }
5365
5366 static void prep_parse_options(struct diff_options *options)
5367 {
5368         struct option parseopts[] = {
5369                 OPT_GROUP(N_("Diff output format options")),
5370                 OPT_BITOP('p', "patch", &options->output_format,
5371                           N_("generate patch"),
5372                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
5373                 OPT_BIT_F('s', "no-patch", &options->output_format,
5374                           N_("suppress diff output"),
5375                           DIFF_FORMAT_NO_OUTPUT, PARSE_OPT_NONEG),
5376                 OPT_BITOP('u', NULL, &options->output_format,
5377                           N_("generate patch"),
5378                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
5379                 OPT_CALLBACK_F('U', "unified", options, N_("<n>"),
5380                                N_("generate diffs with <n> lines context"),
5381                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_unified),
5382                 OPT_BOOL('W', "function-context", &options->flags.funccontext,
5383                          N_("generate diffs with <n> lines context")),
5384                 OPT_BIT_F(0, "raw", &options->output_format,
5385                           N_("generate the diff in raw format"),
5386                           DIFF_FORMAT_RAW, PARSE_OPT_NONEG),
5387                 OPT_BITOP(0, "patch-with-raw", &options->output_format,
5388                           N_("synonym for '-p --raw'"),
5389                           DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW,
5390                           DIFF_FORMAT_NO_OUTPUT),
5391                 OPT_BITOP(0, "patch-with-stat", &options->output_format,
5392                           N_("synonym for '-p --stat'"),
5393                           DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT,
5394                           DIFF_FORMAT_NO_OUTPUT),
5395                 OPT_BIT_F(0, "numstat", &options->output_format,
5396                           N_("machine friendly --stat"),
5397                           DIFF_FORMAT_NUMSTAT, PARSE_OPT_NONEG),
5398                 OPT_BIT_F(0, "shortstat", &options->output_format,
5399                           N_("output only the last line of --stat"),
5400                           DIFF_FORMAT_SHORTSTAT, PARSE_OPT_NONEG),
5401                 OPT_CALLBACK_F('X', "dirstat", options, N_("<param1,param2>..."),
5402                                N_("output the distribution of relative amount of changes for each sub-directory"),
5403                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5404                                diff_opt_dirstat),
5405                 OPT_CALLBACK_F(0, "cumulative", options, NULL,
5406                                N_("synonym for --dirstat=cumulative"),
5407                                PARSE_OPT_NONEG | PARSE_OPT_NOARG,
5408                                diff_opt_dirstat),
5409                 OPT_CALLBACK_F(0, "dirstat-by-file", options, N_("<param1,param2>..."),
5410                                N_("synonym for --dirstat=files,param1,param2..."),
5411                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5412                                diff_opt_dirstat),
5413                 OPT_BIT_F(0, "check", &options->output_format,
5414                           N_("warn if changes introduce conflict markers or whitespace errors"),
5415                           DIFF_FORMAT_CHECKDIFF, PARSE_OPT_NONEG),
5416                 OPT_BIT_F(0, "summary", &options->output_format,
5417                           N_("condensed summary such as creations, renames and mode changes"),
5418                           DIFF_FORMAT_SUMMARY, PARSE_OPT_NONEG),
5419                 OPT_BIT_F(0, "name-only", &options->output_format,
5420                           N_("show only names of changed files"),
5421                           DIFF_FORMAT_NAME, PARSE_OPT_NONEG),
5422                 OPT_BIT_F(0, "name-status", &options->output_format,
5423                           N_("show only names and status of changed files"),
5424                           DIFF_FORMAT_NAME_STATUS, PARSE_OPT_NONEG),
5425                 OPT_CALLBACK_F(0, "stat", options, N_("<width>[,<name-width>[,<count>]]"),
5426                                N_("generate diffstat"),
5427                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_stat),
5428                 OPT_CALLBACK_F(0, "stat-width", options, N_("<width>"),
5429                                N_("generate diffstat with a given width"),
5430                                PARSE_OPT_NONEG, diff_opt_stat),
5431                 OPT_CALLBACK_F(0, "stat-name-width", options, N_("<width>"),
5432                                N_("generate diffstat with a given name width"),
5433                                PARSE_OPT_NONEG, diff_opt_stat),
5434                 OPT_CALLBACK_F(0, "stat-graph-width", options, N_("<width>"),
5435                                N_("generate diffstat with a given graph width"),
5436                                PARSE_OPT_NONEG, diff_opt_stat),
5437                 OPT_CALLBACK_F(0, "stat-count", options, N_("<count>"),
5438                                N_("generate diffstat with limited lines"),
5439                                PARSE_OPT_NONEG, diff_opt_stat),
5440                 OPT_CALLBACK_F(0, "compact-summary", options, NULL,
5441                                N_("generate compact summary in diffstat"),
5442                                PARSE_OPT_NOARG, diff_opt_compact_summary),
5443                 OPT_CALLBACK_F(0, "binary", options, NULL,
5444                                N_("output a binary diff that can be applied"),
5445                                PARSE_OPT_NONEG | PARSE_OPT_NOARG, diff_opt_binary),
5446                 OPT_BOOL(0, "full-index", &options->flags.full_index,
5447                          N_("show full pre- and post-image object names on the \"index\" lines")),
5448                 OPT_COLOR_FLAG(0, "color", &options->use_color,
5449                                N_("show colored diff")),
5450                 OPT_CALLBACK_F(0, "ws-error-highlight", options, N_("<kind>"),
5451                                N_("highlight whitespace errors in the 'context', 'old' or 'new' lines in the diff"),
5452                                PARSE_OPT_NONEG, diff_opt_ws_error_highlight),
5453                 OPT_SET_INT('z', NULL, &options->line_termination,
5454                             N_("do not munge pathnames and use NULs as output field terminators in --raw or --numstat"),
5455                             0),
5456                 OPT__ABBREV(&options->abbrev),
5457                 OPT_STRING_F(0, "src-prefix", &options->a_prefix, N_("<prefix>"),
5458                              N_("show the given source prefix instead of \"a/\""),
5459                              PARSE_OPT_NONEG),
5460                 OPT_STRING_F(0, "dst-prefix", &options->b_prefix, N_("<prefix>"),
5461                              N_("show the given destination prefix instead of \"b/\""),
5462                              PARSE_OPT_NONEG),
5463                 OPT_CALLBACK_F(0, "line-prefix", options, N_("<prefix>"),
5464                                N_("prepend an additional prefix to every line of output"),
5465                                PARSE_OPT_NONEG, diff_opt_line_prefix),
5466                 OPT_CALLBACK_F(0, "no-prefix", options, NULL,
5467                                N_("do not show any source or destination prefix"),
5468                                PARSE_OPT_NONEG | PARSE_OPT_NOARG, diff_opt_no_prefix),
5469                 OPT_INTEGER_F(0, "inter-hunk-context", &options->interhunkcontext,
5470                               N_("show context between diff hunks up to the specified number of lines"),
5471                               PARSE_OPT_NONEG),
5472                 OPT_CALLBACK_F(0, "output-indicator-new",
5473                                &options->output_indicators[OUTPUT_INDICATOR_NEW],
5474                                N_("<char>"),
5475                                N_("specify the character to indicate a new line instead of '+'"),
5476                                PARSE_OPT_NONEG, diff_opt_char),
5477                 OPT_CALLBACK_F(0, "output-indicator-old",
5478                                &options->output_indicators[OUTPUT_INDICATOR_OLD],
5479                                N_("<char>"),
5480                                N_("specify the character to indicate an old line instead of '-'"),
5481                                PARSE_OPT_NONEG, diff_opt_char),
5482                 OPT_CALLBACK_F(0, "output-indicator-context",
5483                                &options->output_indicators[OUTPUT_INDICATOR_CONTEXT],
5484                                N_("<char>"),
5485                                N_("specify the character to indicate a context instead of ' '"),
5486                                PARSE_OPT_NONEG, diff_opt_char),
5487
5488                 OPT_GROUP(N_("Diff rename options")),
5489                 OPT_CALLBACK_F('B', "break-rewrites", &options->break_opt, N_("<n>[/<m>]"),
5490                                N_("break complete rewrite changes into pairs of delete and create"),
5491                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5492                                diff_opt_break_rewrites),
5493                 OPT_CALLBACK_F('M', "find-renames", options, N_("<n>"),
5494                                N_("detect renames"),
5495                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5496                                diff_opt_find_renames),
5497                 OPT_SET_INT_F('D', "irreversible-delete", &options->irreversible_delete,
5498                               N_("omit the preimage for deletes"),
5499                               1, PARSE_OPT_NONEG),
5500                 OPT_CALLBACK_F('C', "find-copies", options, N_("<n>"),
5501                                N_("detect copies"),
5502                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5503                                diff_opt_find_copies),
5504                 OPT_BOOL(0, "find-copies-harder", &options->flags.find_copies_harder,
5505                          N_("use unmodified files as source to find copies")),
5506                 OPT_SET_INT_F(0, "no-renames", &options->detect_rename,
5507                               N_("disable rename detection"),
5508                               0, PARSE_OPT_NONEG),
5509                 OPT_BOOL(0, "rename-empty", &options->flags.rename_empty,
5510                          N_("use empty blobs as rename source")),
5511                 OPT_CALLBACK_F(0, "follow", options, NULL,
5512                                N_("continue listing the history of a file beyond renames"),
5513                                PARSE_OPT_NOARG, diff_opt_follow),
5514                 OPT_INTEGER('l', NULL, &options->rename_limit,
5515                             N_("prevent rename/copy detection if the number of rename/copy targets exceeds given limit")),
5516
5517                 OPT_GROUP(N_("Diff algorithm options")),
5518                 OPT_BIT(0, "minimal", &options->xdl_opts,
5519                         N_("produce the smallest possible diff"),
5520                         XDF_NEED_MINIMAL),
5521                 OPT_BIT_F('w', "ignore-all-space", &options->xdl_opts,
5522                           N_("ignore whitespace when comparing lines"),
5523                           XDF_IGNORE_WHITESPACE, PARSE_OPT_NONEG),
5524                 OPT_BIT_F('b', "ignore-space-change", &options->xdl_opts,
5525                           N_("ignore changes in amount of whitespace"),
5526                           XDF_IGNORE_WHITESPACE_CHANGE, PARSE_OPT_NONEG),
5527                 OPT_BIT_F(0, "ignore-space-at-eol", &options->xdl_opts,
5528                           N_("ignore changes in whitespace at EOL"),
5529                           XDF_IGNORE_WHITESPACE_AT_EOL, PARSE_OPT_NONEG),
5530                 OPT_BIT_F(0, "ignore-cr-at-eol", &options->xdl_opts,
5531                           N_("ignore carrier-return at the end of line"),
5532                           XDF_IGNORE_CR_AT_EOL, PARSE_OPT_NONEG),
5533                 OPT_BIT_F(0, "ignore-blank-lines", &options->xdl_opts,
5534                           N_("ignore changes whose lines are all blank"),
5535                           XDF_IGNORE_BLANK_LINES, PARSE_OPT_NONEG),
5536                 OPT_CALLBACK_F('I', "ignore-matching-lines", options, N_("<regex>"),
5537                                N_("ignore changes whose all lines match <regex>"),
5538                                0, diff_opt_ignore_regex),
5539                 OPT_BIT(0, "indent-heuristic", &options->xdl_opts,
5540                         N_("heuristic to shift diff hunk boundaries for easy reading"),
5541                         XDF_INDENT_HEURISTIC),
5542                 OPT_CALLBACK_F(0, "patience", options, NULL,
5543                                N_("generate diff using the \"patience diff\" algorithm"),
5544                                PARSE_OPT_NONEG | PARSE_OPT_NOARG,
5545                                diff_opt_patience),
5546                 OPT_BITOP(0, "histogram", &options->xdl_opts,
5547                           N_("generate diff using the \"histogram diff\" algorithm"),
5548                           XDF_HISTOGRAM_DIFF, XDF_DIFF_ALGORITHM_MASK),
5549                 OPT_CALLBACK_F(0, "diff-algorithm", options, N_("<algorithm>"),
5550                                N_("choose a diff algorithm"),
5551                                PARSE_OPT_NONEG, diff_opt_diff_algorithm),
5552                 OPT_CALLBACK_F(0, "anchored", options, N_("<text>"),
5553                                N_("generate diff using the \"anchored diff\" algorithm"),
5554                                PARSE_OPT_NONEG, diff_opt_anchored),
5555                 OPT_CALLBACK_F(0, "word-diff", options, N_("<mode>"),
5556                                N_("show word diff, using <mode> to delimit changed words"),
5557                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_word_diff),
5558                 OPT_CALLBACK_F(0, "word-diff-regex", options, N_("<regex>"),
5559                                N_("use <regex> to decide what a word is"),
5560                                PARSE_OPT_NONEG, diff_opt_word_diff_regex),
5561                 OPT_CALLBACK_F(0, "color-words", options, N_("<regex>"),
5562                                N_("equivalent to --word-diff=color --word-diff-regex=<regex>"),
5563                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_color_words),
5564                 OPT_CALLBACK_F(0, "color-moved", options, N_("<mode>"),
5565                                N_("moved lines of code are colored differently"),
5566                                PARSE_OPT_OPTARG, diff_opt_color_moved),
5567                 OPT_CALLBACK_F(0, "color-moved-ws", options, N_("<mode>"),
5568                                N_("how white spaces are ignored in --color-moved"),
5569                                0, diff_opt_color_moved_ws),
5570
5571                 OPT_GROUP(N_("Other diff options")),
5572                 OPT_CALLBACK_F(0, "relative", options, N_("<prefix>"),
5573                                N_("when run from subdir, exclude changes outside and show relative paths"),
5574                                PARSE_OPT_OPTARG,
5575                                diff_opt_relative),
5576                 OPT_BOOL('a', "text", &options->flags.text,
5577                          N_("treat all files as text")),
5578                 OPT_BOOL('R', NULL, &options->flags.reverse_diff,
5579                          N_("swap two inputs, reverse the diff")),
5580                 OPT_BOOL(0, "exit-code", &options->flags.exit_with_status,
5581                          N_("exit with 1 if there were differences, 0 otherwise")),
5582                 OPT_BOOL(0, "quiet", &options->flags.quick,
5583                          N_("disable all output of the program")),
5584                 OPT_BOOL(0, "ext-diff", &options->flags.allow_external,
5585                          N_("allow an external diff helper to be executed")),
5586                 OPT_CALLBACK_F(0, "textconv", options, NULL,
5587                                N_("run external text conversion filters when comparing binary files"),
5588                                PARSE_OPT_NOARG, diff_opt_textconv),
5589                 OPT_CALLBACK_F(0, "ignore-submodules", options, N_("<when>"),
5590                                N_("ignore changes to submodules in the diff generation"),
5591                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5592                                diff_opt_ignore_submodules),
5593                 OPT_CALLBACK_F(0, "submodule", options, N_("<format>"),
5594                                N_("specify how differences in submodules are shown"),
5595                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5596                                diff_opt_submodule),
5597                 OPT_SET_INT_F(0, "ita-invisible-in-index", &options->ita_invisible_in_index,
5598                               N_("hide 'git add -N' entries from the index"),
5599                               1, PARSE_OPT_NONEG),
5600                 OPT_SET_INT_F(0, "ita-visible-in-index", &options->ita_invisible_in_index,
5601                               N_("treat 'git add -N' entries as real in the index"),
5602                               0, PARSE_OPT_NONEG),
5603                 OPT_CALLBACK_F('S', NULL, options, N_("<string>"),
5604                                N_("look for differences that change the number of occurrences of the specified string"),
5605                                0, diff_opt_pickaxe_string),
5606                 OPT_CALLBACK_F('G', NULL, options, N_("<regex>"),
5607                                N_("look for differences that change the number of occurrences of the specified regex"),
5608                                0, diff_opt_pickaxe_regex),
5609                 OPT_BIT_F(0, "pickaxe-all", &options->pickaxe_opts,
5610                           N_("show all changes in the changeset with -S or -G"),
5611                           DIFF_PICKAXE_ALL, PARSE_OPT_NONEG),
5612                 OPT_BIT_F(0, "pickaxe-regex", &options->pickaxe_opts,
5613                           N_("treat <string> in -S as extended POSIX regular expression"),
5614                           DIFF_PICKAXE_REGEX, PARSE_OPT_NONEG),
5615                 OPT_FILENAME('O', NULL, &options->orderfile,
5616                              N_("control the order in which files appear in the output")),
5617                 OPT_CALLBACK_F(0, "rotate-to", options, N_("<path>"),
5618                                N_("show the change in the specified path first"),
5619                                PARSE_OPT_NONEG, diff_opt_rotate_to),
5620                 OPT_CALLBACK_F(0, "skip-to", options, N_("<path>"),
5621                                N_("skip the output to the specified path"),
5622                                PARSE_OPT_NONEG, diff_opt_rotate_to),
5623                 OPT_CALLBACK_F(0, "find-object", options, N_("<object-id>"),
5624                                N_("look for differences that change the number of occurrences of the specified object"),
5625                                PARSE_OPT_NONEG, diff_opt_find_object),
5626                 OPT_CALLBACK_F(0, "diff-filter", options, N_("[(A|C|D|M|R|T|U|X|B)...[*]]"),
5627                                N_("select files by diff type"),
5628                                PARSE_OPT_NONEG, diff_opt_diff_filter),
5629                 { OPTION_CALLBACK, 0, "output", options, N_("<file>"),
5630                   N_("Output to a specific file"),
5631                   PARSE_OPT_NONEG, NULL, 0, diff_opt_output },
5632
5633                 OPT_END()
5634         };
5635
5636         ALLOC_ARRAY(options->parseopts, ARRAY_SIZE(parseopts));
5637         memcpy(options->parseopts, parseopts, sizeof(parseopts));
5638 }
5639
5640 int diff_opt_parse(struct diff_options *options,
5641                    const char **av, int ac, const char *prefix)
5642 {
5643         if (!prefix)
5644                 prefix = "";
5645
5646         ac = parse_options(ac, av, prefix, options->parseopts, NULL,
5647                            PARSE_OPT_KEEP_DASHDASH |
5648                            PARSE_OPT_KEEP_UNKNOWN |
5649                            PARSE_OPT_NO_INTERNAL_HELP |
5650                            PARSE_OPT_ONE_SHOT |
5651                            PARSE_OPT_STOP_AT_NON_OPTION);
5652
5653         return ac;
5654 }
5655
5656 int parse_rename_score(const char **cp_p)
5657 {
5658         unsigned long num, scale;
5659         int ch, dot;
5660         const char *cp = *cp_p;
5661
5662         num = 0;
5663         scale = 1;
5664         dot = 0;
5665         for (;;) {
5666                 ch = *cp;
5667                 if ( !dot && ch == '.' ) {
5668                         scale = 1;
5669                         dot = 1;
5670                 } else if ( ch == '%' ) {
5671                         scale = dot ? scale*100 : 100;
5672                         cp++;   /* % is always at the end */
5673                         break;
5674                 } else if ( ch >= '0' && ch <= '9' ) {
5675                         if ( scale < 100000 ) {
5676                                 scale *= 10;
5677                                 num = (num*10) + (ch-'0');
5678                         }
5679                 } else {
5680                         break;
5681                 }
5682                 cp++;
5683         }
5684         *cp_p = cp;
5685
5686         /* user says num divided by scale and we say internally that
5687          * is MAX_SCORE * num / scale.
5688          */
5689         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5690 }
5691
5692 struct diff_queue_struct diff_queued_diff;
5693
5694 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5695 {
5696         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5697         queue->queue[queue->nr++] = dp;
5698 }
5699
5700 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5701                                  struct diff_filespec *one,
5702                                  struct diff_filespec *two)
5703 {
5704         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5705         dp->one = one;
5706         dp->two = two;
5707         if (queue)
5708                 diff_q(queue, dp);
5709         return dp;
5710 }
5711
5712 void diff_free_filepair(struct diff_filepair *p)
5713 {
5714         free_filespec(p->one);
5715         free_filespec(p->two);
5716         free(p);
5717 }
5718
5719 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5720 {
5721         int abblen;
5722         const char *abbrev;
5723
5724         /* Do we want all 40 hex characters? */
5725         if (len == the_hash_algo->hexsz)
5726                 return oid_to_hex(oid);
5727
5728         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5729         abbrev = diff_abbrev_oid(oid, len);
5730
5731         if (!print_sha1_ellipsis())
5732                 return abbrev;
5733
5734         abblen = strlen(abbrev);
5735
5736         /*
5737          * In well-behaved cases, where the abbreviated result is the
5738          * same as the requested length, append three dots after the
5739          * abbreviation (hence the whole logic is limited to the case
5740          * where abblen < 37); when the actual abbreviated result is a
5741          * bit longer than the requested length, we reduce the number
5742          * of dots so that they match the well-behaved ones.  However,
5743          * if the actual abbreviation is longer than the requested
5744          * length by more than three, we give up on aligning, and add
5745          * three dots anyway, to indicate that the output is not the
5746          * full object name.  Yes, this may be suboptimal, but this
5747          * appears only in "diff --raw --abbrev" output and it is not
5748          * worth the effort to change it now.  Note that this would
5749          * likely to work fine when the automatic sizing of default
5750          * abbreviation length is used--we would be fed -1 in "len" in
5751          * that case, and will end up always appending three-dots, but
5752          * the automatic sizing is supposed to give abblen that ensures
5753          * uniqueness across all objects (statistically speaking).
5754          */
5755         if (abblen < the_hash_algo->hexsz - 3) {
5756                 static char hex[GIT_MAX_HEXSZ + 1];
5757                 if (len < abblen && abblen <= len + 2)
5758                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5759                 else
5760                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5761                 return hex;
5762         }
5763
5764         return oid_to_hex(oid);
5765 }
5766
5767 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5768 {
5769         int line_termination = opt->line_termination;
5770         int inter_name_termination = line_termination ? '\t' : '\0';
5771
5772         fprintf(opt->file, "%s", diff_line_prefix(opt));
5773         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5774                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5775                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5776                 fprintf(opt->file, "%s ",
5777                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5778         }
5779         if (p->score) {
5780                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5781                         inter_name_termination);
5782         } else {
5783                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5784         }
5785
5786         if (p->status == DIFF_STATUS_COPIED ||
5787             p->status == DIFF_STATUS_RENAMED) {
5788                 const char *name_a, *name_b;
5789                 name_a = p->one->path;
5790                 name_b = p->two->path;
5791                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5792                 write_name_quoted(name_a, opt->file, inter_name_termination);
5793                 write_name_quoted(name_b, opt->file, line_termination);
5794         } else {
5795                 const char *name_a, *name_b;
5796                 name_a = p->one->mode ? p->one->path : p->two->path;
5797                 name_b = NULL;
5798                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5799                 write_name_quoted(name_a, opt->file, line_termination);
5800         }
5801 }
5802
5803 int diff_unmodified_pair(struct diff_filepair *p)
5804 {
5805         /* This function is written stricter than necessary to support
5806          * the currently implemented transformers, but the idea is to
5807          * let transformers to produce diff_filepairs any way they want,
5808          * and filter and clean them up here before producing the output.
5809          */
5810         struct diff_filespec *one = p->one, *two = p->two;
5811
5812         if (DIFF_PAIR_UNMERGED(p))
5813                 return 0; /* unmerged is interesting */
5814
5815         /* deletion, addition, mode or type change
5816          * and rename are all interesting.
5817          */
5818         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5819             DIFF_PAIR_MODE_CHANGED(p) ||
5820             strcmp(one->path, two->path))
5821                 return 0;
5822
5823         /* both are valid and point at the same path.  that is, we are
5824          * dealing with a change.
5825          */
5826         if (one->oid_valid && two->oid_valid &&
5827             oideq(&one->oid, &two->oid) &&
5828             !one->dirty_submodule && !two->dirty_submodule)
5829                 return 1; /* no change */
5830         if (!one->oid_valid && !two->oid_valid)
5831                 return 1; /* both look at the same file on the filesystem. */
5832         return 0;
5833 }
5834
5835 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5836 {
5837         if (diff_unmodified_pair(p))
5838                 return;
5839
5840         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5841             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5842                 return; /* no tree diffs in patch format */
5843
5844         run_diff(p, o);
5845 }
5846
5847 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5848                             struct diffstat_t *diffstat)
5849 {
5850         if (diff_unmodified_pair(p))
5851                 return;
5852
5853         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5854             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5855                 return; /* no useful stat for tree diffs */
5856
5857         run_diffstat(p, o, diffstat);
5858 }
5859
5860 static void diff_flush_checkdiff(struct diff_filepair *p,
5861                 struct diff_options *o)
5862 {
5863         if (diff_unmodified_pair(p))
5864                 return;
5865
5866         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5867             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5868                 return; /* nothing to check in tree diffs */
5869
5870         run_checkdiff(p, o);
5871 }
5872
5873 int diff_queue_is_empty(void)
5874 {
5875         struct diff_queue_struct *q = &diff_queued_diff;
5876         int i;
5877         for (i = 0; i < q->nr; i++)
5878                 if (!diff_unmodified_pair(q->queue[i]))
5879                         return 0;
5880         return 1;
5881 }
5882
5883 #if DIFF_DEBUG
5884 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5885 {
5886         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5887                 x, one ? one : "",
5888                 s->path,
5889                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5890                 s->mode,
5891                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5892         fprintf(stderr, "queue[%d] %s size %lu\n",
5893                 x, one ? one : "",
5894                 s->size);
5895 }
5896
5897 void diff_debug_filepair(const struct diff_filepair *p, int i)
5898 {
5899         diff_debug_filespec(p->one, i, "one");
5900         diff_debug_filespec(p->two, i, "two");
5901         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5902                 p->score, p->status ? p->status : '?',
5903                 p->one->rename_used, p->broken_pair);
5904 }
5905
5906 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5907 {
5908         int i;
5909         if (msg)
5910                 fprintf(stderr, "%s\n", msg);
5911         fprintf(stderr, "q->nr = %d\n", q->nr);
5912         for (i = 0; i < q->nr; i++) {
5913                 struct diff_filepair *p = q->queue[i];
5914                 diff_debug_filepair(p, i);
5915         }
5916 }
5917 #endif
5918
5919 static void diff_resolve_rename_copy(void)
5920 {
5921         int i;
5922         struct diff_filepair *p;
5923         struct diff_queue_struct *q = &diff_queued_diff;
5924
5925         diff_debug_queue("resolve-rename-copy", q);
5926
5927         for (i = 0; i < q->nr; i++) {
5928                 p = q->queue[i];
5929                 p->status = 0; /* undecided */
5930                 if (DIFF_PAIR_UNMERGED(p))
5931                         p->status = DIFF_STATUS_UNMERGED;
5932                 else if (!DIFF_FILE_VALID(p->one))
5933                         p->status = DIFF_STATUS_ADDED;
5934                 else if (!DIFF_FILE_VALID(p->two))
5935                         p->status = DIFF_STATUS_DELETED;
5936                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5937                         p->status = DIFF_STATUS_TYPE_CHANGED;
5938
5939                 /* from this point on, we are dealing with a pair
5940                  * whose both sides are valid and of the same type, i.e.
5941                  * either in-place edit or rename/copy edit.
5942                  */
5943                 else if (DIFF_PAIR_RENAME(p)) {
5944                         /*
5945                          * A rename might have re-connected a broken
5946                          * pair up, causing the pathnames to be the
5947                          * same again. If so, that's not a rename at
5948                          * all, just a modification..
5949                          *
5950                          * Otherwise, see if this source was used for
5951                          * multiple renames, in which case we decrement
5952                          * the count, and call it a copy.
5953                          */
5954                         if (!strcmp(p->one->path, p->two->path))
5955                                 p->status = DIFF_STATUS_MODIFIED;
5956                         else if (--p->one->rename_used > 0)
5957                                 p->status = DIFF_STATUS_COPIED;
5958                         else
5959                                 p->status = DIFF_STATUS_RENAMED;
5960                 }
5961                 else if (!oideq(&p->one->oid, &p->two->oid) ||
5962                          p->one->mode != p->two->mode ||
5963                          p->one->dirty_submodule ||
5964                          p->two->dirty_submodule ||
5965                          is_null_oid(&p->one->oid))
5966                         p->status = DIFF_STATUS_MODIFIED;
5967                 else {
5968                         /* This is a "no-change" entry and should not
5969                          * happen anymore, but prepare for broken callers.
5970                          */
5971                         error("feeding unmodified %s to diffcore",
5972                               p->one->path);
5973                         p->status = DIFF_STATUS_UNKNOWN;
5974                 }
5975         }
5976         diff_debug_queue("resolve-rename-copy done", q);
5977 }
5978
5979 static int check_pair_status(struct diff_filepair *p)
5980 {
5981         switch (p->status) {
5982         case DIFF_STATUS_UNKNOWN:
5983                 return 0;
5984         case 0:
5985                 die("internal error in diff-resolve-rename-copy");
5986         default:
5987                 return 1;
5988         }
5989 }
5990
5991 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5992 {
5993         int fmt = opt->output_format;
5994
5995         if (fmt & DIFF_FORMAT_CHECKDIFF)
5996                 diff_flush_checkdiff(p, opt);
5997         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5998                 diff_flush_raw(p, opt);
5999         else if (fmt & DIFF_FORMAT_NAME) {
6000                 const char *name_a, *name_b;
6001                 name_a = p->two->path;
6002                 name_b = NULL;
6003                 strip_prefix(opt->prefix_length, &name_a, &name_b);
6004                 fprintf(opt->file, "%s", diff_line_prefix(opt));
6005                 write_name_quoted(name_a, opt->file, opt->line_termination);
6006         }
6007 }
6008
6009 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
6010 {
6011         struct strbuf sb = STRBUF_INIT;
6012         if (fs->mode)
6013                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
6014         else
6015                 strbuf_addf(&sb, " %s ", newdelete);
6016
6017         quote_c_style(fs->path, &sb, NULL, 0);
6018         strbuf_addch(&sb, '\n');
6019         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
6020                          sb.buf, sb.len, 0);
6021         strbuf_release(&sb);
6022 }
6023
6024 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
6025                 int show_name)
6026 {
6027         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
6028                 struct strbuf sb = STRBUF_INIT;
6029                 strbuf_addf(&sb, " mode change %06o => %06o",
6030                             p->one->mode, p->two->mode);
6031                 if (show_name) {
6032                         strbuf_addch(&sb, ' ');
6033                         quote_c_style(p->two->path, &sb, NULL, 0);
6034                 }
6035                 strbuf_addch(&sb, '\n');
6036                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
6037                                  sb.buf, sb.len, 0);
6038                 strbuf_release(&sb);
6039         }
6040 }
6041
6042 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
6043                 struct diff_filepair *p)
6044 {
6045         struct strbuf sb = STRBUF_INIT;
6046         struct strbuf names = STRBUF_INIT;
6047
6048         pprint_rename(&names, p->one->path, p->two->path);
6049         strbuf_addf(&sb, " %s %s (%d%%)\n",
6050                     renamecopy, names.buf, similarity_index(p));
6051         strbuf_release(&names);
6052         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
6053                                  sb.buf, sb.len, 0);
6054         show_mode_change(opt, p, 0);
6055         strbuf_release(&sb);
6056 }
6057
6058 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
6059 {
6060         switch(p->status) {
6061         case DIFF_STATUS_DELETED:
6062                 show_file_mode_name(opt, "delete", p->one);
6063                 break;
6064         case DIFF_STATUS_ADDED:
6065                 show_file_mode_name(opt, "create", p->two);
6066                 break;
6067         case DIFF_STATUS_COPIED:
6068                 show_rename_copy(opt, "copy", p);
6069                 break;
6070         case DIFF_STATUS_RENAMED:
6071                 show_rename_copy(opt, "rename", p);
6072                 break;
6073         default:
6074                 if (p->score) {
6075                         struct strbuf sb = STRBUF_INIT;
6076                         strbuf_addstr(&sb, " rewrite ");
6077                         quote_c_style(p->two->path, &sb, NULL, 0);
6078                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
6079                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
6080                                          sb.buf, sb.len, 0);
6081                         strbuf_release(&sb);
6082                 }
6083                 show_mode_change(opt, p, !p->score);
6084                 break;
6085         }
6086 }
6087
6088 struct patch_id_t {
6089         git_hash_ctx *ctx;
6090         int patchlen;
6091 };
6092
6093 static int remove_space(char *line, int len)
6094 {
6095         int i;
6096         char *dst = line;
6097         unsigned char c;
6098
6099         for (i = 0; i < len; i++)
6100                 if (!isspace((c = line[i])))
6101                         *dst++ = c;
6102
6103         return dst - line;
6104 }
6105
6106 void flush_one_hunk(struct object_id *result, git_hash_ctx *ctx)
6107 {
6108         unsigned char hash[GIT_MAX_RAWSZ];
6109         unsigned short carry = 0;
6110         int i;
6111
6112         the_hash_algo->final_fn(hash, ctx);
6113         the_hash_algo->init_fn(ctx);
6114         /* 20-byte sum, with carry */
6115         for (i = 0; i < the_hash_algo->rawsz; ++i) {
6116                 carry += result->hash[i] + hash[i];
6117                 result->hash[i] = carry;
6118                 carry >>= 8;
6119         }
6120 }
6121
6122 static void patch_id_consume(void *priv, char *line, unsigned long len)
6123 {
6124         struct patch_id_t *data = priv;
6125         int new_len;
6126
6127         if (len > 12 && starts_with(line, "\\ "))
6128                 return;
6129         new_len = remove_space(line, len);
6130
6131         the_hash_algo->update_fn(data->ctx, line, new_len);
6132         data->patchlen += new_len;
6133 }
6134
6135 static void patch_id_add_string(git_hash_ctx *ctx, const char *str)
6136 {
6137         the_hash_algo->update_fn(ctx, str, strlen(str));
6138 }
6139
6140 static void patch_id_add_mode(git_hash_ctx *ctx, unsigned mode)
6141 {
6142         /* large enough for 2^32 in octal */
6143         char buf[12];
6144         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
6145         the_hash_algo->update_fn(ctx, buf, len);
6146 }
6147
6148 /* returns 0 upon success, and writes result into oid */
6149 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only, int stable)
6150 {
6151         struct diff_queue_struct *q = &diff_queued_diff;
6152         int i;
6153         git_hash_ctx ctx;
6154         struct patch_id_t data;
6155
6156         the_hash_algo->init_fn(&ctx);
6157         memset(&data, 0, sizeof(struct patch_id_t));
6158         data.ctx = &ctx;
6159         oidclr(oid);
6160
6161         for (i = 0; i < q->nr; i++) {
6162                 xpparam_t xpp;
6163                 xdemitconf_t xecfg;
6164                 mmfile_t mf1, mf2;
6165                 struct diff_filepair *p = q->queue[i];
6166                 int len1, len2;
6167
6168                 memset(&xpp, 0, sizeof(xpp));
6169                 memset(&xecfg, 0, sizeof(xecfg));
6170                 if (p->status == 0)
6171                         return error("internal diff status error");
6172                 if (p->status == DIFF_STATUS_UNKNOWN)
6173                         continue;
6174                 if (diff_unmodified_pair(p))
6175                         continue;
6176                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
6177                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
6178                         continue;
6179                 if (DIFF_PAIR_UNMERGED(p))
6180                         continue;
6181
6182                 diff_fill_oid_info(p->one, options->repo->index);
6183                 diff_fill_oid_info(p->two, options->repo->index);
6184
6185                 len1 = remove_space(p->one->path, strlen(p->one->path));
6186                 len2 = remove_space(p->two->path, strlen(p->two->path));
6187                 patch_id_add_string(&ctx, "diff--git");
6188                 patch_id_add_string(&ctx, "a/");
6189                 the_hash_algo->update_fn(&ctx, p->one->path, len1);
6190                 patch_id_add_string(&ctx, "b/");
6191                 the_hash_algo->update_fn(&ctx, p->two->path, len2);
6192
6193                 if (p->one->mode == 0) {
6194                         patch_id_add_string(&ctx, "newfilemode");
6195                         patch_id_add_mode(&ctx, p->two->mode);
6196                         patch_id_add_string(&ctx, "---/dev/null");
6197                         patch_id_add_string(&ctx, "+++b/");
6198                         the_hash_algo->update_fn(&ctx, p->two->path, len2);
6199                 } else if (p->two->mode == 0) {
6200                         patch_id_add_string(&ctx, "deletedfilemode");
6201                         patch_id_add_mode(&ctx, p->one->mode);
6202                         patch_id_add_string(&ctx, "---a/");
6203                         the_hash_algo->update_fn(&ctx, p->one->path, len1);
6204                         patch_id_add_string(&ctx, "+++/dev/null");
6205                 } else {
6206                         patch_id_add_string(&ctx, "---a/");
6207                         the_hash_algo->update_fn(&ctx, p->one->path, len1);
6208                         patch_id_add_string(&ctx, "+++b/");
6209                         the_hash_algo->update_fn(&ctx, p->two->path, len2);
6210                 }
6211
6212                 if (diff_header_only)
6213                         continue;
6214
6215                 if (fill_mmfile(options->repo, &mf1, p->one) < 0 ||
6216                     fill_mmfile(options->repo, &mf2, p->two) < 0)
6217                         return error("unable to read files to diff");
6218
6219                 if (diff_filespec_is_binary(options->repo, p->one) ||
6220                     diff_filespec_is_binary(options->repo, p->two)) {
6221                         the_hash_algo->update_fn(&ctx, oid_to_hex(&p->one->oid),
6222                                         the_hash_algo->hexsz);
6223                         the_hash_algo->update_fn(&ctx, oid_to_hex(&p->two->oid),
6224                                         the_hash_algo->hexsz);
6225                         continue;
6226                 }
6227
6228                 xpp.flags = 0;
6229                 xecfg.ctxlen = 3;
6230                 xecfg.flags = 0;
6231                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
6232                                   patch_id_consume, &data, &xpp, &xecfg))
6233                         return error("unable to generate patch-id diff for %s",
6234                                      p->one->path);
6235
6236                 if (stable)
6237                         flush_one_hunk(oid, &ctx);
6238         }
6239
6240         if (!stable)
6241                 the_hash_algo->final_oid_fn(oid, &ctx);
6242
6243         return 0;
6244 }
6245
6246 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only, int stable)
6247 {
6248         struct diff_queue_struct *q = &diff_queued_diff;
6249         int i;
6250         int result = diff_get_patch_id(options, oid, diff_header_only, stable);
6251
6252         for (i = 0; i < q->nr; i++)
6253                 diff_free_filepair(q->queue[i]);
6254
6255         free(q->queue);
6256         DIFF_QUEUE_CLEAR(q);
6257
6258         return result;
6259 }
6260
6261 static int is_summary_empty(const struct diff_queue_struct *q)
6262 {
6263         int i;
6264
6265         for (i = 0; i < q->nr; i++) {
6266                 const struct diff_filepair *p = q->queue[i];
6267
6268                 switch (p->status) {
6269                 case DIFF_STATUS_DELETED:
6270                 case DIFF_STATUS_ADDED:
6271                 case DIFF_STATUS_COPIED:
6272                 case DIFF_STATUS_RENAMED:
6273                         return 0;
6274                 default:
6275                         if (p->score)
6276                                 return 0;
6277                         if (p->one->mode && p->two->mode &&
6278                             p->one->mode != p->two->mode)
6279                                 return 0;
6280                         break;
6281                 }
6282         }
6283         return 1;
6284 }
6285
6286 static const char rename_limit_warning[] =
6287 N_("inexact rename detection was skipped due to too many files.");
6288
6289 static const char degrade_cc_to_c_warning[] =
6290 N_("only found copies from modified paths due to too many files.");
6291
6292 static const char rename_limit_advice[] =
6293 N_("you may want to set your %s variable to at least "
6294    "%d and retry the command.");
6295
6296 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
6297 {
6298         fflush(stdout);
6299         if (degraded_cc)
6300                 warning(_(degrade_cc_to_c_warning));
6301         else if (needed)
6302                 warning(_(rename_limit_warning));
6303         else
6304                 return;
6305         if (0 < needed)
6306                 warning(_(rename_limit_advice), varname, needed);
6307 }
6308
6309 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
6310 {
6311         int i;
6312         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
6313         struct diff_queue_struct *q = &diff_queued_diff;
6314
6315         if (WSEH_NEW & WS_RULE_MASK)
6316                 BUG("WS rules bit mask overlaps with diff symbol flags");
6317
6318         if (o->color_moved)
6319                 o->emitted_symbols = &esm;
6320
6321         for (i = 0; i < q->nr; i++) {
6322                 struct diff_filepair *p = q->queue[i];
6323                 if (check_pair_status(p))
6324                         diff_flush_patch(p, o);
6325         }
6326
6327         if (o->emitted_symbols) {
6328                 if (o->color_moved) {
6329                         struct hashmap add_lines, del_lines;
6330
6331                         if (o->color_moved_ws_handling &
6332                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
6333                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
6334
6335                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
6336                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
6337
6338                         add_lines_to_move_detection(o, &add_lines, &del_lines);
6339                         mark_color_as_moved(o, &add_lines, &del_lines);
6340                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
6341                                 dim_moved_lines(o);
6342
6343                         hashmap_clear_and_free(&add_lines, struct moved_entry,
6344                                                 ent);
6345                         hashmap_clear_and_free(&del_lines, struct moved_entry,
6346                                                 ent);
6347                 }
6348
6349                 for (i = 0; i < esm.nr; i++)
6350                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
6351
6352                 for (i = 0; i < esm.nr; i++)
6353                         free((void *)esm.buf[i].line);
6354                 esm.nr = 0;
6355
6356                 o->emitted_symbols = NULL;
6357         }
6358 }
6359
6360 static void diff_free_file(struct diff_options *options)
6361 {
6362         if (options->close_file)
6363                 fclose(options->file);
6364 }
6365
6366 static void diff_free_ignore_regex(struct diff_options *options)
6367 {
6368         int i;
6369
6370         for (i = 0; i < options->ignore_regex_nr; i++) {
6371                 regfree(options->ignore_regex[i]);
6372                 free(options->ignore_regex[i]);
6373         }
6374         free(options->ignore_regex);
6375 }
6376
6377 void diff_free(struct diff_options *options)
6378 {
6379         if (options->no_free)
6380                 return;
6381
6382         diff_free_file(options);
6383         diff_free_ignore_regex(options);
6384 }
6385
6386 void diff_flush(struct diff_options *options)
6387 {
6388         struct diff_queue_struct *q = &diff_queued_diff;
6389         int i, output_format = options->output_format;
6390         int separator = 0;
6391         int dirstat_by_line = 0;
6392
6393         /*
6394          * Order: raw, stat, summary, patch
6395          * or:    name/name-status/checkdiff (other bits clear)
6396          */
6397         if (!q->nr)
6398                 goto free_queue;
6399
6400         if (output_format & (DIFF_FORMAT_RAW |
6401                              DIFF_FORMAT_NAME |
6402                              DIFF_FORMAT_NAME_STATUS |
6403                              DIFF_FORMAT_CHECKDIFF)) {
6404                 for (i = 0; i < q->nr; i++) {
6405                         struct diff_filepair *p = q->queue[i];
6406                         if (check_pair_status(p))
6407                                 flush_one_pair(p, options);
6408                 }
6409                 separator++;
6410         }
6411
6412         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
6413                 dirstat_by_line = 1;
6414
6415         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
6416             dirstat_by_line) {
6417                 struct diffstat_t diffstat;
6418
6419                 compute_diffstat(options, &diffstat, q);
6420                 if (output_format & DIFF_FORMAT_NUMSTAT)
6421                         show_numstat(&diffstat, options);
6422                 if (output_format & DIFF_FORMAT_DIFFSTAT)
6423                         show_stats(&diffstat, options);
6424                 if (output_format & DIFF_FORMAT_SHORTSTAT)
6425                         show_shortstats(&diffstat, options);
6426                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
6427                         show_dirstat_by_line(&diffstat, options);
6428                 free_diffstat_info(&diffstat);
6429                 separator++;
6430         }
6431         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
6432                 show_dirstat(options);
6433
6434         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
6435                 for (i = 0; i < q->nr; i++) {
6436                         diff_summary(options, q->queue[i]);
6437                 }
6438                 separator++;
6439         }
6440
6441         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
6442             options->flags.exit_with_status &&
6443             options->flags.diff_from_contents) {
6444                 /*
6445                  * run diff_flush_patch for the exit status. setting
6446                  * options->file to /dev/null should be safe, because we
6447                  * aren't supposed to produce any output anyway.
6448                  */
6449                 diff_free_file(options);
6450                 options->file = xfopen("/dev/null", "w");
6451                 options->close_file = 1;
6452                 options->color_moved = 0;
6453                 for (i = 0; i < q->nr; i++) {
6454                         struct diff_filepair *p = q->queue[i];
6455                         if (check_pair_status(p))
6456                                 diff_flush_patch(p, options);
6457                         if (options->found_changes)
6458                                 break;
6459                 }
6460         }
6461
6462         if (output_format & DIFF_FORMAT_PATCH) {
6463                 if (separator) {
6464                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
6465                         if (options->stat_sep)
6466                                 /* attach patch instead of inline */
6467                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
6468                                                  NULL, 0, 0);
6469                 }
6470
6471                 diff_flush_patch_all_file_pairs(options);
6472         }
6473
6474         if (output_format & DIFF_FORMAT_CALLBACK)
6475                 options->format_callback(q, options, options->format_callback_data);
6476
6477         for (i = 0; i < q->nr; i++)
6478                 diff_free_filepair(q->queue[i]);
6479 free_queue:
6480         free(q->queue);
6481         DIFF_QUEUE_CLEAR(q);
6482         diff_free(options);
6483
6484         /*
6485          * Report the content-level differences with HAS_CHANGES;
6486          * diff_addremove/diff_change does not set the bit when
6487          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
6488          */
6489         if (options->flags.diff_from_contents) {
6490                 if (options->found_changes)
6491                         options->flags.has_changes = 1;
6492                 else
6493                         options->flags.has_changes = 0;
6494         }
6495 }
6496
6497 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
6498 {
6499         return (((p->status == DIFF_STATUS_MODIFIED) &&
6500                  ((p->score &&
6501                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
6502                   (!p->score &&
6503                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
6504                 ((p->status != DIFF_STATUS_MODIFIED) &&
6505                  filter_bit_tst(p->status, options)));
6506 }
6507
6508 static void diffcore_apply_filter(struct diff_options *options)
6509 {
6510         int i;
6511         struct diff_queue_struct *q = &diff_queued_diff;
6512         struct diff_queue_struct outq;
6513
6514         DIFF_QUEUE_CLEAR(&outq);
6515
6516         if (!options->filter)
6517                 return;
6518
6519         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
6520                 int found;
6521                 for (i = found = 0; !found && i < q->nr; i++) {
6522                         if (match_filter(options, q->queue[i]))
6523                                 found++;
6524                 }
6525                 if (found)
6526                         return;
6527
6528                 /* otherwise we will clear the whole queue
6529                  * by copying the empty outq at the end of this
6530                  * function, but first clear the current entries
6531                  * in the queue.
6532                  */
6533                 for (i = 0; i < q->nr; i++)
6534                         diff_free_filepair(q->queue[i]);
6535         }
6536         else {
6537                 /* Only the matching ones */
6538                 for (i = 0; i < q->nr; i++) {
6539                         struct diff_filepair *p = q->queue[i];
6540                         if (match_filter(options, p))
6541                                 diff_q(&outq, p);
6542                         else
6543                                 diff_free_filepair(p);
6544                 }
6545         }
6546         free(q->queue);
6547         *q = outq;
6548 }
6549
6550 /* Check whether two filespecs with the same mode and size are identical */
6551 static int diff_filespec_is_identical(struct repository *r,
6552                                       struct diff_filespec *one,
6553                                       struct diff_filespec *two)
6554 {
6555         if (S_ISGITLINK(one->mode))
6556                 return 0;
6557         if (diff_populate_filespec(r, one, NULL))
6558                 return 0;
6559         if (diff_populate_filespec(r, two, NULL))
6560                 return 0;
6561         return !memcmp(one->data, two->data, one->size);
6562 }
6563
6564 static int diff_filespec_check_stat_unmatch(struct repository *r,
6565                                             struct diff_filepair *p)
6566 {
6567         struct diff_populate_filespec_options dpf_options = {
6568                 .check_size_only = 1,
6569                 .missing_object_cb = diff_queued_diff_prefetch,
6570                 .missing_object_data = r,
6571         };
6572
6573         if (p->done_skip_stat_unmatch)
6574                 return p->skip_stat_unmatch_result;
6575
6576         p->done_skip_stat_unmatch = 1;
6577         p->skip_stat_unmatch_result = 0;
6578         /*
6579          * 1. Entries that come from stat info dirtiness
6580          *    always have both sides (iow, not create/delete),
6581          *    one side of the object name is unknown, with
6582          *    the same mode and size.  Keep the ones that
6583          *    do not match these criteria.  They have real
6584          *    differences.
6585          *
6586          * 2. At this point, the file is known to be modified,
6587          *    with the same mode and size, and the object
6588          *    name of one side is unknown.  Need to inspect
6589          *    the identical contents.
6590          */
6591         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6592             !DIFF_FILE_VALID(p->two) ||
6593             (p->one->oid_valid && p->two->oid_valid) ||
6594             (p->one->mode != p->two->mode) ||
6595             diff_populate_filespec(r, p->one, &dpf_options) ||
6596             diff_populate_filespec(r, p->two, &dpf_options) ||
6597             (p->one->size != p->two->size) ||
6598             !diff_filespec_is_identical(r, p->one, p->two)) /* (2) */
6599                 p->skip_stat_unmatch_result = 1;
6600         return p->skip_stat_unmatch_result;
6601 }
6602
6603 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6604 {
6605         int i;
6606         struct diff_queue_struct *q = &diff_queued_diff;
6607         struct diff_queue_struct outq;
6608         DIFF_QUEUE_CLEAR(&outq);
6609
6610         for (i = 0; i < q->nr; i++) {
6611                 struct diff_filepair *p = q->queue[i];
6612
6613                 if (diff_filespec_check_stat_unmatch(diffopt->repo, p))
6614                         diff_q(&outq, p);
6615                 else {
6616                         /*
6617                          * The caller can subtract 1 from skip_stat_unmatch
6618                          * to determine how many paths were dirty only
6619                          * due to stat info mismatch.
6620                          */
6621                         if (!diffopt->flags.no_index)
6622                                 diffopt->skip_stat_unmatch++;
6623                         diff_free_filepair(p);
6624                 }
6625         }
6626         free(q->queue);
6627         *q = outq;
6628 }
6629
6630 static int diffnamecmp(const void *a_, const void *b_)
6631 {
6632         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6633         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6634         const char *name_a, *name_b;
6635
6636         name_a = a->one ? a->one->path : a->two->path;
6637         name_b = b->one ? b->one->path : b->two->path;
6638         return strcmp(name_a, name_b);
6639 }
6640
6641 void diffcore_fix_diff_index(void)
6642 {
6643         struct diff_queue_struct *q = &diff_queued_diff;
6644         QSORT(q->queue, q->nr, diffnamecmp);
6645 }
6646
6647 void diff_add_if_missing(struct repository *r,
6648                          struct oid_array *to_fetch,
6649                          const struct diff_filespec *filespec)
6650 {
6651         if (filespec && filespec->oid_valid &&
6652             !S_ISGITLINK(filespec->mode) &&
6653             oid_object_info_extended(r, &filespec->oid, NULL,
6654                                      OBJECT_INFO_FOR_PREFETCH))
6655                 oid_array_append(to_fetch, &filespec->oid);
6656 }
6657
6658 void diff_queued_diff_prefetch(void *repository)
6659 {
6660         struct repository *repo = repository;
6661         int i;
6662         struct diff_queue_struct *q = &diff_queued_diff;
6663         struct oid_array to_fetch = OID_ARRAY_INIT;
6664
6665         for (i = 0; i < q->nr; i++) {
6666                 struct diff_filepair *p = q->queue[i];
6667                 diff_add_if_missing(repo, &to_fetch, p->one);
6668                 diff_add_if_missing(repo, &to_fetch, p->two);
6669         }
6670
6671         /*
6672          * NEEDSWORK: Consider deduplicating the OIDs sent.
6673          */
6674         promisor_remote_get_direct(repo, to_fetch.oid, to_fetch.nr);
6675
6676         oid_array_clear(&to_fetch);
6677 }
6678
6679 void diffcore_std(struct diff_options *options)
6680 {
6681         int output_formats_to_prefetch = DIFF_FORMAT_DIFFSTAT |
6682                 DIFF_FORMAT_NUMSTAT |
6683                 DIFF_FORMAT_PATCH |
6684                 DIFF_FORMAT_SHORTSTAT |
6685                 DIFF_FORMAT_DIRSTAT;
6686
6687         /*
6688          * Check if the user requested a blob-data-requiring diff output and/or
6689          * break-rewrite detection (which requires blob data). If yes, prefetch
6690          * the diff pairs.
6691          *
6692          * If no prefetching occurs, diffcore_rename() will prefetch if it
6693          * decides that it needs inexact rename detection.
6694          */
6695         if (options->repo == the_repository && has_promisor_remote() &&
6696             (options->output_format & output_formats_to_prefetch ||
6697              options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
6698                 diff_queued_diff_prefetch(options->repo);
6699
6700         /* NOTE please keep the following in sync with diff_tree_combined() */
6701         if (options->skip_stat_unmatch)
6702                 diffcore_skip_stat_unmatch(options);
6703         if (!options->found_follow) {
6704                 /* See try_to_follow_renames() in tree-diff.c */
6705                 if (options->break_opt != -1)
6706                         diffcore_break(options->repo,
6707                                        options->break_opt);
6708                 if (options->detect_rename)
6709                         diffcore_rename(options);
6710                 if (options->break_opt != -1)
6711                         diffcore_merge_broken();
6712         }
6713         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6714                 diffcore_pickaxe(options);
6715         if (options->orderfile)
6716                 diffcore_order(options->orderfile);
6717         if (options->rotate_to)
6718                 diffcore_rotate(options);
6719         if (!options->found_follow)
6720                 /* See try_to_follow_renames() in tree-diff.c */
6721                 diff_resolve_rename_copy();
6722         diffcore_apply_filter(options);
6723
6724         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6725                 options->flags.has_changes = 1;
6726         else
6727                 options->flags.has_changes = 0;
6728
6729         options->found_follow = 0;
6730 }
6731
6732 int diff_result_code(struct diff_options *opt, int status)
6733 {
6734         int result = 0;
6735
6736         diff_warn_rename_limit("diff.renameLimit",
6737                                opt->needed_rename_limit,
6738                                opt->degraded_cc_to_c);
6739         if (!opt->flags.exit_with_status &&
6740             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6741                 return status;
6742         if (opt->flags.exit_with_status &&
6743             opt->flags.has_changes)
6744                 result |= 01;
6745         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6746             opt->flags.check_failed)
6747                 result |= 02;
6748         return result;
6749 }
6750
6751 int diff_can_quit_early(struct diff_options *opt)
6752 {
6753         return (opt->flags.quick &&
6754                 !opt->filter &&
6755                 opt->flags.has_changes);
6756 }
6757
6758 /*
6759  * Shall changes to this submodule be ignored?
6760  *
6761  * Submodule changes can be configured to be ignored separately for each path,
6762  * but that configuration can be overridden from the command line.
6763  */
6764 static int is_submodule_ignored(const char *path, struct diff_options *options)
6765 {
6766         int ignored = 0;
6767         struct diff_flags orig_flags = options->flags;
6768         if (!options->flags.override_submodule_config)
6769                 set_diffopt_flags_from_submodule_config(options, path);
6770         if (options->flags.ignore_submodules)
6771                 ignored = 1;
6772         options->flags = orig_flags;
6773         return ignored;
6774 }
6775
6776 void compute_diffstat(struct diff_options *options,
6777                       struct diffstat_t *diffstat,
6778                       struct diff_queue_struct *q)
6779 {
6780         int i;
6781
6782         memset(diffstat, 0, sizeof(struct diffstat_t));
6783         for (i = 0; i < q->nr; i++) {
6784                 struct diff_filepair *p = q->queue[i];
6785                 if (check_pair_status(p))
6786                         diff_flush_stat(p, options, diffstat);
6787         }
6788 }
6789
6790 void diff_addremove(struct diff_options *options,
6791                     int addremove, unsigned mode,
6792                     const struct object_id *oid,
6793                     int oid_valid,
6794                     const char *concatpath, unsigned dirty_submodule)
6795 {
6796         struct diff_filespec *one, *two;
6797
6798         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6799                 return;
6800
6801         /* This may look odd, but it is a preparation for
6802          * feeding "there are unchanged files which should
6803          * not produce diffs, but when you are doing copy
6804          * detection you would need them, so here they are"
6805          * entries to the diff-core.  They will be prefixed
6806          * with something like '=' or '*' (I haven't decided
6807          * which but should not make any difference).
6808          * Feeding the same new and old to diff_change()
6809          * also has the same effect.
6810          * Before the final output happens, they are pruned after
6811          * merged into rename/copy pairs as appropriate.
6812          */
6813         if (options->flags.reverse_diff)
6814                 addremove = (addremove == '+' ? '-' :
6815                              addremove == '-' ? '+' : addremove);
6816
6817         if (options->prefix &&
6818             strncmp(concatpath, options->prefix, options->prefix_length))
6819                 return;
6820
6821         one = alloc_filespec(concatpath);
6822         two = alloc_filespec(concatpath);
6823
6824         if (addremove != '+')
6825                 fill_filespec(one, oid, oid_valid, mode);
6826         if (addremove != '-') {
6827                 fill_filespec(two, oid, oid_valid, mode);
6828                 two->dirty_submodule = dirty_submodule;
6829         }
6830
6831         diff_queue(&diff_queued_diff, one, two);
6832         if (!options->flags.diff_from_contents)
6833                 options->flags.has_changes = 1;
6834 }
6835
6836 void diff_change(struct diff_options *options,
6837                  unsigned old_mode, unsigned new_mode,
6838                  const struct object_id *old_oid,
6839                  const struct object_id *new_oid,
6840                  int old_oid_valid, int new_oid_valid,
6841                  const char *concatpath,
6842                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6843 {
6844         struct diff_filespec *one, *two;
6845         struct diff_filepair *p;
6846
6847         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6848             is_submodule_ignored(concatpath, options))
6849                 return;
6850
6851         if (options->flags.reverse_diff) {
6852                 SWAP(old_mode, new_mode);
6853                 SWAP(old_oid, new_oid);
6854                 SWAP(old_oid_valid, new_oid_valid);
6855                 SWAP(old_dirty_submodule, new_dirty_submodule);
6856         }
6857
6858         if (options->prefix &&
6859             strncmp(concatpath, options->prefix, options->prefix_length))
6860                 return;
6861
6862         one = alloc_filespec(concatpath);
6863         two = alloc_filespec(concatpath);
6864         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6865         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6866         one->dirty_submodule = old_dirty_submodule;
6867         two->dirty_submodule = new_dirty_submodule;
6868         p = diff_queue(&diff_queued_diff, one, two);
6869
6870         if (options->flags.diff_from_contents)
6871                 return;
6872
6873         if (options->flags.quick && options->skip_stat_unmatch &&
6874             !diff_filespec_check_stat_unmatch(options->repo, p)) {
6875                 diff_free_filespec_data(p->one);
6876                 diff_free_filespec_data(p->two);
6877                 return;
6878         }
6879
6880         options->flags.has_changes = 1;
6881 }
6882
6883 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6884 {
6885         struct diff_filepair *pair;
6886         struct diff_filespec *one, *two;
6887
6888         if (options->prefix &&
6889             strncmp(path, options->prefix, options->prefix_length))
6890                 return NULL;
6891
6892         one = alloc_filespec(path);
6893         two = alloc_filespec(path);
6894         pair = diff_queue(&diff_queued_diff, one, two);
6895         pair->is_unmerged = 1;
6896         return pair;
6897 }
6898
6899 static char *run_textconv(struct repository *r,
6900                           const char *pgm,
6901                           struct diff_filespec *spec,
6902                           size_t *outsize)
6903 {
6904         struct diff_tempfile *temp;
6905         const char *argv[3];
6906         const char **arg = argv;
6907         struct child_process child = CHILD_PROCESS_INIT;
6908         struct strbuf buf = STRBUF_INIT;
6909         int err = 0;
6910
6911         temp = prepare_temp_file(r, spec->path, spec);
6912         *arg++ = pgm;
6913         *arg++ = temp->name;
6914         *arg = NULL;
6915
6916         child.use_shell = 1;
6917         child.argv = argv;
6918         child.out = -1;
6919         if (start_command(&child)) {
6920                 remove_tempfile();
6921                 return NULL;
6922         }
6923
6924         if (strbuf_read(&buf, child.out, 0) < 0)
6925                 err = error("error reading from textconv command '%s'", pgm);
6926         close(child.out);
6927
6928         if (finish_command(&child) || err) {
6929                 strbuf_release(&buf);
6930                 remove_tempfile();
6931                 return NULL;
6932         }
6933         remove_tempfile();
6934
6935         return strbuf_detach(&buf, outsize);
6936 }
6937
6938 size_t fill_textconv(struct repository *r,
6939                      struct userdiff_driver *driver,
6940                      struct diff_filespec *df,
6941                      char **outbuf)
6942 {
6943         size_t size;
6944
6945         if (!driver) {
6946                 if (!DIFF_FILE_VALID(df)) {
6947                         *outbuf = "";
6948                         return 0;
6949                 }
6950                 if (diff_populate_filespec(r, df, NULL))
6951                         die("unable to read files to diff");
6952                 *outbuf = df->data;
6953                 return df->size;
6954         }
6955
6956         if (!driver->textconv)
6957                 BUG("fill_textconv called with non-textconv driver");
6958
6959         if (driver->textconv_cache && df->oid_valid) {
6960                 *outbuf = notes_cache_get(driver->textconv_cache,
6961                                           &df->oid,
6962                                           &size);
6963                 if (*outbuf)
6964                         return size;
6965         }
6966
6967         *outbuf = run_textconv(r, driver->textconv, df, &size);
6968         if (!*outbuf)
6969                 die("unable to read files to diff");
6970
6971         if (driver->textconv_cache && df->oid_valid) {
6972                 /* ignore errors, as we might be in a readonly repository */
6973                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6974                                 size);
6975                 /*
6976                  * we could save up changes and flush them all at the end,
6977                  * but we would need an extra call after all diffing is done.
6978                  * Since generating a cache entry is the slow path anyway,
6979                  * this extra overhead probably isn't a big deal.
6980                  */
6981                 notes_cache_write(driver->textconv_cache);
6982         }
6983
6984         return size;
6985 }
6986
6987 int textconv_object(struct repository *r,
6988                     const char *path,
6989                     unsigned mode,
6990                     const struct object_id *oid,
6991                     int oid_valid,
6992                     char **buf,
6993                     unsigned long *buf_size)
6994 {
6995         struct diff_filespec *df;
6996         struct userdiff_driver *textconv;
6997
6998         df = alloc_filespec(path);
6999         fill_filespec(df, oid, oid_valid, mode);
7000         textconv = get_textconv(r, df);
7001         if (!textconv) {
7002                 free_filespec(df);
7003                 return 0;
7004         }
7005
7006         *buf_size = fill_textconv(r, textconv, df, buf);
7007         free_filespec(df);
7008         return 1;
7009 }
7010
7011 void setup_diff_pager(struct diff_options *opt)
7012 {
7013         /*
7014          * If the user asked for our exit code, then either they want --quiet
7015          * or --exit-code. We should definitely not bother with a pager in the
7016          * former case, as we will generate no output. Since we still properly
7017          * report our exit code even when a pager is run, we _could_ run a
7018          * pager with --exit-code. But since we have not done so historically,
7019          * and because it is easy to find people oneline advising "git diff
7020          * --exit-code" in hooks and other scripts, we do not do so.
7021          */
7022         if (!opt->flags.exit_with_status &&
7023             check_pager_config("diff") != 0)
7024                 setup_pager();
7025 }