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 "userdiff.h"
  17 #include "submodule-config.h"
  18 #include "submodule.h"
  19 #include "hashmap.h"
  20 #include "ll-merge.h"
  21 #include "string-list.h"
  22 #include "argv-array.h"
  23 #include "graph.h"
  24 #include "packfile.h"
  25
  26 #ifdef NO_FAST_WORKING_DIRECTORY
  27 #define FAST_WORKING_DIRECTORY 0
  28 #else
  29 #define FAST_WORKING_DIRECTORY 1
  30 #endif
  31
  32 static int diff_detect_rename_default;
  33 static int diff_indent_heuristic = 1;
  34 static int diff_rename_limit_default = 400;
  35 static int diff_suppress_blank_empty;
  36 static int diff_use_color_default = -1;
  37 static int diff_color_moved_default;
  38 static int diff_context_default = 3;
  39 static int diff_interhunk_context_default;
  40 static const char *diff_word_regex_cfg;
  41 static const char *external_diff_cmd_cfg;
  42 static const char *diff_order_file_cfg;
  43 int diff_auto_refresh_index = 1;
  44 static int diff_mnemonic_prefix;
  45 static int diff_no_prefix;
  46 static int diff_stat_graph_width;
  47 static int diff_dirstat_permille_default = 30;
  48 static struct diff_options default_diff_options;
  49 static long diff_algorithm;
  50 static unsigned ws_error_highlight_default = WSEH_NEW;
  51
  52 static char diff_colors[][COLOR_MAXLEN] = {
  53         GIT_COLOR_RESET,
  54         GIT_COLOR_NORMAL,       /* CONTEXT */
  55         GIT_COLOR_BOLD,         /* METAINFO */
  56         GIT_COLOR_CYAN,         /* FRAGINFO */
  57         GIT_COLOR_RED,          /* OLD */
  58         GIT_COLOR_GREEN,        /* NEW */
  59         GIT_COLOR_YELLOW,       /* COMMIT */
  60         GIT_COLOR_BG_RED,       /* WHITESPACE */
  61         GIT_COLOR_NORMAL,       /* FUNCINFO */
  62         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
  63         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
  64         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
  65         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
  66         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
  67         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
  68         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
  69         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
  70 };
  71
  72 static NORETURN void die_want_option(const char *option_name)
  73 {
  74         die(_("option '%s' requires a value"), option_name);
  75 }
  76
  77 static int parse_diff_color_slot(const char *var)
  78 {
  79         if (!strcasecmp(var, "context") || !strcasecmp(var, "plain"))
  80                 return DIFF_CONTEXT;
  81         if (!strcasecmp(var, "meta"))
  82                 return DIFF_METAINFO;
  83         if (!strcasecmp(var, "frag"))
  84                 return DIFF_FRAGINFO;
  85         if (!strcasecmp(var, "old"))
  86                 return DIFF_FILE_OLD;
  87         if (!strcasecmp(var, "new"))
  88                 return DIFF_FILE_NEW;
  89         if (!strcasecmp(var, "commit"))
  90                 return DIFF_COMMIT;
  91         if (!strcasecmp(var, "whitespace"))
  92                 return DIFF_WHITESPACE;
  93         if (!strcasecmp(var, "func"))
  94                 return DIFF_FUNCINFO;
  95         if (!strcasecmp(var, "oldmoved"))
  96                 return DIFF_FILE_OLD_MOVED;
  97         if (!strcasecmp(var, "oldmovedalternative"))
  98                 return DIFF_FILE_OLD_MOVED_ALT;
  99         if (!strcasecmp(var, "oldmoveddimmed"))
 100                 return DIFF_FILE_OLD_MOVED_DIM;
 101         if (!strcasecmp(var, "oldmovedalternativedimmed"))
 102                 return DIFF_FILE_OLD_MOVED_ALT_DIM;
 103         if (!strcasecmp(var, "newmoved"))
 104                 return DIFF_FILE_NEW_MOVED;
 105         if (!strcasecmp(var, "newmovedalternative"))
 106                 return DIFF_FILE_NEW_MOVED_ALT;
 107         if (!strcasecmp(var, "newmoveddimmed"))
 108                 return DIFF_FILE_NEW_MOVED_DIM;
 109         if (!strcasecmp(var, "newmovedalternativedimmed"))
 110                 return DIFF_FILE_NEW_MOVED_ALT_DIM;
 111         return -1;
 112 }
 113
 114 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 115                                 struct strbuf *errmsg)
 116 {
 117         char *params_copy = xstrdup(params_string);
 118         struct string_list params = STRING_LIST_INIT_NODUP;
 119         int ret = 0;
 120         int i;
 121
 122         if (*params_copy)
 123                 string_list_split_in_place(&params, params_copy, ',', -1);
 124         for (i = 0; i < params.nr; i++) {
 125                 const char *p = params.items[i].string;
 126                 if (!strcmp(p, "changes")) {
 127                         options->flags.dirstat_by_line = 0;
 128                         options->flags.dirstat_by_file = 0;
 129                 } else if (!strcmp(p, "lines")) {
 130                         options->flags.dirstat_by_line = 1;
 131                         options->flags.dirstat_by_file = 0;
 132                 } else if (!strcmp(p, "files")) {
 133                         options->flags.dirstat_by_line = 0;
 134                         options->flags.dirstat_by_file = 1;
 135                 } else if (!strcmp(p, "noncumulative")) {
 136                         options->flags.dirstat_cumulative = 0;
 137                 } else if (!strcmp(p, "cumulative")) {
 138                         options->flags.dirstat_cumulative = 1;
 139                 } else if (isdigit(*p)) {
 140                         char *end;
 141                         int permille = strtoul(p, &end, 10) * 10;
 142                         if (*end == '.' && isdigit(*++end)) {
 143                                 /* only use first digit */
 144                                 permille += *end - '0';
 145                                 /* .. and ignore any further digits */
 146                                 while (isdigit(*++end))
 147                                         ; /* nothing */
 148                         }
 149                         if (!*end)
 150                                 options->dirstat_permille = permille;
 151                         else {
 152                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 153                                             p);
 154                                 ret++;
 155                         }
 156                 } else {
 157                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 158                         ret++;
 159                 }
 160
 161         }
 162         string_list_clear(&params, 0);
 163         free(params_copy);
 164         return ret;
 165 }
 166
 167 static int parse_submodule_params(struct diff_options *options, const char *value)
 168 {
 169         if (!strcmp(value, "log"))
 170                 options->submodule_format = DIFF_SUBMODULE_LOG;
 171         else if (!strcmp(value, "short"))
 172                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 173         else if (!strcmp(value, "diff"))
 174                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 175         else
 176                 return -1;
 177         return 0;
 178 }
 179
 180 static int git_config_rename(const char *var, const char *value)
 181 {
 182         if (!value)
 183                 return DIFF_DETECT_RENAME;
 184         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 185                 return  DIFF_DETECT_COPY;
 186         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 187 }
 188
 189 long parse_algorithm_value(const char *value)
 190 {
 191         if (!value)
 192                 return -1;
 193         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 194                 return 0;
 195         else if (!strcasecmp(value, "minimal"))
 196                 return XDF_NEED_MINIMAL;
 197         else if (!strcasecmp(value, "patience"))
 198                 return XDF_PATIENCE_DIFF;
 199         else if (!strcasecmp(value, "histogram"))
 200                 return XDF_HISTOGRAM_DIFF;
 201         return -1;
 202 }
 203
 204 static int parse_one_token(const char **arg, const char *token)
 205 {
 206         const char *rest;
 207         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 208                 *arg = rest;
 209                 return 1;
 210         }
 211         return 0;
 212 }
 213
 214 static int parse_ws_error_highlight(const char *arg)
 215 {
 216         const char *orig_arg = arg;
 217         unsigned val = 0;
 218
 219         while (*arg) {
 220                 if (parse_one_token(&arg, "none"))
 221                         val = 0;
 222                 else if (parse_one_token(&arg, "default"))
 223                         val = WSEH_NEW;
 224                 else if (parse_one_token(&arg, "all"))
 225                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 226                 else if (parse_one_token(&arg, "new"))
 227                         val |= WSEH_NEW;
 228                 else if (parse_one_token(&arg, "old"))
 229                         val |= WSEH_OLD;
 230                 else if (parse_one_token(&arg, "context"))
 231                         val |= WSEH_CONTEXT;
 232                 else {
 233                         return -1 - (int)(arg - orig_arg);
 234                 }
 235                 if (*arg)
 236                         arg++;
 237         }
 238         return val;
 239 }
 240
 241 /*
 242  * These are to give UI layer defaults.
 243  * The core-level commands such as git-diff-files should
 244  * never be affected by the setting of diff.renames
 245  * the user happens to have in the configuration file.
 246  */
 247 void init_diff_ui_defaults(void)
 248 {
 249         diff_detect_rename_default = DIFF_DETECT_RENAME;
 250 }
 251
 252 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 253 {
 254         if (!strcmp(var, "diff.indentheuristic"))
 255                 diff_indent_heuristic = git_config_bool(var, value);
 256         return 0;
 257 }
 258
 259 static int parse_color_moved(const char *arg)
 260 {
 261         switch (git_parse_maybe_bool(arg)) {
 262         case 0:
 263                 return COLOR_MOVED_NO;
 264         case 1:
 265                 return COLOR_MOVED_DEFAULT;
 266         default:
 267                 break;
 268         }
 269
 270         if (!strcmp(arg, "no"))
 271                 return COLOR_MOVED_NO;
 272         else if (!strcmp(arg, "plain"))
 273                 return COLOR_MOVED_PLAIN;
 274         else if (!strcmp(arg, "blocks"))
 275                 return COLOR_MOVED_BLOCKS;
 276         else if (!strcmp(arg, "zebra"))
 277                 return COLOR_MOVED_ZEBRA;
 278         else if (!strcmp(arg, "default"))
 279                 return COLOR_MOVED_DEFAULT;
 280         else if (!strcmp(arg, "dimmed_zebra"))
 281                 return COLOR_MOVED_ZEBRA_DIM;
 282         else
 283                 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed_zebra', 'plain'"));
 284 }
 285
 286 static int parse_color_moved_ws(const char *arg)
 287 {
 288         int ret = 0;
 289         struct string_list l = STRING_LIST_INIT_DUP;
 290         struct string_list_item *i;
 291
 292         string_list_split(&l, arg, ',', -1);
 293
 294         for_each_string_list_item(i, &l) {
 295                 struct strbuf sb = STRBUF_INIT;
 296                 strbuf_addstr(&sb, i->string);
 297                 strbuf_trim(&sb);
 298
 299                 if (!strcmp(sb.buf, "ignore-space-change"))
 300                         ret |= XDF_IGNORE_WHITESPACE_CHANGE;
 301                 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
 302                         ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
 303                 else if (!strcmp(sb.buf, "ignore-all-space"))
 304                         ret |= XDF_IGNORE_WHITESPACE;
 305                 else if (!strcmp(sb.buf, "allow-indentation-change"))
 306                         ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
 307                 else
 308                         error(_("ignoring unknown color-moved-ws mode '%s'"), sb.buf);
 309
 310                 strbuf_release(&sb);
 311         }
 312
 313         if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
 314             (ret & XDF_WHITESPACE_FLAGS))
 315                 die(_("color-moved-ws: allow-indentation-change cannot be combined with other white space modes"));
 316
 317         string_list_clear(&l, 0);
 318
 319         return ret;
 320 }
 321
 322 int git_diff_ui_config(const char *var, const char *value, void *cb)
 323 {
 324         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 325                 diff_use_color_default = git_config_colorbool(var, value);
 326                 return 0;
 327         }
 328         if (!strcmp(var, "diff.colormoved")) {
 329                 int cm = parse_color_moved(value);
 330                 if (cm < 0)
 331                         return -1;
 332                 diff_color_moved_default = cm;
 333                 return 0;
 334         }
 335         if (!strcmp(var, "diff.context")) {
 336                 diff_context_default = git_config_int(var, value);
 337                 if (diff_context_default < 0)
 338                         return -1;
 339                 return 0;
 340         }
 341         if (!strcmp(var, "diff.interhunkcontext")) {
 342                 diff_interhunk_context_default = git_config_int(var, value);
 343                 if (diff_interhunk_context_default < 0)
 344                         return -1;
 345                 return 0;
 346         }
 347         if (!strcmp(var, "diff.renames")) {
 348                 diff_detect_rename_default = git_config_rename(var, value);
 349                 return 0;
 350         }
 351         if (!strcmp(var, "diff.autorefreshindex")) {
 352                 diff_auto_refresh_index = git_config_bool(var, value);
 353                 return 0;
 354         }
 355         if (!strcmp(var, "diff.mnemonicprefix")) {
 356                 diff_mnemonic_prefix = git_config_bool(var, value);
 357                 return 0;
 358         }
 359         if (!strcmp(var, "diff.noprefix")) {
 360                 diff_no_prefix = git_config_bool(var, value);
 361                 return 0;
 362         }
 363         if (!strcmp(var, "diff.statgraphwidth")) {
 364                 diff_stat_graph_width = git_config_int(var, value);
 365                 return 0;
 366         }
 367         if (!strcmp(var, "diff.external"))
 368                 return git_config_string(&external_diff_cmd_cfg, var, value);
 369         if (!strcmp(var, "diff.wordregex"))
 370                 return git_config_string(&diff_word_regex_cfg, var, value);
 371         if (!strcmp(var, "diff.orderfile"))
 372                 return git_config_pathname(&diff_order_file_cfg, var, value);
 373
 374         if (!strcmp(var, "diff.ignoresubmodules"))
 375                 handle_ignore_submodules_arg(&default_diff_options, value);
 376
 377         if (!strcmp(var, "diff.submodule")) {
 378                 if (parse_submodule_params(&default_diff_options, value))
 379                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 380                                 value);
 381                 return 0;
 382         }
 383
 384         if (!strcmp(var, "diff.algorithm")) {
 385                 diff_algorithm = parse_algorithm_value(value);
 386                 if (diff_algorithm < 0)
 387                         return -1;
 388                 return 0;
 389         }
 390
 391         if (!strcmp(var, "diff.wserrorhighlight")) {
 392                 int val = parse_ws_error_highlight(value);
 393                 if (val < 0)
 394                         return -1;
 395                 ws_error_highlight_default = val;
 396                 return 0;
 397         }
 398
 399         if (git_color_config(var, value, cb) < 0)
 400                 return -1;
 401
 402         return git_diff_basic_config(var, value, cb);
 403 }
 404
 405 int git_diff_basic_config(const char *var, const char *value, void *cb)
 406 {
 407         const char *name;
 408
 409         if (!strcmp(var, "diff.renamelimit")) {
 410                 diff_rename_limit_default = git_config_int(var, value);
 411                 return 0;
 412         }
 413
 414         if (userdiff_config(var, value) < 0)
 415                 return -1;
 416
 417         if (skip_prefix(var, "diff.color.", &name) ||
 418             skip_prefix(var, "color.diff.", &name)) {
 419                 int slot = parse_diff_color_slot(name);
 420                 if (slot < 0)
 421                         return 0;
 422                 if (!value)
 423                         return config_error_nonbool(var);
 424                 return color_parse(value, diff_colors[slot]);
 425         }
 426
 427         /* like GNU diff's --suppress-blank-empty option  */
 428         if (!strcmp(var, "diff.suppressblankempty") ||
 429                         /* for backwards compatibility */
 430                         !strcmp(var, "diff.suppress-blank-empty")) {
 431                 diff_suppress_blank_empty = git_config_bool(var, value);
 432                 return 0;
 433         }
 434
 435         if (!strcmp(var, "diff.dirstat")) {
 436                 struct strbuf errmsg = STRBUF_INIT;
 437                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 438                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 439                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 440                                 errmsg.buf);
 441                 strbuf_release(&errmsg);
 442                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 443                 return 0;
 444         }
 445
 446         if (git_diff_heuristic_config(var, value, cb) < 0)
 447                 return -1;
 448
 449         return git_default_config(var, value, cb);
 450 }
 451
 452 static char *quote_two(const char *one, const char *two)
 453 {
 454         int need_one = quote_c_style(one, NULL, NULL, 1);
 455         int need_two = quote_c_style(two, NULL, NULL, 1);
 456         struct strbuf res = STRBUF_INIT;
 457
 458         if (need_one + need_two) {
 459                 strbuf_addch(&res, '"');
 460                 quote_c_style(one, &res, NULL, 1);
 461                 quote_c_style(two, &res, NULL, 1);
 462                 strbuf_addch(&res, '"');
 463         } else {
 464                 strbuf_addstr(&res, one);
 465                 strbuf_addstr(&res, two);
 466         }
 467         return strbuf_detach(&res, NULL);
 468 }
 469
 470 static const char *external_diff(void)
 471 {
 472         static const char *external_diff_cmd = NULL;
 473         static int done_preparing = 0;
 474
 475         if (done_preparing)
 476                 return external_diff_cmd;
 477         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 478         if (!external_diff_cmd)
 479                 external_diff_cmd = external_diff_cmd_cfg;
 480         done_preparing = 1;
 481         return external_diff_cmd;
 482 }
 483
 484 /*
 485  * Keep track of files used for diffing. Sometimes such an entry
 486  * refers to a temporary file, sometimes to an existing file, and
 487  * sometimes to "/dev/null".
 488  */
 489 static struct diff_tempfile {
 490         /*
 491          * filename external diff should read from, or NULL if this
 492          * entry is currently not in use:
 493          */
 494         const char *name;
 495
 496         char hex[GIT_MAX_HEXSZ + 1];
 497         char mode[10];
 498
 499         /*
 500          * If this diff_tempfile instance refers to a temporary file,
 501          * this tempfile object is used to manage its lifetime.
 502          */
 503         struct tempfile *tempfile;
 504 } diff_temp[2];
 505
 506 struct emit_callback {
 507         int color_diff;
 508         unsigned ws_rule;
 509         int blank_at_eof_in_preimage;
 510         int blank_at_eof_in_postimage;
 511         int lno_in_preimage;
 512         int lno_in_postimage;
 513         const char **label_path;
 514         struct diff_words_data *diff_words;
 515         struct diff_options *opt;
 516         struct strbuf *header;
 517 };
 518
 519 static int count_lines(const char *data, int size)
 520 {
 521         int count, ch, completely_empty = 1, nl_just_seen = 0;
 522         count = 0;
 523         while (0 < size--) {
 524                 ch = *data++;
 525                 if (ch == '\n') {
 526                         count++;
 527                         nl_just_seen = 1;
 528                         completely_empty = 0;
 529                 }
 530                 else {
 531                         nl_just_seen = 0;
 532                         completely_empty = 0;
 533                 }
 534         }
 535         if (completely_empty)
 536                 return 0;
 537         if (!nl_just_seen)
 538                 count++; /* no trailing newline */
 539         return count;
 540 }
 541
 542 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 543 {
 544         if (!DIFF_FILE_VALID(one)) {
 545                 mf->ptr = (char *)""; /* does not matter */
 546                 mf->size = 0;
 547                 return 0;
 548         }
 549         else if (diff_populate_filespec(one, 0))
 550                 return -1;
 551
 552         mf->ptr = one->data;
 553         mf->size = one->size;
 554         return 0;
 555 }
 556
 557 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 558 static unsigned long diff_filespec_size(struct diff_filespec *one)
 559 {
 560         if (!DIFF_FILE_VALID(one))
 561                 return 0;
 562         diff_populate_filespec(one, CHECK_SIZE_ONLY);
 563         return one->size;
 564 }
 565
 566 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 567 {
 568         char *ptr = mf->ptr;
 569         long size = mf->size;
 570         int cnt = 0;
 571
 572         if (!size)
 573                 return cnt;
 574         ptr += size - 1; /* pointing at the very end */
 575         if (*ptr != '\n')
 576                 ; /* incomplete line */
 577         else
 578                 ptr--; /* skip the last LF */
 579         while (mf->ptr < ptr) {
 580                 char *prev_eol;
 581                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 582                         if (*prev_eol == '\n')
 583                                 break;
 584                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 585                         break;
 586                 cnt++;
 587                 ptr = prev_eol - 1;
 588         }
 589         return cnt;
 590 }
 591
 592 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 593                                struct emit_callback *ecbdata)
 594 {
 595         int l1, l2, at;
 596         unsigned ws_rule = ecbdata->ws_rule;
 597         l1 = count_trailing_blank(mf1, ws_rule);
 598         l2 = count_trailing_blank(mf2, ws_rule);
 599         if (l2 <= l1) {
 600                 ecbdata->blank_at_eof_in_preimage = 0;
 601                 ecbdata->blank_at_eof_in_postimage = 0;
 602                 return;
 603         }
 604         at = count_lines(mf1->ptr, mf1->size);
 605         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 606
 607         at = count_lines(mf2->ptr, mf2->size);
 608         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 609 }
 610
 611 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 612                         int first, const char *line, int len)
 613 {
 614         int has_trailing_newline, has_trailing_carriage_return;
 615         int nofirst;
 616         FILE *file = o->file;
 617
 618         fputs(diff_line_prefix(o), file);
 619
 620         if (len == 0) {
 621                 has_trailing_newline = (first == '\n');
 622                 has_trailing_carriage_return = (!has_trailing_newline &&
 623                                                 (first == '\r'));
 624                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 625         } else {
 626                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 627                 if (has_trailing_newline)
 628                         len--;
 629                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 630                 if (has_trailing_carriage_return)
 631                         len--;
 632                 nofirst = 0;
 633         }
 634
 635         if (len || !nofirst) {
 636                 fputs(set, file);
 637                 if (!nofirst)
 638                         fputc(first, file);
 639                 fwrite(line, len, 1, file);
 640                 fputs(reset, file);
 641         }
 642         if (has_trailing_carriage_return)
 643                 fputc('\r', file);
 644         if (has_trailing_newline)
 645                 fputc('\n', file);
 646 }
 647
 648 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 649                       const char *line, int len)
 650 {
 651         emit_line_0(o, set, reset, line[0], line+1, len-1);
 652 }
 653
 654 enum diff_symbol {
 655         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 656         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 657         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 658         DIFF_SYMBOL_BINARY_DIFF_BODY,
 659         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 660         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 661         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 662         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 663         DIFF_SYMBOL_STATS_LINE,
 664         DIFF_SYMBOL_WORD_DIFF,
 665         DIFF_SYMBOL_STAT_SEP,
 666         DIFF_SYMBOL_SUMMARY,
 667         DIFF_SYMBOL_SUBMODULE_ADD,
 668         DIFF_SYMBOL_SUBMODULE_DEL,
 669         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 670         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 671         DIFF_SYMBOL_SUBMODULE_HEADER,
 672         DIFF_SYMBOL_SUBMODULE_ERROR,
 673         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 674         DIFF_SYMBOL_REWRITE_DIFF,
 675         DIFF_SYMBOL_BINARY_FILES,
 676         DIFF_SYMBOL_HEADER,
 677         DIFF_SYMBOL_FILEPAIR_PLUS,
 678         DIFF_SYMBOL_FILEPAIR_MINUS,
 679         DIFF_SYMBOL_WORDS_PORCELAIN,
 680         DIFF_SYMBOL_WORDS,
 681         DIFF_SYMBOL_CONTEXT,
 682         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 683         DIFF_SYMBOL_PLUS,
 684         DIFF_SYMBOL_MINUS,
 685         DIFF_SYMBOL_NO_LF_EOF,
 686         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 687         DIFF_SYMBOL_CONTEXT_MARKER,
 688         DIFF_SYMBOL_SEPARATOR
 689 };
 690 /*
 691  * Flags for content lines:
 692  * 0..12 are whitespace rules
 693  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 694  * 16 is marking if the line is blank at EOF
 695  */
 696 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 697 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 698 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 699 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 700 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 701
 702 /*
 703  * This struct is used when we need to buffer the output of the diff output.
 704  *
 705  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 706  * into the pre/post image file. This pointer could be a union with the
 707  * line pointer. By storing an offset into the file instead of the literal line,
 708  * we can decrease the memory footprint for the buffered output. At first we
 709  * may want to only have indirection for the content lines, but we could also
 710  * enhance the state for emitting prefabricated lines, e.g. the similarity
 711  * score line or hunk/file headers would only need to store a number or path
 712  * and then the output can be constructed later on depending on state.
 713  */
 714 struct emitted_diff_symbol {
 715         const char *line;
 716         int len;
 717         int flags;
 718         enum diff_symbol s;
 719 };
 720 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 721
 722 struct emitted_diff_symbols {
 723         struct emitted_diff_symbol *buf;
 724         int nr, alloc;
 725 };
 726 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 727
 728 static void append_emitted_diff_symbol(struct diff_options *o,
 729                                        struct emitted_diff_symbol *e)
 730 {
 731         struct emitted_diff_symbol *f;
 732
 733         ALLOC_GROW(o->emitted_symbols->buf,
 734                    o->emitted_symbols->nr + 1,
 735                    o->emitted_symbols->alloc);
 736         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 737
 738         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 739         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 740 }
 741
 742 struct moved_entry {
 743         struct hashmap_entry ent;
 744         const struct emitted_diff_symbol *es;
 745         struct moved_entry *next_line;
 746         struct ws_delta *wsd;
 747 };
 748
 749 /**
 750  * The struct ws_delta holds white space differences between moved lines, i.e.
 751  * between '+' and '-' lines that have been detected to be a move.
 752  * The string contains the difference in leading white spaces, before the
 753  * rest of the line is compared using the white space config for move
 754  * coloring. The current_longer indicates if the first string in the
 755  * comparision is longer than the second.
 756  */
 757 struct ws_delta {
 758         char *string;
 759         unsigned int current_longer : 1;
 760 };
 761 #define WS_DELTA_INIT { NULL, 0 }
 762
 763 static int compute_ws_delta(const struct emitted_diff_symbol *a,
 764                              const struct emitted_diff_symbol *b,
 765                              struct ws_delta *out)
 766 {
 767         const struct emitted_diff_symbol *longer =  a->len > b->len ? a : b;
 768         const struct emitted_diff_symbol *shorter = a->len > b->len ? b : a;
 769         int d = longer->len - shorter->len;
 770
 771         out->string = xmemdupz(longer->line, d);
 772         out->current_longer = (a == longer);
 773
 774         return !strncmp(longer->line + d, shorter->line, shorter->len);
 775 }
 776
 777 static int cmp_in_block_with_wsd(const struct diff_options *o,
 778                                  const struct moved_entry *cur,
 779                                  const struct moved_entry *match,
 780                                  struct moved_entry *pmb,
 781                                  int n)
 782 {
 783         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 784         int al = cur->es->len, cl = l->len;
 785         const char *a = cur->es->line,
 786                    *b = match->es->line,
 787                    *c = l->line;
 788
 789         int wslen;
 790
 791         /*
 792          * We need to check if 'cur' is equal to 'match'.
 793          * As those are from the same (+/-) side, we do not need to adjust for
 794          * indent changes. However these were found using fuzzy matching
 795          * so we do have to check if they are equal.
 796          */
 797         if (strcmp(a, b))
 798                 return 1;
 799
 800         if (!pmb->wsd)
 801                 /*
 802                  * No white space delta was carried forward? This can happen
 803                  * when we exit early in this function and do not carry
 804                  * forward ws.
 805                  */
 806                 return 1;
 807
 808         /*
 809          * The indent changes of the block are known and carried forward in
 810          * pmb->wsd; however we need to check if the indent changes of the
 811          * current line are still the same as before.
 812          *
 813          * To do so we need to compare 'l' to 'cur', adjusting the
 814          * one of them for the white spaces, depending which was longer.
 815          */
 816
 817         wslen = strlen(pmb->wsd->string);
 818         if (pmb->wsd->current_longer) {
 819                 c += wslen;
 820                 cl -= wslen;
 821         } else {
 822                 a += wslen;
 823                 al -= wslen;
 824         }
 825
 826         if (strcmp(a, c))
 827                 return 1;
 828
 829         return 0;
 830 }
 831
 832 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
 833                            const void *entry,
 834                            const void *entry_or_key,
 835                            const void *keydata)
 836 {
 837         const struct diff_options *diffopt = hashmap_cmp_fn_data;
 838         const struct moved_entry *a = entry;
 839         const struct moved_entry *b = entry_or_key;
 840         unsigned flags = diffopt->color_moved_ws_handling
 841                          & XDF_WHITESPACE_FLAGS;
 842
 843         if (diffopt->color_moved_ws_handling &
 844             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 845                 /*
 846                  * As there is not specific white space config given,
 847                  * we'd need to check for a new block, so ignore all
 848                  * white space. The setup of the white space
 849                  * configuration for the next block is done else where
 850                  */
 851                 flags |= XDF_IGNORE_WHITESPACE;
 852
 853         return !xdiff_compare_lines(a->es->line, a->es->len,
 854                                     b->es->line, b->es->len,
 855                                     flags);
 856 }
 857
 858 static struct moved_entry *prepare_entry(struct diff_options *o,
 859                                          int line_no)
 860 {
 861         struct moved_entry *ret = xmalloc(sizeof(*ret));
 862         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 863         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 864
 865         ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
 866         ret->es = l;
 867         ret->next_line = NULL;
 868         ret->wsd = NULL;
 869
 870         return ret;
 871 }
 872
 873 static void add_lines_to_move_detection(struct diff_options *o,
 874                                         struct hashmap *add_lines,
 875                                         struct hashmap *del_lines)
 876 {
 877         struct moved_entry *prev_line = NULL;
 878
 879         int n;
 880         for (n = 0; n < o->emitted_symbols->nr; n++) {
 881                 struct hashmap *hm;
 882                 struct moved_entry *key;
 883
 884                 switch (o->emitted_symbols->buf[n].s) {
 885                 case DIFF_SYMBOL_PLUS:
 886                         hm = add_lines;
 887                         break;
 888                 case DIFF_SYMBOL_MINUS:
 889                         hm = del_lines;
 890                         break;
 891                 default:
 892                         prev_line = NULL;
 893                         continue;
 894                 }
 895
 896                 key = prepare_entry(o, n);
 897                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
 898                         prev_line->next_line = key;
 899
 900                 hashmap_add(hm, key);
 901                 prev_line = key;
 902         }
 903 }
 904
 905 static void pmb_advance_or_null(struct diff_options *o,
 906                                 struct moved_entry *match,
 907                                 struct hashmap *hm,
 908                                 struct moved_entry **pmb,
 909                                 int pmb_nr)
 910 {
 911         int i;
 912         for (i = 0; i < pmb_nr; i++) {
 913                 struct moved_entry *prev = pmb[i];
 914                 struct moved_entry *cur = (prev && prev->next_line) ?
 915                                 prev->next_line : NULL;
 916                 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
 917                         pmb[i] = cur;
 918                 } else {
 919                         pmb[i] = NULL;
 920                 }
 921         }
 922 }
 923
 924 static void pmb_advance_or_null_multi_match(struct diff_options *o,
 925                                             struct moved_entry *match,
 926                                             struct hashmap *hm,
 927                                             struct moved_entry **pmb,
 928                                             int pmb_nr, int n)
 929 {
 930         int i;
 931         char *got_match = xcalloc(1, pmb_nr);
 932
 933         for (; match; match = hashmap_get_next(hm, match)) {
 934                 for (i = 0; i < pmb_nr; i++) {
 935                         struct moved_entry *prev = pmb[i];
 936                         struct moved_entry *cur = (prev && prev->next_line) ?
 937                                         prev->next_line : NULL;
 938                         if (!cur)
 939                                 continue;
 940                         if (!cmp_in_block_with_wsd(o, cur, match, pmb[i], n))
 941                                 got_match[i] |= 1;
 942                 }
 943         }
 944
 945         for (i = 0; i < pmb_nr; i++) {
 946                 if (got_match[i]) {
 947                         /* Carry the white space delta forward */
 948                         pmb[i]->next_line->wsd = pmb[i]->wsd;
 949                         pmb[i] = pmb[i]->next_line;
 950                 } else
 951                         pmb[i] = NULL;
 952         }
 953 }
 954
 955 static int shrink_potential_moved_blocks(struct moved_entry **pmb,
 956                                          int pmb_nr)
 957 {
 958         int lp, rp;
 959
 960         /* Shrink the set of potential block to the remaining running */
 961         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
 962                 while (lp < pmb_nr && pmb[lp])
 963                         lp++;
 964                 /* lp points at the first NULL now */
 965
 966                 while (rp > -1 && !pmb[rp])
 967                         rp--;
 968                 /* rp points at the last non-NULL */
 969
 970                 if (lp < pmb_nr && rp > -1 && lp < rp) {
 971                         pmb[lp] = pmb[rp];
 972                         if (pmb[rp]->wsd) {
 973                                 free(pmb[rp]->wsd->string);
 974                                 FREE_AND_NULL(pmb[rp]->wsd);
 975                         }
 976                         pmb[rp] = NULL;
 977                         rp--;
 978                         lp++;
 979                 }
 980         }
 981
 982         /* Remember the number of running sets */
 983         return rp + 1;
 984 }
 985
 986 /*
 987  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
 988  *
 989  * Otherwise, if the last block has fewer alphanumeric characters than
 990  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
 991  * that block.
 992  *
 993  * The last block consists of the (n - block_length)'th line up to but not
 994  * including the nth line.
 995  *
 996  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
 997  * Think of a way to unify them.
 998  */
 999 static void adjust_last_block(struct diff_options *o, int n, int block_length)
1000 {
1001         int i, alnum_count = 0;
1002         if (o->color_moved == COLOR_MOVED_PLAIN)
1003                 return;
1004         for (i = 1; i < block_length + 1; i++) {
1005                 const char *c = o->emitted_symbols->buf[n - i].line;
1006                 for (; *c; c++) {
1007                         if (!isalnum(*c))
1008                                 continue;
1009                         alnum_count++;
1010                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1011                                 return;
1012                 }
1013         }
1014         for (i = 1; i < block_length + 1; i++)
1015                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1016 }
1017
1018 /* Find blocks of moved code, delegate actual coloring decision to helper */
1019 static void mark_color_as_moved(struct diff_options *o,
1020                                 struct hashmap *add_lines,
1021                                 struct hashmap *del_lines)
1022 {
1023         struct moved_entry **pmb = NULL; /* potentially moved blocks */
1024         int pmb_nr = 0, pmb_alloc = 0;
1025         int n, flipped_block = 1, block_length = 0;
1026
1027
1028         for (n = 0; n < o->emitted_symbols->nr; n++) {
1029                 struct hashmap *hm = NULL;
1030                 struct moved_entry *key;
1031                 struct moved_entry *match = NULL;
1032                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1033
1034                 switch (l->s) {
1035                 case DIFF_SYMBOL_PLUS:
1036                         hm = del_lines;
1037                         key = prepare_entry(o, n);
1038                         match = hashmap_get(hm, key, NULL);
1039                         free(key);
1040                         break;
1041                 case DIFF_SYMBOL_MINUS:
1042                         hm = add_lines;
1043                         key = prepare_entry(o, n);
1044                         match = hashmap_get(hm, key, NULL);
1045                         free(key);
1046                         break;
1047                 default:
1048                         flipped_block = 1;
1049                 }
1050
1051                 if (!match) {
1052                         adjust_last_block(o, n, block_length);
1053                         pmb_nr = 0;
1054                         block_length = 0;
1055                         continue;
1056                 }
1057
1058                 l->flags |= DIFF_SYMBOL_MOVED_LINE;
1059
1060                 if (o->color_moved == COLOR_MOVED_PLAIN)
1061                         continue;
1062
1063                 if (o->color_moved_ws_handling &
1064                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1065                         pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1066                 else
1067                         pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1068
1069                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1070
1071                 if (pmb_nr == 0) {
1072                         /*
1073                          * The current line is the start of a new block.
1074                          * Setup the set of potential blocks.
1075                          */
1076                         for (; match; match = hashmap_get_next(hm, match)) {
1077                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1078                                 if (o->color_moved_ws_handling &
1079                                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1080                                         struct ws_delta *wsd = xmalloc(sizeof(*match->wsd));
1081                                         if (compute_ws_delta(l, match->es, wsd)) {
1082                                                 match->wsd = wsd;
1083                                                 pmb[pmb_nr++] = match;
1084                                         } else
1085                                                 free(wsd);
1086                                 } else {
1087                                         pmb[pmb_nr++] = match;
1088                                 }
1089                         }
1090
1091                         flipped_block = (flipped_block + 1) % 2;
1092
1093                         adjust_last_block(o, n, block_length);
1094                         block_length = 0;
1095                 }
1096
1097                 block_length++;
1098
1099                 if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1100                         l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1101         }
1102         adjust_last_block(o, n, block_length);
1103
1104         free(pmb);
1105 }
1106
1107 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1108   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1109 static void dim_moved_lines(struct diff_options *o)
1110 {
1111         int n;
1112         for (n = 0; n < o->emitted_symbols->nr; n++) {
1113                 struct emitted_diff_symbol *prev = (n != 0) ?
1114                                 &o->emitted_symbols->buf[n - 1] : NULL;
1115                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1116                 struct emitted_diff_symbol *next =
1117                                 (n < o->emitted_symbols->nr - 1) ?
1118                                 &o->emitted_symbols->buf[n + 1] : NULL;
1119
1120                 /* Not a plus or minus line? */
1121                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1122                         continue;
1123
1124                 /* Not a moved line? */
1125                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1126                         continue;
1127
1128                 /*
1129                  * If prev or next are not a plus or minus line,
1130                  * pretend they don't exist
1131                  */
1132                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1133                             prev->s != DIFF_SYMBOL_MINUS)
1134                         prev = NULL;
1135                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1136                             next->s != DIFF_SYMBOL_MINUS)
1137                         next = NULL;
1138
1139                 /* Inside a block? */
1140                 if ((prev &&
1141                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1142                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1143                     (next &&
1144                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1145                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1146                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1147                         continue;
1148                 }
1149
1150                 /* Check if we are at an interesting bound: */
1151                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1152                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1153                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1154                         continue;
1155                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1156                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1157                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1158                         continue;
1159
1160                 /*
1161                  * The boundary to prev and next are not interesting,
1162                  * so this line is not interesting as a whole
1163                  */
1164                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1165         }
1166 }
1167
1168 static void emit_line_ws_markup(struct diff_options *o,
1169                                 const char *set, const char *reset,
1170                                 const char *line, int len, char sign,
1171                                 unsigned ws_rule, int blank_at_eof)
1172 {
1173         const char *ws = NULL;
1174
1175         if (o->ws_error_highlight & ws_rule) {
1176                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1177                 if (!*ws)
1178                         ws = NULL;
1179         }
1180
1181         if (!ws)
1182                 emit_line_0(o, set, reset, sign, line, len);
1183         else if (blank_at_eof)
1184                 /* Blank line at EOF - paint '+' as well */
1185                 emit_line_0(o, ws, reset, sign, line, len);
1186         else {
1187                 /* Emit just the prefix, then the rest. */
1188                 emit_line_0(o, set, reset, sign, "", 0);
1189                 ws_check_emit(line, len, ws_rule,
1190                               o->file, set, reset, ws);
1191         }
1192 }
1193
1194 static void emit_diff_symbol_from_struct(struct diff_options *o,
1195                                          struct emitted_diff_symbol *eds)
1196 {
1197         static const char *nneof = " No newline at end of file\n";
1198         const char *context, *reset, *set, *meta, *fraginfo;
1199         struct strbuf sb = STRBUF_INIT;
1200
1201         enum diff_symbol s = eds->s;
1202         const char *line = eds->line;
1203         int len = eds->len;
1204         unsigned flags = eds->flags;
1205
1206         switch (s) {
1207         case DIFF_SYMBOL_NO_LF_EOF:
1208                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1209                 reset = diff_get_color_opt(o, DIFF_RESET);
1210                 putc('\n', o->file);
1211                 emit_line_0(o, context, reset, '\\',
1212                             nneof, strlen(nneof));
1213                 break;
1214         case DIFF_SYMBOL_SUBMODULE_HEADER:
1215         case DIFF_SYMBOL_SUBMODULE_ERROR:
1216         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1217         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1218         case DIFF_SYMBOL_SUMMARY:
1219         case DIFF_SYMBOL_STATS_LINE:
1220         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1221         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1222                 emit_line(o, "", "", line, len);
1223                 break;
1224         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1225         case DIFF_SYMBOL_CONTEXT_MARKER:
1226                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1227                 reset = diff_get_color_opt(o, DIFF_RESET);
1228                 emit_line(o, context, reset, line, len);
1229                 break;
1230         case DIFF_SYMBOL_SEPARATOR:
1231                 fprintf(o->file, "%s%c",
1232                         diff_line_prefix(o),
1233                         o->line_termination);
1234                 break;
1235         case DIFF_SYMBOL_CONTEXT:
1236                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1237                 reset = diff_get_color_opt(o, DIFF_RESET);
1238                 emit_line_ws_markup(o, set, reset, line, len, ' ',
1239                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1240                 break;
1241         case DIFF_SYMBOL_PLUS:
1242                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1243                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1244                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1245                 case DIFF_SYMBOL_MOVED_LINE |
1246                      DIFF_SYMBOL_MOVED_LINE_ALT |
1247                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1248                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1249                         break;
1250                 case DIFF_SYMBOL_MOVED_LINE |
1251                      DIFF_SYMBOL_MOVED_LINE_ALT:
1252                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1253                         break;
1254                 case DIFF_SYMBOL_MOVED_LINE |
1255                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1256                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1257                         break;
1258                 case DIFF_SYMBOL_MOVED_LINE:
1259                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1260                         break;
1261                 default:
1262                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1263                 }
1264                 reset = diff_get_color_opt(o, DIFF_RESET);
1265                 emit_line_ws_markup(o, set, reset, line, len, '+',
1266                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1267                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1268                 break;
1269         case DIFF_SYMBOL_MINUS:
1270                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1271                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1272                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1273                 case DIFF_SYMBOL_MOVED_LINE |
1274                      DIFF_SYMBOL_MOVED_LINE_ALT |
1275                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1276                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1277                         break;
1278                 case DIFF_SYMBOL_MOVED_LINE |
1279                      DIFF_SYMBOL_MOVED_LINE_ALT:
1280                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1281                         break;
1282                 case DIFF_SYMBOL_MOVED_LINE |
1283                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1284                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1285                         break;
1286                 case DIFF_SYMBOL_MOVED_LINE:
1287                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1288                         break;
1289                 default:
1290                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1291                 }
1292                 reset = diff_get_color_opt(o, DIFF_RESET);
1293                 emit_line_ws_markup(o, set, reset, line, len, '-',
1294                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1295                 break;
1296         case DIFF_SYMBOL_WORDS_PORCELAIN:
1297                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1298                 reset = diff_get_color_opt(o, DIFF_RESET);
1299                 emit_line(o, context, reset, line, len);
1300                 fputs("~\n", o->file);
1301                 break;
1302         case DIFF_SYMBOL_WORDS:
1303                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1304                 reset = diff_get_color_opt(o, DIFF_RESET);
1305                 /*
1306                  * Skip the prefix character, if any.  With
1307                  * diff_suppress_blank_empty, there may be
1308                  * none.
1309                  */
1310                 if (line[0] != '\n') {
1311                         line++;
1312                         len--;
1313                 }
1314                 emit_line(o, context, reset, line, len);
1315                 break;
1316         case DIFF_SYMBOL_FILEPAIR_PLUS:
1317                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1318                 reset = diff_get_color_opt(o, DIFF_RESET);
1319                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1320                         line, reset,
1321                         strchr(line, ' ') ? "\t" : "");
1322                 break;
1323         case DIFF_SYMBOL_FILEPAIR_MINUS:
1324                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1325                 reset = diff_get_color_opt(o, DIFF_RESET);
1326                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1327                         line, reset,
1328                         strchr(line, ' ') ? "\t" : "");
1329                 break;
1330         case DIFF_SYMBOL_BINARY_FILES:
1331         case DIFF_SYMBOL_HEADER:
1332                 fprintf(o->file, "%s", line);
1333                 break;
1334         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1335                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1336                 break;
1337         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1338                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1339                 break;
1340         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1341                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1342                 break;
1343         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1344                 fputs(diff_line_prefix(o), o->file);
1345                 fputc('\n', o->file);
1346                 break;
1347         case DIFF_SYMBOL_REWRITE_DIFF:
1348                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1349                 reset = diff_get_color_opt(o, DIFF_RESET);
1350                 emit_line(o, fraginfo, reset, line, len);
1351                 break;
1352         case DIFF_SYMBOL_SUBMODULE_ADD:
1353                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1354                 reset = diff_get_color_opt(o, DIFF_RESET);
1355                 emit_line(o, set, reset, line, len);
1356                 break;
1357         case DIFF_SYMBOL_SUBMODULE_DEL:
1358                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1359                 reset = diff_get_color_opt(o, DIFF_RESET);
1360                 emit_line(o, set, reset, line, len);
1361                 break;
1362         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1363                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1364                         diff_line_prefix(o), line);
1365                 break;
1366         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1367                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1368                         diff_line_prefix(o), line);
1369                 break;
1370         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1371                 emit_line(o, "", "", " 0 files changed\n",
1372                           strlen(" 0 files changed\n"));
1373                 break;
1374         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1375                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1376                 break;
1377         case DIFF_SYMBOL_WORD_DIFF:
1378                 fprintf(o->file, "%.*s", len, line);
1379                 break;
1380         case DIFF_SYMBOL_STAT_SEP:
1381                 fputs(o->stat_sep, o->file);
1382                 break;
1383         default:
1384                 die("BUG: unknown diff symbol");
1385         }
1386         strbuf_release(&sb);
1387 }
1388
1389 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1390                              const char *line, int len, unsigned flags)
1391 {
1392         struct emitted_diff_symbol e = {line, len, flags, s};
1393
1394         if (o->emitted_symbols)
1395                 append_emitted_diff_symbol(o, &e);
1396         else
1397                 emit_diff_symbol_from_struct(o, &e);
1398 }
1399
1400 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1401 {
1402         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1403 }
1404
1405 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1406 {
1407         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1408 }
1409
1410 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1411 {
1412         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1413                          path, strlen(path), 0);
1414 }
1415
1416 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1417 {
1418         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1419                          path, strlen(path), 0);
1420 }
1421
1422 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1423 {
1424         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1425                          header, strlen(header), 0);
1426 }
1427
1428 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1429 {
1430         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1431 }
1432
1433 void diff_emit_submodule_pipethrough(struct diff_options *o,
1434                                      const char *line, int len)
1435 {
1436         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1437 }
1438
1439 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1440 {
1441         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1442               ecbdata->blank_at_eof_in_preimage &&
1443               ecbdata->blank_at_eof_in_postimage &&
1444               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1445               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1446                 return 0;
1447         return ws_blank_line(line, len, ecbdata->ws_rule);
1448 }
1449
1450 static void emit_add_line(const char *reset,
1451                           struct emit_callback *ecbdata,
1452                           const char *line, int len)
1453 {
1454         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1455         if (new_blank_line_at_eof(ecbdata, line, len))
1456                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1457
1458         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1459 }
1460
1461 static void emit_del_line(const char *reset,
1462                           struct emit_callback *ecbdata,
1463                           const char *line, int len)
1464 {
1465         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1466         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1467 }
1468
1469 static void emit_context_line(const char *reset,
1470                               struct emit_callback *ecbdata,
1471                               const char *line, int len)
1472 {
1473         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1474         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1475 }
1476
1477 static void emit_hunk_header(struct emit_callback *ecbdata,
1478                              const char *line, int len)
1479 {
1480         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1481         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1482         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1483         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1484         static const char atat[2] = { '@', '@' };
1485         const char *cp, *ep;
1486         struct strbuf msgbuf = STRBUF_INIT;
1487         int org_len = len;
1488         int i = 1;
1489
1490         /*
1491          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1492          * it always is at least 10 bytes long.
1493          */
1494         if (len < 10 ||
1495             memcmp(line, atat, 2) ||
1496             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1497                 emit_diff_symbol(ecbdata->opt,
1498                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1499                 return;
1500         }
1501         ep += 2; /* skip over @@ */
1502
1503         /* The hunk header in fraginfo color */
1504         strbuf_addstr(&msgbuf, frag);
1505         strbuf_add(&msgbuf, line, ep - line);
1506         strbuf_addstr(&msgbuf, reset);
1507
1508         /*
1509          * trailing "\r\n"
1510          */
1511         for ( ; i < 3; i++)
1512                 if (line[len - i] == '\r' || line[len - i] == '\n')
1513                         len--;
1514
1515         /* blank before the func header */
1516         for (cp = ep; ep - line < len; ep++)
1517                 if (*ep != ' ' && *ep != '\t')
1518                         break;
1519         if (ep != cp) {
1520                 strbuf_addstr(&msgbuf, context);
1521                 strbuf_add(&msgbuf, cp, ep - cp);
1522                 strbuf_addstr(&msgbuf, reset);
1523         }
1524
1525         if (ep < line + len) {
1526                 strbuf_addstr(&msgbuf, func);
1527                 strbuf_add(&msgbuf, ep, line + len - ep);
1528                 strbuf_addstr(&msgbuf, reset);
1529         }
1530
1531         strbuf_add(&msgbuf, line + len, org_len - len);
1532         strbuf_complete_line(&msgbuf);
1533         emit_diff_symbol(ecbdata->opt,
1534                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1535         strbuf_release(&msgbuf);
1536 }
1537
1538 static struct diff_tempfile *claim_diff_tempfile(void) {
1539         int i;
1540         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1541                 if (!diff_temp[i].name)
1542                         return diff_temp + i;
1543         die("BUG: diff is failing to clean up its tempfiles");
1544 }
1545
1546 static void remove_tempfile(void)
1547 {
1548         int i;
1549         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1550                 if (is_tempfile_active(diff_temp[i].tempfile))
1551                         delete_tempfile(&diff_temp[i].tempfile);
1552                 diff_temp[i].name = NULL;
1553         }
1554 }
1555
1556 static void add_line_count(struct strbuf *out, int count)
1557 {
1558         switch (count) {
1559         case 0:
1560                 strbuf_addstr(out, "0,0");
1561                 break;
1562         case 1:
1563                 strbuf_addstr(out, "1");
1564                 break;
1565         default:
1566                 strbuf_addf(out, "1,%d", count);
1567                 break;
1568         }
1569 }
1570
1571 static void emit_rewrite_lines(struct emit_callback *ecb,
1572                                int prefix, const char *data, int size)
1573 {
1574         const char *endp = NULL;
1575         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1576
1577         while (0 < size) {
1578                 int len;
1579
1580                 endp = memchr(data, '\n', size);
1581                 len = endp ? (endp - data + 1) : size;
1582                 if (prefix != '+') {
1583                         ecb->lno_in_preimage++;
1584                         emit_del_line(reset, ecb, data, len);
1585                 } else {
1586                         ecb->lno_in_postimage++;
1587                         emit_add_line(reset, ecb, data, len);
1588                 }
1589                 size -= len;
1590                 data += len;
1591         }
1592         if (!endp)
1593                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1594 }
1595
1596 static void emit_rewrite_diff(const char *name_a,
1597                               const char *name_b,
1598                               struct diff_filespec *one,
1599                               struct diff_filespec *two,
1600                               struct userdiff_driver *textconv_one,
1601                               struct userdiff_driver *textconv_two,
1602                               struct diff_options *o)
1603 {
1604         int lc_a, lc_b;
1605         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1606         const char *a_prefix, *b_prefix;
1607         char *data_one, *data_two;
1608         size_t size_one, size_two;
1609         struct emit_callback ecbdata;
1610         struct strbuf out = STRBUF_INIT;
1611
1612         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1613                 a_prefix = o->b_prefix;
1614                 b_prefix = o->a_prefix;
1615         } else {
1616                 a_prefix = o->a_prefix;
1617                 b_prefix = o->b_prefix;
1618         }
1619
1620         name_a += (*name_a == '/');
1621         name_b += (*name_b == '/');
1622
1623         strbuf_reset(&a_name);
1624         strbuf_reset(&b_name);
1625         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1626         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1627
1628         size_one = fill_textconv(textconv_one, one, &data_one);
1629         size_two = fill_textconv(textconv_two, two, &data_two);
1630
1631         memset(&ecbdata, 0, sizeof(ecbdata));
1632         ecbdata.color_diff = want_color(o->use_color);
1633         ecbdata.ws_rule = whitespace_rule(name_b);
1634         ecbdata.opt = o;
1635         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1636                 mmfile_t mf1, mf2;
1637                 mf1.ptr = (char *)data_one;
1638                 mf2.ptr = (char *)data_two;
1639                 mf1.size = size_one;
1640                 mf2.size = size_two;
1641                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1642         }
1643         ecbdata.lno_in_preimage = 1;
1644         ecbdata.lno_in_postimage = 1;
1645
1646         lc_a = count_lines(data_one, size_one);
1647         lc_b = count_lines(data_two, size_two);
1648
1649         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1650                          a_name.buf, a_name.len, 0);
1651         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1652                          b_name.buf, b_name.len, 0);
1653
1654         strbuf_addstr(&out, "@@ -");
1655         if (!o->irreversible_delete)
1656                 add_line_count(&out, lc_a);
1657         else
1658                 strbuf_addstr(&out, "?,?");
1659         strbuf_addstr(&out, " +");
1660         add_line_count(&out, lc_b);
1661         strbuf_addstr(&out, " @@\n");
1662         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1663         strbuf_release(&out);
1664
1665         if (lc_a && !o->irreversible_delete)
1666                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1667         if (lc_b)
1668                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1669         if (textconv_one)
1670                 free((char *)data_one);
1671         if (textconv_two)
1672                 free((char *)data_two);
1673 }
1674
1675 struct diff_words_buffer {
1676         mmfile_t text;
1677         unsigned long alloc;
1678         struct diff_words_orig {
1679                 const char *begin, *end;
1680         } *orig;
1681         int orig_nr, orig_alloc;
1682 };
1683
1684 static void diff_words_append(char *line, unsigned long len,
1685                 struct diff_words_buffer *buffer)
1686 {
1687         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1688         line++;
1689         len--;
1690         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1691         buffer->text.size += len;
1692         buffer->text.ptr[buffer->text.size] = '\0';
1693 }
1694
1695 struct diff_words_style_elem {
1696         const char *prefix;
1697         const char *suffix;
1698         const char *color; /* NULL; filled in by the setup code if
1699                             * color is enabled */
1700 };
1701
1702 struct diff_words_style {
1703         enum diff_words_type type;
1704         struct diff_words_style_elem new_word, old_word, ctx;
1705         const char *newline;
1706 };
1707
1708 static struct diff_words_style diff_words_styles[] = {
1709         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1710         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1711         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1712 };
1713
1714 struct diff_words_data {
1715         struct diff_words_buffer minus, plus;
1716         const char *current_plus;
1717         int last_minus;
1718         struct diff_options *opt;
1719         regex_t *word_regex;
1720         enum diff_words_type type;
1721         struct diff_words_style *style;
1722 };
1723
1724 static int fn_out_diff_words_write_helper(struct diff_options *o,
1725                                           struct diff_words_style_elem *st_el,
1726                                           const char *newline,
1727                                           size_t count, const char *buf)
1728 {
1729         int print = 0;
1730         struct strbuf sb = STRBUF_INIT;
1731
1732         while (count) {
1733                 char *p = memchr(buf, '\n', count);
1734                 if (print)
1735                         strbuf_addstr(&sb, diff_line_prefix(o));
1736
1737                 if (p != buf) {
1738                         const char *reset = st_el->color && *st_el->color ?
1739                                             GIT_COLOR_RESET : NULL;
1740                         if (st_el->color && *st_el->color)
1741                                 strbuf_addstr(&sb, st_el->color);
1742                         strbuf_addstr(&sb, st_el->prefix);
1743                         strbuf_add(&sb, buf, p ? p - buf : count);
1744                         strbuf_addstr(&sb, st_el->suffix);
1745                         if (reset)
1746                                 strbuf_addstr(&sb, reset);
1747                 }
1748                 if (!p)
1749                         goto out;
1750
1751                 strbuf_addstr(&sb, newline);
1752                 count -= p + 1 - buf;
1753                 buf = p + 1;
1754                 print = 1;
1755                 if (count) {
1756                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1757                                          sb.buf, sb.len, 0);
1758                         strbuf_reset(&sb);
1759                 }
1760         }
1761
1762 out:
1763         if (sb.len)
1764                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1765                                  sb.buf, sb.len, 0);
1766         strbuf_release(&sb);
1767         return 0;
1768 }
1769
1770 /*
1771  * '--color-words' algorithm can be described as:
1772  *
1773  *   1. collect the minus/plus lines of a diff hunk, divided into
1774  *      minus-lines and plus-lines;
1775  *
1776  *   2. break both minus-lines and plus-lines into words and
1777  *      place them into two mmfile_t with one word for each line;
1778  *
1779  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1780  *
1781  * And for the common parts of the both file, we output the plus side text.
1782  * diff_words->current_plus is used to trace the current position of the plus file
1783  * which printed. diff_words->last_minus is used to trace the last minus word
1784  * printed.
1785  *
1786  * For '--graph' to work with '--color-words', we need to output the graph prefix
1787  * on each line of color words output. Generally, there are two conditions on
1788  * which we should output the prefix.
1789  *
1790  *   1. diff_words->last_minus == 0 &&
1791  *      diff_words->current_plus == diff_words->plus.text.ptr
1792  *
1793  *      that is: the plus text must start as a new line, and if there is no minus
1794  *      word printed, a graph prefix must be printed.
1795  *
1796  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1797  *      *(diff_words->current_plus - 1) == '\n'
1798  *
1799  *      that is: a graph prefix must be printed following a '\n'
1800  */
1801 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1802 {
1803         if ((diff_words->last_minus == 0 &&
1804                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1805                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1806                 *(diff_words->current_plus - 1) == '\n')) {
1807                 return 1;
1808         } else {
1809                 return 0;
1810         }
1811 }
1812
1813 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
1814 {
1815         struct diff_words_data *diff_words = priv;
1816         struct diff_words_style *style = diff_words->style;
1817         int minus_first, minus_len, plus_first, plus_len;
1818         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1819         struct diff_options *opt = diff_words->opt;
1820         const char *line_prefix;
1821
1822         if (line[0] != '@' || parse_hunk_header(line, len,
1823                         &minus_first, &minus_len, &plus_first, &plus_len))
1824                 return;
1825
1826         assert(opt);
1827         line_prefix = diff_line_prefix(opt);
1828
1829         /* POSIX requires that first be decremented by one if len == 0... */
1830         if (minus_len) {
1831                 minus_begin = diff_words->minus.orig[minus_first].begin;
1832                 minus_end =
1833                         diff_words->minus.orig[minus_first + minus_len - 1].end;
1834         } else
1835                 minus_begin = minus_end =
1836                         diff_words->minus.orig[minus_first].end;
1837
1838         if (plus_len) {
1839                 plus_begin = diff_words->plus.orig[plus_first].begin;
1840                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1841         } else
1842                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1843
1844         if (color_words_output_graph_prefix(diff_words)) {
1845                 fputs(line_prefix, diff_words->opt->file);
1846         }
1847         if (diff_words->current_plus != plus_begin) {
1848                 fn_out_diff_words_write_helper(diff_words->opt,
1849                                 &style->ctx, style->newline,
1850                                 plus_begin - diff_words->current_plus,
1851                                 diff_words->current_plus);
1852         }
1853         if (minus_begin != minus_end) {
1854                 fn_out_diff_words_write_helper(diff_words->opt,
1855                                 &style->old_word, style->newline,
1856                                 minus_end - minus_begin, minus_begin);
1857         }
1858         if (plus_begin != plus_end) {
1859                 fn_out_diff_words_write_helper(diff_words->opt,
1860                                 &style->new_word, style->newline,
1861                                 plus_end - plus_begin, plus_begin);
1862         }
1863
1864         diff_words->current_plus = plus_end;
1865         diff_words->last_minus = minus_first;
1866 }
1867
1868 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1869 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1870                 int *begin, int *end)
1871 {
1872         if (word_regex && *begin < buffer->size) {
1873                 regmatch_t match[1];
1874                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1875                                  buffer->size - *begin, 1, match, 0)) {
1876                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1877                                         '\n', match[0].rm_eo - match[0].rm_so);
1878                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1879                         *begin += match[0].rm_so;
1880                         return *begin >= *end;
1881                 }
1882                 return -1;
1883         }
1884
1885         /* find the next word */
1886         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1887                 (*begin)++;
1888         if (*begin >= buffer->size)
1889                 return -1;
1890
1891         /* find the end of the word */
1892         *end = *begin + 1;
1893         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1894                 (*end)++;
1895
1896         return 0;
1897 }
1898
1899 /*
1900  * This function splits the words in buffer->text, stores the list with
1901  * newline separator into out, and saves the offsets of the original words
1902  * in buffer->orig.
1903  */
1904 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
1905                 regex_t *word_regex)
1906 {
1907         int i, j;
1908         long alloc = 0;
1909
1910         out->size = 0;
1911         out->ptr = NULL;
1912
1913         /* fake an empty "0th" word */
1914         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
1915         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
1916         buffer->orig_nr = 1;
1917
1918         for (i = 0; i < buffer->text.size; i++) {
1919                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
1920                         return;
1921
1922                 /* store original boundaries */
1923                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
1924                                 buffer->orig_alloc);
1925                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
1926                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
1927                 buffer->orig_nr++;
1928
1929                 /* store one word */
1930                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
1931                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1932                 out->ptr[out->size + j - i] = '\n';
1933                 out->size += j - i + 1;
1934
1935                 i = j - 1;
1936         }
1937 }
1938
1939 /* this executes the word diff on the accumulated buffers */
1940 static void diff_words_show(struct diff_words_data *diff_words)
1941 {
1942         xpparam_t xpp;
1943         xdemitconf_t xecfg;
1944         mmfile_t minus, plus;
1945         struct diff_words_style *style = diff_words->style;
1946
1947         struct diff_options *opt = diff_words->opt;
1948         const char *line_prefix;
1949
1950         assert(opt);
1951         line_prefix = diff_line_prefix(opt);
1952
1953         /* special case: only removal */
1954         if (!diff_words->plus.text.size) {
1955                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
1956                                  line_prefix, strlen(line_prefix), 0);
1957                 fn_out_diff_words_write_helper(diff_words->opt,
1958                         &style->old_word, style->newline,
1959                         diff_words->minus.text.size,
1960                         diff_words->minus.text.ptr);
1961                 diff_words->minus.text.size = 0;
1962                 return;
1963         }
1964
1965         diff_words->current_plus = diff_words->plus.text.ptr;
1966         diff_words->last_minus = 0;
1967
1968         memset(&xpp, 0, sizeof(xpp));
1969         memset(&xecfg, 0, sizeof(xecfg));
1970         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1971         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1972         xpp.flags = 0;
1973         /* as only the hunk header will be parsed, we need a 0-context */
1974         xecfg.ctxlen = 0;
1975         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1976                           &xpp, &xecfg))
1977                 die("unable to generate word diff");
1978         free(minus.ptr);
1979         free(plus.ptr);
1980         if (diff_words->current_plus != diff_words->plus.text.ptr +
1981                         diff_words->plus.text.size) {
1982                 if (color_words_output_graph_prefix(diff_words))
1983                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
1984                                          line_prefix, strlen(line_prefix), 0);
1985                 fn_out_diff_words_write_helper(diff_words->opt,
1986                         &style->ctx, style->newline,
1987                         diff_words->plus.text.ptr + diff_words->plus.text.size
1988                         - diff_words->current_plus, diff_words->current_plus);
1989         }
1990         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1991 }
1992
1993 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1994 static void diff_words_flush(struct emit_callback *ecbdata)
1995 {
1996         struct diff_options *wo = ecbdata->diff_words->opt;
1997
1998         if (ecbdata->diff_words->minus.text.size ||
1999             ecbdata->diff_words->plus.text.size)
2000                 diff_words_show(ecbdata->diff_words);
2001
2002         if (wo->emitted_symbols) {
2003                 struct diff_options *o = ecbdata->opt;
2004                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2005                 int i;
2006
2007                 /*
2008                  * NEEDSWORK:
2009                  * Instead of appending each, concat all words to a line?
2010                  */
2011                 for (i = 0; i < wol->nr; i++)
2012                         append_emitted_diff_symbol(o, &wol->buf[i]);
2013
2014                 for (i = 0; i < wol->nr; i++)
2015                         free((void *)wol->buf[i].line);
2016
2017                 wol->nr = 0;
2018         }
2019 }
2020
2021 static void diff_filespec_load_driver(struct diff_filespec *one)
2022 {
2023         /* Use already-loaded driver */
2024         if (one->driver)
2025                 return;
2026
2027         if (S_ISREG(one->mode))
2028                 one->driver = userdiff_find_by_path(one->path);
2029
2030         /* Fallback to default settings */
2031         if (!one->driver)
2032                 one->driver = userdiff_find_by_name("default");
2033 }
2034
2035 static const char *userdiff_word_regex(struct diff_filespec *one)
2036 {
2037         diff_filespec_load_driver(one);
2038         return one->driver->word_regex;
2039 }
2040
2041 static void init_diff_words_data(struct emit_callback *ecbdata,
2042                                  struct diff_options *orig_opts,
2043                                  struct diff_filespec *one,
2044                                  struct diff_filespec *two)
2045 {
2046         int i;
2047         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2048         memcpy(o, orig_opts, sizeof(struct diff_options));
2049
2050         ecbdata->diff_words =
2051                 xcalloc(1, sizeof(struct diff_words_data));
2052         ecbdata->diff_words->type = o->word_diff;
2053         ecbdata->diff_words->opt = o;
2054
2055         if (orig_opts->emitted_symbols)
2056                 o->emitted_symbols =
2057                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2058
2059         if (!o->word_regex)
2060                 o->word_regex = userdiff_word_regex(one);
2061         if (!o->word_regex)
2062                 o->word_regex = userdiff_word_regex(two);
2063         if (!o->word_regex)
2064                 o->word_regex = diff_word_regex_cfg;
2065         if (o->word_regex) {
2066                 ecbdata->diff_words->word_regex = (regex_t *)
2067                         xmalloc(sizeof(regex_t));
2068                 if (regcomp(ecbdata->diff_words->word_regex,
2069                             o->word_regex,
2070                             REG_EXTENDED | REG_NEWLINE))
2071                         die ("Invalid regular expression: %s",
2072                              o->word_regex);
2073         }
2074         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2075                 if (o->word_diff == diff_words_styles[i].type) {
2076                         ecbdata->diff_words->style =
2077                                 &diff_words_styles[i];
2078                         break;
2079                 }
2080         }
2081         if (want_color(o->use_color)) {
2082                 struct diff_words_style *st = ecbdata->diff_words->style;
2083                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2084                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2085                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2086         }
2087 }
2088
2089 static void free_diff_words_data(struct emit_callback *ecbdata)
2090 {
2091         if (ecbdata->diff_words) {
2092                 diff_words_flush(ecbdata);
2093                 free (ecbdata->diff_words->opt->emitted_symbols);
2094                 free (ecbdata->diff_words->opt);
2095                 free (ecbdata->diff_words->minus.text.ptr);
2096                 free (ecbdata->diff_words->minus.orig);
2097                 free (ecbdata->diff_words->plus.text.ptr);
2098                 free (ecbdata->diff_words->plus.orig);
2099                 if (ecbdata->diff_words->word_regex) {
2100                         regfree(ecbdata->diff_words->word_regex);
2101                         free(ecbdata->diff_words->word_regex);
2102                 }
2103                 FREE_AND_NULL(ecbdata->diff_words);
2104         }
2105 }
2106
2107 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2108 {
2109         if (want_color(diff_use_color))
2110                 return diff_colors[ix];
2111         return "";
2112 }
2113
2114 const char *diff_line_prefix(struct diff_options *opt)
2115 {
2116         struct strbuf *msgbuf;
2117         if (!opt->output_prefix)
2118                 return "";
2119
2120         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2121         return msgbuf->buf;
2122 }
2123
2124 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
2125 {
2126         const char *cp;
2127         unsigned long allot;
2128         size_t l = len;
2129
2130         cp = line;
2131         allot = l;
2132         while (0 < l) {
2133                 (void) utf8_width(&cp, &l);
2134                 if (!cp)
2135                         break; /* truncated in the middle? */
2136         }
2137         return allot - l;
2138 }
2139
2140 static void find_lno(const char *line, struct emit_callback *ecbdata)
2141 {
2142         const char *p;
2143         ecbdata->lno_in_preimage = 0;
2144         ecbdata->lno_in_postimage = 0;
2145         p = strchr(line, '-');
2146         if (!p)
2147                 return; /* cannot happen */
2148         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2149         p = strchr(p, '+');
2150         if (!p)
2151                 return; /* cannot happen */
2152         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2153 }
2154
2155 static void fn_out_consume(void *priv, char *line, unsigned long len)
2156 {
2157         struct emit_callback *ecbdata = priv;
2158         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
2159         struct diff_options *o = ecbdata->opt;
2160
2161         o->found_changes = 1;
2162
2163         if (ecbdata->header) {
2164                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2165                                  ecbdata->header->buf, ecbdata->header->len, 0);
2166                 strbuf_reset(ecbdata->header);
2167                 ecbdata->header = NULL;
2168         }
2169
2170         if (ecbdata->label_path[0]) {
2171                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2172                                  ecbdata->label_path[0],
2173                                  strlen(ecbdata->label_path[0]), 0);
2174                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2175                                  ecbdata->label_path[1],
2176                                  strlen(ecbdata->label_path[1]), 0);
2177                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2178         }
2179
2180         if (diff_suppress_blank_empty
2181             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2182                 line[0] = '\n';
2183                 len = 1;
2184         }
2185
2186         if (line[0] == '@') {
2187                 if (ecbdata->diff_words)
2188                         diff_words_flush(ecbdata);
2189                 len = sane_truncate_line(ecbdata, line, len);
2190                 find_lno(line, ecbdata);
2191                 emit_hunk_header(ecbdata, line, len);
2192                 return;
2193         }
2194
2195         if (ecbdata->diff_words) {
2196                 enum diff_symbol s =
2197                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2198                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2199                 if (line[0] == '-') {
2200                         diff_words_append(line, len,
2201                                           &ecbdata->diff_words->minus);
2202                         return;
2203                 } else if (line[0] == '+') {
2204                         diff_words_append(line, len,
2205                                           &ecbdata->diff_words->plus);
2206                         return;
2207                 } else if (starts_with(line, "\\ ")) {
2208                         /*
2209                          * Eat the "no newline at eof" marker as if we
2210                          * saw a "+" or "-" line with nothing on it,
2211                          * and return without diff_words_flush() to
2212                          * defer processing. If this is the end of
2213                          * preimage, more "+" lines may come after it.
2214                          */
2215                         return;
2216                 }
2217                 diff_words_flush(ecbdata);
2218                 emit_diff_symbol(o, s, line, len, 0);
2219                 return;
2220         }
2221
2222         switch (line[0]) {
2223         case '+':
2224                 ecbdata->lno_in_postimage++;
2225                 emit_add_line(reset, ecbdata, line + 1, len - 1);
2226                 break;
2227         case '-':
2228                 ecbdata->lno_in_preimage++;
2229                 emit_del_line(reset, ecbdata, line + 1, len - 1);
2230                 break;
2231         case ' ':
2232                 ecbdata->lno_in_postimage++;
2233                 ecbdata->lno_in_preimage++;
2234                 emit_context_line(reset, ecbdata, line + 1, len - 1);
2235                 break;
2236         default:
2237                 /* incomplete line at the end */
2238                 ecbdata->lno_in_preimage++;
2239                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2240                                  line, len, 0);
2241                 break;
2242         }
2243 }
2244
2245 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2246 {
2247         const char *old_name = a;
2248         const char *new_name = b;
2249         int pfx_length, sfx_length;
2250         int pfx_adjust_for_slash;
2251         int len_a = strlen(a);
2252         int len_b = strlen(b);
2253         int a_midlen, b_midlen;
2254         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2255         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2256
2257         if (qlen_a || qlen_b) {
2258                 quote_c_style(a, name, NULL, 0);
2259                 strbuf_addstr(name, " => ");
2260                 quote_c_style(b, name, NULL, 0);
2261                 return;
2262         }
2263
2264         /* Find common prefix */
2265         pfx_length = 0;
2266         while (*old_name && *new_name && *old_name == *new_name) {
2267                 if (*old_name == '/')
2268                         pfx_length = old_name - a + 1;
2269                 old_name++;
2270                 new_name++;
2271         }
2272
2273         /* Find common suffix */
2274         old_name = a + len_a;
2275         new_name = b + len_b;
2276         sfx_length = 0;
2277         /*
2278          * If there is a common prefix, it must end in a slash.  In
2279          * that case we let this loop run 1 into the prefix to see the
2280          * same slash.
2281          *
2282          * If there is no common prefix, we cannot do this as it would
2283          * underrun the input strings.
2284          */
2285         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2286         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2287                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2288                *old_name == *new_name) {
2289                 if (*old_name == '/')
2290                         sfx_length = len_a - (old_name - a);
2291                 old_name--;
2292                 new_name--;
2293         }
2294
2295         /*
2296          * pfx{mid-a => mid-b}sfx
2297          * {pfx-a => pfx-b}sfx
2298          * pfx{sfx-a => sfx-b}
2299          * name-a => name-b
2300          */
2301         a_midlen = len_a - pfx_length - sfx_length;
2302         b_midlen = len_b - pfx_length - sfx_length;
2303         if (a_midlen < 0)
2304                 a_midlen = 0;
2305         if (b_midlen < 0)
2306                 b_midlen = 0;
2307
2308         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2309         if (pfx_length + sfx_length) {
2310                 strbuf_add(name, a, pfx_length);
2311                 strbuf_addch(name, '{');
2312         }
2313         strbuf_add(name, a + pfx_length, a_midlen);
2314         strbuf_addstr(name, " => ");
2315         strbuf_add(name, b + pfx_length, b_midlen);
2316         if (pfx_length + sfx_length) {
2317                 strbuf_addch(name, '}');
2318                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2319         }
2320 }
2321
2322 struct diffstat_t {
2323         int nr;
2324         int alloc;
2325         struct diffstat_file {
2326                 char *from_name;
2327                 char *name;
2328                 char *print_name;
2329                 const char *comments;
2330                 unsigned is_unmerged:1;
2331                 unsigned is_binary:1;
2332                 unsigned is_renamed:1;
2333                 unsigned is_interesting:1;
2334                 uintmax_t added, deleted;
2335         } **files;
2336 };
2337
2338 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2339                                           const char *name_a,
2340                                           const char *name_b)
2341 {
2342         struct diffstat_file *x;
2343         x = xcalloc(1, sizeof(*x));
2344         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2345         diffstat->files[diffstat->nr++] = x;
2346         if (name_b) {
2347                 x->from_name = xstrdup(name_a);
2348                 x->name = xstrdup(name_b);
2349                 x->is_renamed = 1;
2350         }
2351         else {
2352                 x->from_name = NULL;
2353                 x->name = xstrdup(name_a);
2354         }
2355         return x;
2356 }
2357
2358 static void diffstat_consume(void *priv, char *line, unsigned long len)
2359 {
2360         struct diffstat_t *diffstat = priv;
2361         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2362
2363         if (line[0] == '+')
2364                 x->added++;
2365         else if (line[0] == '-')
2366                 x->deleted++;
2367 }
2368
2369 const char mime_boundary_leader[] = "------------";
2370
2371 static int scale_linear(int it, int width, int max_change)
2372 {
2373         if (!it)
2374                 return 0;
2375         /*
2376          * make sure that at least one '-' or '+' is printed if
2377          * there is any change to this path. The easiest way is to
2378          * scale linearly as if the alloted width is one column shorter
2379          * than it is, and then add 1 to the result.
2380          */
2381         return 1 + (it * (width - 1) / max_change);
2382 }
2383
2384 static void show_graph(struct strbuf *out, char ch, int cnt,
2385                        const char *set, const char *reset)
2386 {
2387         if (cnt <= 0)
2388                 return;
2389         strbuf_addstr(out, set);
2390         strbuf_addchars(out, ch, cnt);
2391         strbuf_addstr(out, reset);
2392 }
2393
2394 static void fill_print_name(struct diffstat_file *file)
2395 {
2396         struct strbuf pname = STRBUF_INIT;
2397
2398         if (file->print_name)
2399                 return;
2400
2401         if (file->is_renamed)
2402                 pprint_rename(&pname, file->from_name, file->name);
2403         else
2404                 quote_c_style(file->name, &pname, NULL, 0);
2405
2406         if (file->comments)
2407                 strbuf_addf(&pname, " (%s)", file->comments);
2408
2409         file->print_name = strbuf_detach(&pname, NULL);
2410 }
2411
2412 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2413                 int files, int insertions, int deletions)
2414 {
2415         struct strbuf sb = STRBUF_INIT;
2416
2417         if (!files) {
2418                 assert(insertions == 0 && deletions == 0);
2419                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2420                                  NULL, 0, 0);
2421                 return;
2422         }
2423
2424         strbuf_addf(&sb,
2425                     (files == 1) ? " %d file changed" : " %d files changed",
2426                     files);
2427
2428         /*
2429          * For binary diff, the caller may want to print "x files
2430          * changed" with insertions == 0 && deletions == 0.
2431          *
2432          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2433          * is probably less confusing (i.e skip over "2 files changed
2434          * but nothing about added/removed lines? Is this a bug in Git?").
2435          */
2436         if (insertions || deletions == 0) {
2437                 strbuf_addf(&sb,
2438                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2439                             insertions);
2440         }
2441
2442         if (deletions || insertions == 0) {
2443                 strbuf_addf(&sb,
2444                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2445                             deletions);
2446         }
2447         strbuf_addch(&sb, '\n');
2448         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2449                          sb.buf, sb.len, 0);
2450         strbuf_release(&sb);
2451 }
2452
2453 void print_stat_summary(FILE *fp, int files,
2454                         int insertions, int deletions)
2455 {
2456         struct diff_options o;
2457         memset(&o, 0, sizeof(o));
2458         o.file = fp;
2459
2460         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2461 }
2462
2463 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2464 {
2465         int i, len, add, del, adds = 0, dels = 0;
2466         uintmax_t max_change = 0, max_len = 0;
2467         int total_files = data->nr, count;
2468         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2469         const char *reset, *add_c, *del_c;
2470         int extra_shown = 0;
2471         const char *line_prefix = diff_line_prefix(options);
2472         struct strbuf out = STRBUF_INIT;
2473
2474         if (data->nr == 0)
2475                 return;
2476
2477         count = options->stat_count ? options->stat_count : data->nr;
2478
2479         reset = diff_get_color_opt(options, DIFF_RESET);
2480         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2481         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2482
2483         /*
2484          * Find the longest filename and max number of changes
2485          */
2486         for (i = 0; (i < count) && (i < data->nr); i++) {
2487                 struct diffstat_file *file = data->files[i];
2488                 uintmax_t change = file->added + file->deleted;
2489
2490                 if (!file->is_interesting && (change == 0)) {
2491                         count++; /* not shown == room for one more */
2492                         continue;
2493                 }
2494                 fill_print_name(file);
2495                 len = strlen(file->print_name);
2496                 if (max_len < len)
2497                         max_len = len;
2498
2499                 if (file->is_unmerged) {
2500                         /* "Unmerged" is 8 characters */
2501                         bin_width = bin_width < 8 ? 8 : bin_width;
2502                         continue;
2503                 }
2504                 if (file->is_binary) {
2505                         /* "Bin XXX -> YYY bytes" */
2506                         int w = 14 + decimal_width(file->added)
2507                                 + decimal_width(file->deleted);
2508                         bin_width = bin_width < w ? w : bin_width;
2509                         /* Display change counts aligned with "Bin" */
2510                         number_width = 3;
2511                         continue;
2512                 }
2513
2514                 if (max_change < change)
2515                         max_change = change;
2516         }
2517         count = i; /* where we can stop scanning in data->files[] */
2518
2519         /*
2520          * We have width = stat_width or term_columns() columns total.
2521          * We want a maximum of min(max_len, stat_name_width) for the name part.
2522          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2523          * We also need 1 for " " and 4 + decimal_width(max_change)
2524          * for " | NNNN " and one the empty column at the end, altogether
2525          * 6 + decimal_width(max_change).
2526          *
2527          * If there's not enough space, we will use the smaller of
2528          * stat_name_width (if set) and 5/8*width for the filename,
2529          * and the rest for constant elements + graph part, but no more
2530          * than stat_graph_width for the graph part.
2531          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2532          * for the standard terminal size).
2533          *
2534          * In other words: stat_width limits the maximum width, and
2535          * stat_name_width fixes the maximum width of the filename,
2536          * and is also used to divide available columns if there
2537          * aren't enough.
2538          *
2539          * Binary files are displayed with "Bin XXX -> YYY bytes"
2540          * instead of the change count and graph. This part is treated
2541          * similarly to the graph part, except that it is not
2542          * "scaled". If total width is too small to accommodate the
2543          * guaranteed minimum width of the filename part and the
2544          * separators and this message, this message will "overflow"
2545          * making the line longer than the maximum width.
2546          */
2547
2548         if (options->stat_width == -1)
2549                 width = term_columns() - strlen(line_prefix);
2550         else
2551                 width = options->stat_width ? options->stat_width : 80;
2552         number_width = decimal_width(max_change) > number_width ?
2553                 decimal_width(max_change) : number_width;
2554
2555         if (options->stat_graph_width == -1)
2556                 options->stat_graph_width = diff_stat_graph_width;
2557
2558         /*
2559          * Guarantee 3/8*16==6 for the graph part
2560          * and 5/8*16==10 for the filename part
2561          */
2562         if (width < 16 + 6 + number_width)
2563                 width = 16 + 6 + number_width;
2564
2565         /*
2566          * First assign sizes that are wanted, ignoring available width.
2567          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2568          * starting from "XXX" should fit in graph_width.
2569          */
2570         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2571         if (options->stat_graph_width &&
2572             options->stat_graph_width < graph_width)
2573                 graph_width = options->stat_graph_width;
2574
2575         name_width = (options->stat_name_width > 0 &&
2576                       options->stat_name_width < max_len) ?
2577                 options->stat_name_width : max_len;
2578
2579         /*
2580          * Adjust adjustable widths not to exceed maximum width
2581          */
2582         if (name_width + number_width + 6 + graph_width > width) {
2583                 if (graph_width > width * 3/8 - number_width - 6) {
2584                         graph_width = width * 3/8 - number_width - 6;
2585                         if (graph_width < 6)
2586                                 graph_width = 6;
2587                 }
2588
2589                 if (options->stat_graph_width &&
2590                     graph_width > options->stat_graph_width)
2591                         graph_width = options->stat_graph_width;
2592                 if (name_width > width - number_width - 6 - graph_width)
2593                         name_width = width - number_width - 6 - graph_width;
2594                 else
2595                         graph_width = width - number_width - 6 - name_width;
2596         }
2597
2598         /*
2599          * From here name_width is the width of the name area,
2600          * and graph_width is the width of the graph area.
2601          * max_change is used to scale graph properly.
2602          */
2603         for (i = 0; i < count; i++) {
2604                 const char *prefix = "";
2605                 struct diffstat_file *file = data->files[i];
2606                 char *name = file->print_name;
2607                 uintmax_t added = file->added;
2608                 uintmax_t deleted = file->deleted;
2609                 int name_len;
2610
2611                 if (!file->is_interesting && (added + deleted == 0))
2612                         continue;
2613
2614                 /*
2615                  * "scale" the filename
2616                  */
2617                 len = name_width;
2618                 name_len = strlen(name);
2619                 if (name_width < name_len) {
2620                         char *slash;
2621                         prefix = "...";
2622                         len -= 3;
2623                         name += name_len - len;
2624                         slash = strchr(name, '/');
2625                         if (slash)
2626                                 name = slash;
2627                 }
2628
2629                 if (file->is_binary) {
2630                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2631                         strbuf_addf(&out, " %*s", number_width, "Bin");
2632                         if (!added && !deleted) {
2633                                 strbuf_addch(&out, '\n');
2634                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2635                                                  out.buf, out.len, 0);
2636                                 strbuf_reset(&out);
2637                                 continue;
2638                         }
2639                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2640                                 del_c, deleted, reset);
2641                         strbuf_addstr(&out, " -> ");
2642                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2643                                 add_c, added, reset);
2644                         strbuf_addstr(&out, " bytes\n");
2645                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2646                                          out.buf, out.len, 0);
2647                         strbuf_reset(&out);
2648                         continue;
2649                 }
2650                 else if (file->is_unmerged) {
2651                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2652                         strbuf_addstr(&out, " Unmerged\n");
2653                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2654                                          out.buf, out.len, 0);
2655                         strbuf_reset(&out);
2656                         continue;
2657                 }
2658
2659                 /*
2660                  * scale the add/delete
2661                  */
2662                 add = added;
2663                 del = deleted;
2664
2665                 if (graph_width <= max_change) {
2666                         int total = scale_linear(add + del, graph_width, max_change);
2667                         if (total < 2 && add && del)
2668                                 /* width >= 2 due to the sanity check */
2669                                 total = 2;
2670                         if (add < del) {
2671                                 add = scale_linear(add, graph_width, max_change);
2672                                 del = total - add;
2673                         } else {
2674                                 del = scale_linear(del, graph_width, max_change);
2675                                 add = total - del;
2676                         }
2677                 }
2678                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2679                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2680                         number_width, added + deleted,
2681                         added + deleted ? " " : "");
2682                 show_graph(&out, '+', add, add_c, reset);
2683                 show_graph(&out, '-', del, del_c, reset);
2684                 strbuf_addch(&out, '\n');
2685                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2686                                  out.buf, out.len, 0);
2687                 strbuf_reset(&out);
2688         }
2689
2690         for (i = 0; i < data->nr; i++) {
2691                 struct diffstat_file *file = data->files[i];
2692                 uintmax_t added = file->added;
2693                 uintmax_t deleted = file->deleted;
2694
2695                 if (file->is_unmerged ||
2696                     (!file->is_interesting && (added + deleted == 0))) {
2697                         total_files--;
2698                         continue;
2699                 }
2700
2701                 if (!file->is_binary) {
2702                         adds += added;
2703                         dels += deleted;
2704                 }
2705                 if (i < count)
2706                         continue;
2707                 if (!extra_shown)
2708                         emit_diff_symbol(options,
2709                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2710                                          NULL, 0, 0);
2711                 extra_shown = 1;
2712         }
2713
2714         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2715         strbuf_release(&out);
2716 }
2717
2718 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2719 {
2720         int i, adds = 0, dels = 0, total_files = data->nr;
2721
2722         if (data->nr == 0)
2723                 return;
2724
2725         for (i = 0; i < data->nr; i++) {
2726                 int added = data->files[i]->added;
2727                 int deleted = data->files[i]->deleted;
2728
2729                 if (data->files[i]->is_unmerged ||
2730                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2731                         total_files--;
2732                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2733                         adds += added;
2734                         dels += deleted;
2735                 }
2736         }
2737         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2738 }
2739
2740 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2741 {
2742         int i;
2743
2744         if (data->nr == 0)
2745                 return;
2746
2747         for (i = 0; i < data->nr; i++) {
2748                 struct diffstat_file *file = data->files[i];
2749
2750                 fprintf(options->file, "%s", diff_line_prefix(options));
2751
2752                 if (file->is_binary)
2753                         fprintf(options->file, "-\t-\t");
2754                 else
2755                         fprintf(options->file,
2756                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2757                                 file->added, file->deleted);
2758                 if (options->line_termination) {
2759                         fill_print_name(file);
2760                         if (!file->is_renamed)
2761                                 write_name_quoted(file->name, options->file,
2762                                                   options->line_termination);
2763                         else {
2764                                 fputs(file->print_name, options->file);
2765                                 putc(options->line_termination, options->file);
2766                         }
2767                 } else {
2768                         if (file->is_renamed) {
2769                                 putc('\0', options->file);
2770                                 write_name_quoted(file->from_name, options->file, '\0');
2771                         }
2772                         write_name_quoted(file->name, options->file, '\0');
2773                 }
2774         }
2775 }
2776
2777 struct dirstat_file {
2778         const char *name;
2779         unsigned long changed;
2780 };
2781
2782 struct dirstat_dir {
2783         struct dirstat_file *files;
2784         int alloc, nr, permille, cumulative;
2785 };
2786
2787 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2788                 unsigned long changed, const char *base, int baselen)
2789 {
2790         unsigned long sum_changes = 0;
2791         unsigned int sources = 0;
2792         const char *line_prefix = diff_line_prefix(opt);
2793
2794         while (dir->nr) {
2795                 struct dirstat_file *f = dir->files;
2796                 int namelen = strlen(f->name);
2797                 unsigned long changes;
2798                 char *slash;
2799
2800                 if (namelen < baselen)
2801                         break;
2802                 if (memcmp(f->name, base, baselen))
2803                         break;
2804                 slash = strchr(f->name + baselen, '/');
2805                 if (slash) {
2806                         int newbaselen = slash + 1 - f->name;
2807                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2808                         sources++;
2809                 } else {
2810                         changes = f->changed;
2811                         dir->files++;
2812                         dir->nr--;
2813                         sources += 2;
2814                 }
2815                 sum_changes += changes;
2816         }
2817
2818         /*
2819          * We don't report dirstat's for
2820          *  - the top level
2821          *  - or cases where everything came from a single directory
2822          *    under this directory (sources == 1).
2823          */
2824         if (baselen && sources != 1) {
2825                 if (sum_changes) {
2826                         int permille = sum_changes * 1000 / changed;
2827                         if (permille >= dir->permille) {
2828                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2829                                         permille / 10, permille % 10, baselen, base);
2830                                 if (!dir->cumulative)
2831                                         return 0;
2832                         }
2833                 }
2834         }
2835         return sum_changes;
2836 }
2837
2838 static int dirstat_compare(const void *_a, const void *_b)
2839 {
2840         const struct dirstat_file *a = _a;
2841         const struct dirstat_file *b = _b;
2842         return strcmp(a->name, b->name);
2843 }
2844
2845 static void show_dirstat(struct diff_options *options)
2846 {
2847         int i;
2848         unsigned long changed;
2849         struct dirstat_dir dir;
2850         struct diff_queue_struct *q = &diff_queued_diff;
2851
2852         dir.files = NULL;
2853         dir.alloc = 0;
2854         dir.nr = 0;
2855         dir.permille = options->dirstat_permille;
2856         dir.cumulative = options->flags.dirstat_cumulative;
2857
2858         changed = 0;
2859         for (i = 0; i < q->nr; i++) {
2860                 struct diff_filepair *p = q->queue[i];
2861                 const char *name;
2862                 unsigned long copied, added, damage;
2863                 int content_changed;
2864
2865                 name = p->two->path ? p->two->path : p->one->path;
2866
2867                 if (p->one->oid_valid && p->two->oid_valid)
2868                         content_changed = oidcmp(&p->one->oid, &p->two->oid);
2869                 else
2870                         content_changed = 1;
2871
2872                 if (!content_changed) {
2873                         /*
2874                          * The SHA1 has not changed, so pre-/post-content is
2875                          * identical. We can therefore skip looking at the
2876                          * file contents altogether.
2877                          */
2878                         damage = 0;
2879                         goto found_damage;
2880                 }
2881
2882                 if (options->flags.dirstat_by_file) {
2883                         /*
2884                          * In --dirstat-by-file mode, we don't really need to
2885                          * look at the actual file contents at all.
2886                          * The fact that the SHA1 changed is enough for us to
2887                          * add this file to the list of results
2888                          * (with each file contributing equal damage).
2889                          */
2890                         damage = 1;
2891                         goto found_damage;
2892                 }
2893
2894                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2895                         diff_populate_filespec(p->one, 0);
2896                         diff_populate_filespec(p->two, 0);
2897                         diffcore_count_changes(p->one, p->two, NULL, NULL,
2898                                                &copied, &added);
2899                         diff_free_filespec_data(p->one);
2900                         diff_free_filespec_data(p->two);
2901                 } else if (DIFF_FILE_VALID(p->one)) {
2902                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
2903                         copied = added = 0;
2904                         diff_free_filespec_data(p->one);
2905                 } else if (DIFF_FILE_VALID(p->two)) {
2906                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
2907                         copied = 0;
2908                         added = p->two->size;
2909                         diff_free_filespec_data(p->two);
2910                 } else
2911                         continue;
2912
2913                 /*
2914                  * Original minus copied is the removed material,
2915                  * added is the new material.  They are both damages
2916                  * made to the preimage.
2917                  * If the resulting damage is zero, we know that
2918                  * diffcore_count_changes() considers the two entries to
2919                  * be identical, but since content_changed is true, we
2920                  * know that there must have been _some_ kind of change,
2921                  * so we force all entries to have damage > 0.
2922                  */
2923                 damage = (p->one->size - copied) + added;
2924                 if (!damage)
2925                         damage = 1;
2926
2927 found_damage:
2928                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2929                 dir.files[dir.nr].name = name;
2930                 dir.files[dir.nr].changed = damage;
2931                 changed += damage;
2932                 dir.nr++;
2933         }
2934
2935         /* This can happen even with many files, if everything was renames */
2936         if (!changed)
2937                 return;
2938
2939         /* Show all directories with more than x% of the changes */
2940         QSORT(dir.files, dir.nr, dirstat_compare);
2941         gather_dirstat(options, &dir, changed, "", 0);
2942 }
2943
2944 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2945 {
2946         int i;
2947         unsigned long changed;
2948         struct dirstat_dir dir;
2949
2950         if (data->nr == 0)
2951                 return;
2952
2953         dir.files = NULL;
2954         dir.alloc = 0;
2955         dir.nr = 0;
2956         dir.permille = options->dirstat_permille;
2957         dir.cumulative = options->flags.dirstat_cumulative;
2958
2959         changed = 0;
2960         for (i = 0; i < data->nr; i++) {
2961                 struct diffstat_file *file = data->files[i];
2962                 unsigned long damage = file->added + file->deleted;
2963                 if (file->is_binary)
2964                         /*
2965                          * binary files counts bytes, not lines. Must find some
2966                          * way to normalize binary bytes vs. textual lines.
2967                          * The following heuristic assumes that there are 64
2968                          * bytes per "line".
2969                          * This is stupid and ugly, but very cheap...
2970                          */
2971                         damage = DIV_ROUND_UP(damage, 64);
2972                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2973                 dir.files[dir.nr].name = file->name;
2974                 dir.files[dir.nr].changed = damage;
2975                 changed += damage;
2976                 dir.nr++;
2977         }
2978
2979         /* This can happen even with many files, if everything was renames */
2980         if (!changed)
2981                 return;
2982
2983         /* Show all directories with more than x% of the changes */
2984         QSORT(dir.files, dir.nr, dirstat_compare);
2985         gather_dirstat(options, &dir, changed, "", 0);
2986 }
2987
2988 static void free_diffstat_info(struct diffstat_t *diffstat)
2989 {
2990         int i;
2991         for (i = 0; i < diffstat->nr; i++) {
2992                 struct diffstat_file *f = diffstat->files[i];
2993                 free(f->print_name);
2994                 free(f->name);
2995                 free(f->from_name);
2996                 free(f);
2997         }
2998         free(diffstat->files);
2999 }
3000
3001 struct checkdiff_t {
3002         const char *filename;
3003         int lineno;
3004         int conflict_marker_size;
3005         struct diff_options *o;
3006         unsigned ws_rule;
3007         unsigned status;
3008 };
3009
3010 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3011 {
3012         char firstchar;
3013         int cnt;
3014
3015         if (len < marker_size + 1)
3016                 return 0;
3017         firstchar = line[0];
3018         switch (firstchar) {
3019         case '=': case '>': case '<': case '|':
3020                 break;
3021         default:
3022                 return 0;
3023         }
3024         for (cnt = 1; cnt < marker_size; cnt++)
3025                 if (line[cnt] != firstchar)
3026                         return 0;
3027         /* line[1] thru line[marker_size-1] are same as firstchar */
3028         if (len < marker_size + 1 || !isspace(line[marker_size]))
3029                 return 0;
3030         return 1;
3031 }
3032
3033 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3034 {
3035         struct checkdiff_t *data = priv;
3036         int marker_size = data->conflict_marker_size;
3037         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3038         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3039         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3040         char *err;
3041         const char *line_prefix;
3042
3043         assert(data->o);
3044         line_prefix = diff_line_prefix(data->o);
3045
3046         if (line[0] == '+') {
3047                 unsigned bad;
3048                 data->lineno++;
3049                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3050                         data->status |= 1;
3051                         fprintf(data->o->file,
3052                                 "%s%s:%d: leftover conflict marker\n",
3053                                 line_prefix, data->filename, data->lineno);
3054                 }
3055                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3056                 if (!bad)
3057                         return;
3058                 data->status |= bad;
3059                 err = whitespace_error_string(bad);
3060                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3061                         line_prefix, data->filename, data->lineno, err);
3062                 free(err);
3063                 emit_line(data->o, set, reset, line, 1);
3064                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3065                               data->o->file, set, reset, ws);
3066         } else if (line[0] == ' ') {
3067                 data->lineno++;
3068         } else if (line[0] == '@') {
3069                 char *plus = strchr(line, '+');
3070                 if (plus)
3071                         data->lineno = strtol(plus, NULL, 10) - 1;
3072                 else
3073                         die("invalid diff");
3074         }
3075 }
3076
3077 static unsigned char *deflate_it(char *data,
3078                                  unsigned long size,
3079                                  unsigned long *result_size)
3080 {
3081         int bound;
3082         unsigned char *deflated;
3083         git_zstream stream;
3084
3085         git_deflate_init(&stream, zlib_compression_level);
3086         bound = git_deflate_bound(&stream, size);
3087         deflated = xmalloc(bound);
3088         stream.next_out = deflated;
3089         stream.avail_out = bound;
3090
3091         stream.next_in = (unsigned char *)data;
3092         stream.avail_in = size;
3093         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3094                 ; /* nothing */
3095         git_deflate_end(&stream);
3096         *result_size = stream.total_out;
3097         return deflated;
3098 }
3099
3100 static void emit_binary_diff_body(struct diff_options *o,
3101                                   mmfile_t *one, mmfile_t *two)
3102 {
3103         void *cp;
3104         void *delta;
3105         void *deflated;
3106         void *data;
3107         unsigned long orig_size;
3108         unsigned long delta_size;
3109         unsigned long deflate_size;
3110         unsigned long data_size;
3111
3112         /* We could do deflated delta, or we could do just deflated two,
3113          * whichever is smaller.
3114          */
3115         delta = NULL;
3116         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3117         if (one->size && two->size) {
3118                 delta = diff_delta(one->ptr, one->size,
3119                                    two->ptr, two->size,
3120                                    &delta_size, deflate_size);
3121                 if (delta) {
3122                         void *to_free = delta;
3123                         orig_size = delta_size;
3124                         delta = deflate_it(delta, delta_size, &delta_size);
3125                         free(to_free);
3126                 }
3127         }
3128
3129         if (delta && delta_size < deflate_size) {
3130                 char *s = xstrfmt("%lu", orig_size);
3131                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3132                                  s, strlen(s), 0);
3133                 free(s);
3134                 free(deflated);
3135                 data = delta;
3136                 data_size = delta_size;
3137         } else {
3138                 char *s = xstrfmt("%lu", two->size);
3139                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3140                                  s, strlen(s), 0);
3141                 free(s);
3142                 free(delta);
3143                 data = deflated;
3144                 data_size = deflate_size;
3145         }
3146
3147         /* emit data encoded in base85 */
3148         cp = data;
3149         while (data_size) {
3150                 int len;
3151                 int bytes = (52 < data_size) ? 52 : data_size;
3152                 char line[71];
3153                 data_size -= bytes;
3154                 if (bytes <= 26)
3155                         line[0] = bytes + 'A' - 1;
3156                 else
3157                         line[0] = bytes - 26 + 'a' - 1;
3158                 encode_85(line + 1, cp, bytes);
3159                 cp = (char *) cp + bytes;
3160
3161                 len = strlen(line);
3162                 line[len++] = '\n';
3163                 line[len] = '\0';
3164
3165                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3166                                  line, len, 0);
3167         }
3168         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3169         free(data);
3170 }
3171
3172 static void emit_binary_diff(struct diff_options *o,
3173                              mmfile_t *one, mmfile_t *two)
3174 {
3175         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3176         emit_binary_diff_body(o, one, two);
3177         emit_binary_diff_body(o, two, one);
3178 }
3179
3180 int diff_filespec_is_binary(struct diff_filespec *one)
3181 {
3182         if (one->is_binary == -1) {
3183                 diff_filespec_load_driver(one);
3184                 if (one->driver->binary != -1)
3185                         one->is_binary = one->driver->binary;
3186                 else {
3187                         if (!one->data && DIFF_FILE_VALID(one))
3188                                 diff_populate_filespec(one, CHECK_BINARY);
3189                         if (one->is_binary == -1 && one->data)
3190                                 one->is_binary = buffer_is_binary(one->data,
3191                                                 one->size);
3192                         if (one->is_binary == -1)
3193                                 one->is_binary = 0;
3194                 }
3195         }
3196         return one->is_binary;
3197 }
3198
3199 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
3200 {
3201         diff_filespec_load_driver(one);
3202         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3203 }
3204
3205 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3206 {
3207         if (!options->a_prefix)
3208                 options->a_prefix = a;
3209         if (!options->b_prefix)
3210                 options->b_prefix = b;
3211 }
3212
3213 struct userdiff_driver *get_textconv(struct diff_filespec *one)
3214 {
3215         if (!DIFF_FILE_VALID(one))
3216                 return NULL;
3217
3218         diff_filespec_load_driver(one);
3219         return userdiff_get_textconv(one->driver);
3220 }
3221
3222 static void builtin_diff(const char *name_a,
3223                          const char *name_b,
3224                          struct diff_filespec *one,
3225                          struct diff_filespec *two,
3226                          const char *xfrm_msg,
3227                          int must_show_header,
3228                          struct diff_options *o,
3229                          int complete_rewrite)
3230 {
3231         mmfile_t mf1, mf2;
3232         const char *lbl[2];
3233         char *a_one, *b_two;
3234         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3235         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3236         const char *a_prefix, *b_prefix;
3237         struct userdiff_driver *textconv_one = NULL;
3238         struct userdiff_driver *textconv_two = NULL;
3239         struct strbuf header = STRBUF_INIT;
3240         const char *line_prefix = diff_line_prefix(o);
3241
3242         diff_set_mnemonic_prefix(o, "a/", "b/");
3243         if (o->flags.reverse_diff) {
3244                 a_prefix = o->b_prefix;
3245                 b_prefix = o->a_prefix;
3246         } else {
3247                 a_prefix = o->a_prefix;
3248                 b_prefix = o->b_prefix;
3249         }
3250
3251         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3252             (!one->mode || S_ISGITLINK(one->mode)) &&
3253             (!two->mode || S_ISGITLINK(two->mode))) {
3254                 show_submodule_summary(o, one->path ? one->path : two->path,
3255                                 &one->oid, &two->oid,
3256                                 two->dirty_submodule);
3257                 return;
3258         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3259                    (!one->mode || S_ISGITLINK(one->mode)) &&
3260                    (!two->mode || S_ISGITLINK(two->mode))) {
3261                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3262                                 &one->oid, &two->oid,
3263                                 two->dirty_submodule);
3264                 return;
3265         }
3266
3267         if (o->flags.allow_textconv) {
3268                 textconv_one = get_textconv(one);
3269                 textconv_two = get_textconv(two);
3270         }
3271
3272         /* Never use a non-valid filename anywhere if at all possible */
3273         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3274         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3275
3276         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3277         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3278         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3279         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3280         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3281         if (lbl[0][0] == '/') {
3282                 /* /dev/null */
3283                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3284                 if (xfrm_msg)
3285                         strbuf_addstr(&header, xfrm_msg);
3286                 must_show_header = 1;
3287         }
3288         else if (lbl[1][0] == '/') {
3289                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3290                 if (xfrm_msg)
3291                         strbuf_addstr(&header, xfrm_msg);
3292                 must_show_header = 1;
3293         }
3294         else {
3295                 if (one->mode != two->mode) {
3296                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3297                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3298                         must_show_header = 1;
3299                 }
3300                 if (xfrm_msg)
3301                         strbuf_addstr(&header, xfrm_msg);
3302
3303                 /*
3304                  * we do not run diff between different kind
3305                  * of objects.
3306                  */
3307                 if ((one->mode ^ two->mode) & S_IFMT)
3308                         goto free_ab_and_return;
3309                 if (complete_rewrite &&
3310                     (textconv_one || !diff_filespec_is_binary(one)) &&
3311                     (textconv_two || !diff_filespec_is_binary(two))) {
3312                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3313                                          header.buf, header.len, 0);
3314                         strbuf_reset(&header);
3315                         emit_rewrite_diff(name_a, name_b, one, two,
3316                                                 textconv_one, textconv_two, o);
3317                         o->found_changes = 1;
3318                         goto free_ab_and_return;
3319                 }
3320         }
3321
3322         if (o->irreversible_delete && lbl[1][0] == '/') {
3323                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3324                                  header.len, 0);
3325                 strbuf_reset(&header);
3326                 goto free_ab_and_return;
3327         } else if (!o->flags.text &&
3328             ( (!textconv_one && diff_filespec_is_binary(one)) ||
3329               (!textconv_two && diff_filespec_is_binary(two)) )) {
3330                 struct strbuf sb = STRBUF_INIT;
3331                 if (!one->data && !two->data &&
3332                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3333                     !o->flags.binary) {
3334                         if (!oidcmp(&one->oid, &two->oid)) {
3335                                 if (must_show_header)
3336                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3337                                                          header.buf, header.len,
3338                                                          0);
3339                                 goto free_ab_and_return;
3340                         }
3341                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3342                                          header.buf, header.len, 0);
3343                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3344                                     diff_line_prefix(o), lbl[0], lbl[1]);
3345                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3346                                          sb.buf, sb.len, 0);
3347                         strbuf_release(&sb);
3348                         goto free_ab_and_return;
3349                 }
3350                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3351                         die("unable to read files to diff");
3352                 /* Quite common confusing case */
3353                 if (mf1.size == mf2.size &&
3354                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3355                         if (must_show_header)
3356                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3357                                                  header.buf, header.len, 0);
3358                         goto free_ab_and_return;
3359                 }
3360                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3361                 strbuf_reset(&header);
3362                 if (o->flags.binary)
3363                         emit_binary_diff(o, &mf1, &mf2);
3364                 else {
3365                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3366                                     diff_line_prefix(o), lbl[0], lbl[1]);
3367                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3368                                          sb.buf, sb.len, 0);
3369                         strbuf_release(&sb);
3370                 }
3371                 o->found_changes = 1;
3372         } else {
3373                 /* Crazy xdl interfaces.. */
3374                 const char *diffopts = getenv("GIT_DIFF_OPTS");
3375                 const char *v;
3376                 xpparam_t xpp;
3377                 xdemitconf_t xecfg;
3378                 struct emit_callback ecbdata;
3379                 const struct userdiff_funcname *pe;
3380
3381                 if (must_show_header) {
3382                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3383                                          header.buf, header.len, 0);
3384                         strbuf_reset(&header);
3385                 }
3386
3387                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
3388                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
3389
3390                 pe = diff_funcname_pattern(one);
3391                 if (!pe)
3392                         pe = diff_funcname_pattern(two);
3393
3394                 memset(&xpp, 0, sizeof(xpp));
3395                 memset(&xecfg, 0, sizeof(xecfg));
3396                 memset(&ecbdata, 0, sizeof(ecbdata));
3397                 ecbdata.label_path = lbl;
3398                 ecbdata.color_diff = want_color(o->use_color);
3399                 ecbdata.ws_rule = whitespace_rule(name_b);
3400                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3401                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3402                 ecbdata.opt = o;
3403                 ecbdata.header = header.len ? &header : NULL;
3404                 xpp.flags = o->xdl_opts;
3405                 xpp.anchors = o->anchors;
3406                 xpp.anchors_nr = o->anchors_nr;
3407                 xecfg.ctxlen = o->context;
3408                 xecfg.interhunkctxlen = o->interhunkcontext;
3409                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3410                 if (o->flags.funccontext)
3411                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3412                 if (pe)
3413                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3414                 if (!diffopts)
3415                         ;
3416                 else if (skip_prefix(diffopts, "--unified=", &v))
3417                         xecfg.ctxlen = strtoul(v, NULL, 10);
3418                 else if (skip_prefix(diffopts, "-u", &v))
3419                         xecfg.ctxlen = strtoul(v, NULL, 10);
3420                 if (o->word_diff)
3421                         init_diff_words_data(&ecbdata, o, one, two);
3422                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
3423                                   &xpp, &xecfg))
3424                         die("unable to generate diff for %s", one->path);
3425                 if (o->word_diff)
3426                         free_diff_words_data(&ecbdata);
3427                 if (textconv_one)
3428                         free(mf1.ptr);
3429                 if (textconv_two)
3430                         free(mf2.ptr);
3431                 xdiff_clear_find_func(&xecfg);
3432         }
3433
3434  free_ab_and_return:
3435         strbuf_release(&header);
3436         diff_free_filespec_data(one);
3437         diff_free_filespec_data(two);
3438         free(a_one);
3439         free(b_two);
3440         return;
3441 }
3442
3443 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3444 {
3445         if (!is_renamed) {
3446                 if (p->status == DIFF_STATUS_ADDED) {
3447                         if (S_ISLNK(p->two->mode))
3448                                 return "new +l";
3449                         else if ((p->two->mode & 0777) == 0755)
3450                                 return "new +x";
3451                         else
3452                                 return "new";
3453                 } else if (p->status == DIFF_STATUS_DELETED)
3454                         return "gone";
3455         }
3456         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3457                 return "mode -l";
3458         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3459                 return "mode +l";
3460         else if ((p->one->mode & 0777) == 0644 &&
3461                  (p->two->mode & 0777) == 0755)
3462                 return "mode +x";
3463         else if ((p->one->mode & 0777) == 0755 &&
3464                  (p->two->mode & 0777) == 0644)
3465                 return "mode -x";
3466         return NULL;
3467 }
3468
3469 static void builtin_diffstat(const char *name_a, const char *name_b,
3470                              struct diff_filespec *one,
3471                              struct diff_filespec *two,
3472                              struct diffstat_t *diffstat,
3473                              struct diff_options *o,
3474                              struct diff_filepair *p)
3475 {
3476         mmfile_t mf1, mf2;
3477         struct diffstat_file *data;
3478         int same_contents;
3479         int complete_rewrite = 0;
3480
3481         if (!DIFF_PAIR_UNMERGED(p)) {
3482                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3483                         complete_rewrite = 1;
3484         }
3485
3486         data = diffstat_add(diffstat, name_a, name_b);
3487         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3488         if (o->flags.stat_with_summary)
3489                 data->comments = get_compact_summary(p, data->is_renamed);
3490
3491         if (!one || !two) {
3492                 data->is_unmerged = 1;
3493                 return;
3494         }
3495
3496         same_contents = !oidcmp(&one->oid, &two->oid);
3497
3498         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
3499                 data->is_binary = 1;
3500                 if (same_contents) {
3501                         data->added = 0;
3502                         data->deleted = 0;
3503                 } else {
3504                         data->added = diff_filespec_size(two);
3505                         data->deleted = diff_filespec_size(one);
3506                 }
3507         }
3508
3509         else if (complete_rewrite) {
3510                 diff_populate_filespec(one, 0);
3511                 diff_populate_filespec(two, 0);
3512                 data->deleted = count_lines(one->data, one->size);
3513                 data->added = count_lines(two->data, two->size);
3514         }
3515
3516         else if (!same_contents) {
3517                 /* Crazy xdl interfaces.. */
3518                 xpparam_t xpp;
3519                 xdemitconf_t xecfg;
3520
3521                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3522                         die("unable to read files to diff");
3523
3524                 memset(&xpp, 0, sizeof(xpp));
3525                 memset(&xecfg, 0, sizeof(xecfg));
3526                 xpp.flags = o->xdl_opts;
3527                 xpp.anchors = o->anchors;
3528                 xpp.anchors_nr = o->anchors_nr;
3529                 xecfg.ctxlen = o->context;
3530                 xecfg.interhunkctxlen = o->interhunkcontext;
3531                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
3532                                   &xpp, &xecfg))
3533                         die("unable to generate diffstat for %s", one->path);
3534         }
3535
3536         diff_free_filespec_data(one);
3537         diff_free_filespec_data(two);
3538 }
3539
3540 static void builtin_checkdiff(const char *name_a, const char *name_b,
3541                               const char *attr_path,
3542                               struct diff_filespec *one,
3543                               struct diff_filespec *two,
3544                               struct diff_options *o)
3545 {
3546         mmfile_t mf1, mf2;
3547         struct checkdiff_t data;
3548
3549         if (!two)
3550                 return;
3551
3552         memset(&data, 0, sizeof(data));
3553         data.filename = name_b ? name_b : name_a;
3554         data.lineno = 0;
3555         data.o = o;
3556         data.ws_rule = whitespace_rule(attr_path);
3557         data.conflict_marker_size = ll_merge_marker_size(attr_path);
3558
3559         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3560                 die("unable to read files to diff");
3561
3562         /*
3563          * All the other codepaths check both sides, but not checking
3564          * the "old" side here is deliberate.  We are checking the newly
3565          * introduced changes, and as long as the "new" side is text, we
3566          * can and should check what it introduces.
3567          */
3568         if (diff_filespec_is_binary(two))
3569                 goto free_and_return;
3570         else {
3571                 /* Crazy xdl interfaces.. */
3572                 xpparam_t xpp;
3573                 xdemitconf_t xecfg;
3574
3575                 memset(&xpp, 0, sizeof(xpp));
3576                 memset(&xecfg, 0, sizeof(xecfg));
3577                 xecfg.ctxlen = 1; /* at least one context line */
3578                 xpp.flags = 0;
3579                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
3580                                   &xpp, &xecfg))
3581                         die("unable to generate checkdiff for %s", one->path);
3582
3583                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3584                         struct emit_callback ecbdata;
3585                         int blank_at_eof;
3586
3587                         ecbdata.ws_rule = data.ws_rule;
3588                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3589                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3590
3591                         if (blank_at_eof) {
3592                                 static char *err;
3593                                 if (!err)
3594                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3595                                 fprintf(o->file, "%s:%d: %s.\n",
3596                                         data.filename, blank_at_eof, err);
3597                                 data.status = 1; /* report errors */
3598                         }
3599                 }
3600         }
3601  free_and_return:
3602         diff_free_filespec_data(one);
3603         diff_free_filespec_data(two);
3604         if (data.status)
3605                 o->flags.check_failed = 1;
3606 }
3607
3608 struct diff_filespec *alloc_filespec(const char *path)
3609 {
3610         struct diff_filespec *spec;
3611
3612         FLEXPTR_ALLOC_STR(spec, path, path);
3613         spec->count = 1;
3614         spec->is_binary = -1;
3615         return spec;
3616 }
3617
3618 void free_filespec(struct diff_filespec *spec)
3619 {
3620         if (!--spec->count) {
3621                 diff_free_filespec_data(spec);
3622                 free(spec);
3623         }
3624 }
3625
3626 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3627                    int oid_valid, unsigned short mode)
3628 {
3629         if (mode) {
3630                 spec->mode = canon_mode(mode);
3631                 oidcpy(&spec->oid, oid);
3632                 spec->oid_valid = oid_valid;
3633         }
3634 }
3635
3636 /*
3637  * Given a name and sha1 pair, if the index tells us the file in
3638  * the work tree has that object contents, return true, so that
3639  * prepare_temp_file() does not have to inflate and extract.
3640  */
3641 static int reuse_worktree_file(const char *name, const struct object_id *oid, int want_file)
3642 {
3643         const struct cache_entry *ce;
3644         struct stat st;
3645         int pos, len;
3646
3647         /*
3648          * We do not read the cache ourselves here, because the
3649          * benchmark with my previous version that always reads cache
3650          * shows that it makes things worse for diff-tree comparing
3651          * two linux-2.6 kernel trees in an already checked out work
3652          * tree.  This is because most diff-tree comparisons deal with
3653          * only a small number of files, while reading the cache is
3654          * expensive for a large project, and its cost outweighs the
3655          * savings we get by not inflating the object to a temporary
3656          * file.  Practically, this code only helps when we are used
3657          * by diff-cache --cached, which does read the cache before
3658          * calling us.
3659          */
3660         if (!active_cache)
3661                 return 0;
3662
3663         /* We want to avoid the working directory if our caller
3664          * doesn't need the data in a normal file, this system
3665          * is rather slow with its stat/open/mmap/close syscalls,
3666          * and the object is contained in a pack file.  The pack
3667          * is probably already open and will be faster to obtain
3668          * the data through than the working directory.  Loose
3669          * objects however would tend to be slower as they need
3670          * to be individually opened and inflated.
3671          */
3672         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(oid->hash))
3673                 return 0;
3674
3675         /*
3676          * Similarly, if we'd have to convert the file contents anyway, that
3677          * makes the optimization not worthwhile.
3678          */
3679         if (!want_file && would_convert_to_git(&the_index, name))
3680                 return 0;
3681
3682         len = strlen(name);
3683         pos = cache_name_pos(name, len);
3684         if (pos < 0)
3685                 return 0;
3686         ce = active_cache[pos];
3687
3688         /*
3689          * This is not the sha1 we are looking for, or
3690          * unreusable because it is not a regular file.
3691          */
3692         if (oidcmp(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3693                 return 0;
3694
3695         /*
3696          * If ce is marked as "assume unchanged", there is no
3697          * guarantee that work tree matches what we are looking for.
3698          */
3699         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3700                 return 0;
3701
3702         /*
3703          * If ce matches the file in the work tree, we can reuse it.
3704          */
3705         if (ce_uptodate(ce) ||
3706             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
3707                 return 1;
3708
3709         return 0;
3710 }
3711
3712 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3713 {
3714         struct strbuf buf = STRBUF_INIT;
3715         char *dirty = "";
3716
3717         /* Are we looking at the work tree? */
3718         if (s->dirty_submodule)
3719                 dirty = "-dirty";
3720
3721         strbuf_addf(&buf, "Subproject commit %s%s\n",
3722                     oid_to_hex(&s->oid), dirty);
3723         s->size = buf.len;
3724         if (size_only) {
3725                 s->data = NULL;
3726                 strbuf_release(&buf);
3727         } else {
3728                 s->data = strbuf_detach(&buf, NULL);
3729                 s->should_free = 1;
3730         }
3731         return 0;
3732 }
3733
3734 /*
3735  * While doing rename detection and pickaxe operation, we may need to
3736  * grab the data for the blob (or file) for our own in-core comparison.
3737  * diff_filespec has data and size fields for this purpose.
3738  */
3739 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
3740 {
3741         int size_only = flags & CHECK_SIZE_ONLY;
3742         int err = 0;
3743         int conv_flags = global_conv_flags_eol;
3744         /*
3745          * demote FAIL to WARN to allow inspecting the situation
3746          * instead of refusing.
3747          */
3748         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3749                 conv_flags = CONV_EOL_RNDTRP_WARN;
3750
3751         if (!DIFF_FILE_VALID(s))
3752                 die("internal error: asking to populate invalid file.");
3753         if (S_ISDIR(s->mode))
3754                 return -1;
3755
3756         if (s->data)
3757                 return 0;
3758
3759         if (size_only && 0 < s->size)
3760                 return 0;
3761
3762         if (S_ISGITLINK(s->mode))
3763                 return diff_populate_gitlink(s, size_only);
3764
3765         if (!s->oid_valid ||
3766             reuse_worktree_file(s->path, &s->oid, 0)) {
3767                 struct strbuf buf = STRBUF_INIT;
3768                 struct stat st;
3769                 int fd;
3770
3771                 if (lstat(s->path, &st) < 0) {
3772                 err_empty:
3773                         err = -1;
3774                 empty:
3775                         s->data = (char *)"";
3776                         s->size = 0;
3777                         return err;
3778                 }
3779                 s->size = xsize_t(st.st_size);
3780                 if (!s->size)
3781                         goto empty;
3782                 if (S_ISLNK(st.st_mode)) {
3783                         struct strbuf sb = STRBUF_INIT;
3784
3785                         if (strbuf_readlink(&sb, s->path, s->size))
3786                                 goto err_empty;
3787                         s->size = sb.len;
3788                         s->data = strbuf_detach(&sb, NULL);
3789                         s->should_free = 1;
3790                         return 0;
3791                 }
3792
3793                 /*
3794                  * Even if the caller would be happy with getting
3795                  * only the size, we cannot return early at this
3796                  * point if the path requires us to run the content
3797                  * conversion.
3798                  */
3799                 if (size_only && !would_convert_to_git(&the_index, s->path))
3800                         return 0;
3801
3802                 /*
3803                  * Note: this check uses xsize_t(st.st_size) that may
3804                  * not be the true size of the blob after it goes
3805                  * through convert_to_git().  This may not strictly be
3806                  * correct, but the whole point of big_file_threshold
3807                  * and is_binary check being that we want to avoid
3808                  * opening the file and inspecting the contents, this
3809                  * is probably fine.
3810                  */
3811                 if ((flags & CHECK_BINARY) &&
3812                     s->size > big_file_threshold && s->is_binary == -1) {
3813                         s->is_binary = 1;
3814                         return 0;
3815                 }
3816                 fd = open(s->path, O_RDONLY);
3817                 if (fd < 0)
3818                         goto err_empty;
3819                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
3820                 close(fd);
3821                 s->should_munmap = 1;
3822
3823                 /*
3824                  * Convert from working tree format to canonical git format
3825                  */
3826                 if (convert_to_git(&the_index, s->path, s->data, s->size, &buf, conv_flags)) {
3827                         size_t size = 0;
3828                         munmap(s->data, s->size);
3829                         s->should_munmap = 0;
3830                         s->data = strbuf_detach(&buf, &size);
3831                         s->size = size;
3832                         s->should_free = 1;
3833                 }
3834         }
3835         else {
3836                 enum object_type type;
3837                 if (size_only || (flags & CHECK_BINARY)) {
3838                         type = oid_object_info(&s->oid, &s->size);
3839                         if (type < 0)
3840                                 die("unable to read %s",
3841                                     oid_to_hex(&s->oid));
3842                         if (size_only)
3843                                 return 0;
3844                         if (s->size > big_file_threshold && s->is_binary == -1) {
3845                                 s->is_binary = 1;
3846                                 return 0;
3847                         }
3848                 }
3849                 s->data = read_object_file(&s->oid, &type, &s->size);
3850                 if (!s->data)
3851                         die("unable to read %s", oid_to_hex(&s->oid));
3852                 s->should_free = 1;
3853         }
3854         return 0;
3855 }
3856
3857 void diff_free_filespec_blob(struct diff_filespec *s)
3858 {
3859         if (s->should_free)
3860                 free(s->data);
3861         else if (s->should_munmap)
3862                 munmap(s->data, s->size);
3863
3864         if (s->should_free || s->should_munmap) {
3865                 s->should_free = s->should_munmap = 0;
3866                 s->data = NULL;
3867         }
3868 }
3869
3870 void diff_free_filespec_data(struct diff_filespec *s)
3871 {
3872         diff_free_filespec_blob(s);
3873         FREE_AND_NULL(s->cnt_data);
3874 }
3875
3876 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
3877                            void *blob,
3878                            unsigned long size,
3879                            const struct object_id *oid,
3880                            int mode)
3881 {
3882         struct strbuf buf = STRBUF_INIT;
3883         struct strbuf tempfile = STRBUF_INIT;
3884         char *path_dup = xstrdup(path);
3885         const char *base = basename(path_dup);
3886
3887         /* Generate "XXXXXX_basename.ext" */
3888         strbuf_addstr(&tempfile, "XXXXXX_");
3889         strbuf_addstr(&tempfile, base);
3890
3891         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
3892         if (!temp->tempfile)
3893                 die_errno("unable to create temp-file");
3894         if (convert_to_working_tree(path,
3895                         (const char *)blob, (size_t)size, &buf)) {
3896                 blob = buf.buf;
3897                 size = buf.len;
3898         }
3899         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
3900             close_tempfile_gently(temp->tempfile))
3901                 die_errno("unable to write temp-file");
3902         temp->name = get_tempfile_path(temp->tempfile);
3903         oid_to_hex_r(temp->hex, oid);
3904         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
3905         strbuf_release(&buf);
3906         strbuf_release(&tempfile);
3907         free(path_dup);
3908 }
3909
3910 static struct diff_tempfile *prepare_temp_file(const char *name,
3911                 struct diff_filespec *one)
3912 {
3913         struct diff_tempfile *temp = claim_diff_tempfile();
3914
3915         if (!DIFF_FILE_VALID(one)) {
3916         not_a_valid_file:
3917                 /* A '-' entry produces this for file-2, and
3918                  * a '+' entry produces this for file-1.
3919                  */
3920                 temp->name = "/dev/null";
3921                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
3922                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
3923                 return temp;
3924         }
3925
3926         if (!S_ISGITLINK(one->mode) &&
3927             (!one->oid_valid ||
3928              reuse_worktree_file(name, &one->oid, 1))) {
3929                 struct stat st;
3930                 if (lstat(name, &st) < 0) {
3931                         if (errno == ENOENT)
3932                                 goto not_a_valid_file;
3933                         die_errno("stat(%s)", name);
3934                 }
3935                 if (S_ISLNK(st.st_mode)) {
3936                         struct strbuf sb = STRBUF_INIT;
3937                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
3938                                 die_errno("readlink(%s)", name);
3939                         prep_temp_blob(name, temp, sb.buf, sb.len,
3940                                        (one->oid_valid ?
3941                                         &one->oid : &null_oid),
3942                                        (one->oid_valid ?
3943                                         one->mode : S_IFLNK));
3944                         strbuf_release(&sb);
3945                 }
3946                 else {
3947                         /* we can borrow from the file in the work tree */
3948                         temp->name = name;
3949                         if (!one->oid_valid)
3950                                 oid_to_hex_r(temp->hex, &null_oid);
3951                         else
3952                                 oid_to_hex_r(temp->hex, &one->oid);
3953                         /* Even though we may sometimes borrow the
3954                          * contents from the work tree, we always want
3955                          * one->mode.  mode is trustworthy even when
3956                          * !(one->oid_valid), as long as
3957                          * DIFF_FILE_VALID(one).
3958                          */
3959                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
3960                 }
3961                 return temp;
3962         }
3963         else {
3964                 if (diff_populate_filespec(one, 0))
3965                         die("cannot read data blob for %s", one->path);
3966                 prep_temp_blob(name, temp, one->data, one->size,
3967                                &one->oid, one->mode);
3968         }
3969         return temp;
3970 }
3971
3972 static void add_external_diff_name(struct argv_array *argv,
3973                                    const char *name,
3974                                    struct diff_filespec *df)
3975 {
3976         struct diff_tempfile *temp = prepare_temp_file(name, df);
3977         argv_array_push(argv, temp->name);
3978         argv_array_push(argv, temp->hex);
3979         argv_array_push(argv, temp->mode);
3980 }
3981
3982 /* An external diff command takes:
3983  *
3984  * diff-cmd name infile1 infile1-sha1 infile1-mode \
3985  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
3986  *
3987  */
3988 static void run_external_diff(const char *pgm,
3989                               const char *name,
3990                               const char *other,
3991                               struct diff_filespec *one,
3992                               struct diff_filespec *two,
3993                               const char *xfrm_msg,
3994                               int complete_rewrite,
3995                               struct diff_options *o)
3996 {
3997         struct argv_array argv = ARGV_ARRAY_INIT;
3998         struct argv_array env = ARGV_ARRAY_INIT;
3999         struct diff_queue_struct *q = &diff_queued_diff;
4000
4001         argv_array_push(&argv, pgm);
4002         argv_array_push(&argv, name);
4003
4004         if (one && two) {
4005                 add_external_diff_name(&argv, name, one);
4006                 if (!other)
4007                         add_external_diff_name(&argv, name, two);
4008                 else {
4009                         add_external_diff_name(&argv, other, two);
4010                         argv_array_push(&argv, other);
4011                         argv_array_push(&argv, xfrm_msg);
4012                 }
4013         }
4014
4015         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4016         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4017
4018         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
4019                 die(_("external diff died, stopping at %s"), name);
4020
4021         remove_tempfile();
4022         argv_array_clear(&argv);
4023         argv_array_clear(&env);
4024 }
4025
4026 static int similarity_index(struct diff_filepair *p)
4027 {
4028         return p->score * 100 / MAX_SCORE;
4029 }
4030
4031 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4032 {
4033         if (startup_info->have_repository)
4034                 return find_unique_abbrev(oid, abbrev);
4035         else {
4036                 char *hex = oid_to_hex(oid);
4037                 if (abbrev < 0)
4038                         abbrev = FALLBACK_DEFAULT_ABBREV;
4039                 if (abbrev > GIT_SHA1_HEXSZ)
4040                         die("BUG: oid abbreviation out of range: %d", abbrev);
4041                 if (abbrev)
4042                         hex[abbrev] = '\0';
4043                 return hex;
4044         }
4045 }
4046
4047 static void fill_metainfo(struct strbuf *msg,
4048                           const char *name,
4049                           const char *other,
4050                           struct diff_filespec *one,
4051                           struct diff_filespec *two,
4052                           struct diff_options *o,
4053                           struct diff_filepair *p,
4054                           int *must_show_header,
4055                           int use_color)
4056 {
4057         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4058         const char *reset = diff_get_color(use_color, DIFF_RESET);
4059         const char *line_prefix = diff_line_prefix(o);
4060
4061         *must_show_header = 1;
4062         strbuf_init(msg, PATH_MAX * 2 + 300);
4063         switch (p->status) {
4064         case DIFF_STATUS_COPIED:
4065                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4066                             line_prefix, set, similarity_index(p));
4067                 strbuf_addf(msg, "%s\n%s%scopy from ",
4068                             reset,  line_prefix, set);
4069                 quote_c_style(name, msg, NULL, 0);
4070                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4071                 quote_c_style(other, msg, NULL, 0);
4072                 strbuf_addf(msg, "%s\n", reset);
4073                 break;
4074         case DIFF_STATUS_RENAMED:
4075                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4076                             line_prefix, set, similarity_index(p));
4077                 strbuf_addf(msg, "%s\n%s%srename from ",
4078                             reset, line_prefix, set);
4079                 quote_c_style(name, msg, NULL, 0);
4080                 strbuf_addf(msg, "%s\n%s%srename to ",
4081                             reset, line_prefix, set);
4082                 quote_c_style(other, msg, NULL, 0);
4083                 strbuf_addf(msg, "%s\n", reset);
4084                 break;
4085         case DIFF_STATUS_MODIFIED:
4086                 if (p->score) {
4087                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4088                                     line_prefix,
4089                                     set, similarity_index(p), reset);
4090                         break;
4091                 }
4092                 /* fallthru */
4093         default:
4094                 *must_show_header = 0;
4095         }
4096         if (one && two && oidcmp(&one->oid, &two->oid)) {
4097                 int abbrev = o->flags.full_index ? 40 : DEFAULT_ABBREV;
4098
4099                 if (o->flags.binary) {
4100                         mmfile_t mf;
4101                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
4102                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
4103                                 abbrev = 40;
4104                 }
4105                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4106                             diff_abbrev_oid(&one->oid, abbrev),
4107                             diff_abbrev_oid(&two->oid, abbrev));
4108                 if (one->mode == two->mode)
4109                         strbuf_addf(msg, " %06o", one->mode);
4110                 strbuf_addf(msg, "%s\n", reset);
4111         }
4112 }
4113
4114 static void run_diff_cmd(const char *pgm,
4115                          const char *name,
4116                          const char *other,
4117                          const char *attr_path,
4118                          struct diff_filespec *one,
4119                          struct diff_filespec *two,
4120                          struct strbuf *msg,
4121                          struct diff_options *o,
4122                          struct diff_filepair *p)
4123 {
4124         const char *xfrm_msg = NULL;
4125         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4126         int must_show_header = 0;
4127
4128
4129         if (o->flags.allow_external) {
4130                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
4131                 if (drv && drv->external)
4132                         pgm = drv->external;
4133         }
4134
4135         if (msg) {
4136                 /*
4137                  * don't use colors when the header is intended for an
4138                  * external diff driver
4139                  */
4140                 fill_metainfo(msg, name, other, one, two, o, p,
4141                               &must_show_header,
4142                               want_color(o->use_color) && !pgm);
4143                 xfrm_msg = msg->len ? msg->buf : NULL;
4144         }
4145
4146         if (pgm) {
4147                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
4148                                   complete_rewrite, o);
4149                 return;
4150         }
4151         if (one && two)
4152                 builtin_diff(name, other ? other : name,
4153                              one, two, xfrm_msg, must_show_header,
4154                              o, complete_rewrite);
4155         else
4156                 fprintf(o->file, "* Unmerged path %s\n", name);
4157 }
4158
4159 static void diff_fill_oid_info(struct diff_filespec *one)
4160 {
4161         if (DIFF_FILE_VALID(one)) {
4162                 if (!one->oid_valid) {
4163                         struct stat st;
4164                         if (one->is_stdin) {
4165                                 oidclr(&one->oid);
4166                                 return;
4167                         }
4168                         if (lstat(one->path, &st) < 0)
4169                                 die_errno("stat '%s'", one->path);
4170                         if (index_path(&one->oid, one->path, &st, 0))
4171                                 die("cannot hash %s", one->path);
4172                 }
4173         }
4174         else
4175                 oidclr(&one->oid);
4176 }
4177
4178 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4179 {
4180         /* Strip the prefix but do not molest /dev/null and absolute paths */
4181         if (*namep && **namep != '/') {
4182                 *namep += prefix_length;
4183                 if (**namep == '/')
4184                         ++*namep;
4185         }
4186         if (*otherp && **otherp != '/') {
4187                 *otherp += prefix_length;
4188                 if (**otherp == '/')
4189                         ++*otherp;
4190         }
4191 }
4192
4193 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4194 {
4195         const char *pgm = external_diff();
4196         struct strbuf msg;
4197         struct diff_filespec *one = p->one;
4198         struct diff_filespec *two = p->two;
4199         const char *name;
4200         const char *other;
4201         const char *attr_path;
4202
4203         name  = one->path;
4204         other = (strcmp(name, two->path) ? two->path : NULL);
4205         attr_path = name;
4206         if (o->prefix_length)
4207                 strip_prefix(o->prefix_length, &name, &other);
4208
4209         if (!o->flags.allow_external)
4210                 pgm = NULL;
4211
4212         if (DIFF_PAIR_UNMERGED(p)) {
4213                 run_diff_cmd(pgm, name, NULL, attr_path,
4214                              NULL, NULL, NULL, o, p);
4215                 return;
4216         }
4217
4218         diff_fill_oid_info(one);
4219         diff_fill_oid_info(two);
4220
4221         if (!pgm &&
4222             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4223             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4224                 /*
4225                  * a filepair that changes between file and symlink
4226                  * needs to be split into deletion and creation.
4227                  */
4228                 struct diff_filespec *null = alloc_filespec(two->path);
4229                 run_diff_cmd(NULL, name, other, attr_path,
4230                              one, null, &msg, o, p);
4231                 free(null);
4232                 strbuf_release(&msg);
4233
4234                 null = alloc_filespec(one->path);
4235                 run_diff_cmd(NULL, name, other, attr_path,
4236                              null, two, &msg, o, p);
4237                 free(null);
4238         }
4239         else
4240                 run_diff_cmd(pgm, name, other, attr_path,
4241                              one, two, &msg, o, p);
4242
4243         strbuf_release(&msg);
4244 }
4245
4246 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4247                          struct diffstat_t *diffstat)
4248 {
4249         const char *name;
4250         const char *other;
4251
4252         if (DIFF_PAIR_UNMERGED(p)) {
4253                 /* unmerged */
4254                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
4255                 return;
4256         }
4257
4258         name = p->one->path;
4259         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4260
4261         if (o->prefix_length)
4262                 strip_prefix(o->prefix_length, &name, &other);
4263
4264         diff_fill_oid_info(p->one);
4265         diff_fill_oid_info(p->two);
4266
4267         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
4268 }
4269
4270 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4271 {
4272         const char *name;
4273         const char *other;
4274         const char *attr_path;
4275
4276         if (DIFF_PAIR_UNMERGED(p)) {
4277                 /* unmerged */
4278                 return;
4279         }
4280
4281         name = p->one->path;
4282         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4283         attr_path = other ? other : name;
4284
4285         if (o->prefix_length)
4286                 strip_prefix(o->prefix_length, &name, &other);
4287
4288         diff_fill_oid_info(p->one);
4289         diff_fill_oid_info(p->two);
4290
4291         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4292 }
4293
4294 void diff_setup(struct diff_options *options)
4295 {
4296         memcpy(options, &default_diff_options, sizeof(*options));
4297
4298         options->file = stdout;
4299
4300         options->abbrev = DEFAULT_ABBREV;
4301         options->line_termination = '\n';
4302         options->break_opt = -1;
4303         options->rename_limit = -1;
4304         options->dirstat_permille = diff_dirstat_permille_default;
4305         options->context = diff_context_default;
4306         options->interhunkcontext = diff_interhunk_context_default;
4307         options->ws_error_highlight = ws_error_highlight_default;
4308         options->flags.rename_empty = 1;
4309         options->objfind = NULL;
4310
4311         /* pathchange left =NULL by default */
4312         options->change = diff_change;
4313         options->add_remove = diff_addremove;
4314         options->use_color = diff_use_color_default;
4315         options->detect_rename = diff_detect_rename_default;
4316         options->xdl_opts |= diff_algorithm;
4317         if (diff_indent_heuristic)
4318                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4319
4320         options->orderfile = diff_order_file_cfg;
4321
4322         if (diff_no_prefix) {
4323                 options->a_prefix = options->b_prefix = "";
4324         } else if (!diff_mnemonic_prefix) {
4325                 options->a_prefix = "a/";
4326                 options->b_prefix = "b/";
4327         }
4328
4329         options->color_moved = diff_color_moved_default;
4330 }
4331
4332 void diff_setup_done(struct diff_options *options)
4333 {
4334         unsigned check_mask = DIFF_FORMAT_NAME |
4335                               DIFF_FORMAT_NAME_STATUS |
4336                               DIFF_FORMAT_CHECKDIFF |
4337                               DIFF_FORMAT_NO_OUTPUT;
4338
4339         if (options->set_default)
4340                 options->set_default(options);
4341
4342         if (HAS_MULTI_BITS(options->output_format & check_mask))
4343                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4344
4345         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4346                 die(_("-G, -S and --find-object are mutually exclusive"));
4347
4348         /*
4349          * Most of the time we can say "there are changes"
4350          * only by checking if there are changed paths, but
4351          * --ignore-whitespace* options force us to look
4352          * inside contents.
4353          */
4354
4355         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4356                 options->flags.diff_from_contents = 1;
4357         else
4358                 options->flags.diff_from_contents = 0;
4359
4360         if (options->flags.find_copies_harder)
4361                 options->detect_rename = DIFF_DETECT_COPY;
4362
4363         if (!options->flags.relative_name)
4364                 options->prefix = NULL;
4365         if (options->prefix)
4366                 options->prefix_length = strlen(options->prefix);
4367         else
4368                 options->prefix_length = 0;
4369
4370         if (options->output_format & (DIFF_FORMAT_NAME |
4371                                       DIFF_FORMAT_NAME_STATUS |
4372                                       DIFF_FORMAT_CHECKDIFF |
4373                                       DIFF_FORMAT_NO_OUTPUT))
4374                 options->output_format &= ~(DIFF_FORMAT_RAW |
4375                                             DIFF_FORMAT_NUMSTAT |
4376                                             DIFF_FORMAT_DIFFSTAT |
4377                                             DIFF_FORMAT_SHORTSTAT |
4378                                             DIFF_FORMAT_DIRSTAT |
4379                                             DIFF_FORMAT_SUMMARY |
4380                                             DIFF_FORMAT_PATCH);
4381
4382         /*
4383          * These cases always need recursive; we do not drop caller-supplied
4384          * recursive bits for other formats here.
4385          */
4386         if (options->output_format & (DIFF_FORMAT_PATCH |
4387                                       DIFF_FORMAT_NUMSTAT |
4388                                       DIFF_FORMAT_DIFFSTAT |
4389                                       DIFF_FORMAT_SHORTSTAT |
4390                                       DIFF_FORMAT_DIRSTAT |
4391                                       DIFF_FORMAT_SUMMARY |
4392                                       DIFF_FORMAT_CHECKDIFF))
4393                 options->flags.recursive = 1;
4394         /*
4395          * Also pickaxe would not work very well if you do not say recursive
4396          */
4397         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4398                 options->flags.recursive = 1;
4399         /*
4400          * When patches are generated, submodules diffed against the work tree
4401          * must be checked for dirtiness too so it can be shown in the output
4402          */
4403         if (options->output_format & DIFF_FORMAT_PATCH)
4404                 options->flags.dirty_submodules = 1;
4405
4406         if (options->detect_rename && options->rename_limit < 0)
4407                 options->rename_limit = diff_rename_limit_default;
4408         if (options->setup & DIFF_SETUP_USE_CACHE) {
4409                 if (!active_cache)
4410                         /* read-cache does not die even when it fails
4411                          * so it is safe for us to do this here.  Also
4412                          * it does not smudge active_cache or active_nr
4413                          * when it fails, so we do not have to worry about
4414                          * cleaning it up ourselves either.
4415                          */
4416                         read_cache();
4417         }
4418         if (40 < options->abbrev)
4419                 options->abbrev = 40; /* full */
4420
4421         /*
4422          * It does not make sense to show the first hit we happened
4423          * to have found.  It does not make sense not to return with
4424          * exit code in such a case either.
4425          */
4426         if (options->flags.quick) {
4427                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4428                 options->flags.exit_with_status = 1;
4429         }
4430
4431         options->diff_path_counter = 0;
4432
4433         if (options->flags.follow_renames && options->pathspec.nr != 1)
4434                 die(_("--follow requires exactly one pathspec"));
4435
4436         if (!options->use_color || external_diff())
4437                 options->color_moved = 0;
4438 }
4439
4440 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4441 {
4442         char c, *eq;
4443         int len;
4444
4445         if (*arg != '-')
4446                 return 0;
4447         c = *++arg;
4448         if (!c)
4449                 return 0;
4450         if (c == arg_short) {
4451                 c = *++arg;
4452                 if (!c)
4453                         return 1;
4454                 if (val && isdigit(c)) {
4455                         char *end;
4456                         int n = strtoul(arg, &end, 10);
4457                         if (*end)
4458                                 return 0;
4459                         *val = n;
4460                         return 1;
4461                 }
4462                 return 0;
4463         }
4464         if (c != '-')
4465                 return 0;
4466         arg++;
4467         eq = strchrnul(arg, '=');
4468         len = eq - arg;
4469         if (!len || strncmp(arg, arg_long, len))
4470                 return 0;
4471         if (*eq) {
4472                 int n;
4473                 char *end;
4474                 if (!isdigit(*++eq))
4475                         return 0;
4476                 n = strtoul(eq, &end, 10);
4477                 if (*end)
4478                         return 0;
4479                 *val = n;
4480         }
4481         return 1;
4482 }
4483
4484 static int diff_scoreopt_parse(const char *opt);
4485
4486 static inline int short_opt(char opt, const char **argv,
4487                             const char **optarg)
4488 {
4489         const char *arg = argv[0];
4490         if (arg[0] != '-' || arg[1] != opt)
4491                 return 0;
4492         if (arg[2] != '\0') {
4493                 *optarg = arg + 2;
4494                 return 1;
4495         }
4496         if (!argv[1])
4497                 die("Option '%c' requires a value", opt);
4498         *optarg = argv[1];
4499         return 2;
4500 }
4501
4502 int parse_long_opt(const char *opt, const char **argv,
4503                    const char **optarg)
4504 {
4505         const char *arg = argv[0];
4506         if (!skip_prefix(arg, "--", &arg))
4507                 return 0;
4508         if (!skip_prefix(arg, opt, &arg))
4509                 return 0;
4510         if (*arg == '=') { /* stuck form: --option=value */
4511                 *optarg = arg + 1;
4512                 return 1;
4513         }
4514         if (*arg != '\0')
4515                 return 0;
4516         /* separate form: --option value */
4517         if (!argv[1])
4518                 die("Option '--%s' requires a value", opt);
4519         *optarg = argv[1];
4520         return 2;
4521 }
4522
4523 static int stat_opt(struct diff_options *options, const char **av)
4524 {
4525         const char *arg = av[0];
4526         char *end;
4527         int width = options->stat_width;
4528         int name_width = options->stat_name_width;
4529         int graph_width = options->stat_graph_width;
4530         int count = options->stat_count;
4531         int argcount = 1;
4532
4533         if (!skip_prefix(arg, "--stat", &arg))
4534                 die("BUG: stat option does not begin with --stat: %s", arg);
4535         end = (char *)arg;
4536
4537         switch (*arg) {
4538         case '-':
4539                 if (skip_prefix(arg, "-width", &arg)) {
4540                         if (*arg == '=')
4541                                 width = strtoul(arg + 1, &end, 10);
4542                         else if (!*arg && !av[1])
4543                                 die_want_option("--stat-width");
4544                         else if (!*arg) {
4545                                 width = strtoul(av[1], &end, 10);
4546                                 argcount = 2;
4547                         }
4548                 } else if (skip_prefix(arg, "-name-width", &arg)) {
4549                         if (*arg == '=')
4550                                 name_width = strtoul(arg + 1, &end, 10);
4551                         else if (!*arg && !av[1])
4552                                 die_want_option("--stat-name-width");
4553                         else if (!*arg) {
4554                                 name_width = strtoul(av[1], &end, 10);
4555                                 argcount = 2;
4556                         }
4557                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
4558                         if (*arg == '=')
4559                                 graph_width = strtoul(arg + 1, &end, 10);
4560                         else if (!*arg && !av[1])
4561                                 die_want_option("--stat-graph-width");
4562                         else if (!*arg) {
4563                                 graph_width = strtoul(av[1], &end, 10);
4564                                 argcount = 2;
4565                         }
4566                 } else if (skip_prefix(arg, "-count", &arg)) {
4567                         if (*arg == '=')
4568                                 count = strtoul(arg + 1, &end, 10);
4569                         else if (!*arg && !av[1])
4570                                 die_want_option("--stat-count");
4571                         else if (!*arg) {
4572                                 count = strtoul(av[1], &end, 10);
4573                                 argcount = 2;
4574                         }
4575                 }
4576                 break;
4577         case '=':
4578                 width = strtoul(arg+1, &end, 10);
4579                 if (*end == ',')
4580                         name_width = strtoul(end+1, &end, 10);
4581                 if (*end == ',')
4582                         count = strtoul(end+1, &end, 10);
4583         }
4584
4585         /* Important! This checks all the error cases! */
4586         if (*end)
4587                 return 0;
4588         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4589         options->stat_name_width = name_width;
4590         options->stat_graph_width = graph_width;
4591         options->stat_width = width;
4592         options->stat_count = count;
4593         return argcount;
4594 }
4595
4596 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4597 {
4598         struct strbuf errmsg = STRBUF_INIT;
4599         if (parse_dirstat_params(options, params, &errmsg))
4600                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4601                     errmsg.buf);
4602         strbuf_release(&errmsg);
4603         /*
4604          * The caller knows a dirstat-related option is given from the command
4605          * line; allow it to say "return this_function();"
4606          */
4607         options->output_format |= DIFF_FORMAT_DIRSTAT;
4608         return 1;
4609 }
4610
4611 static int parse_submodule_opt(struct diff_options *options, const char *value)
4612 {
4613         if (parse_submodule_params(options, value))
4614                 die(_("Failed to parse --submodule option parameter: '%s'"),
4615                         value);
4616         return 1;
4617 }
4618
4619 static const char diff_status_letters[] = {
4620         DIFF_STATUS_ADDED,
4621         DIFF_STATUS_COPIED,
4622         DIFF_STATUS_DELETED,
4623         DIFF_STATUS_MODIFIED,
4624         DIFF_STATUS_RENAMED,
4625         DIFF_STATUS_TYPE_CHANGED,
4626         DIFF_STATUS_UNKNOWN,
4627         DIFF_STATUS_UNMERGED,
4628         DIFF_STATUS_FILTER_AON,
4629         DIFF_STATUS_FILTER_BROKEN,
4630         '\0',
4631 };
4632
4633 static unsigned int filter_bit['Z' + 1];
4634
4635 static void prepare_filter_bits(void)
4636 {
4637         int i;
4638
4639         if (!filter_bit[DIFF_STATUS_ADDED]) {
4640                 for (i = 0; diff_status_letters[i]; i++)
4641                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4642         }
4643 }
4644
4645 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4646 {
4647         return opt->filter & filter_bit[(int) status];
4648 }
4649
4650 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4651 {
4652         int i, optch;
4653
4654         prepare_filter_bits();
4655
4656         /*
4657          * If there is a negation e.g. 'd' in the input, and we haven't
4658          * initialized the filter field with another --diff-filter, start
4659          * from full set of bits, except for AON.
4660          */
4661         if (!opt->filter) {
4662                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4663                         if (optch < 'a' || 'z' < optch)
4664                                 continue;
4665                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4666                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4667                         break;
4668                 }
4669         }
4670
4671         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4672                 unsigned int bit;
4673                 int negate;
4674
4675                 if ('a' <= optch && optch <= 'z') {
4676                         negate = 1;
4677                         optch = toupper(optch);
4678                 } else {
4679                         negate = 0;
4680                 }
4681
4682                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4683                 if (!bit)
4684                         return optarg[i];
4685                 if (negate)
4686                         opt->filter &= ~bit;
4687                 else
4688                         opt->filter |= bit;
4689         }
4690         return 0;
4691 }
4692
4693 static void enable_patch_output(int *fmt) {
4694         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4695         *fmt |= DIFF_FORMAT_PATCH;
4696 }
4697
4698 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4699 {
4700         int val = parse_ws_error_highlight(arg);
4701
4702         if (val < 0) {
4703                 error("unknown value after ws-error-highlight=%.*s",
4704                       -1 - val, arg);
4705                 return 0;
4706         }
4707         opt->ws_error_highlight = val;
4708         return 1;
4709 }
4710
4711 static int parse_objfind_opt(struct diff_options *opt, const char *arg)
4712 {
4713         struct object_id oid;
4714
4715         if (get_oid(arg, &oid))
4716                 return error("unable to resolve '%s'", arg);
4717
4718         if (!opt->objfind)
4719                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4720
4721         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4722         opt->flags.recursive = 1;
4723         opt->flags.tree_in_recursive = 1;
4724         oidset_insert(opt->objfind, &oid);
4725         return 1;
4726 }
4727
4728 int diff_opt_parse(struct diff_options *options,
4729                    const char **av, int ac, const char *prefix)
4730 {
4731         const char *arg = av[0];
4732         const char *optarg;
4733         int argcount;
4734
4735         if (!prefix)
4736                 prefix = "";
4737
4738         /* Output format options */
4739         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
4740             || opt_arg(arg, 'U', "unified", &options->context))
4741                 enable_patch_output(&options->output_format);
4742         else if (!strcmp(arg, "--raw"))
4743                 options->output_format |= DIFF_FORMAT_RAW;
4744         else if (!strcmp(arg, "--patch-with-raw")) {
4745                 enable_patch_output(&options->output_format);
4746                 options->output_format |= DIFF_FORMAT_RAW;
4747         } else if (!strcmp(arg, "--numstat"))
4748                 options->output_format |= DIFF_FORMAT_NUMSTAT;
4749         else if (!strcmp(arg, "--shortstat"))
4750                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
4751         else if (skip_prefix(arg, "-X", &arg) ||
4752                  skip_to_optional_arg(arg, "--dirstat", &arg))
4753                 return parse_dirstat_opt(options, arg);
4754         else if (!strcmp(arg, "--cumulative"))
4755                 return parse_dirstat_opt(options, "cumulative");
4756         else if (skip_to_optional_arg(arg, "--dirstat-by-file", &arg)) {
4757                 parse_dirstat_opt(options, "files");
4758                 return parse_dirstat_opt(options, arg);
4759         }
4760         else if (!strcmp(arg, "--check"))
4761                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
4762         else if (!strcmp(arg, "--summary"))
4763                 options->output_format |= DIFF_FORMAT_SUMMARY;
4764         else if (!strcmp(arg, "--patch-with-stat")) {
4765                 enable_patch_output(&options->output_format);
4766                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4767         } else if (!strcmp(arg, "--name-only"))
4768                 options->output_format |= DIFF_FORMAT_NAME;
4769         else if (!strcmp(arg, "--name-status"))
4770                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
4771         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
4772                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
4773         else if (starts_with(arg, "--stat"))
4774                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
4775                 return stat_opt(options, av);
4776         else if (!strcmp(arg, "--compact-summary")) {
4777                  options->flags.stat_with_summary = 1;
4778                  options->output_format |= DIFF_FORMAT_DIFFSTAT;
4779         } else if (!strcmp(arg, "--no-compact-summary"))
4780                  options->flags.stat_with_summary = 0;
4781
4782         /* renames options */
4783         else if (starts_with(arg, "-B") ||
4784                  skip_to_optional_arg(arg, "--break-rewrites", NULL)) {
4785                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
4786                         return error("invalid argument to -B: %s", arg+2);
4787         }
4788         else if (starts_with(arg, "-M") ||
4789                  skip_to_optional_arg(arg, "--find-renames", NULL)) {
4790                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4791                         return error("invalid argument to -M: %s", arg+2);
4792                 options->detect_rename = DIFF_DETECT_RENAME;
4793         }
4794         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
4795                 options->irreversible_delete = 1;
4796         }
4797         else if (starts_with(arg, "-C") ||
4798                  skip_to_optional_arg(arg, "--find-copies", NULL)) {
4799                 if (options->detect_rename == DIFF_DETECT_COPY)
4800                         options->flags.find_copies_harder = 1;
4801                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4802                         return error("invalid argument to -C: %s", arg+2);
4803                 options->detect_rename = DIFF_DETECT_COPY;
4804         }
4805         else if (!strcmp(arg, "--no-renames"))
4806                 options->detect_rename = 0;
4807         else if (!strcmp(arg, "--rename-empty"))
4808                 options->flags.rename_empty = 1;
4809         else if (!strcmp(arg, "--no-rename-empty"))
4810                 options->flags.rename_empty = 0;
4811         else if (skip_to_optional_arg_default(arg, "--relative", &arg, NULL)) {
4812                 options->flags.relative_name = 1;
4813                 if (arg)
4814                         options->prefix = arg;
4815         }
4816
4817         /* xdiff options */
4818         else if (!strcmp(arg, "--minimal"))
4819                 DIFF_XDL_SET(options, NEED_MINIMAL);
4820         else if (!strcmp(arg, "--no-minimal"))
4821                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4822         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
4823                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
4824         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
4825                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
4826         else if (!strcmp(arg, "--ignore-space-at-eol"))
4827                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
4828         else if (!strcmp(arg, "--ignore-cr-at-eol"))
4829                 DIFF_XDL_SET(options, IGNORE_CR_AT_EOL);
4830         else if (!strcmp(arg, "--ignore-blank-lines"))
4831                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
4832         else if (!strcmp(arg, "--indent-heuristic"))
4833                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4834         else if (!strcmp(arg, "--no-indent-heuristic"))
4835                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
4836         else if (!strcmp(arg, "--patience")) {
4837                 int i;
4838                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4839                 /*
4840                  * Both --patience and --anchored use PATIENCE_DIFF
4841                  * internally, so remove any anchors previously
4842                  * specified.
4843                  */
4844                 for (i = 0; i < options->anchors_nr; i++)
4845                         free(options->anchors[i]);
4846                 options->anchors_nr = 0;
4847         } else if (!strcmp(arg, "--histogram"))
4848                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
4849         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
4850                 long value = parse_algorithm_value(optarg);
4851                 if (value < 0)
4852                         return error("option diff-algorithm accepts \"myers\", "
4853                                      "\"minimal\", \"patience\" and \"histogram\"");
4854                 /* clear out previous settings */
4855                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4856                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
4857                 options->xdl_opts |= value;
4858                 return argcount;
4859         } else if (skip_prefix(arg, "--anchored=", &arg)) {
4860                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4861                 ALLOC_GROW(options->anchors, options->anchors_nr + 1,
4862                            options->anchors_alloc);
4863                 options->anchors[options->anchors_nr++] = xstrdup(arg);
4864         }
4865
4866         /* flags options */
4867         else if (!strcmp(arg, "--binary")) {
4868                 enable_patch_output(&options->output_format);
4869                 options->flags.binary = 1;
4870         }
4871         else if (!strcmp(arg, "--full-index"))
4872                 options->flags.full_index = 1;
4873         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
4874                 options->flags.text = 1;
4875         else if (!strcmp(arg, "-R"))
4876                 options->flags.reverse_diff = 1;
4877         else if (!strcmp(arg, "--find-copies-harder"))
4878                 options->flags.find_copies_harder = 1;
4879         else if (!strcmp(arg, "--follow"))
4880                 options->flags.follow_renames = 1;
4881         else if (!strcmp(arg, "--no-follow")) {
4882                 options->flags.follow_renames = 0;
4883                 options->flags.default_follow_renames = 0;
4884         } else if (skip_to_optional_arg_default(arg, "--color", &arg, "always")) {
4885                 int value = git_config_colorbool(NULL, arg);
4886                 if (value < 0)
4887                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
4888                 options->use_color = value;
4889         }
4890         else if (!strcmp(arg, "--no-color"))
4891                 options->use_color = 0;
4892         else if (!strcmp(arg, "--color-moved")) {
4893                 if (diff_color_moved_default)
4894                         options->color_moved = diff_color_moved_default;
4895                 if (options->color_moved == COLOR_MOVED_NO)
4896                         options->color_moved = COLOR_MOVED_DEFAULT;
4897         } else if (!strcmp(arg, "--no-color-moved"))
4898                 options->color_moved = COLOR_MOVED_NO;
4899         else if (skip_prefix(arg, "--color-moved=", &arg)) {
4900                 int cm = parse_color_moved(arg);
4901                 if (cm < 0)
4902                         die("bad --color-moved argument: %s", arg);
4903                 options->color_moved = cm;
4904         } else if (skip_prefix(arg, "--color-moved-ws=", &arg)) {
4905                 options->color_moved_ws_handling = parse_color_moved_ws(arg);
4906         } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
4907                 options->use_color = 1;
4908                 options->word_diff = DIFF_WORDS_COLOR;
4909         }
4910         else if (!strcmp(arg, "--word-diff")) {
4911                 if (options->word_diff == DIFF_WORDS_NONE)
4912                         options->word_diff = DIFF_WORDS_PLAIN;
4913         }
4914         else if (skip_prefix(arg, "--word-diff=", &arg)) {
4915                 if (!strcmp(arg, "plain"))
4916                         options->word_diff = DIFF_WORDS_PLAIN;
4917                 else if (!strcmp(arg, "color")) {
4918                         options->use_color = 1;
4919                         options->word_diff = DIFF_WORDS_COLOR;
4920                 }
4921                 else if (!strcmp(arg, "porcelain"))
4922                         options->word_diff = DIFF_WORDS_PORCELAIN;
4923                 else if (!strcmp(arg, "none"))
4924                         options->word_diff = DIFF_WORDS_NONE;
4925                 else
4926                         die("bad --word-diff argument: %s", arg);
4927         }
4928         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
4929                 if (options->word_diff == DIFF_WORDS_NONE)
4930                         options->word_diff = DIFF_WORDS_PLAIN;
4931                 options->word_regex = optarg;
4932                 return argcount;
4933         }
4934         else if (!strcmp(arg, "--exit-code"))
4935                 options->flags.exit_with_status = 1;
4936         else if (!strcmp(arg, "--quiet"))
4937                 options->flags.quick = 1;
4938         else if (!strcmp(arg, "--ext-diff"))
4939                 options->flags.allow_external = 1;
4940         else if (!strcmp(arg, "--no-ext-diff"))
4941                 options->flags.allow_external = 0;
4942         else if (!strcmp(arg, "--textconv")) {
4943                 options->flags.allow_textconv = 1;
4944                 options->flags.textconv_set_via_cmdline = 1;
4945         } else if (!strcmp(arg, "--no-textconv"))
4946                 options->flags.allow_textconv = 0;
4947         else if (skip_to_optional_arg_default(arg, "--ignore-submodules", &arg, "all")) {
4948                 options->flags.override_submodule_config = 1;
4949                 handle_ignore_submodules_arg(options, arg);
4950         } else if (skip_to_optional_arg_default(arg, "--submodule", &arg, "log"))
4951                 return parse_submodule_opt(options, arg);
4952         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
4953                 return parse_ws_error_highlight_opt(options, arg);
4954         else if (!strcmp(arg, "--ita-invisible-in-index"))
4955                 options->ita_invisible_in_index = 1;
4956         else if (!strcmp(arg, "--ita-visible-in-index"))
4957                 options->ita_invisible_in_index = 0;
4958
4959         /* misc options */
4960         else if (!strcmp(arg, "-z"))
4961                 options->line_termination = 0;
4962         else if ((argcount = short_opt('l', av, &optarg))) {
4963                 options->rename_limit = strtoul(optarg, NULL, 10);
4964                 return argcount;
4965         }
4966         else if ((argcount = short_opt('S', av, &optarg))) {
4967                 options->pickaxe = optarg;
4968                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
4969                 return argcount;
4970         } else if ((argcount = short_opt('G', av, &optarg))) {
4971                 options->pickaxe = optarg;
4972                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
4973                 return argcount;
4974         }
4975         else if (!strcmp(arg, "--pickaxe-all"))
4976                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
4977         else if (!strcmp(arg, "--pickaxe-regex"))
4978                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
4979         else if ((argcount = short_opt('O', av, &optarg))) {
4980                 options->orderfile = prefix_filename(prefix, optarg);
4981                 return argcount;
4982         } else if (skip_prefix(arg, "--find-object=", &arg))
4983                 return parse_objfind_opt(options, arg);
4984         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
4985                 int offending = parse_diff_filter_opt(optarg, options);
4986                 if (offending)
4987                         die("unknown change class '%c' in --diff-filter=%s",
4988                             offending, optarg);
4989                 return argcount;
4990         }
4991         else if (!strcmp(arg, "--no-abbrev"))
4992                 options->abbrev = 0;
4993         else if (!strcmp(arg, "--abbrev"))
4994                 options->abbrev = DEFAULT_ABBREV;
4995         else if (skip_prefix(arg, "--abbrev=", &arg)) {
4996                 options->abbrev = strtoul(arg, NULL, 10);
4997                 if (options->abbrev < MINIMUM_ABBREV)
4998                         options->abbrev = MINIMUM_ABBREV;
4999                 else if (40 < options->abbrev)
5000                         options->abbrev = 40;
5001         }
5002         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
5003                 options->a_prefix = optarg;
5004                 return argcount;
5005         }
5006         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
5007                 options->line_prefix = optarg;
5008                 options->line_prefix_length = strlen(options->line_prefix);
5009                 graph_setup_line_prefix(options);
5010                 return argcount;
5011         }
5012         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
5013                 options->b_prefix = optarg;
5014                 return argcount;
5015         }
5016         else if (!strcmp(arg, "--no-prefix"))
5017                 options->a_prefix = options->b_prefix = "";
5018         else if (opt_arg(arg, '\0', "inter-hunk-context",
5019                          &options->interhunkcontext))
5020                 ;
5021         else if (!strcmp(arg, "-W"))
5022                 options->flags.funccontext = 1;
5023         else if (!strcmp(arg, "--function-context"))
5024                 options->flags.funccontext = 1;
5025         else if (!strcmp(arg, "--no-function-context"))
5026                 options->flags.funccontext = 0;
5027         else if ((argcount = parse_long_opt("output", av, &optarg))) {
5028                 char *path = prefix_filename(prefix, optarg);
5029                 options->file = xfopen(path, "w");
5030                 options->close_file = 1;
5031                 if (options->use_color != GIT_COLOR_ALWAYS)
5032                         options->use_color = GIT_COLOR_NEVER;
5033                 free(path);
5034                 return argcount;
5035         } else
5036                 return 0;
5037         return 1;
5038 }
5039
5040 int parse_rename_score(const char **cp_p)
5041 {
5042         unsigned long num, scale;
5043         int ch, dot;
5044         const char *cp = *cp_p;
5045
5046         num = 0;
5047         scale = 1;
5048         dot = 0;
5049         for (;;) {
5050                 ch = *cp;
5051                 if ( !dot && ch == '.' ) {
5052                         scale = 1;
5053                         dot = 1;
5054                 } else if ( ch == '%' ) {
5055                         scale = dot ? scale*100 : 100;
5056                         cp++;   /* % is always at the end */
5057                         break;
5058                 } else if ( ch >= '0' && ch <= '9' ) {
5059                         if ( scale < 100000 ) {
5060                                 scale *= 10;
5061                                 num = (num*10) + (ch-'0');
5062                         }
5063                 } else {
5064                         break;
5065                 }
5066                 cp++;
5067         }
5068         *cp_p = cp;
5069
5070         /* user says num divided by scale and we say internally that
5071          * is MAX_SCORE * num / scale.
5072          */
5073         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5074 }
5075
5076 static int diff_scoreopt_parse(const char *opt)
5077 {
5078         int opt1, opt2, cmd;
5079
5080         if (*opt++ != '-')
5081                 return -1;
5082         cmd = *opt++;
5083         if (cmd == '-') {
5084                 /* convert the long-form arguments into short-form versions */
5085                 if (skip_prefix(opt, "break-rewrites", &opt)) {
5086                         if (*opt == 0 || *opt++ == '=')
5087                                 cmd = 'B';
5088                 } else if (skip_prefix(opt, "find-copies", &opt)) {
5089                         if (*opt == 0 || *opt++ == '=')
5090                                 cmd = 'C';
5091                 } else if (skip_prefix(opt, "find-renames", &opt)) {
5092                         if (*opt == 0 || *opt++ == '=')
5093                                 cmd = 'M';
5094                 }
5095         }
5096         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
5097                 return -1; /* that is not a -M, -C, or -B option */
5098
5099         opt1 = parse_rename_score(&opt);
5100         if (cmd != 'B')
5101                 opt2 = 0;
5102         else {
5103                 if (*opt == 0)
5104                         opt2 = 0;
5105                 else if (*opt != '/')
5106                         return -1; /* we expect -B80/99 or -B80 */
5107                 else {
5108                         opt++;
5109                         opt2 = parse_rename_score(&opt);
5110                 }
5111         }
5112         if (*opt != 0)
5113                 return -1;
5114         return opt1 | (opt2 << 16);
5115 }
5116
5117 struct diff_queue_struct diff_queued_diff;
5118
5119 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5120 {
5121         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5122         queue->queue[queue->nr++] = dp;
5123 }
5124
5125 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5126                                  struct diff_filespec *one,
5127                                  struct diff_filespec *two)
5128 {
5129         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5130         dp->one = one;
5131         dp->two = two;
5132         if (queue)
5133                 diff_q(queue, dp);
5134         return dp;
5135 }
5136
5137 void diff_free_filepair(struct diff_filepair *p)
5138 {
5139         free_filespec(p->one);
5140         free_filespec(p->two);
5141         free(p);
5142 }
5143
5144 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5145 {
5146         int abblen;
5147         const char *abbrev;
5148
5149         /* Do we want all 40 hex characters? */
5150         if (len == GIT_SHA1_HEXSZ)
5151                 return oid_to_hex(oid);
5152
5153         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5154         abbrev = diff_abbrev_oid(oid, len);
5155
5156         if (!print_sha1_ellipsis())
5157                 return abbrev;
5158
5159         abblen = strlen(abbrev);
5160
5161         /*
5162          * In well-behaved cases, where the abbreviated result is the
5163          * same as the requested length, append three dots after the
5164          * abbreviation (hence the whole logic is limited to the case
5165          * where abblen < 37); when the actual abbreviated result is a
5166          * bit longer than the requested length, we reduce the number
5167          * of dots so that they match the well-behaved ones.  However,
5168          * if the actual abbreviation is longer than the requested
5169          * length by more than three, we give up on aligning, and add
5170          * three dots anyway, to indicate that the output is not the
5171          * full object name.  Yes, this may be suboptimal, but this
5172          * appears only in "diff --raw --abbrev" output and it is not
5173          * worth the effort to change it now.  Note that this would
5174          * likely to work fine when the automatic sizing of default
5175          * abbreviation length is used--we would be fed -1 in "len" in
5176          * that case, and will end up always appending three-dots, but
5177          * the automatic sizing is supposed to give abblen that ensures
5178          * uniqueness across all objects (statistically speaking).
5179          */
5180         if (abblen < GIT_SHA1_HEXSZ - 3) {
5181                 static char hex[GIT_MAX_HEXSZ + 1];
5182                 if (len < abblen && abblen <= len + 2)
5183                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5184                 else
5185                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5186                 return hex;
5187         }
5188
5189         return oid_to_hex(oid);
5190 }
5191
5192 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5193 {
5194         int line_termination = opt->line_termination;
5195         int inter_name_termination = line_termination ? '\t' : '\0';
5196
5197         fprintf(opt->file, "%s", diff_line_prefix(opt));
5198         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5199                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5200                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5201                 fprintf(opt->file, "%s ",
5202                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5203         }
5204         if (p->score) {
5205                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5206                         inter_name_termination);
5207         } else {
5208                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5209         }
5210
5211         if (p->status == DIFF_STATUS_COPIED ||
5212             p->status == DIFF_STATUS_RENAMED) {
5213                 const char *name_a, *name_b;
5214                 name_a = p->one->path;
5215                 name_b = p->two->path;
5216                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5217                 write_name_quoted(name_a, opt->file, inter_name_termination);
5218                 write_name_quoted(name_b, opt->file, line_termination);
5219         } else {
5220                 const char *name_a, *name_b;
5221                 name_a = p->one->mode ? p->one->path : p->two->path;
5222                 name_b = NULL;
5223                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5224                 write_name_quoted(name_a, opt->file, line_termination);
5225         }
5226 }
5227
5228 int diff_unmodified_pair(struct diff_filepair *p)
5229 {
5230         /* This function is written stricter than necessary to support
5231          * the currently implemented transformers, but the idea is to
5232          * let transformers to produce diff_filepairs any way they want,
5233          * and filter and clean them up here before producing the output.
5234          */
5235         struct diff_filespec *one = p->one, *two = p->two;
5236
5237         if (DIFF_PAIR_UNMERGED(p))
5238                 return 0; /* unmerged is interesting */
5239
5240         /* deletion, addition, mode or type change
5241          * and rename are all interesting.
5242          */
5243         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5244             DIFF_PAIR_MODE_CHANGED(p) ||
5245             strcmp(one->path, two->path))
5246                 return 0;
5247
5248         /* both are valid and point at the same path.  that is, we are
5249          * dealing with a change.
5250          */
5251         if (one->oid_valid && two->oid_valid &&
5252             !oidcmp(&one->oid, &two->oid) &&
5253             !one->dirty_submodule && !two->dirty_submodule)
5254                 return 1; /* no change */
5255         if (!one->oid_valid && !two->oid_valid)
5256                 return 1; /* both look at the same file on the filesystem. */
5257         return 0;
5258 }
5259
5260 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5261 {
5262         if (diff_unmodified_pair(p))
5263                 return;
5264
5265         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5266             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5267                 return; /* no tree diffs in patch format */
5268
5269         run_diff(p, o);
5270 }
5271
5272 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5273                             struct diffstat_t *diffstat)
5274 {
5275         if (diff_unmodified_pair(p))
5276                 return;
5277
5278         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5279             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5280                 return; /* no useful stat for tree diffs */
5281
5282         run_diffstat(p, o, diffstat);
5283 }
5284
5285 static void diff_flush_checkdiff(struct diff_filepair *p,
5286                 struct diff_options *o)
5287 {
5288         if (diff_unmodified_pair(p))
5289                 return;
5290
5291         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5292             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5293                 return; /* nothing to check in tree diffs */
5294
5295         run_checkdiff(p, o);
5296 }
5297
5298 int diff_queue_is_empty(void)
5299 {
5300         struct diff_queue_struct *q = &diff_queued_diff;
5301         int i;
5302         for (i = 0; i < q->nr; i++)
5303                 if (!diff_unmodified_pair(q->queue[i]))
5304                         return 0;
5305         return 1;
5306 }
5307
5308 #if DIFF_DEBUG
5309 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5310 {
5311         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5312                 x, one ? one : "",
5313                 s->path,
5314                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5315                 s->mode,
5316                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5317         fprintf(stderr, "queue[%d] %s size %lu\n",
5318                 x, one ? one : "",
5319                 s->size);
5320 }
5321
5322 void diff_debug_filepair(const struct diff_filepair *p, int i)
5323 {
5324         diff_debug_filespec(p->one, i, "one");
5325         diff_debug_filespec(p->two, i, "two");
5326         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5327                 p->score, p->status ? p->status : '?',
5328                 p->one->rename_used, p->broken_pair);
5329 }
5330
5331 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5332 {
5333         int i;
5334         if (msg)
5335                 fprintf(stderr, "%s\n", msg);
5336         fprintf(stderr, "q->nr = %d\n", q->nr);
5337         for (i = 0; i < q->nr; i++) {
5338                 struct diff_filepair *p = q->queue[i];
5339                 diff_debug_filepair(p, i);
5340         }
5341 }
5342 #endif
5343
5344 static void diff_resolve_rename_copy(void)
5345 {
5346         int i;
5347         struct diff_filepair *p;
5348         struct diff_queue_struct *q = &diff_queued_diff;
5349
5350         diff_debug_queue("resolve-rename-copy", q);
5351
5352         for (i = 0; i < q->nr; i++) {
5353                 p = q->queue[i];
5354                 p->status = 0; /* undecided */
5355                 if (DIFF_PAIR_UNMERGED(p))
5356                         p->status = DIFF_STATUS_UNMERGED;
5357                 else if (!DIFF_FILE_VALID(p->one))
5358                         p->status = DIFF_STATUS_ADDED;
5359                 else if (!DIFF_FILE_VALID(p->two))
5360                         p->status = DIFF_STATUS_DELETED;
5361                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5362                         p->status = DIFF_STATUS_TYPE_CHANGED;
5363
5364                 /* from this point on, we are dealing with a pair
5365                  * whose both sides are valid and of the same type, i.e.
5366                  * either in-place edit or rename/copy edit.
5367                  */
5368                 else if (DIFF_PAIR_RENAME(p)) {
5369                         /*
5370                          * A rename might have re-connected a broken
5371                          * pair up, causing the pathnames to be the
5372                          * same again. If so, that's not a rename at
5373                          * all, just a modification..
5374                          *
5375                          * Otherwise, see if this source was used for
5376                          * multiple renames, in which case we decrement
5377                          * the count, and call it a copy.
5378                          */
5379                         if (!strcmp(p->one->path, p->two->path))
5380                                 p->status = DIFF_STATUS_MODIFIED;
5381                         else if (--p->one->rename_used > 0)
5382                                 p->status = DIFF_STATUS_COPIED;
5383                         else
5384                                 p->status = DIFF_STATUS_RENAMED;
5385                 }
5386                 else if (oidcmp(&p->one->oid, &p->two->oid) ||
5387                          p->one->mode != p->two->mode ||
5388                          p->one->dirty_submodule ||
5389                          p->two->dirty_submodule ||
5390                          is_null_oid(&p->one->oid))
5391                         p->status = DIFF_STATUS_MODIFIED;
5392                 else {
5393                         /* This is a "no-change" entry and should not
5394                          * happen anymore, but prepare for broken callers.
5395                          */
5396                         error("feeding unmodified %s to diffcore",
5397                               p->one->path);
5398                         p->status = DIFF_STATUS_UNKNOWN;
5399                 }
5400         }
5401         diff_debug_queue("resolve-rename-copy done", q);
5402 }
5403
5404 static int check_pair_status(struct diff_filepair *p)
5405 {
5406         switch (p->status) {
5407         case DIFF_STATUS_UNKNOWN:
5408                 return 0;
5409         case 0:
5410                 die("internal error in diff-resolve-rename-copy");
5411         default:
5412                 return 1;
5413         }
5414 }
5415
5416 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5417 {
5418         int fmt = opt->output_format;
5419
5420         if (fmt & DIFF_FORMAT_CHECKDIFF)
5421                 diff_flush_checkdiff(p, opt);
5422         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5423                 diff_flush_raw(p, opt);
5424         else if (fmt & DIFF_FORMAT_NAME) {
5425                 const char *name_a, *name_b;
5426                 name_a = p->two->path;
5427                 name_b = NULL;
5428                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5429                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5430                 write_name_quoted(name_a, opt->file, opt->line_termination);
5431         }
5432 }
5433
5434 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5435 {
5436         struct strbuf sb = STRBUF_INIT;
5437         if (fs->mode)
5438                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5439         else
5440                 strbuf_addf(&sb, " %s ", newdelete);
5441
5442         quote_c_style(fs->path, &sb, NULL, 0);
5443         strbuf_addch(&sb, '\n');
5444         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5445                          sb.buf, sb.len, 0);
5446         strbuf_release(&sb);
5447 }
5448
5449 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5450                 int show_name)
5451 {
5452         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5453                 struct strbuf sb = STRBUF_INIT;
5454                 strbuf_addf(&sb, " mode change %06o => %06o",
5455                             p->one->mode, p->two->mode);
5456                 if (show_name) {
5457                         strbuf_addch(&sb, ' ');
5458                         quote_c_style(p->two->path, &sb, NULL, 0);
5459                 }
5460                 strbuf_addch(&sb, '\n');
5461                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5462                                  sb.buf, sb.len, 0);
5463                 strbuf_release(&sb);
5464         }
5465 }
5466
5467 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5468                 struct diff_filepair *p)
5469 {
5470         struct strbuf sb = STRBUF_INIT;
5471         struct strbuf names = STRBUF_INIT;
5472
5473         pprint_rename(&names, p->one->path, p->two->path);
5474         strbuf_addf(&sb, " %s %s (%d%%)\n",
5475                     renamecopy, names.buf, similarity_index(p));
5476         strbuf_release(&names);
5477         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5478                                  sb.buf, sb.len, 0);
5479         show_mode_change(opt, p, 0);
5480         strbuf_release(&sb);
5481 }
5482
5483 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5484 {
5485         switch(p->status) {
5486         case DIFF_STATUS_DELETED:
5487                 show_file_mode_name(opt, "delete", p->one);
5488                 break;
5489         case DIFF_STATUS_ADDED:
5490                 show_file_mode_name(opt, "create", p->two);
5491                 break;
5492         case DIFF_STATUS_COPIED:
5493                 show_rename_copy(opt, "copy", p);
5494                 break;
5495         case DIFF_STATUS_RENAMED:
5496                 show_rename_copy(opt, "rename", p);
5497                 break;
5498         default:
5499                 if (p->score) {
5500                         struct strbuf sb = STRBUF_INIT;
5501                         strbuf_addstr(&sb, " rewrite ");
5502                         quote_c_style(p->two->path, &sb, NULL, 0);
5503                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5504                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5505                                          sb.buf, sb.len, 0);
5506                         strbuf_release(&sb);
5507                 }
5508                 show_mode_change(opt, p, !p->score);
5509                 break;
5510         }
5511 }
5512
5513 struct patch_id_t {
5514         git_SHA_CTX *ctx;
5515         int patchlen;
5516 };
5517
5518 static int remove_space(char *line, int len)
5519 {
5520         int i;
5521         char *dst = line;
5522         unsigned char c;
5523
5524         for (i = 0; i < len; i++)
5525                 if (!isspace((c = line[i])))
5526                         *dst++ = c;
5527
5528         return dst - line;
5529 }
5530
5531 static void patch_id_consume(void *priv, char *line, unsigned long len)
5532 {
5533         struct patch_id_t *data = priv;
5534         int new_len;
5535
5536         /* Ignore line numbers when computing the SHA1 of the patch */
5537         if (starts_with(line, "@@ -"))
5538                 return;
5539
5540         new_len = remove_space(line, len);
5541
5542         git_SHA1_Update(data->ctx, line, new_len);
5543         data->patchlen += new_len;
5544 }
5545
5546 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5547 {
5548         git_SHA1_Update(ctx, str, strlen(str));
5549 }
5550
5551 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5552 {
5553         /* large enough for 2^32 in octal */
5554         char buf[12];
5555         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5556         git_SHA1_Update(ctx, buf, len);
5557 }
5558
5559 /* returns 0 upon success, and writes result into sha1 */
5560 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5561 {
5562         struct diff_queue_struct *q = &diff_queued_diff;
5563         int i;
5564         git_SHA_CTX ctx;
5565         struct patch_id_t data;
5566
5567         git_SHA1_Init(&ctx);
5568         memset(&data, 0, sizeof(struct patch_id_t));
5569         data.ctx = &ctx;
5570
5571         for (i = 0; i < q->nr; i++) {
5572                 xpparam_t xpp;
5573                 xdemitconf_t xecfg;
5574                 mmfile_t mf1, mf2;
5575                 struct diff_filepair *p = q->queue[i];
5576                 int len1, len2;
5577
5578                 memset(&xpp, 0, sizeof(xpp));
5579                 memset(&xecfg, 0, sizeof(xecfg));
5580                 if (p->status == 0)
5581                         return error("internal diff status error");
5582                 if (p->status == DIFF_STATUS_UNKNOWN)
5583                         continue;
5584                 if (diff_unmodified_pair(p))
5585                         continue;
5586                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5587                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5588                         continue;
5589                 if (DIFF_PAIR_UNMERGED(p))
5590                         continue;
5591
5592                 diff_fill_oid_info(p->one);
5593                 diff_fill_oid_info(p->two);
5594
5595                 len1 = remove_space(p->one->path, strlen(p->one->path));
5596                 len2 = remove_space(p->two->path, strlen(p->two->path));
5597                 patch_id_add_string(&ctx, "diff--git");
5598                 patch_id_add_string(&ctx, "a/");
5599                 git_SHA1_Update(&ctx, p->one->path, len1);
5600                 patch_id_add_string(&ctx, "b/");
5601                 git_SHA1_Update(&ctx, p->two->path, len2);
5602
5603                 if (p->one->mode == 0) {
5604                         patch_id_add_string(&ctx, "newfilemode");
5605                         patch_id_add_mode(&ctx, p->two->mode);
5606                         patch_id_add_string(&ctx, "---/dev/null");
5607                         patch_id_add_string(&ctx, "+++b/");
5608                         git_SHA1_Update(&ctx, p->two->path, len2);
5609                 } else if (p->two->mode == 0) {
5610                         patch_id_add_string(&ctx, "deletedfilemode");
5611                         patch_id_add_mode(&ctx, p->one->mode);
5612                         patch_id_add_string(&ctx, "---a/");
5613                         git_SHA1_Update(&ctx, p->one->path, len1);
5614                         patch_id_add_string(&ctx, "+++/dev/null");
5615                 } else {
5616                         patch_id_add_string(&ctx, "---a/");
5617                         git_SHA1_Update(&ctx, p->one->path, len1);
5618                         patch_id_add_string(&ctx, "+++b/");
5619                         git_SHA1_Update(&ctx, p->two->path, len2);
5620                 }
5621
5622                 if (diff_header_only)
5623                         continue;
5624
5625                 if (fill_mmfile(&mf1, p->one) < 0 ||
5626                     fill_mmfile(&mf2, p->two) < 0)
5627                         return error("unable to read files to diff");
5628
5629                 if (diff_filespec_is_binary(p->one) ||
5630                     diff_filespec_is_binary(p->two)) {
5631                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5632                                         GIT_SHA1_HEXSZ);
5633                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5634                                         GIT_SHA1_HEXSZ);
5635                         continue;
5636                 }
5637
5638                 xpp.flags = 0;
5639                 xecfg.ctxlen = 3;
5640                 xecfg.flags = 0;
5641                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
5642                                   &xpp, &xecfg))
5643                         return error("unable to generate patch-id diff for %s",
5644                                      p->one->path);
5645         }
5646
5647         git_SHA1_Final(oid->hash, &ctx);
5648         return 0;
5649 }
5650
5651 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5652 {
5653         struct diff_queue_struct *q = &diff_queued_diff;
5654         int i;
5655         int result = diff_get_patch_id(options, oid, diff_header_only);
5656
5657         for (i = 0; i < q->nr; i++)
5658                 diff_free_filepair(q->queue[i]);
5659
5660         free(q->queue);
5661         DIFF_QUEUE_CLEAR(q);
5662
5663         return result;
5664 }
5665
5666 static int is_summary_empty(const struct diff_queue_struct *q)
5667 {
5668         int i;
5669
5670         for (i = 0; i < q->nr; i++) {
5671                 const struct diff_filepair *p = q->queue[i];
5672
5673                 switch (p->status) {
5674                 case DIFF_STATUS_DELETED:
5675                 case DIFF_STATUS_ADDED:
5676                 case DIFF_STATUS_COPIED:
5677                 case DIFF_STATUS_RENAMED:
5678                         return 0;
5679                 default:
5680                         if (p->score)
5681                                 return 0;
5682                         if (p->one->mode && p->two->mode &&
5683                             p->one->mode != p->two->mode)
5684                                 return 0;
5685                         break;
5686                 }
5687         }
5688         return 1;
5689 }
5690
5691 static const char rename_limit_warning[] =
5692 N_("inexact rename detection was skipped due to too many files.");
5693
5694 static const char degrade_cc_to_c_warning[] =
5695 N_("only found copies from modified paths due to too many files.");
5696
5697 static const char rename_limit_advice[] =
5698 N_("you may want to set your %s variable to at least "
5699    "%d and retry the command.");
5700
5701 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5702 {
5703         fflush(stdout);
5704         if (degraded_cc)
5705                 warning(_(degrade_cc_to_c_warning));
5706         else if (needed)
5707                 warning(_(rename_limit_warning));
5708         else
5709                 return;
5710         if (0 < needed)
5711                 warning(_(rename_limit_advice), varname, needed);
5712 }
5713
5714 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5715 {
5716         int i;
5717         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5718         struct diff_queue_struct *q = &diff_queued_diff;
5719
5720         if (WSEH_NEW & WS_RULE_MASK)
5721                 die("BUG: WS rules bit mask overlaps with diff symbol flags");
5722
5723         if (o->color_moved)
5724                 o->emitted_symbols = &esm;
5725
5726         for (i = 0; i < q->nr; i++) {
5727                 struct diff_filepair *p = q->queue[i];
5728                 if (check_pair_status(p))
5729                         diff_flush_patch(p, o);
5730         }
5731
5732         if (o->emitted_symbols) {
5733                 if (o->color_moved) {
5734                         struct hashmap add_lines, del_lines;
5735
5736                         if (o->color_moved_ws_handling &
5737                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
5738                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
5739
5740                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
5741                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
5742
5743                         add_lines_to_move_detection(o, &add_lines, &del_lines);
5744                         mark_color_as_moved(o, &add_lines, &del_lines);
5745                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
5746                                 dim_moved_lines(o);
5747
5748                         hashmap_free(&add_lines, 0);
5749                         hashmap_free(&del_lines, 0);
5750                 }
5751
5752                 for (i = 0; i < esm.nr; i++)
5753                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
5754
5755                 for (i = 0; i < esm.nr; i++)
5756                         free((void *)esm.buf[i].line);
5757         }
5758         esm.nr = 0;
5759 }
5760
5761 void diff_flush(struct diff_options *options)
5762 {
5763         struct diff_queue_struct *q = &diff_queued_diff;
5764         int i, output_format = options->output_format;
5765         int separator = 0;
5766         int dirstat_by_line = 0;
5767
5768         /*
5769          * Order: raw, stat, summary, patch
5770          * or:    name/name-status/checkdiff (other bits clear)
5771          */
5772         if (!q->nr)
5773                 goto free_queue;
5774
5775         if (output_format & (DIFF_FORMAT_RAW |
5776                              DIFF_FORMAT_NAME |
5777                              DIFF_FORMAT_NAME_STATUS |
5778                              DIFF_FORMAT_CHECKDIFF)) {
5779                 for (i = 0; i < q->nr; i++) {
5780                         struct diff_filepair *p = q->queue[i];
5781                         if (check_pair_status(p))
5782                                 flush_one_pair(p, options);
5783                 }
5784                 separator++;
5785         }
5786
5787         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
5788                 dirstat_by_line = 1;
5789
5790         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
5791             dirstat_by_line) {
5792                 struct diffstat_t diffstat;
5793
5794                 memset(&diffstat, 0, sizeof(struct diffstat_t));
5795                 for (i = 0; i < q->nr; i++) {
5796                         struct diff_filepair *p = q->queue[i];
5797                         if (check_pair_status(p))
5798                                 diff_flush_stat(p, options, &diffstat);
5799                 }
5800                 if (output_format & DIFF_FORMAT_NUMSTAT)
5801                         show_numstat(&diffstat, options);
5802                 if (output_format & DIFF_FORMAT_DIFFSTAT)
5803                         show_stats(&diffstat, options);
5804                 if (output_format & DIFF_FORMAT_SHORTSTAT)
5805                         show_shortstats(&diffstat, options);
5806                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
5807                         show_dirstat_by_line(&diffstat, options);
5808                 free_diffstat_info(&diffstat);
5809                 separator++;
5810         }
5811         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
5812                 show_dirstat(options);
5813
5814         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
5815                 for (i = 0; i < q->nr; i++) {
5816                         diff_summary(options, q->queue[i]);
5817                 }
5818                 separator++;
5819         }
5820
5821         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
5822             options->flags.exit_with_status &&
5823             options->flags.diff_from_contents) {
5824                 /*
5825                  * run diff_flush_patch for the exit status. setting
5826                  * options->file to /dev/null should be safe, because we
5827                  * aren't supposed to produce any output anyway.
5828                  */
5829                 if (options->close_file)
5830                         fclose(options->file);
5831                 options->file = xfopen("/dev/null", "w");
5832                 options->close_file = 1;
5833                 options->color_moved = 0;
5834                 for (i = 0; i < q->nr; i++) {
5835                         struct diff_filepair *p = q->queue[i];
5836                         if (check_pair_status(p))
5837                                 diff_flush_patch(p, options);
5838                         if (options->found_changes)
5839                                 break;
5840                 }
5841         }
5842
5843         if (output_format & DIFF_FORMAT_PATCH) {
5844                 if (separator) {
5845                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
5846                         if (options->stat_sep)
5847                                 /* attach patch instead of inline */
5848                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
5849                                                  NULL, 0, 0);
5850                 }
5851
5852                 diff_flush_patch_all_file_pairs(options);
5853         }
5854
5855         if (output_format & DIFF_FORMAT_CALLBACK)
5856                 options->format_callback(q, options, options->format_callback_data);
5857
5858         for (i = 0; i < q->nr; i++)
5859                 diff_free_filepair(q->queue[i]);
5860 free_queue:
5861         free(q->queue);
5862         DIFF_QUEUE_CLEAR(q);
5863         if (options->close_file)
5864                 fclose(options->file);
5865
5866         /*
5867          * Report the content-level differences with HAS_CHANGES;
5868          * diff_addremove/diff_change does not set the bit when
5869          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
5870          */
5871         if (options->flags.diff_from_contents) {
5872                 if (options->found_changes)
5873                         options->flags.has_changes = 1;
5874                 else
5875                         options->flags.has_changes = 0;
5876         }
5877 }
5878
5879 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
5880 {
5881         return (((p->status == DIFF_STATUS_MODIFIED) &&
5882                  ((p->score &&
5883                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
5884                   (!p->score &&
5885                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
5886                 ((p->status != DIFF_STATUS_MODIFIED) &&
5887                  filter_bit_tst(p->status, options)));
5888 }
5889
5890 static void diffcore_apply_filter(struct diff_options *options)
5891 {
5892         int i;
5893         struct diff_queue_struct *q = &diff_queued_diff;
5894         struct diff_queue_struct outq;
5895
5896         DIFF_QUEUE_CLEAR(&outq);
5897
5898         if (!options->filter)
5899                 return;
5900
5901         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
5902                 int found;
5903                 for (i = found = 0; !found && i < q->nr; i++) {
5904                         if (match_filter(options, q->queue[i]))
5905                                 found++;
5906                 }
5907                 if (found)
5908                         return;
5909
5910                 /* otherwise we will clear the whole queue
5911                  * by copying the empty outq at the end of this
5912                  * function, but first clear the current entries
5913                  * in the queue.
5914                  */
5915                 for (i = 0; i < q->nr; i++)
5916                         diff_free_filepair(q->queue[i]);
5917         }
5918         else {
5919                 /* Only the matching ones */
5920                 for (i = 0; i < q->nr; i++) {
5921                         struct diff_filepair *p = q->queue[i];
5922                         if (match_filter(options, p))
5923                                 diff_q(&outq, p);
5924                         else
5925                                 diff_free_filepair(p);
5926                 }
5927         }
5928         free(q->queue);
5929         *q = outq;
5930 }
5931
5932 /* Check whether two filespecs with the same mode and size are identical */
5933 static int diff_filespec_is_identical(struct diff_filespec *one,
5934                                       struct diff_filespec *two)
5935 {
5936         if (S_ISGITLINK(one->mode))
5937                 return 0;
5938         if (diff_populate_filespec(one, 0))
5939                 return 0;
5940         if (diff_populate_filespec(two, 0))
5941                 return 0;
5942         return !memcmp(one->data, two->data, one->size);
5943 }
5944
5945 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
5946 {
5947         if (p->done_skip_stat_unmatch)
5948                 return p->skip_stat_unmatch_result;
5949
5950         p->done_skip_stat_unmatch = 1;
5951         p->skip_stat_unmatch_result = 0;
5952         /*
5953          * 1. Entries that come from stat info dirtiness
5954          *    always have both sides (iow, not create/delete),
5955          *    one side of the object name is unknown, with
5956          *    the same mode and size.  Keep the ones that
5957          *    do not match these criteria.  They have real
5958          *    differences.
5959          *
5960          * 2. At this point, the file is known to be modified,
5961          *    with the same mode and size, and the object
5962          *    name of one side is unknown.  Need to inspect
5963          *    the identical contents.
5964          */
5965         if (!DIFF_FILE_VALID(p->one) || /* (1) */
5966             !DIFF_FILE_VALID(p->two) ||
5967             (p->one->oid_valid && p->two->oid_valid) ||
5968             (p->one->mode != p->two->mode) ||
5969             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
5970             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
5971             (p->one->size != p->two->size) ||
5972             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
5973                 p->skip_stat_unmatch_result = 1;
5974         return p->skip_stat_unmatch_result;
5975 }
5976
5977 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
5978 {
5979         int i;
5980         struct diff_queue_struct *q = &diff_queued_diff;
5981         struct diff_queue_struct outq;
5982         DIFF_QUEUE_CLEAR(&outq);
5983
5984         for (i = 0; i < q->nr; i++) {
5985                 struct diff_filepair *p = q->queue[i];
5986
5987                 if (diff_filespec_check_stat_unmatch(p))
5988                         diff_q(&outq, p);
5989                 else {
5990                         /*
5991                          * The caller can subtract 1 from skip_stat_unmatch
5992                          * to determine how many paths were dirty only
5993                          * due to stat info mismatch.
5994                          */
5995                         if (!diffopt->flags.no_index)
5996                                 diffopt->skip_stat_unmatch++;
5997                         diff_free_filepair(p);
5998                 }
5999         }
6000         free(q->queue);
6001         *q = outq;
6002 }
6003
6004 static int diffnamecmp(const void *a_, const void *b_)
6005 {
6006         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6007         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6008         const char *name_a, *name_b;
6009
6010         name_a = a->one ? a->one->path : a->two->path;
6011         name_b = b->one ? b->one->path : b->two->path;
6012         return strcmp(name_a, name_b);
6013 }
6014
6015 void diffcore_fix_diff_index(struct diff_options *options)
6016 {
6017         struct diff_queue_struct *q = &diff_queued_diff;
6018         QSORT(q->queue, q->nr, diffnamecmp);
6019 }
6020
6021 void diffcore_std(struct diff_options *options)
6022 {
6023         /* NOTE please keep the following in sync with diff_tree_combined() */
6024         if (options->skip_stat_unmatch)
6025                 diffcore_skip_stat_unmatch(options);
6026         if (!options->found_follow) {
6027                 /* See try_to_follow_renames() in tree-diff.c */
6028                 if (options->break_opt != -1)
6029                         diffcore_break(options->break_opt);
6030                 if (options->detect_rename)
6031                         diffcore_rename(options);
6032                 if (options->break_opt != -1)
6033                         diffcore_merge_broken();
6034         }
6035         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6036                 diffcore_pickaxe(options);
6037         if (options->orderfile)
6038                 diffcore_order(options->orderfile);
6039         if (!options->found_follow)
6040                 /* See try_to_follow_renames() in tree-diff.c */
6041                 diff_resolve_rename_copy();
6042         diffcore_apply_filter(options);
6043
6044         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6045                 options->flags.has_changes = 1;
6046         else
6047                 options->flags.has_changes = 0;
6048
6049         options->found_follow = 0;
6050 }
6051
6052 int diff_result_code(struct diff_options *opt, int status)
6053 {
6054         int result = 0;
6055
6056         diff_warn_rename_limit("diff.renameLimit",
6057                                opt->needed_rename_limit,
6058                                opt->degraded_cc_to_c);
6059         if (!opt->flags.exit_with_status &&
6060             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6061                 return status;
6062         if (opt->flags.exit_with_status &&
6063             opt->flags.has_changes)
6064                 result |= 01;
6065         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6066             opt->flags.check_failed)
6067                 result |= 02;
6068         return result;
6069 }
6070
6071 int diff_can_quit_early(struct diff_options *opt)
6072 {
6073         return (opt->flags.quick &&
6074                 !opt->filter &&
6075                 opt->flags.has_changes);
6076 }
6077
6078 /*
6079  * Shall changes to this submodule be ignored?
6080  *
6081  * Submodule changes can be configured to be ignored separately for each path,
6082  * but that configuration can be overridden from the command line.
6083  */
6084 static int is_submodule_ignored(const char *path, struct diff_options *options)
6085 {
6086         int ignored = 0;
6087         struct diff_flags orig_flags = options->flags;
6088         if (!options->flags.override_submodule_config)
6089                 set_diffopt_flags_from_submodule_config(options, path);
6090         if (options->flags.ignore_submodules)
6091                 ignored = 1;
6092         options->flags = orig_flags;
6093         return ignored;
6094 }
6095
6096 void diff_addremove(struct diff_options *options,
6097                     int addremove, unsigned mode,
6098                     const struct object_id *oid,
6099                     int oid_valid,
6100                     const char *concatpath, unsigned dirty_submodule)
6101 {
6102         struct diff_filespec *one, *two;
6103
6104         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6105                 return;
6106
6107         /* This may look odd, but it is a preparation for
6108          * feeding "there are unchanged files which should
6109          * not produce diffs, but when you are doing copy
6110          * detection you would need them, so here they are"
6111          * entries to the diff-core.  They will be prefixed
6112          * with something like '=' or '*' (I haven't decided
6113          * which but should not make any difference).
6114          * Feeding the same new and old to diff_change()
6115          * also has the same effect.
6116          * Before the final output happens, they are pruned after
6117          * merged into rename/copy pairs as appropriate.
6118          */
6119         if (options->flags.reverse_diff)
6120                 addremove = (addremove == '+' ? '-' :
6121                              addremove == '-' ? '+' : addremove);
6122
6123         if (options->prefix &&
6124             strncmp(concatpath, options->prefix, options->prefix_length))
6125                 return;
6126
6127         one = alloc_filespec(concatpath);
6128         two = alloc_filespec(concatpath);
6129
6130         if (addremove != '+')
6131                 fill_filespec(one, oid, oid_valid, mode);
6132         if (addremove != '-') {
6133                 fill_filespec(two, oid, oid_valid, mode);
6134                 two->dirty_submodule = dirty_submodule;
6135         }
6136
6137         diff_queue(&diff_queued_diff, one, two);
6138         if (!options->flags.diff_from_contents)
6139                 options->flags.has_changes = 1;
6140 }
6141
6142 void diff_change(struct diff_options *options,
6143                  unsigned old_mode, unsigned new_mode,
6144                  const struct object_id *old_oid,
6145                  const struct object_id *new_oid,
6146                  int old_oid_valid, int new_oid_valid,
6147                  const char *concatpath,
6148                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6149 {
6150         struct diff_filespec *one, *two;
6151         struct diff_filepair *p;
6152
6153         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6154             is_submodule_ignored(concatpath, options))
6155                 return;
6156
6157         if (options->flags.reverse_diff) {
6158                 SWAP(old_mode, new_mode);
6159                 SWAP(old_oid, new_oid);
6160                 SWAP(old_oid_valid, new_oid_valid);
6161                 SWAP(old_dirty_submodule, new_dirty_submodule);
6162         }
6163
6164         if (options->prefix &&
6165             strncmp(concatpath, options->prefix, options->prefix_length))
6166                 return;
6167
6168         one = alloc_filespec(concatpath);
6169         two = alloc_filespec(concatpath);
6170         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6171         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6172         one->dirty_submodule = old_dirty_submodule;
6173         two->dirty_submodule = new_dirty_submodule;
6174         p = diff_queue(&diff_queued_diff, one, two);
6175
6176         if (options->flags.diff_from_contents)
6177                 return;
6178
6179         if (options->flags.quick && options->skip_stat_unmatch &&
6180             !diff_filespec_check_stat_unmatch(p))
6181                 return;
6182
6183         options->flags.has_changes = 1;
6184 }
6185
6186 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6187 {
6188         struct diff_filepair *pair;
6189         struct diff_filespec *one, *two;
6190
6191         if (options->prefix &&
6192             strncmp(path, options->prefix, options->prefix_length))
6193                 return NULL;
6194
6195         one = alloc_filespec(path);
6196         two = alloc_filespec(path);
6197         pair = diff_queue(&diff_queued_diff, one, two);
6198         pair->is_unmerged = 1;
6199         return pair;
6200 }
6201
6202 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
6203                 size_t *outsize)
6204 {
6205         struct diff_tempfile *temp;
6206         const char *argv[3];
6207         const char **arg = argv;
6208         struct child_process child = CHILD_PROCESS_INIT;
6209         struct strbuf buf = STRBUF_INIT;
6210         int err = 0;
6211
6212         temp = prepare_temp_file(spec->path, spec);
6213         *arg++ = pgm;
6214         *arg++ = temp->name;
6215         *arg = NULL;
6216
6217         child.use_shell = 1;
6218         child.argv = argv;
6219         child.out = -1;
6220         if (start_command(&child)) {
6221                 remove_tempfile();
6222                 return NULL;
6223         }
6224
6225         if (strbuf_read(&buf, child.out, 0) < 0)
6226                 err = error("error reading from textconv command '%s'", pgm);
6227         close(child.out);
6228
6229         if (finish_command(&child) || err) {
6230                 strbuf_release(&buf);
6231                 remove_tempfile();
6232                 return NULL;
6233         }
6234         remove_tempfile();
6235
6236         return strbuf_detach(&buf, outsize);
6237 }
6238
6239 size_t fill_textconv(struct userdiff_driver *driver,
6240                      struct diff_filespec *df,
6241                      char **outbuf)
6242 {
6243         size_t size;
6244
6245         if (!driver) {
6246                 if (!DIFF_FILE_VALID(df)) {
6247                         *outbuf = "";
6248                         return 0;
6249                 }
6250                 if (diff_populate_filespec(df, 0))
6251                         die("unable to read files to diff");
6252                 *outbuf = df->data;
6253                 return df->size;
6254         }
6255
6256         if (!driver->textconv)
6257                 die("BUG: fill_textconv called with non-textconv driver");
6258
6259         if (driver->textconv_cache && df->oid_valid) {
6260                 *outbuf = notes_cache_get(driver->textconv_cache,
6261                                           &df->oid,
6262                                           &size);
6263                 if (*outbuf)
6264                         return size;
6265         }
6266
6267         *outbuf = run_textconv(driver->textconv, df, &size);
6268         if (!*outbuf)
6269                 die("unable to read files to diff");
6270
6271         if (driver->textconv_cache && df->oid_valid) {
6272                 /* ignore errors, as we might be in a readonly repository */
6273                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6274                                 size);
6275                 /*
6276                  * we could save up changes and flush them all at the end,
6277                  * but we would need an extra call after all diffing is done.
6278                  * Since generating a cache entry is the slow path anyway,
6279                  * this extra overhead probably isn't a big deal.
6280                  */
6281                 notes_cache_write(driver->textconv_cache);
6282         }
6283
6284         return size;
6285 }
6286
6287 int textconv_object(const char *path,
6288                     unsigned mode,
6289                     const struct object_id *oid,
6290                     int oid_valid,
6291                     char **buf,
6292                     unsigned long *buf_size)
6293 {
6294         struct diff_filespec *df;
6295         struct userdiff_driver *textconv;
6296
6297         df = alloc_filespec(path);
6298         fill_filespec(df, oid, oid_valid, mode);
6299         textconv = get_textconv(df);
6300         if (!textconv) {
6301                 free_filespec(df);
6302                 return 0;
6303         }
6304
6305         *buf_size = fill_textconv(textconv, df, buf);
6306         free_filespec(df);
6307         return 1;
6308 }
6309
6310 void setup_diff_pager(struct diff_options *opt)
6311 {
6312         /*
6313          * If the user asked for our exit code, then either they want --quiet
6314          * or --exit-code. We should definitely not bother with a pager in the
6315          * former case, as we will generate no output. Since we still properly
6316          * report our exit code even when a pager is run, we _could_ run a
6317          * pager with --exit-code. But since we have not done so historically,
6318          * and because it is easy to find people oneline advising "git diff
6319          * --exit-code" in hooks and other scripts, we do not do so.
6320          */
6321         if (!opt->flags.exit_with_status &&
6322             check_pager_config("diff") != 0)
6323                 setup_pager();
6324 }