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

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "quote.h"
   6 #include "diff.h"
   7 #include "diffcore.h"
   8 #include "delta.h"
   9 #include "xdiff-interface.h"
  10 #include "color.h"
  11 #include "attr.h"
  12 #include "run-command.h"
  13 #include "utf8.h"
  14 #include "userdiff.h"
  15 #include "sigchain.h"
  16 #include "submodule-config.h"
  17 #include "submodule.h"
  18 #include "ll-merge.h"
  19 #include "string-list.h"
  20 #include "argv-array.h"
  21
  22 #ifdef NO_FAST_WORKING_DIRECTORY
  23 #define FAST_WORKING_DIRECTORY 0
  24 #else
  25 #define FAST_WORKING_DIRECTORY 1
  26 #endif
  27
  28 static int diff_detect_rename_default;
  29 static int diff_rename_limit_default = 400;
  30 static int diff_suppress_blank_empty;
  31 static int diff_use_color_default = -1;
  32 static int diff_context_default = 3;
  33 static const char *diff_word_regex_cfg;
  34 static const char *external_diff_cmd_cfg;
  35 static const char *diff_order_file_cfg;
  36 int diff_auto_refresh_index = 1;
  37 static int diff_mnemonic_prefix;
  38 static int diff_no_prefix;
  39 static int diff_stat_graph_width;
  40 static int diff_dirstat_permille_default = 30;
  41 static struct diff_options default_diff_options;
  42 static long diff_algorithm;
  43
  44 static char diff_colors[][COLOR_MAXLEN] = {
  45         GIT_COLOR_RESET,
  46         GIT_COLOR_NORMAL,       /* CONTEXT */
  47         GIT_COLOR_BOLD,         /* METAINFO */
  48         GIT_COLOR_CYAN,         /* FRAGINFO */
  49         GIT_COLOR_RED,          /* OLD */
  50         GIT_COLOR_GREEN,        /* NEW */
  51         GIT_COLOR_YELLOW,       /* COMMIT */
  52         GIT_COLOR_BG_RED,       /* WHITESPACE */
  53         GIT_COLOR_NORMAL,       /* FUNCINFO */
  54 };
  55
  56 static int parse_diff_color_slot(const char *var)
  57 {
  58         if (!strcasecmp(var, "context") || !strcasecmp(var, "plain"))
  59                 return DIFF_CONTEXT;
  60         if (!strcasecmp(var, "meta"))
  61                 return DIFF_METAINFO;
  62         if (!strcasecmp(var, "frag"))
  63                 return DIFF_FRAGINFO;
  64         if (!strcasecmp(var, "old"))
  65                 return DIFF_FILE_OLD;
  66         if (!strcasecmp(var, "new"))
  67                 return DIFF_FILE_NEW;
  68         if (!strcasecmp(var, "commit"))
  69                 return DIFF_COMMIT;
  70         if (!strcasecmp(var, "whitespace"))
  71                 return DIFF_WHITESPACE;
  72         if (!strcasecmp(var, "func"))
  73                 return DIFF_FUNCINFO;
  74         return -1;
  75 }
  76
  77 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
  78                                 struct strbuf *errmsg)
  79 {
  80         char *params_copy = xstrdup(params_string);
  81         struct string_list params = STRING_LIST_INIT_NODUP;
  82         int ret = 0;
  83         int i;
  84
  85         if (*params_copy)
  86                 string_list_split_in_place(&params, params_copy, ',', -1);
  87         for (i = 0; i < params.nr; i++) {
  88                 const char *p = params.items[i].string;
  89                 if (!strcmp(p, "changes")) {
  90                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  91                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
  92                 } else if (!strcmp(p, "lines")) {
  93                         DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
  94                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
  95                 } else if (!strcmp(p, "files")) {
  96                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  97                         DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
  98                 } else if (!strcmp(p, "noncumulative")) {
  99                         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
 100                 } else if (!strcmp(p, "cumulative")) {
 101                         DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
 102                 } else if (isdigit(*p)) {
 103                         char *end;
 104                         int permille = strtoul(p, &end, 10) * 10;
 105                         if (*end == '.' && isdigit(*++end)) {
 106                                 /* only use first digit */
 107                                 permille += *end - '0';
 108                                 /* .. and ignore any further digits */
 109                                 while (isdigit(*++end))
 110                                         ; /* nothing */
 111                         }
 112                         if (!*end)
 113                                 options->dirstat_permille = permille;
 114                         else {
 115                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 116                                             p);
 117                                 ret++;
 118                         }
 119                 } else {
 120                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 121                         ret++;
 122                 }
 123
 124         }
 125         string_list_clear(&params, 0);
 126         free(params_copy);
 127         return ret;
 128 }
 129
 130 static int parse_submodule_params(struct diff_options *options, const char *value)
 131 {
 132         if (!strcmp(value, "log"))
 133                 DIFF_OPT_SET(options, SUBMODULE_LOG);
 134         else if (!strcmp(value, "short"))
 135                 DIFF_OPT_CLR(options, SUBMODULE_LOG);
 136         else
 137                 return -1;
 138         return 0;
 139 }
 140
 141 static int git_config_rename(const char *var, const char *value)
 142 {
 143         if (!value)
 144                 return DIFF_DETECT_RENAME;
 145         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 146                 return  DIFF_DETECT_COPY;
 147         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 148 }
 149
 150 long parse_algorithm_value(const char *value)
 151 {
 152         if (!value)
 153                 return -1;
 154         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 155                 return 0;
 156         else if (!strcasecmp(value, "minimal"))
 157                 return XDF_NEED_MINIMAL;
 158         else if (!strcasecmp(value, "patience"))
 159                 return XDF_PATIENCE_DIFF;
 160         else if (!strcasecmp(value, "histogram"))
 161                 return XDF_HISTOGRAM_DIFF;
 162         return -1;
 163 }
 164
 165 /*
 166  * These are to give UI layer defaults.
 167  * The core-level commands such as git-diff-files should
 168  * never be affected by the setting of diff.renames
 169  * the user happens to have in the configuration file.
 170  */
 171 int git_diff_ui_config(const char *var, const char *value, void *cb)
 172 {
 173         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 174                 diff_use_color_default = git_config_colorbool(var, value);
 175                 return 0;
 176         }
 177         if (!strcmp(var, "diff.context")) {
 178                 diff_context_default = git_config_int(var, value);
 179                 if (diff_context_default < 0)
 180                         return -1;
 181                 return 0;
 182         }
 183         if (!strcmp(var, "diff.renames")) {
 184                 diff_detect_rename_default = git_config_rename(var, value);
 185                 return 0;
 186         }
 187         if (!strcmp(var, "diff.autorefreshindex")) {
 188                 diff_auto_refresh_index = git_config_bool(var, value);
 189                 return 0;
 190         }
 191         if (!strcmp(var, "diff.mnemonicprefix")) {
 192                 diff_mnemonic_prefix = git_config_bool(var, value);
 193                 return 0;
 194         }
 195         if (!strcmp(var, "diff.noprefix")) {
 196                 diff_no_prefix = git_config_bool(var, value);
 197                 return 0;
 198         }
 199         if (!strcmp(var, "diff.statgraphwidth")) {
 200                 diff_stat_graph_width = git_config_int(var, value);
 201                 return 0;
 202         }
 203         if (!strcmp(var, "diff.external"))
 204                 return git_config_string(&external_diff_cmd_cfg, var, value);
 205         if (!strcmp(var, "diff.wordregex"))
 206                 return git_config_string(&diff_word_regex_cfg, var, value);
 207         if (!strcmp(var, "diff.orderfile"))
 208                 return git_config_pathname(&diff_order_file_cfg, var, value);
 209
 210         if (!strcmp(var, "diff.ignoresubmodules"))
 211                 handle_ignore_submodules_arg(&default_diff_options, value);
 212
 213         if (!strcmp(var, "diff.submodule")) {
 214                 if (parse_submodule_params(&default_diff_options, value))
 215                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 216                                 value);
 217                 return 0;
 218         }
 219
 220         if (!strcmp(var, "diff.algorithm")) {
 221                 diff_algorithm = parse_algorithm_value(value);
 222                 if (diff_algorithm < 0)
 223                         return -1;
 224                 return 0;
 225         }
 226
 227         if (git_color_config(var, value, cb) < 0)
 228                 return -1;
 229
 230         return git_diff_basic_config(var, value, cb);
 231 }
 232
 233 int git_diff_basic_config(const char *var, const char *value, void *cb)
 234 {
 235         const char *name;
 236
 237         if (!strcmp(var, "diff.renamelimit")) {
 238                 diff_rename_limit_default = git_config_int(var, value);
 239                 return 0;
 240         }
 241
 242         if (userdiff_config(var, value) < 0)
 243                 return -1;
 244
 245         if (skip_prefix(var, "diff.color.", &name) ||
 246             skip_prefix(var, "color.diff.", &name)) {
 247                 int slot = parse_diff_color_slot(name);
 248                 if (slot < 0)
 249                         return 0;
 250                 if (!value)
 251                         return config_error_nonbool(var);
 252                 return color_parse(value, diff_colors[slot]);
 253         }
 254
 255         /* like GNU diff's --suppress-blank-empty option  */
 256         if (!strcmp(var, "diff.suppressblankempty") ||
 257                         /* for backwards compatibility */
 258                         !strcmp(var, "diff.suppress-blank-empty")) {
 259                 diff_suppress_blank_empty = git_config_bool(var, value);
 260                 return 0;
 261         }
 262
 263         if (!strcmp(var, "diff.dirstat")) {
 264                 struct strbuf errmsg = STRBUF_INIT;
 265                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 266                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 267                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 268                                 errmsg.buf);
 269                 strbuf_release(&errmsg);
 270                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 271                 return 0;
 272         }
 273
 274         if (starts_with(var, "submodule."))
 275                 return parse_submodule_config_option(var, value);
 276
 277         return git_default_config(var, value, cb);
 278 }
 279
 280 static char *quote_two(const char *one, const char *two)
 281 {
 282         int need_one = quote_c_style(one, NULL, NULL, 1);
 283         int need_two = quote_c_style(two, NULL, NULL, 1);
 284         struct strbuf res = STRBUF_INIT;
 285
 286         if (need_one + need_two) {
 287                 strbuf_addch(&res, '"');
 288                 quote_c_style(one, &res, NULL, 1);
 289                 quote_c_style(two, &res, NULL, 1);
 290                 strbuf_addch(&res, '"');
 291         } else {
 292                 strbuf_addstr(&res, one);
 293                 strbuf_addstr(&res, two);
 294         }
 295         return strbuf_detach(&res, NULL);
 296 }
 297
 298 static const char *external_diff(void)
 299 {
 300         static const char *external_diff_cmd = NULL;
 301         static int done_preparing = 0;
 302
 303         if (done_preparing)
 304                 return external_diff_cmd;
 305         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 306         if (!external_diff_cmd)
 307                 external_diff_cmd = external_diff_cmd_cfg;
 308         done_preparing = 1;
 309         return external_diff_cmd;
 310 }
 311
 312 static struct diff_tempfile {
 313         const char *name; /* filename external diff should read from */
 314         char hex[41];
 315         char mode[10];
 316         char tmp_path[PATH_MAX];
 317 } diff_temp[2];
 318
 319 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 320
 321 struct emit_callback {
 322         int color_diff;
 323         unsigned ws_rule;
 324         int blank_at_eof_in_preimage;
 325         int blank_at_eof_in_postimage;
 326         int lno_in_preimage;
 327         int lno_in_postimage;
 328         sane_truncate_fn truncate;
 329         const char **label_path;
 330         struct diff_words_data *diff_words;
 331         struct diff_options *opt;
 332         int *found_changesp;
 333         struct strbuf *header;
 334 };
 335
 336 static int count_lines(const char *data, int size)
 337 {
 338         int count, ch, completely_empty = 1, nl_just_seen = 0;
 339         count = 0;
 340         while (0 < size--) {
 341                 ch = *data++;
 342                 if (ch == '\n') {
 343                         count++;
 344                         nl_just_seen = 1;
 345                         completely_empty = 0;
 346                 }
 347                 else {
 348                         nl_just_seen = 0;
 349                         completely_empty = 0;
 350                 }
 351         }
 352         if (completely_empty)
 353                 return 0;
 354         if (!nl_just_seen)
 355                 count++; /* no trailing newline */
 356         return count;
 357 }
 358
 359 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 360 {
 361         if (!DIFF_FILE_VALID(one)) {
 362                 mf->ptr = (char *)""; /* does not matter */
 363                 mf->size = 0;
 364                 return 0;
 365         }
 366         else if (diff_populate_filespec(one, 0))
 367                 return -1;
 368
 369         mf->ptr = one->data;
 370         mf->size = one->size;
 371         return 0;
 372 }
 373
 374 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 375 static unsigned long diff_filespec_size(struct diff_filespec *one)
 376 {
 377         if (!DIFF_FILE_VALID(one))
 378                 return 0;
 379         diff_populate_filespec(one, CHECK_SIZE_ONLY);
 380         return one->size;
 381 }
 382
 383 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 384 {
 385         char *ptr = mf->ptr;
 386         long size = mf->size;
 387         int cnt = 0;
 388
 389         if (!size)
 390                 return cnt;
 391         ptr += size - 1; /* pointing at the very end */
 392         if (*ptr != '\n')
 393                 ; /* incomplete line */
 394         else
 395                 ptr--; /* skip the last LF */
 396         while (mf->ptr < ptr) {
 397                 char *prev_eol;
 398                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 399                         if (*prev_eol == '\n')
 400                                 break;
 401                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 402                         break;
 403                 cnt++;
 404                 ptr = prev_eol - 1;
 405         }
 406         return cnt;
 407 }
 408
 409 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 410                                struct emit_callback *ecbdata)
 411 {
 412         int l1, l2, at;
 413         unsigned ws_rule = ecbdata->ws_rule;
 414         l1 = count_trailing_blank(mf1, ws_rule);
 415         l2 = count_trailing_blank(mf2, ws_rule);
 416         if (l2 <= l1) {
 417                 ecbdata->blank_at_eof_in_preimage = 0;
 418                 ecbdata->blank_at_eof_in_postimage = 0;
 419                 return;
 420         }
 421         at = count_lines(mf1->ptr, mf1->size);
 422         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 423
 424         at = count_lines(mf2->ptr, mf2->size);
 425         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 426 }
 427
 428 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 429                         int first, const char *line, int len)
 430 {
 431         int has_trailing_newline, has_trailing_carriage_return;
 432         int nofirst;
 433         FILE *file = o->file;
 434
 435         fputs(diff_line_prefix(o), file);
 436
 437         if (len == 0) {
 438                 has_trailing_newline = (first == '\n');
 439                 has_trailing_carriage_return = (!has_trailing_newline &&
 440                                                 (first == '\r'));
 441                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 442         } else {
 443                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 444                 if (has_trailing_newline)
 445                         len--;
 446                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 447                 if (has_trailing_carriage_return)
 448                         len--;
 449                 nofirst = 0;
 450         }
 451
 452         if (len || !nofirst) {
 453                 fputs(set, file);
 454                 if (!nofirst)
 455                         fputc(first, file);
 456                 fwrite(line, len, 1, file);
 457                 fputs(reset, file);
 458         }
 459         if (has_trailing_carriage_return)
 460                 fputc('\r', file);
 461         if (has_trailing_newline)
 462                 fputc('\n', file);
 463 }
 464
 465 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 466                       const char *line, int len)
 467 {
 468         emit_line_0(o, set, reset, line[0], line+1, len-1);
 469 }
 470
 471 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
 472 {
 473         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
 474               ecbdata->blank_at_eof_in_preimage &&
 475               ecbdata->blank_at_eof_in_postimage &&
 476               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
 477               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
 478                 return 0;
 479         return ws_blank_line(line, len, ecbdata->ws_rule);
 480 }
 481
 482 static void emit_line_checked(const char *reset,
 483                               struct emit_callback *ecbdata,
 484                               const char *line, int len,
 485                               enum color_diff color,
 486                               unsigned ws_error_highlight,
 487                               char sign)
 488 {
 489         const char *set = diff_get_color(ecbdata->color_diff, color);
 490         const char *ws = NULL;
 491
 492         if (ecbdata->opt->ws_error_highlight & ws_error_highlight) {
 493                 ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 494                 if (!*ws)
 495                         ws = NULL;
 496         }
 497
 498         if (!ws)
 499                 emit_line_0(ecbdata->opt, set, reset, sign, line, len);
 500         else if (sign == '+' && new_blank_line_at_eof(ecbdata, line, len))
 501                 /* Blank line at EOF - paint '+' as well */
 502                 emit_line_0(ecbdata->opt, ws, reset, sign, line, len);
 503         else {
 504                 /* Emit just the prefix, then the rest. */
 505                 emit_line_0(ecbdata->opt, set, reset, sign, "", 0);
 506                 ws_check_emit(line, len, ecbdata->ws_rule,
 507                               ecbdata->opt->file, set, reset, ws);
 508         }
 509 }
 510
 511 static void emit_add_line(const char *reset,
 512                           struct emit_callback *ecbdata,
 513                           const char *line, int len)
 514 {
 515         emit_line_checked(reset, ecbdata, line, len,
 516                           DIFF_FILE_NEW, WSEH_NEW, '+');
 517 }
 518
 519 static void emit_del_line(const char *reset,
 520                           struct emit_callback *ecbdata,
 521                           const char *line, int len)
 522 {
 523         emit_line_checked(reset, ecbdata, line, len,
 524                           DIFF_FILE_OLD, WSEH_OLD, '-');
 525 }
 526
 527 static void emit_context_line(const char *reset,
 528                               struct emit_callback *ecbdata,
 529                               const char *line, int len)
 530 {
 531         emit_line_checked(reset, ecbdata, line, len,
 532                           DIFF_CONTEXT, WSEH_CONTEXT, ' ');
 533 }
 534
 535 static void emit_hunk_header(struct emit_callback *ecbdata,
 536                              const char *line, int len)
 537 {
 538         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
 539         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
 540         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
 541         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 542         static const char atat[2] = { '@', '@' };
 543         const char *cp, *ep;
 544         struct strbuf msgbuf = STRBUF_INIT;
 545         int org_len = len;
 546         int i = 1;
 547
 548         /*
 549          * As a hunk header must begin with "@@ -<old>, +<new> @@",
 550          * it always is at least 10 bytes long.
 551          */
 552         if (len < 10 ||
 553             memcmp(line, atat, 2) ||
 554             !(ep = memmem(line + 2, len - 2, atat, 2))) {
 555                 emit_line(ecbdata->opt, context, reset, line, len);
 556                 return;
 557         }
 558         ep += 2; /* skip over @@ */
 559
 560         /* The hunk header in fraginfo color */
 561         strbuf_addstr(&msgbuf, frag);
 562         strbuf_add(&msgbuf, line, ep - line);
 563         strbuf_addstr(&msgbuf, reset);
 564
 565         /*
 566          * trailing "\r\n"
 567          */
 568         for ( ; i < 3; i++)
 569                 if (line[len - i] == '\r' || line[len - i] == '\n')
 570                         len--;
 571
 572         /* blank before the func header */
 573         for (cp = ep; ep - line < len; ep++)
 574                 if (*ep != ' ' && *ep != '\t')
 575                         break;
 576         if (ep != cp) {
 577                 strbuf_addstr(&msgbuf, context);
 578                 strbuf_add(&msgbuf, cp, ep - cp);
 579                 strbuf_addstr(&msgbuf, reset);
 580         }
 581
 582         if (ep < line + len) {
 583                 strbuf_addstr(&msgbuf, func);
 584                 strbuf_add(&msgbuf, ep, line + len - ep);
 585                 strbuf_addstr(&msgbuf, reset);
 586         }
 587
 588         strbuf_add(&msgbuf, line + len, org_len - len);
 589         emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
 590         strbuf_release(&msgbuf);
 591 }
 592
 593 static struct diff_tempfile *claim_diff_tempfile(void) {
 594         int i;
 595         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
 596                 if (!diff_temp[i].name)
 597                         return diff_temp + i;
 598         die("BUG: diff is failing to clean up its tempfiles");
 599 }
 600
 601 static int remove_tempfile_installed;
 602
 603 static void remove_tempfile(void)
 604 {
 605         int i;
 606         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
 607                 if (diff_temp[i].name == diff_temp[i].tmp_path)
 608                         unlink_or_warn(diff_temp[i].name);
 609                 diff_temp[i].name = NULL;
 610         }
 611 }
 612
 613 static void remove_tempfile_on_signal(int signo)
 614 {
 615         remove_tempfile();
 616         sigchain_pop(signo);
 617         raise(signo);
 618 }
 619
 620 static void print_line_count(FILE *file, int count)
 621 {
 622         switch (count) {
 623         case 0:
 624                 fprintf(file, "0,0");
 625                 break;
 626         case 1:
 627                 fprintf(file, "1");
 628                 break;
 629         default:
 630                 fprintf(file, "1,%d", count);
 631                 break;
 632         }
 633 }
 634
 635 static void emit_rewrite_lines(struct emit_callback *ecb,
 636                                int prefix, const char *data, int size)
 637 {
 638         const char *endp = NULL;
 639         static const char *nneof = " No newline at end of file\n";
 640         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
 641
 642         while (0 < size) {
 643                 int len;
 644
 645                 endp = memchr(data, '\n', size);
 646                 len = endp ? (endp - data + 1) : size;
 647                 if (prefix != '+') {
 648                         ecb->lno_in_preimage++;
 649                         emit_del_line(reset, ecb, data, len);
 650                 } else {
 651                         ecb->lno_in_postimage++;
 652                         emit_add_line(reset, ecb, data, len);
 653                 }
 654                 size -= len;
 655                 data += len;
 656         }
 657         if (!endp) {
 658                 const char *context = diff_get_color(ecb->color_diff,
 659                                                      DIFF_CONTEXT);
 660                 putc('\n', ecb->opt->file);
 661                 emit_line_0(ecb->opt, context, reset, '\\',
 662                             nneof, strlen(nneof));
 663         }
 664 }
 665
 666 static void emit_rewrite_diff(const char *name_a,
 667                               const char *name_b,
 668                               struct diff_filespec *one,
 669                               struct diff_filespec *two,
 670                               struct userdiff_driver *textconv_one,
 671                               struct userdiff_driver *textconv_two,
 672                               struct diff_options *o)
 673 {
 674         int lc_a, lc_b;
 675         const char *name_a_tab, *name_b_tab;
 676         const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
 677         const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
 678         const char *reset = diff_get_color(o->use_color, DIFF_RESET);
 679         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 680         const char *a_prefix, *b_prefix;
 681         char *data_one, *data_two;
 682         size_t size_one, size_two;
 683         struct emit_callback ecbdata;
 684         const char *line_prefix = diff_line_prefix(o);
 685
 686         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 687                 a_prefix = o->b_prefix;
 688                 b_prefix = o->a_prefix;
 689         } else {
 690                 a_prefix = o->a_prefix;
 691                 b_prefix = o->b_prefix;
 692         }
 693
 694         name_a += (*name_a == '/');
 695         name_b += (*name_b == '/');
 696         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 697         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 698
 699         strbuf_reset(&a_name);
 700         strbuf_reset(&b_name);
 701         quote_two_c_style(&a_name, a_prefix, name_a, 0);
 702         quote_two_c_style(&b_name, b_prefix, name_b, 0);
 703
 704         size_one = fill_textconv(textconv_one, one, &data_one);
 705         size_two = fill_textconv(textconv_two, two, &data_two);
 706
 707         memset(&ecbdata, 0, sizeof(ecbdata));
 708         ecbdata.color_diff = want_color(o->use_color);
 709         ecbdata.found_changesp = &o->found_changes;
 710         ecbdata.ws_rule = whitespace_rule(name_b);
 711         ecbdata.opt = o;
 712         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
 713                 mmfile_t mf1, mf2;
 714                 mf1.ptr = (char *)data_one;
 715                 mf2.ptr = (char *)data_two;
 716                 mf1.size = size_one;
 717                 mf2.size = size_two;
 718                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
 719         }
 720         ecbdata.lno_in_preimage = 1;
 721         ecbdata.lno_in_postimage = 1;
 722
 723         lc_a = count_lines(data_one, size_one);
 724         lc_b = count_lines(data_two, size_two);
 725         fprintf(o->file,
 726                 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
 727                 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
 728                 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
 729                 line_prefix, fraginfo);
 730         if (!o->irreversible_delete)
 731                 print_line_count(o->file, lc_a);
 732         else
 733                 fprintf(o->file, "?,?");
 734         fprintf(o->file, " +");
 735         print_line_count(o->file, lc_b);
 736         fprintf(o->file, " @@%s\n", reset);
 737         if (lc_a && !o->irreversible_delete)
 738                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
 739         if (lc_b)
 740                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
 741         if (textconv_one)
 742                 free((char *)data_one);
 743         if (textconv_two)
 744                 free((char *)data_two);
 745 }
 746
 747 struct diff_words_buffer {
 748         mmfile_t text;
 749         long alloc;
 750         struct diff_words_orig {
 751                 const char *begin, *end;
 752         } *orig;
 753         int orig_nr, orig_alloc;
 754 };
 755
 756 static void diff_words_append(char *line, unsigned long len,
 757                 struct diff_words_buffer *buffer)
 758 {
 759         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
 760         line++;
 761         len--;
 762         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 763         buffer->text.size += len;
 764         buffer->text.ptr[buffer->text.size] = '\0';
 765 }
 766
 767 struct diff_words_style_elem {
 768         const char *prefix;
 769         const char *suffix;
 770         const char *color; /* NULL; filled in by the setup code if
 771                             * color is enabled */
 772 };
 773
 774 struct diff_words_style {
 775         enum diff_words_type type;
 776         struct diff_words_style_elem new, old, ctx;
 777         const char *newline;
 778 };
 779
 780 static struct diff_words_style diff_words_styles[] = {
 781         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
 782         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
 783         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
 784 };
 785
 786 struct diff_words_data {
 787         struct diff_words_buffer minus, plus;
 788         const char *current_plus;
 789         int last_minus;
 790         struct diff_options *opt;
 791         regex_t *word_regex;
 792         enum diff_words_type type;
 793         struct diff_words_style *style;
 794 };
 795
 796 static int fn_out_diff_words_write_helper(FILE *fp,
 797                                           struct diff_words_style_elem *st_el,
 798                                           const char *newline,
 799                                           size_t count, const char *buf,
 800                                           const char *line_prefix)
 801 {
 802         int print = 0;
 803
 804         while (count) {
 805                 char *p = memchr(buf, '\n', count);
 806                 if (print)
 807                         fputs(line_prefix, fp);
 808                 if (p != buf) {
 809                         if (st_el->color && fputs(st_el->color, fp) < 0)
 810                                 return -1;
 811                         if (fputs(st_el->prefix, fp) < 0 ||
 812                             fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
 813                             fputs(st_el->suffix, fp) < 0)
 814                                 return -1;
 815                         if (st_el->color && *st_el->color
 816                             && fputs(GIT_COLOR_RESET, fp) < 0)
 817                                 return -1;
 818                 }
 819                 if (!p)
 820                         return 0;
 821                 if (fputs(newline, fp) < 0)
 822                         return -1;
 823                 count -= p + 1 - buf;
 824                 buf = p + 1;
 825                 print = 1;
 826         }
 827         return 0;
 828 }
 829
 830 /*
 831  * '--color-words' algorithm can be described as:
 832  *
 833  *   1. collect a the minus/plus lines of a diff hunk, divided into
 834  *      minus-lines and plus-lines;
 835  *
 836  *   2. break both minus-lines and plus-lines into words and
 837  *      place them into two mmfile_t with one word for each line;
 838  *
 839  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
 840  *
 841  * And for the common parts of the both file, we output the plus side text.
 842  * diff_words->current_plus is used to trace the current position of the plus file
 843  * which printed. diff_words->last_minus is used to trace the last minus word
 844  * printed.
 845  *
 846  * For '--graph' to work with '--color-words', we need to output the graph prefix
 847  * on each line of color words output. Generally, there are two conditions on
 848  * which we should output the prefix.
 849  *
 850  *   1. diff_words->last_minus == 0 &&
 851  *      diff_words->current_plus == diff_words->plus.text.ptr
 852  *
 853  *      that is: the plus text must start as a new line, and if there is no minus
 854  *      word printed, a graph prefix must be printed.
 855  *
 856  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
 857  *      *(diff_words->current_plus - 1) == '\n'
 858  *
 859  *      that is: a graph prefix must be printed following a '\n'
 860  */
 861 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
 862 {
 863         if ((diff_words->last_minus == 0 &&
 864                 diff_words->current_plus == diff_words->plus.text.ptr) ||
 865                 (diff_words->current_plus > diff_words->plus.text.ptr &&
 866                 *(diff_words->current_plus - 1) == '\n')) {
 867                 return 1;
 868         } else {
 869                 return 0;
 870         }
 871 }
 872
 873 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 874 {
 875         struct diff_words_data *diff_words = priv;
 876         struct diff_words_style *style = diff_words->style;
 877         int minus_first, minus_len, plus_first, plus_len;
 878         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 879         struct diff_options *opt = diff_words->opt;
 880         const char *line_prefix;
 881
 882         if (line[0] != '@' || parse_hunk_header(line, len,
 883                         &minus_first, &minus_len, &plus_first, &plus_len))
 884                 return;
 885
 886         assert(opt);
 887         line_prefix = diff_line_prefix(opt);
 888
 889         /* POSIX requires that first be decremented by one if len == 0... */
 890         if (minus_len) {
 891                 minus_begin = diff_words->minus.orig[minus_first].begin;
 892                 minus_end =
 893                         diff_words->minus.orig[minus_first + minus_len - 1].end;
 894         } else
 895                 minus_begin = minus_end =
 896                         diff_words->minus.orig[minus_first].end;
 897
 898         if (plus_len) {
 899                 plus_begin = diff_words->plus.orig[plus_first].begin;
 900                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
 901         } else
 902                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
 903
 904         if (color_words_output_graph_prefix(diff_words)) {
 905                 fputs(line_prefix, diff_words->opt->file);
 906         }
 907         if (diff_words->current_plus != plus_begin) {
 908                 fn_out_diff_words_write_helper(diff_words->opt->file,
 909                                 &style->ctx, style->newline,
 910                                 plus_begin - diff_words->current_plus,
 911                                 diff_words->current_plus, line_prefix);
 912                 if (*(plus_begin - 1) == '\n')
 913                         fputs(line_prefix, diff_words->opt->file);
 914         }
 915         if (minus_begin != minus_end) {
 916                 fn_out_diff_words_write_helper(diff_words->opt->file,
 917                                 &style->old, style->newline,
 918                                 minus_end - minus_begin, minus_begin,
 919                                 line_prefix);
 920         }
 921         if (plus_begin != plus_end) {
 922                 fn_out_diff_words_write_helper(diff_words->opt->file,
 923                                 &style->new, style->newline,
 924                                 plus_end - plus_begin, plus_begin,
 925                                 line_prefix);
 926         }
 927
 928         diff_words->current_plus = plus_end;
 929         diff_words->last_minus = minus_first;
 930 }
 931
 932 /* This function starts looking at *begin, and returns 0 iff a word was found. */
 933 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
 934                 int *begin, int *end)
 935 {
 936         if (word_regex && *begin < buffer->size) {
 937                 regmatch_t match[1];
 938                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
 939                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
 940                                         '\n', match[0].rm_eo - match[0].rm_so);
 941                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
 942                         *begin += match[0].rm_so;
 943                         return *begin >= *end;
 944                 }
 945                 return -1;
 946         }
 947
 948         /* find the next word */
 949         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
 950                 (*begin)++;
 951         if (*begin >= buffer->size)
 952                 return -1;
 953
 954         /* find the end of the word */
 955         *end = *begin + 1;
 956         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
 957                 (*end)++;
 958
 959         return 0;
 960 }
 961
 962 /*
 963  * This function splits the words in buffer->text, stores the list with
 964  * newline separator into out, and saves the offsets of the original words
 965  * in buffer->orig.
 966  */
 967 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
 968                 regex_t *word_regex)
 969 {
 970         int i, j;
 971         long alloc = 0;
 972
 973         out->size = 0;
 974         out->ptr = NULL;
 975
 976         /* fake an empty "0th" word */
 977         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
 978         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
 979         buffer->orig_nr = 1;
 980
 981         for (i = 0; i < buffer->text.size; i++) {
 982                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
 983                         return;
 984
 985                 /* store original boundaries */
 986                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
 987                                 buffer->orig_alloc);
 988                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
 989                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
 990                 buffer->orig_nr++;
 991
 992                 /* store one word */
 993                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
 994                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
 995                 out->ptr[out->size + j - i] = '\n';
 996                 out->size += j - i + 1;
 997
 998                 i = j - 1;
 999         }
1000 }
1001
1002 /* this executes the word diff on the accumulated buffers */
1003 static void diff_words_show(struct diff_words_data *diff_words)
1004 {
1005         xpparam_t xpp;
1006         xdemitconf_t xecfg;
1007         mmfile_t minus, plus;
1008         struct diff_words_style *style = diff_words->style;
1009
1010         struct diff_options *opt = diff_words->opt;
1011         const char *line_prefix;
1012
1013         assert(opt);
1014         line_prefix = diff_line_prefix(opt);
1015
1016         /* special case: only removal */
1017         if (!diff_words->plus.text.size) {
1018                 fputs(line_prefix, diff_words->opt->file);
1019                 fn_out_diff_words_write_helper(diff_words->opt->file,
1020                         &style->old, style->newline,
1021                         diff_words->minus.text.size,
1022                         diff_words->minus.text.ptr, line_prefix);
1023                 diff_words->minus.text.size = 0;
1024                 return;
1025         }
1026
1027         diff_words->current_plus = diff_words->plus.text.ptr;
1028         diff_words->last_minus = 0;
1029
1030         memset(&xpp, 0, sizeof(xpp));
1031         memset(&xecfg, 0, sizeof(xecfg));
1032         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1033         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1034         xpp.flags = 0;
1035         /* as only the hunk header will be parsed, we need a 0-context */
1036         xecfg.ctxlen = 0;
1037         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1038                       &xpp, &xecfg);
1039         free(minus.ptr);
1040         free(plus.ptr);
1041         if (diff_words->current_plus != diff_words->plus.text.ptr +
1042                         diff_words->plus.text.size) {
1043                 if (color_words_output_graph_prefix(diff_words))
1044                         fputs(line_prefix, diff_words->opt->file);
1045                 fn_out_diff_words_write_helper(diff_words->opt->file,
1046                         &style->ctx, style->newline,
1047                         diff_words->plus.text.ptr + diff_words->plus.text.size
1048                         - diff_words->current_plus, diff_words->current_plus,
1049                         line_prefix);
1050         }
1051         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1052 }
1053
1054 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1055 static void diff_words_flush(struct emit_callback *ecbdata)
1056 {
1057         if (ecbdata->diff_words->minus.text.size ||
1058             ecbdata->diff_words->plus.text.size)
1059                 diff_words_show(ecbdata->diff_words);
1060 }
1061
1062 static void diff_filespec_load_driver(struct diff_filespec *one)
1063 {
1064         /* Use already-loaded driver */
1065         if (one->driver)
1066                 return;
1067
1068         if (S_ISREG(one->mode))
1069                 one->driver = userdiff_find_by_path(one->path);
1070
1071         /* Fallback to default settings */
1072         if (!one->driver)
1073                 one->driver = userdiff_find_by_name("default");
1074 }
1075
1076 static const char *userdiff_word_regex(struct diff_filespec *one)
1077 {
1078         diff_filespec_load_driver(one);
1079         return one->driver->word_regex;
1080 }
1081
1082 static void init_diff_words_data(struct emit_callback *ecbdata,
1083                                  struct diff_options *orig_opts,
1084                                  struct diff_filespec *one,
1085                                  struct diff_filespec *two)
1086 {
1087         int i;
1088         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1089         memcpy(o, orig_opts, sizeof(struct diff_options));
1090
1091         ecbdata->diff_words =
1092                 xcalloc(1, sizeof(struct diff_words_data));
1093         ecbdata->diff_words->type = o->word_diff;
1094         ecbdata->diff_words->opt = o;
1095         if (!o->word_regex)
1096                 o->word_regex = userdiff_word_regex(one);
1097         if (!o->word_regex)
1098                 o->word_regex = userdiff_word_regex(two);
1099         if (!o->word_regex)
1100                 o->word_regex = diff_word_regex_cfg;
1101         if (o->word_regex) {
1102                 ecbdata->diff_words->word_regex = (regex_t *)
1103                         xmalloc(sizeof(regex_t));
1104                 if (regcomp(ecbdata->diff_words->word_regex,
1105                             o->word_regex,
1106                             REG_EXTENDED | REG_NEWLINE))
1107                         die ("Invalid regular expression: %s",
1108                              o->word_regex);
1109         }
1110         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1111                 if (o->word_diff == diff_words_styles[i].type) {
1112                         ecbdata->diff_words->style =
1113                                 &diff_words_styles[i];
1114                         break;
1115                 }
1116         }
1117         if (want_color(o->use_color)) {
1118                 struct diff_words_style *st = ecbdata->diff_words->style;
1119                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1120                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1121                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
1122         }
1123 }
1124
1125 static void free_diff_words_data(struct emit_callback *ecbdata)
1126 {
1127         if (ecbdata->diff_words) {
1128                 diff_words_flush(ecbdata);
1129                 free (ecbdata->diff_words->opt);
1130                 free (ecbdata->diff_words->minus.text.ptr);
1131                 free (ecbdata->diff_words->minus.orig);
1132                 free (ecbdata->diff_words->plus.text.ptr);
1133                 free (ecbdata->diff_words->plus.orig);
1134                 if (ecbdata->diff_words->word_regex) {
1135                         regfree(ecbdata->diff_words->word_regex);
1136                         free(ecbdata->diff_words->word_regex);
1137                 }
1138                 free(ecbdata->diff_words);
1139                 ecbdata->diff_words = NULL;
1140         }
1141 }
1142
1143 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1144 {
1145         if (want_color(diff_use_color))
1146                 return diff_colors[ix];
1147         return "";
1148 }
1149
1150 const char *diff_line_prefix(struct diff_options *opt)
1151 {
1152         struct strbuf *msgbuf;
1153         if (!opt->output_prefix)
1154                 return "";
1155
1156         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1157         return msgbuf->buf;
1158 }
1159
1160 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1161 {
1162         const char *cp;
1163         unsigned long allot;
1164         size_t l = len;
1165
1166         if (ecb->truncate)
1167                 return ecb->truncate(line, len);
1168         cp = line;
1169         allot = l;
1170         while (0 < l) {
1171                 (void) utf8_width(&cp, &l);
1172                 if (!cp)
1173                         break; /* truncated in the middle? */
1174         }
1175         return allot - l;
1176 }
1177
1178 static void find_lno(const char *line, struct emit_callback *ecbdata)
1179 {
1180         const char *p;
1181         ecbdata->lno_in_preimage = 0;
1182         ecbdata->lno_in_postimage = 0;
1183         p = strchr(line, '-');
1184         if (!p)
1185                 return; /* cannot happen */
1186         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1187         p = strchr(p, '+');
1188         if (!p)
1189                 return; /* cannot happen */
1190         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1191 }
1192
1193 static void fn_out_consume(void *priv, char *line, unsigned long len)
1194 {
1195         struct emit_callback *ecbdata = priv;
1196         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1197         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1198         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1199         struct diff_options *o = ecbdata->opt;
1200         const char *line_prefix = diff_line_prefix(o);
1201
1202         if (ecbdata->header) {
1203                 fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
1204                 strbuf_reset(ecbdata->header);
1205                 ecbdata->header = NULL;
1206         }
1207         *(ecbdata->found_changesp) = 1;
1208
1209         if (ecbdata->label_path[0]) {
1210                 const char *name_a_tab, *name_b_tab;
1211
1212                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1213                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1214
1215                 fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1216                         line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1217                 fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1218                         line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1219                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1220         }
1221
1222         if (diff_suppress_blank_empty
1223             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1224                 line[0] = '\n';
1225                 len = 1;
1226         }
1227
1228         if (line[0] == '@') {
1229                 if (ecbdata->diff_words)
1230                         diff_words_flush(ecbdata);
1231                 len = sane_truncate_line(ecbdata, line, len);
1232                 find_lno(line, ecbdata);
1233                 emit_hunk_header(ecbdata, line, len);
1234                 if (line[len-1] != '\n')
1235                         putc('\n', ecbdata->opt->file);
1236                 return;
1237         }
1238
1239         if (len < 1) {
1240                 emit_line(ecbdata->opt, reset, reset, line, len);
1241                 if (ecbdata->diff_words
1242                     && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1243                         fputs("~\n", ecbdata->opt->file);
1244                 return;
1245         }
1246
1247         if (ecbdata->diff_words) {
1248                 if (line[0] == '-') {
1249                         diff_words_append(line, len,
1250                                           &ecbdata->diff_words->minus);
1251                         return;
1252                 } else if (line[0] == '+') {
1253                         diff_words_append(line, len,
1254                                           &ecbdata->diff_words->plus);
1255                         return;
1256                 } else if (starts_with(line, "\\ ")) {
1257                         /*
1258                          * Eat the "no newline at eof" marker as if we
1259                          * saw a "+" or "-" line with nothing on it,
1260                          * and return without diff_words_flush() to
1261                          * defer processing. If this is the end of
1262                          * preimage, more "+" lines may come after it.
1263                          */
1264                         return;
1265                 }
1266                 diff_words_flush(ecbdata);
1267                 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1268                         emit_line(ecbdata->opt, context, reset, line, len);
1269                         fputs("~\n", ecbdata->opt->file);
1270                 } else {
1271                         /*
1272                          * Skip the prefix character, if any.  With
1273                          * diff_suppress_blank_empty, there may be
1274                          * none.
1275                          */
1276                         if (line[0] != '\n') {
1277                               line++;
1278                               len--;
1279                         }
1280                         emit_line(ecbdata->opt, context, reset, line, len);
1281                 }
1282                 return;
1283         }
1284
1285         switch (line[0]) {
1286         case '+':
1287                 ecbdata->lno_in_postimage++;
1288                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1289                 break;
1290         case '-':
1291                 ecbdata->lno_in_preimage++;
1292                 emit_del_line(reset, ecbdata, line + 1, len - 1);
1293                 break;
1294         case ' ':
1295                 ecbdata->lno_in_postimage++;
1296                 ecbdata->lno_in_preimage++;
1297                 emit_context_line(reset, ecbdata, line + 1, len - 1);
1298                 break;
1299         default:
1300                 /* incomplete line at the end */
1301                 ecbdata->lno_in_preimage++;
1302                 emit_line(ecbdata->opt,
1303                           diff_get_color(ecbdata->color_diff, DIFF_CONTEXT),
1304                           reset, line, len);
1305                 break;
1306         }
1307 }
1308
1309 static char *pprint_rename(const char *a, const char *b)
1310 {
1311         const char *old = a;
1312         const char *new = b;
1313         struct strbuf name = STRBUF_INIT;
1314         int pfx_length, sfx_length;
1315         int pfx_adjust_for_slash;
1316         int len_a = strlen(a);
1317         int len_b = strlen(b);
1318         int a_midlen, b_midlen;
1319         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1320         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1321
1322         if (qlen_a || qlen_b) {
1323                 quote_c_style(a, &name, NULL, 0);
1324                 strbuf_addstr(&name, " => ");
1325                 quote_c_style(b, &name, NULL, 0);
1326                 return strbuf_detach(&name, NULL);
1327         }
1328
1329         /* Find common prefix */
1330         pfx_length = 0;
1331         while (*old && *new && *old == *new) {
1332                 if (*old == '/')
1333                         pfx_length = old - a + 1;
1334                 old++;
1335                 new++;
1336         }
1337
1338         /* Find common suffix */
1339         old = a + len_a;
1340         new = b + len_b;
1341         sfx_length = 0;
1342         /*
1343          * If there is a common prefix, it must end in a slash.  In
1344          * that case we let this loop run 1 into the prefix to see the
1345          * same slash.
1346          *
1347          * If there is no common prefix, we cannot do this as it would
1348          * underrun the input strings.
1349          */
1350         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1351         while (a + pfx_length - pfx_adjust_for_slash <= old &&
1352                b + pfx_length - pfx_adjust_for_slash <= new &&
1353                *old == *new) {
1354                 if (*old == '/')
1355                         sfx_length = len_a - (old - a);
1356                 old--;
1357                 new--;
1358         }
1359
1360         /*
1361          * pfx{mid-a => mid-b}sfx
1362          * {pfx-a => pfx-b}sfx
1363          * pfx{sfx-a => sfx-b}
1364          * name-a => name-b
1365          */
1366         a_midlen = len_a - pfx_length - sfx_length;
1367         b_midlen = len_b - pfx_length - sfx_length;
1368         if (a_midlen < 0)
1369                 a_midlen = 0;
1370         if (b_midlen < 0)
1371                 b_midlen = 0;
1372
1373         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1374         if (pfx_length + sfx_length) {
1375                 strbuf_add(&name, a, pfx_length);
1376                 strbuf_addch(&name, '{');
1377         }
1378         strbuf_add(&name, a + pfx_length, a_midlen);
1379         strbuf_addstr(&name, " => ");
1380         strbuf_add(&name, b + pfx_length, b_midlen);
1381         if (pfx_length + sfx_length) {
1382                 strbuf_addch(&name, '}');
1383                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1384         }
1385         return strbuf_detach(&name, NULL);
1386 }
1387
1388 struct diffstat_t {
1389         int nr;
1390         int alloc;
1391         struct diffstat_file {
1392                 char *from_name;
1393                 char *name;
1394                 char *print_name;
1395                 unsigned is_unmerged:1;
1396                 unsigned is_binary:1;
1397                 unsigned is_renamed:1;
1398                 unsigned is_interesting:1;
1399                 uintmax_t added, deleted;
1400         } **files;
1401 };
1402
1403 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1404                                           const char *name_a,
1405                                           const char *name_b)
1406 {
1407         struct diffstat_file *x;
1408         x = xcalloc(1, sizeof(*x));
1409         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
1410         diffstat->files[diffstat->nr++] = x;
1411         if (name_b) {
1412                 x->from_name = xstrdup(name_a);
1413                 x->name = xstrdup(name_b);
1414                 x->is_renamed = 1;
1415         }
1416         else {
1417                 x->from_name = NULL;
1418                 x->name = xstrdup(name_a);
1419         }
1420         return x;
1421 }
1422
1423 static void diffstat_consume(void *priv, char *line, unsigned long len)
1424 {
1425         struct diffstat_t *diffstat = priv;
1426         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1427
1428         if (line[0] == '+')
1429                 x->added++;
1430         else if (line[0] == '-')
1431                 x->deleted++;
1432 }
1433
1434 const char mime_boundary_leader[] = "------------";
1435
1436 static int scale_linear(int it, int width, int max_change)
1437 {
1438         if (!it)
1439                 return 0;
1440         /*
1441          * make sure that at least one '-' or '+' is printed if
1442          * there is any change to this path. The easiest way is to
1443          * scale linearly as if the alloted width is one column shorter
1444          * than it is, and then add 1 to the result.
1445          */
1446         return 1 + (it * (width - 1) / max_change);
1447 }
1448
1449 static void show_name(FILE *file,
1450                       const char *prefix, const char *name, int len)
1451 {
1452         fprintf(file, " %s%-*s |", prefix, len, name);
1453 }
1454
1455 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1456 {
1457         if (cnt <= 0)
1458                 return;
1459         fprintf(file, "%s", set);
1460         while (cnt--)
1461                 putc(ch, file);
1462         fprintf(file, "%s", reset);
1463 }
1464
1465 static void fill_print_name(struct diffstat_file *file)
1466 {
1467         char *pname;
1468
1469         if (file->print_name)
1470                 return;
1471
1472         if (!file->is_renamed) {
1473                 struct strbuf buf = STRBUF_INIT;
1474                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1475                         pname = strbuf_detach(&buf, NULL);
1476                 } else {
1477                         pname = file->name;
1478                         strbuf_release(&buf);
1479                 }
1480         } else {
1481                 pname = pprint_rename(file->from_name, file->name);
1482         }
1483         file->print_name = pname;
1484 }
1485
1486 int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1487 {
1488         struct strbuf sb = STRBUF_INIT;
1489         int ret;
1490
1491         if (!files) {
1492                 assert(insertions == 0 && deletions == 0);
1493                 return fprintf(fp, "%s\n", " 0 files changed");
1494         }
1495
1496         strbuf_addf(&sb,
1497                     (files == 1) ? " %d file changed" : " %d files changed",
1498                     files);
1499
1500         /*
1501          * For binary diff, the caller may want to print "x files
1502          * changed" with insertions == 0 && deletions == 0.
1503          *
1504          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1505          * is probably less confusing (i.e skip over "2 files changed
1506          * but nothing about added/removed lines? Is this a bug in Git?").
1507          */
1508         if (insertions || deletions == 0) {
1509                 strbuf_addf(&sb,
1510                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1511                             insertions);
1512         }
1513
1514         if (deletions || insertions == 0) {
1515                 strbuf_addf(&sb,
1516                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1517                             deletions);
1518         }
1519         strbuf_addch(&sb, '\n');
1520         ret = fputs(sb.buf, fp);
1521         strbuf_release(&sb);
1522         return ret;
1523 }
1524
1525 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1526 {
1527         int i, len, add, del, adds = 0, dels = 0;
1528         uintmax_t max_change = 0, max_len = 0;
1529         int total_files = data->nr, count;
1530         int width, name_width, graph_width, number_width = 0, bin_width = 0;
1531         const char *reset, *add_c, *del_c;
1532         const char *line_prefix = "";
1533         int extra_shown = 0;
1534
1535         if (data->nr == 0)
1536                 return;
1537
1538         line_prefix = diff_line_prefix(options);
1539         count = options->stat_count ? options->stat_count : data->nr;
1540
1541         reset = diff_get_color_opt(options, DIFF_RESET);
1542         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1543         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1544
1545         /*
1546          * Find the longest filename and max number of changes
1547          */
1548         for (i = 0; (i < count) && (i < data->nr); i++) {
1549                 struct diffstat_file *file = data->files[i];
1550                 uintmax_t change = file->added + file->deleted;
1551
1552                 if (!file->is_interesting && (change == 0)) {
1553                         count++; /* not shown == room for one more */
1554                         continue;
1555                 }
1556                 fill_print_name(file);
1557                 len = strlen(file->print_name);
1558                 if (max_len < len)
1559                         max_len = len;
1560
1561                 if (file->is_unmerged) {
1562                         /* "Unmerged" is 8 characters */
1563                         bin_width = bin_width < 8 ? 8 : bin_width;
1564                         continue;
1565                 }
1566                 if (file->is_binary) {
1567                         /* "Bin XXX -> YYY bytes" */
1568                         int w = 14 + decimal_width(file->added)
1569                                 + decimal_width(file->deleted);
1570                         bin_width = bin_width < w ? w : bin_width;
1571                         /* Display change counts aligned with "Bin" */
1572                         number_width = 3;
1573                         continue;
1574                 }
1575
1576                 if (max_change < change)
1577                         max_change = change;
1578         }
1579         count = i; /* where we can stop scanning in data->files[] */
1580
1581         /*
1582          * We have width = stat_width or term_columns() columns total.
1583          * We want a maximum of min(max_len, stat_name_width) for the name part.
1584          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1585          * We also need 1 for " " and 4 + decimal_width(max_change)
1586          * for " | NNNN " and one the empty column at the end, altogether
1587          * 6 + decimal_width(max_change).
1588          *
1589          * If there's not enough space, we will use the smaller of
1590          * stat_name_width (if set) and 5/8*width for the filename,
1591          * and the rest for constant elements + graph part, but no more
1592          * than stat_graph_width for the graph part.
1593          * (5/8 gives 50 for filename and 30 for the constant parts + graph
1594          * for the standard terminal size).
1595          *
1596          * In other words: stat_width limits the maximum width, and
1597          * stat_name_width fixes the maximum width of the filename,
1598          * and is also used to divide available columns if there
1599          * aren't enough.
1600          *
1601          * Binary files are displayed with "Bin XXX -> YYY bytes"
1602          * instead of the change count and graph. This part is treated
1603          * similarly to the graph part, except that it is not
1604          * "scaled". If total width is too small to accommodate the
1605          * guaranteed minimum width of the filename part and the
1606          * separators and this message, this message will "overflow"
1607          * making the line longer than the maximum width.
1608          */
1609
1610         if (options->stat_width == -1)
1611                 width = term_columns() - options->output_prefix_length;
1612         else
1613                 width = options->stat_width ? options->stat_width : 80;
1614         number_width = decimal_width(max_change) > number_width ?
1615                 decimal_width(max_change) : number_width;
1616
1617         if (options->stat_graph_width == -1)
1618                 options->stat_graph_width = diff_stat_graph_width;
1619
1620         /*
1621          * Guarantee 3/8*16==6 for the graph part
1622          * and 5/8*16==10 for the filename part
1623          */
1624         if (width < 16 + 6 + number_width)
1625                 width = 16 + 6 + number_width;
1626
1627         /*
1628          * First assign sizes that are wanted, ignoring available width.
1629          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1630          * starting from "XXX" should fit in graph_width.
1631          */
1632         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1633         if (options->stat_graph_width &&
1634             options->stat_graph_width < graph_width)
1635                 graph_width = options->stat_graph_width;
1636
1637         name_width = (options->stat_name_width > 0 &&
1638                       options->stat_name_width < max_len) ?
1639                 options->stat_name_width : max_len;
1640
1641         /*
1642          * Adjust adjustable widths not to exceed maximum width
1643          */
1644         if (name_width + number_width + 6 + graph_width > width) {
1645                 if (graph_width > width * 3/8 - number_width - 6) {
1646                         graph_width = width * 3/8 - number_width - 6;
1647                         if (graph_width < 6)
1648                                 graph_width = 6;
1649                 }
1650
1651                 if (options->stat_graph_width &&
1652                     graph_width > options->stat_graph_width)
1653                         graph_width = options->stat_graph_width;
1654                 if (name_width > width - number_width - 6 - graph_width)
1655                         name_width = width - number_width - 6 - graph_width;
1656                 else
1657                         graph_width = width - number_width - 6 - name_width;
1658         }
1659
1660         /*
1661          * From here name_width is the width of the name area,
1662          * and graph_width is the width of the graph area.
1663          * max_change is used to scale graph properly.
1664          */
1665         for (i = 0; i < count; i++) {
1666                 const char *prefix = "";
1667                 struct diffstat_file *file = data->files[i];
1668                 char *name = file->print_name;
1669                 uintmax_t added = file->added;
1670                 uintmax_t deleted = file->deleted;
1671                 int name_len;
1672
1673                 if (!file->is_interesting && (added + deleted == 0))
1674                         continue;
1675
1676                 /*
1677                  * "scale" the filename
1678                  */
1679                 len = name_width;
1680                 name_len = strlen(name);
1681                 if (name_width < name_len) {
1682                         char *slash;
1683                         prefix = "...";
1684                         len -= 3;
1685                         name += name_len - len;
1686                         slash = strchr(name, '/');
1687                         if (slash)
1688                                 name = slash;
1689                 }
1690
1691                 if (file->is_binary) {
1692                         fprintf(options->file, "%s", line_prefix);
1693                         show_name(options->file, prefix, name, len);
1694                         fprintf(options->file, " %*s", number_width, "Bin");
1695                         if (!added && !deleted) {
1696                                 putc('\n', options->file);
1697                                 continue;
1698                         }
1699                         fprintf(options->file, " %s%"PRIuMAX"%s",
1700                                 del_c, deleted, reset);
1701                         fprintf(options->file, " -> ");
1702                         fprintf(options->file, "%s%"PRIuMAX"%s",
1703                                 add_c, added, reset);
1704                         fprintf(options->file, " bytes");
1705                         fprintf(options->file, "\n");
1706                         continue;
1707                 }
1708                 else if (file->is_unmerged) {
1709                         fprintf(options->file, "%s", line_prefix);
1710                         show_name(options->file, prefix, name, len);
1711                         fprintf(options->file, " Unmerged\n");
1712                         continue;
1713                 }
1714
1715                 /*
1716                  * scale the add/delete
1717                  */
1718                 add = added;
1719                 del = deleted;
1720
1721                 if (graph_width <= max_change) {
1722                         int total = scale_linear(add + del, graph_width, max_change);
1723                         if (total < 2 && add && del)
1724                                 /* width >= 2 due to the sanity check */
1725                                 total = 2;
1726                         if (add < del) {
1727                                 add = scale_linear(add, graph_width, max_change);
1728                                 del = total - add;
1729                         } else {
1730                                 del = scale_linear(del, graph_width, max_change);
1731                                 add = total - del;
1732                         }
1733                 }
1734                 fprintf(options->file, "%s", line_prefix);
1735                 show_name(options->file, prefix, name, len);
1736                 fprintf(options->file, " %*"PRIuMAX"%s",
1737                         number_width, added + deleted,
1738                         added + deleted ? " " : "");
1739                 show_graph(options->file, '+', add, add_c, reset);
1740                 show_graph(options->file, '-', del, del_c, reset);
1741                 fprintf(options->file, "\n");
1742         }
1743
1744         for (i = 0; i < data->nr; i++) {
1745                 struct diffstat_file *file = data->files[i];
1746                 uintmax_t added = file->added;
1747                 uintmax_t deleted = file->deleted;
1748
1749                 if (file->is_unmerged ||
1750                     (!file->is_interesting && (added + deleted == 0))) {
1751                         total_files--;
1752                         continue;
1753                 }
1754
1755                 if (!file->is_binary) {
1756                         adds += added;
1757                         dels += deleted;
1758                 }
1759                 if (i < count)
1760                         continue;
1761                 if (!extra_shown)
1762                         fprintf(options->file, "%s ...\n", line_prefix);
1763                 extra_shown = 1;
1764         }
1765         fprintf(options->file, "%s", line_prefix);
1766         print_stat_summary(options->file, total_files, adds, dels);
1767 }
1768
1769 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1770 {
1771         int i, adds = 0, dels = 0, total_files = data->nr;
1772
1773         if (data->nr == 0)
1774                 return;
1775
1776         for (i = 0; i < data->nr; i++) {
1777                 int added = data->files[i]->added;
1778                 int deleted= data->files[i]->deleted;
1779
1780                 if (data->files[i]->is_unmerged ||
1781                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
1782                         total_files--;
1783                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
1784                         adds += added;
1785                         dels += deleted;
1786                 }
1787         }
1788         fprintf(options->file, "%s", diff_line_prefix(options));
1789         print_stat_summary(options->file, total_files, adds, dels);
1790 }
1791
1792 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1793 {
1794         int i;
1795
1796         if (data->nr == 0)
1797                 return;
1798
1799         for (i = 0; i < data->nr; i++) {
1800                 struct diffstat_file *file = data->files[i];
1801
1802                 fprintf(options->file, "%s", diff_line_prefix(options));
1803
1804                 if (file->is_binary)
1805                         fprintf(options->file, "-\t-\t");
1806                 else
1807                         fprintf(options->file,
1808                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1809                                 file->added, file->deleted);
1810                 if (options->line_termination) {
1811                         fill_print_name(file);
1812                         if (!file->is_renamed)
1813                                 write_name_quoted(file->name, options->file,
1814                                                   options->line_termination);
1815                         else {
1816                                 fputs(file->print_name, options->file);
1817                                 putc(options->line_termination, options->file);
1818                         }
1819                 } else {
1820                         if (file->is_renamed) {
1821                                 putc('\0', options->file);
1822                                 write_name_quoted(file->from_name, options->file, '\0');
1823                         }
1824                         write_name_quoted(file->name, options->file, '\0');
1825                 }
1826         }
1827 }
1828
1829 struct dirstat_file {
1830         const char *name;
1831         unsigned long changed;
1832 };
1833
1834 struct dirstat_dir {
1835         struct dirstat_file *files;
1836         int alloc, nr, permille, cumulative;
1837 };
1838
1839 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1840                 unsigned long changed, const char *base, int baselen)
1841 {
1842         unsigned long this_dir = 0;
1843         unsigned int sources = 0;
1844         const char *line_prefix = diff_line_prefix(opt);
1845
1846         while (dir->nr) {
1847                 struct dirstat_file *f = dir->files;
1848                 int namelen = strlen(f->name);
1849                 unsigned long this;
1850                 char *slash;
1851
1852                 if (namelen < baselen)
1853                         break;
1854                 if (memcmp(f->name, base, baselen))
1855                         break;
1856                 slash = strchr(f->name + baselen, '/');
1857                 if (slash) {
1858                         int newbaselen = slash + 1 - f->name;
1859                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1860                         sources++;
1861                 } else {
1862                         this = f->changed;
1863                         dir->files++;
1864                         dir->nr--;
1865                         sources += 2;
1866                 }
1867                 this_dir += this;
1868         }
1869
1870         /*
1871          * We don't report dirstat's for
1872          *  - the top level
1873          *  - or cases where everything came from a single directory
1874          *    under this directory (sources == 1).
1875          */
1876         if (baselen && sources != 1) {
1877                 if (this_dir) {
1878                         int permille = this_dir * 1000 / changed;
1879                         if (permille >= dir->permille) {
1880                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1881                                         permille / 10, permille % 10, baselen, base);
1882                                 if (!dir->cumulative)
1883                                         return 0;
1884                         }
1885                 }
1886         }
1887         return this_dir;
1888 }
1889
1890 static int dirstat_compare(const void *_a, const void *_b)
1891 {
1892         const struct dirstat_file *a = _a;
1893         const struct dirstat_file *b = _b;
1894         return strcmp(a->name, b->name);
1895 }
1896
1897 static void show_dirstat(struct diff_options *options)
1898 {
1899         int i;
1900         unsigned long changed;
1901         struct dirstat_dir dir;
1902         struct diff_queue_struct *q = &diff_queued_diff;
1903
1904         dir.files = NULL;
1905         dir.alloc = 0;
1906         dir.nr = 0;
1907         dir.permille = options->dirstat_permille;
1908         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1909
1910         changed = 0;
1911         for (i = 0; i < q->nr; i++) {
1912                 struct diff_filepair *p = q->queue[i];
1913                 const char *name;
1914                 unsigned long copied, added, damage;
1915                 int content_changed;
1916
1917                 name = p->two->path ? p->two->path : p->one->path;
1918
1919                 if (p->one->sha1_valid && p->two->sha1_valid)
1920                         content_changed = hashcmp(p->one->sha1, p->two->sha1);
1921                 else
1922                         content_changed = 1;
1923
1924                 if (!content_changed) {
1925                         /*
1926                          * The SHA1 has not changed, so pre-/post-content is
1927                          * identical. We can therefore skip looking at the
1928                          * file contents altogether.
1929                          */
1930                         damage = 0;
1931                         goto found_damage;
1932                 }
1933
1934                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1935                         /*
1936                          * In --dirstat-by-file mode, we don't really need to
1937                          * look at the actual file contents at all.
1938                          * The fact that the SHA1 changed is enough for us to
1939                          * add this file to the list of results
1940                          * (with each file contributing equal damage).
1941                          */
1942                         damage = 1;
1943                         goto found_damage;
1944                 }
1945
1946                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1947                         diff_populate_filespec(p->one, 0);
1948                         diff_populate_filespec(p->two, 0);
1949                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1950                                                &copied, &added);
1951                         diff_free_filespec_data(p->one);
1952                         diff_free_filespec_data(p->two);
1953                 } else if (DIFF_FILE_VALID(p->one)) {
1954                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
1955                         copied = added = 0;
1956                         diff_free_filespec_data(p->one);
1957                 } else if (DIFF_FILE_VALID(p->two)) {
1958                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
1959                         copied = 0;
1960                         added = p->two->size;
1961                         diff_free_filespec_data(p->two);
1962                 } else
1963                         continue;
1964
1965                 /*
1966                  * Original minus copied is the removed material,
1967                  * added is the new material.  They are both damages
1968                  * made to the preimage.
1969                  * If the resulting damage is zero, we know that
1970                  * diffcore_count_changes() considers the two entries to
1971                  * be identical, but since content_changed is true, we
1972                  * know that there must have been _some_ kind of change,
1973                  * so we force all entries to have damage > 0.
1974                  */
1975                 damage = (p->one->size - copied) + added;
1976                 if (!damage)
1977                         damage = 1;
1978
1979 found_damage:
1980                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1981                 dir.files[dir.nr].name = name;
1982                 dir.files[dir.nr].changed = damage;
1983                 changed += damage;
1984                 dir.nr++;
1985         }
1986
1987         /* This can happen even with many files, if everything was renames */
1988         if (!changed)
1989                 return;
1990
1991         /* Show all directories with more than x% of the changes */
1992         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1993         gather_dirstat(options, &dir, changed, "", 0);
1994 }
1995
1996 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
1997 {
1998         int i;
1999         unsigned long changed;
2000         struct dirstat_dir dir;
2001
2002         if (data->nr == 0)
2003                 return;
2004
2005         dir.files = NULL;
2006         dir.alloc = 0;
2007         dir.nr = 0;
2008         dir.permille = options->dirstat_permille;
2009         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2010
2011         changed = 0;
2012         for (i = 0; i < data->nr; i++) {
2013                 struct diffstat_file *file = data->files[i];
2014                 unsigned long damage = file->added + file->deleted;
2015                 if (file->is_binary)
2016                         /*
2017                          * binary files counts bytes, not lines. Must find some
2018                          * way to normalize binary bytes vs. textual lines.
2019                          * The following heuristic assumes that there are 64
2020                          * bytes per "line".
2021                          * This is stupid and ugly, but very cheap...
2022                          */
2023                         damage = (damage + 63) / 64;
2024                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2025                 dir.files[dir.nr].name = file->name;
2026                 dir.files[dir.nr].changed = damage;
2027                 changed += damage;
2028                 dir.nr++;
2029         }
2030
2031         /* This can happen even with many files, if everything was renames */
2032         if (!changed)
2033                 return;
2034
2035         /* Show all directories with more than x% of the changes */
2036         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
2037         gather_dirstat(options, &dir, changed, "", 0);
2038 }
2039
2040 static void free_diffstat_info(struct diffstat_t *diffstat)
2041 {
2042         int i;
2043         for (i = 0; i < diffstat->nr; i++) {
2044                 struct diffstat_file *f = diffstat->files[i];
2045                 if (f->name != f->print_name)
2046                         free(f->print_name);
2047                 free(f->name);
2048                 free(f->from_name);
2049                 free(f);
2050         }
2051         free(diffstat->files);
2052 }
2053
2054 struct checkdiff_t {
2055         const char *filename;
2056         int lineno;
2057         int conflict_marker_size;
2058         struct diff_options *o;
2059         unsigned ws_rule;
2060         unsigned status;
2061 };
2062
2063 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2064 {
2065         char firstchar;
2066         int cnt;
2067
2068         if (len < marker_size + 1)
2069                 return 0;
2070         firstchar = line[0];
2071         switch (firstchar) {
2072         case '=': case '>': case '<': case '|':
2073                 break;
2074         default:
2075                 return 0;
2076         }
2077         for (cnt = 1; cnt < marker_size; cnt++)
2078                 if (line[cnt] != firstchar)
2079                         return 0;
2080         /* line[1] thru line[marker_size-1] are same as firstchar */
2081         if (len < marker_size + 1 || !isspace(line[marker_size]))
2082                 return 0;
2083         return 1;
2084 }
2085
2086 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2087 {
2088         struct checkdiff_t *data = priv;
2089         int marker_size = data->conflict_marker_size;
2090         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2091         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2092         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2093         char *err;
2094         const char *line_prefix;
2095
2096         assert(data->o);
2097         line_prefix = diff_line_prefix(data->o);
2098
2099         if (line[0] == '+') {
2100                 unsigned bad;
2101                 data->lineno++;
2102                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2103                         data->status |= 1;
2104                         fprintf(data->o->file,
2105                                 "%s%s:%d: leftover conflict marker\n",
2106                                 line_prefix, data->filename, data->lineno);
2107                 }
2108                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2109                 if (!bad)
2110                         return;
2111                 data->status |= bad;
2112                 err = whitespace_error_string(bad);
2113                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2114                         line_prefix, data->filename, data->lineno, err);
2115                 free(err);
2116                 emit_line(data->o, set, reset, line, 1);
2117                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2118                               data->o->file, set, reset, ws);
2119         } else if (line[0] == ' ') {
2120                 data->lineno++;
2121         } else if (line[0] == '@') {
2122                 char *plus = strchr(line, '+');
2123                 if (plus)
2124                         data->lineno = strtol(plus, NULL, 10) - 1;
2125                 else
2126                         die("invalid diff");
2127         }
2128 }
2129
2130 static unsigned char *deflate_it(char *data,
2131                                  unsigned long size,
2132                                  unsigned long *result_size)
2133 {
2134         int bound;
2135         unsigned char *deflated;
2136         git_zstream stream;
2137
2138         git_deflate_init(&stream, zlib_compression_level);
2139         bound = git_deflate_bound(&stream, size);
2140         deflated = xmalloc(bound);
2141         stream.next_out = deflated;
2142         stream.avail_out = bound;
2143
2144         stream.next_in = (unsigned char *)data;
2145         stream.avail_in = size;
2146         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2147                 ; /* nothing */
2148         git_deflate_end(&stream);
2149         *result_size = stream.total_out;
2150         return deflated;
2151 }
2152
2153 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2154                                   const char *prefix)
2155 {
2156         void *cp;
2157         void *delta;
2158         void *deflated;
2159         void *data;
2160         unsigned long orig_size;
2161         unsigned long delta_size;
2162         unsigned long deflate_size;
2163         unsigned long data_size;
2164
2165         /* We could do deflated delta, or we could do just deflated two,
2166          * whichever is smaller.
2167          */
2168         delta = NULL;
2169         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2170         if (one->size && two->size) {
2171                 delta = diff_delta(one->ptr, one->size,
2172                                    two->ptr, two->size,
2173                                    &delta_size, deflate_size);
2174                 if (delta) {
2175                         void *to_free = delta;
2176                         orig_size = delta_size;
2177                         delta = deflate_it(delta, delta_size, &delta_size);
2178                         free(to_free);
2179                 }
2180         }
2181
2182         if (delta && delta_size < deflate_size) {
2183                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2184                 free(deflated);
2185                 data = delta;
2186                 data_size = delta_size;
2187         }
2188         else {
2189                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
2190                 free(delta);
2191                 data = deflated;
2192                 data_size = deflate_size;
2193         }
2194
2195         /* emit data encoded in base85 */
2196         cp = data;
2197         while (data_size) {
2198                 int bytes = (52 < data_size) ? 52 : data_size;
2199                 char line[70];
2200                 data_size -= bytes;
2201                 if (bytes <= 26)
2202                         line[0] = bytes + 'A' - 1;
2203                 else
2204                         line[0] = bytes - 26 + 'a' - 1;
2205                 encode_85(line + 1, cp, bytes);
2206                 cp = (char *) cp + bytes;
2207                 fprintf(file, "%s", prefix);
2208                 fputs(line, file);
2209                 fputc('\n', file);
2210         }
2211         fprintf(file, "%s\n", prefix);
2212         free(data);
2213 }
2214
2215 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2216                              const char *prefix)
2217 {
2218         fprintf(file, "%sGIT binary patch\n", prefix);
2219         emit_binary_diff_body(file, one, two, prefix);
2220         emit_binary_diff_body(file, two, one, prefix);
2221 }
2222
2223 int diff_filespec_is_binary(struct diff_filespec *one)
2224 {
2225         if (one->is_binary == -1) {
2226                 diff_filespec_load_driver(one);
2227                 if (one->driver->binary != -1)
2228                         one->is_binary = one->driver->binary;
2229                 else {
2230                         if (!one->data && DIFF_FILE_VALID(one))
2231                                 diff_populate_filespec(one, CHECK_BINARY);
2232                         if (one->is_binary == -1 && one->data)
2233                                 one->is_binary = buffer_is_binary(one->data,
2234                                                 one->size);
2235                         if (one->is_binary == -1)
2236                                 one->is_binary = 0;
2237                 }
2238         }
2239         return one->is_binary;
2240 }
2241
2242 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2243 {
2244         diff_filespec_load_driver(one);
2245         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2246 }
2247
2248 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2249 {
2250         if (!options->a_prefix)
2251                 options->a_prefix = a;
2252         if (!options->b_prefix)
2253                 options->b_prefix = b;
2254 }
2255
2256 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2257 {
2258         if (!DIFF_FILE_VALID(one))
2259                 return NULL;
2260
2261         diff_filespec_load_driver(one);
2262         return userdiff_get_textconv(one->driver);
2263 }
2264
2265 static void builtin_diff(const char *name_a,
2266                          const char *name_b,
2267                          struct diff_filespec *one,
2268                          struct diff_filespec *two,
2269                          const char *xfrm_msg,
2270                          int must_show_header,
2271                          struct diff_options *o,
2272                          int complete_rewrite)
2273 {
2274         mmfile_t mf1, mf2;
2275         const char *lbl[2];
2276         char *a_one, *b_two;
2277         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2278         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2279         const char *a_prefix, *b_prefix;
2280         struct userdiff_driver *textconv_one = NULL;
2281         struct userdiff_driver *textconv_two = NULL;
2282         struct strbuf header = STRBUF_INIT;
2283         const char *line_prefix = diff_line_prefix(o);
2284
2285         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2286                         (!one->mode || S_ISGITLINK(one->mode)) &&
2287                         (!two->mode || S_ISGITLINK(two->mode))) {
2288                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2289                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2290                 show_submodule_summary(o->file, one->path ? one->path : two->path,
2291                                 line_prefix,
2292                                 one->sha1, two->sha1, two->dirty_submodule,
2293                                 meta, del, add, reset);
2294                 return;
2295         }
2296
2297         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2298                 textconv_one = get_textconv(one);
2299                 textconv_two = get_textconv(two);
2300         }
2301
2302         diff_set_mnemonic_prefix(o, "a/", "b/");
2303         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2304                 a_prefix = o->b_prefix;
2305                 b_prefix = o->a_prefix;
2306         } else {
2307                 a_prefix = o->a_prefix;
2308                 b_prefix = o->b_prefix;
2309         }
2310
2311         /* Never use a non-valid filename anywhere if at all possible */
2312         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2313         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2314
2315         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2316         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2317         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2318         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2319         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
2320         if (lbl[0][0] == '/') {
2321                 /* /dev/null */
2322                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
2323                 if (xfrm_msg)
2324                         strbuf_addstr(&header, xfrm_msg);
2325                 must_show_header = 1;
2326         }
2327         else if (lbl[1][0] == '/') {
2328                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
2329                 if (xfrm_msg)
2330                         strbuf_addstr(&header, xfrm_msg);
2331                 must_show_header = 1;
2332         }
2333         else {
2334                 if (one->mode != two->mode) {
2335                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2336                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
2337                         must_show_header = 1;
2338                 }
2339                 if (xfrm_msg)
2340                         strbuf_addstr(&header, xfrm_msg);
2341
2342                 /*
2343                  * we do not run diff between different kind
2344                  * of objects.
2345                  */
2346                 if ((one->mode ^ two->mode) & S_IFMT)
2347                         goto free_ab_and_return;
2348                 if (complete_rewrite &&
2349                     (textconv_one || !diff_filespec_is_binary(one)) &&
2350                     (textconv_two || !diff_filespec_is_binary(two))) {
2351                         fprintf(o->file, "%s", header.buf);
2352                         strbuf_reset(&header);
2353                         emit_rewrite_diff(name_a, name_b, one, two,
2354                                                 textconv_one, textconv_two, o);
2355                         o->found_changes = 1;
2356                         goto free_ab_and_return;
2357                 }
2358         }
2359
2360         if (o->irreversible_delete && lbl[1][0] == '/') {
2361                 fprintf(o->file, "%s", header.buf);
2362                 strbuf_reset(&header);
2363                 goto free_ab_and_return;
2364         } else if (!DIFF_OPT_TST(o, TEXT) &&
2365             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2366               (!textconv_two && diff_filespec_is_binary(two)) )) {
2367                 if (!one->data && !two->data &&
2368                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
2369                     !DIFF_OPT_TST(o, BINARY)) {
2370                         if (!hashcmp(one->sha1, two->sha1)) {
2371                                 if (must_show_header)
2372                                         fprintf(o->file, "%s", header.buf);
2373                                 goto free_ab_and_return;
2374                         }
2375                         fprintf(o->file, "%s", header.buf);
2376                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2377                                 line_prefix, lbl[0], lbl[1]);
2378                         goto free_ab_and_return;
2379                 }
2380                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2381                         die("unable to read files to diff");
2382                 /* Quite common confusing case */
2383                 if (mf1.size == mf2.size &&
2384                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2385                         if (must_show_header)
2386                                 fprintf(o->file, "%s", header.buf);
2387                         goto free_ab_and_return;
2388                 }
2389                 fprintf(o->file, "%s", header.buf);
2390                 strbuf_reset(&header);
2391                 if (DIFF_OPT_TST(o, BINARY))
2392                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2393                 else
2394                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2395                                 line_prefix, lbl[0], lbl[1]);
2396                 o->found_changes = 1;
2397         } else {
2398                 /* Crazy xdl interfaces.. */
2399                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2400                 const char *v;
2401                 xpparam_t xpp;
2402                 xdemitconf_t xecfg;
2403                 struct emit_callback ecbdata;
2404                 const struct userdiff_funcname *pe;
2405
2406                 if (must_show_header) {
2407                         fprintf(o->file, "%s", header.buf);
2408                         strbuf_reset(&header);
2409                 }
2410
2411                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2412                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2413
2414                 pe = diff_funcname_pattern(one);
2415                 if (!pe)
2416                         pe = diff_funcname_pattern(two);
2417
2418                 memset(&xpp, 0, sizeof(xpp));
2419                 memset(&xecfg, 0, sizeof(xecfg));
2420                 memset(&ecbdata, 0, sizeof(ecbdata));
2421                 ecbdata.label_path = lbl;
2422                 ecbdata.color_diff = want_color(o->use_color);
2423                 ecbdata.found_changesp = &o->found_changes;
2424                 ecbdata.ws_rule = whitespace_rule(name_b);
2425                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2426                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2427                 ecbdata.opt = o;
2428                 ecbdata.header = header.len ? &header : NULL;
2429                 xpp.flags = o->xdl_opts;
2430                 xecfg.ctxlen = o->context;
2431                 xecfg.interhunkctxlen = o->interhunkcontext;
2432                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2433                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2434                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2435                 if (pe)
2436                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2437                 if (!diffopts)
2438                         ;
2439                 else if (skip_prefix(diffopts, "--unified=", &v))
2440                         xecfg.ctxlen = strtoul(v, NULL, 10);
2441                 else if (skip_prefix(diffopts, "-u", &v))
2442                         xecfg.ctxlen = strtoul(v, NULL, 10);
2443                 if (o->word_diff)
2444                         init_diff_words_data(&ecbdata, o, one, two);
2445                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2446                               &xpp, &xecfg);
2447                 if (o->word_diff)
2448                         free_diff_words_data(&ecbdata);
2449                 if (textconv_one)
2450                         free(mf1.ptr);
2451                 if (textconv_two)
2452                         free(mf2.ptr);
2453                 xdiff_clear_find_func(&xecfg);
2454         }
2455
2456  free_ab_and_return:
2457         strbuf_release(&header);
2458         diff_free_filespec_data(one);
2459         diff_free_filespec_data(two);
2460         free(a_one);
2461         free(b_two);
2462         return;
2463 }
2464
2465 static void builtin_diffstat(const char *name_a, const char *name_b,
2466                              struct diff_filespec *one,
2467                              struct diff_filespec *two,
2468                              struct diffstat_t *diffstat,
2469                              struct diff_options *o,
2470                              struct diff_filepair *p)
2471 {
2472         mmfile_t mf1, mf2;
2473         struct diffstat_file *data;
2474         int same_contents;
2475         int complete_rewrite = 0;
2476
2477         if (!DIFF_PAIR_UNMERGED(p)) {
2478                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2479                         complete_rewrite = 1;
2480         }
2481
2482         data = diffstat_add(diffstat, name_a, name_b);
2483         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2484
2485         if (!one || !two) {
2486                 data->is_unmerged = 1;
2487                 return;
2488         }
2489
2490         same_contents = !hashcmp(one->sha1, two->sha1);
2491
2492         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2493                 data->is_binary = 1;
2494                 if (same_contents) {
2495                         data->added = 0;
2496                         data->deleted = 0;
2497                 } else {
2498                         data->added = diff_filespec_size(two);
2499                         data->deleted = diff_filespec_size(one);
2500                 }
2501         }
2502
2503         else if (complete_rewrite) {
2504                 diff_populate_filespec(one, 0);
2505                 diff_populate_filespec(two, 0);
2506                 data->deleted = count_lines(one->data, one->size);
2507                 data->added = count_lines(two->data, two->size);
2508         }
2509
2510         else if (!same_contents) {
2511                 /* Crazy xdl interfaces.. */
2512                 xpparam_t xpp;
2513                 xdemitconf_t xecfg;
2514
2515                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2516                         die("unable to read files to diff");
2517
2518                 memset(&xpp, 0, sizeof(xpp));
2519                 memset(&xecfg, 0, sizeof(xecfg));
2520                 xpp.flags = o->xdl_opts;
2521                 xecfg.ctxlen = o->context;
2522                 xecfg.interhunkctxlen = o->interhunkcontext;
2523                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2524                               &xpp, &xecfg);
2525         }
2526
2527         diff_free_filespec_data(one);
2528         diff_free_filespec_data(two);
2529 }
2530
2531 static void builtin_checkdiff(const char *name_a, const char *name_b,
2532                               const char *attr_path,
2533                               struct diff_filespec *one,
2534                               struct diff_filespec *two,
2535                               struct diff_options *o)
2536 {
2537         mmfile_t mf1, mf2;
2538         struct checkdiff_t data;
2539
2540         if (!two)
2541                 return;
2542
2543         memset(&data, 0, sizeof(data));
2544         data.filename = name_b ? name_b : name_a;
2545         data.lineno = 0;
2546         data.o = o;
2547         data.ws_rule = whitespace_rule(attr_path);
2548         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2549
2550         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2551                 die("unable to read files to diff");
2552
2553         /*
2554          * All the other codepaths check both sides, but not checking
2555          * the "old" side here is deliberate.  We are checking the newly
2556          * introduced changes, and as long as the "new" side is text, we
2557          * can and should check what it introduces.
2558          */
2559         if (diff_filespec_is_binary(two))
2560                 goto free_and_return;
2561         else {
2562                 /* Crazy xdl interfaces.. */
2563                 xpparam_t xpp;
2564                 xdemitconf_t xecfg;
2565
2566                 memset(&xpp, 0, sizeof(xpp));
2567                 memset(&xecfg, 0, sizeof(xecfg));
2568                 xecfg.ctxlen = 1; /* at least one context line */
2569                 xpp.flags = 0;
2570                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2571                               &xpp, &xecfg);
2572
2573                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2574                         struct emit_callback ecbdata;
2575                         int blank_at_eof;
2576
2577                         ecbdata.ws_rule = data.ws_rule;
2578                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2579                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2580
2581                         if (blank_at_eof) {
2582                                 static char *err;
2583                                 if (!err)
2584                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2585                                 fprintf(o->file, "%s:%d: %s.\n",
2586                                         data.filename, blank_at_eof, err);
2587                                 data.status = 1; /* report errors */
2588                         }
2589                 }
2590         }
2591  free_and_return:
2592         diff_free_filespec_data(one);
2593         diff_free_filespec_data(two);
2594         if (data.status)
2595                 DIFF_OPT_SET(o, CHECK_FAILED);
2596 }
2597
2598 struct diff_filespec *alloc_filespec(const char *path)
2599 {
2600         int namelen = strlen(path);
2601         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2602
2603         memset(spec, 0, sizeof(*spec));
2604         spec->path = (char *)(spec + 1);
2605         memcpy(spec->path, path, namelen+1);
2606         spec->count = 1;
2607         spec->is_binary = -1;
2608         return spec;
2609 }
2610
2611 void free_filespec(struct diff_filespec *spec)
2612 {
2613         if (!--spec->count) {
2614                 diff_free_filespec_data(spec);
2615                 free(spec);
2616         }
2617 }
2618
2619 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2620                    int sha1_valid, unsigned short mode)
2621 {
2622         if (mode) {
2623                 spec->mode = canon_mode(mode);
2624                 hashcpy(spec->sha1, sha1);
2625                 spec->sha1_valid = sha1_valid;
2626         }
2627 }
2628
2629 /*
2630  * Given a name and sha1 pair, if the index tells us the file in
2631  * the work tree has that object contents, return true, so that
2632  * prepare_temp_file() does not have to inflate and extract.
2633  */
2634 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2635 {
2636         const struct cache_entry *ce;
2637         struct stat st;
2638         int pos, len;
2639
2640         /*
2641          * We do not read the cache ourselves here, because the
2642          * benchmark with my previous version that always reads cache
2643          * shows that it makes things worse for diff-tree comparing
2644          * two linux-2.6 kernel trees in an already checked out work
2645          * tree.  This is because most diff-tree comparisons deal with
2646          * only a small number of files, while reading the cache is
2647          * expensive for a large project, and its cost outweighs the
2648          * savings we get by not inflating the object to a temporary
2649          * file.  Practically, this code only helps when we are used
2650          * by diff-cache --cached, which does read the cache before
2651          * calling us.
2652          */
2653         if (!active_cache)
2654                 return 0;
2655
2656         /* We want to avoid the working directory if our caller
2657          * doesn't need the data in a normal file, this system
2658          * is rather slow with its stat/open/mmap/close syscalls,
2659          * and the object is contained in a pack file.  The pack
2660          * is probably already open and will be faster to obtain
2661          * the data through than the working directory.  Loose
2662          * objects however would tend to be slower as they need
2663          * to be individually opened and inflated.
2664          */
2665         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2666                 return 0;
2667
2668         len = strlen(name);
2669         pos = cache_name_pos(name, len);
2670         if (pos < 0)
2671                 return 0;
2672         ce = active_cache[pos];
2673
2674         /*
2675          * This is not the sha1 we are looking for, or
2676          * unreusable because it is not a regular file.
2677          */
2678         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2679                 return 0;
2680
2681         /*
2682          * If ce is marked as "assume unchanged", there is no
2683          * guarantee that work tree matches what we are looking for.
2684          */
2685         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2686                 return 0;
2687
2688         /*
2689          * If ce matches the file in the work tree, we can reuse it.
2690          */
2691         if (ce_uptodate(ce) ||
2692             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2693                 return 1;
2694
2695         return 0;
2696 }
2697
2698 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2699 {
2700         int len;
2701         char *data = xmalloc(100), *dirty = "";
2702
2703         /* Are we looking at the work tree? */
2704         if (s->dirty_submodule)
2705                 dirty = "-dirty";
2706
2707         len = snprintf(data, 100,
2708                        "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2709         s->data = data;
2710         s->size = len;
2711         s->should_free = 1;
2712         if (size_only) {
2713                 s->data = NULL;
2714                 free(data);
2715         }
2716         return 0;
2717 }
2718
2719 /*
2720  * While doing rename detection and pickaxe operation, we may need to
2721  * grab the data for the blob (or file) for our own in-core comparison.
2722  * diff_filespec has data and size fields for this purpose.
2723  */
2724 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
2725 {
2726         int size_only = flags & CHECK_SIZE_ONLY;
2727         int err = 0;
2728         /*
2729          * demote FAIL to WARN to allow inspecting the situation
2730          * instead of refusing.
2731          */
2732         enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2733                                     ? SAFE_CRLF_WARN
2734                                     : safe_crlf);
2735
2736         if (!DIFF_FILE_VALID(s))
2737                 die("internal error: asking to populate invalid file.");
2738         if (S_ISDIR(s->mode))
2739                 return -1;
2740
2741         if (s->data)
2742                 return 0;
2743
2744         if (size_only && 0 < s->size)
2745                 return 0;
2746
2747         if (S_ISGITLINK(s->mode))
2748                 return diff_populate_gitlink(s, size_only);
2749
2750         if (!s->sha1_valid ||
2751             reuse_worktree_file(s->path, s->sha1, 0)) {
2752                 struct strbuf buf = STRBUF_INIT;
2753                 struct stat st;
2754                 int fd;
2755
2756                 if (lstat(s->path, &st) < 0) {
2757                         if (errno == ENOENT) {
2758                         err_empty:
2759                                 err = -1;
2760                         empty:
2761                                 s->data = (char *)"";
2762                                 s->size = 0;
2763                                 return err;
2764                         }
2765                 }
2766                 s->size = xsize_t(st.st_size);
2767                 if (!s->size)
2768                         goto empty;
2769                 if (S_ISLNK(st.st_mode)) {
2770                         struct strbuf sb = STRBUF_INIT;
2771
2772                         if (strbuf_readlink(&sb, s->path, s->size))
2773                                 goto err_empty;
2774                         s->size = sb.len;
2775                         s->data = strbuf_detach(&sb, NULL);
2776                         s->should_free = 1;
2777                         return 0;
2778                 }
2779                 if (size_only)
2780                         return 0;
2781                 if ((flags & CHECK_BINARY) &&
2782                     s->size > big_file_threshold && s->is_binary == -1) {
2783                         s->is_binary = 1;
2784                         return 0;
2785                 }
2786                 fd = open(s->path, O_RDONLY);
2787                 if (fd < 0)
2788                         goto err_empty;
2789                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2790                 close(fd);
2791                 s->should_munmap = 1;
2792
2793                 /*
2794                  * Convert from working tree format to canonical git format
2795                  */
2796                 if (convert_to_git(s->path, s->data, s->size, &buf, crlf_warn)) {
2797                         size_t size = 0;
2798                         munmap(s->data, s->size);
2799                         s->should_munmap = 0;
2800                         s->data = strbuf_detach(&buf, &size);
2801                         s->size = size;
2802                         s->should_free = 1;
2803                 }
2804         }
2805         else {
2806                 enum object_type type;
2807                 if (size_only || (flags & CHECK_BINARY)) {
2808                         type = sha1_object_info(s->sha1, &s->size);
2809                         if (type < 0)
2810                                 die("unable to read %s", sha1_to_hex(s->sha1));
2811                         if (size_only)
2812                                 return 0;
2813                         if (s->size > big_file_threshold && s->is_binary == -1) {
2814                                 s->is_binary = 1;
2815                                 return 0;
2816                         }
2817                 }
2818                 s->data = read_sha1_file(s->sha1, &type, &s->size);
2819                 if (!s->data)
2820                         die("unable to read %s", sha1_to_hex(s->sha1));
2821                 s->should_free = 1;
2822         }
2823         return 0;
2824 }
2825
2826 void diff_free_filespec_blob(struct diff_filespec *s)
2827 {
2828         if (s->should_free)
2829                 free(s->data);
2830         else if (s->should_munmap)
2831                 munmap(s->data, s->size);
2832
2833         if (s->should_free || s->should_munmap) {
2834                 s->should_free = s->should_munmap = 0;
2835                 s->data = NULL;
2836         }
2837 }
2838
2839 void diff_free_filespec_data(struct diff_filespec *s)
2840 {
2841         diff_free_filespec_blob(s);
2842         free(s->cnt_data);
2843         s->cnt_data = NULL;
2844 }
2845
2846 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2847                            void *blob,
2848                            unsigned long size,
2849                            const unsigned char *sha1,
2850                            int mode)
2851 {
2852         int fd;
2853         struct strbuf buf = STRBUF_INIT;
2854         struct strbuf template = STRBUF_INIT;
2855         char *path_dup = xstrdup(path);
2856         const char *base = basename(path_dup);
2857
2858         /* Generate "XXXXXX_basename.ext" */
2859         strbuf_addstr(&template, "XXXXXX_");
2860         strbuf_addstr(&template, base);
2861
2862         fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2863                         strlen(base) + 1);
2864         if (fd < 0)
2865                 die_errno("unable to create temp-file");
2866         if (convert_to_working_tree(path,
2867                         (const char *)blob, (size_t)size, &buf)) {
2868                 blob = buf.buf;
2869                 size = buf.len;
2870         }
2871         if (write_in_full(fd, blob, size) != size)
2872                 die_errno("unable to write temp-file");
2873         close(fd);
2874         temp->name = temp->tmp_path;
2875         strcpy(temp->hex, sha1_to_hex(sha1));
2876         temp->hex[40] = 0;
2877         sprintf(temp->mode, "%06o", mode);
2878         strbuf_release(&buf);
2879         strbuf_release(&template);
2880         free(path_dup);
2881 }
2882
2883 static struct diff_tempfile *prepare_temp_file(const char *name,
2884                 struct diff_filespec *one)
2885 {
2886         struct diff_tempfile *temp = claim_diff_tempfile();
2887
2888         if (!DIFF_FILE_VALID(one)) {
2889         not_a_valid_file:
2890                 /* A '-' entry produces this for file-2, and
2891                  * a '+' entry produces this for file-1.
2892                  */
2893                 temp->name = "/dev/null";
2894                 strcpy(temp->hex, ".");
2895                 strcpy(temp->mode, ".");
2896                 return temp;
2897         }
2898
2899         if (!remove_tempfile_installed) {
2900                 atexit(remove_tempfile);
2901                 sigchain_push_common(remove_tempfile_on_signal);
2902                 remove_tempfile_installed = 1;
2903         }
2904
2905         if (!S_ISGITLINK(one->mode) &&
2906             (!one->sha1_valid ||
2907              reuse_worktree_file(name, one->sha1, 1))) {
2908                 struct stat st;
2909                 if (lstat(name, &st) < 0) {
2910                         if (errno == ENOENT)
2911                                 goto not_a_valid_file;
2912                         die_errno("stat(%s)", name);
2913                 }
2914                 if (S_ISLNK(st.st_mode)) {
2915                         struct strbuf sb = STRBUF_INIT;
2916                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2917                                 die_errno("readlink(%s)", name);
2918                         prep_temp_blob(name, temp, sb.buf, sb.len,
2919                                        (one->sha1_valid ?
2920                                         one->sha1 : null_sha1),
2921                                        (one->sha1_valid ?
2922                                         one->mode : S_IFLNK));
2923                         strbuf_release(&sb);
2924                 }
2925                 else {
2926                         /* we can borrow from the file in the work tree */
2927                         temp->name = name;
2928                         if (!one->sha1_valid)
2929                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2930                         else
2931                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2932                         /* Even though we may sometimes borrow the
2933                          * contents from the work tree, we always want
2934                          * one->mode.  mode is trustworthy even when
2935                          * !(one->sha1_valid), as long as
2936                          * DIFF_FILE_VALID(one).
2937                          */
2938                         sprintf(temp->mode, "%06o", one->mode);
2939                 }
2940                 return temp;
2941         }
2942         else {
2943                 if (diff_populate_filespec(one, 0))
2944                         die("cannot read data blob for %s", one->path);
2945                 prep_temp_blob(name, temp, one->data, one->size,
2946                                one->sha1, one->mode);
2947         }
2948         return temp;
2949 }
2950
2951 static void add_external_diff_name(struct argv_array *argv,
2952                                    const char *name,
2953                                    struct diff_filespec *df)
2954 {
2955         struct diff_tempfile *temp = prepare_temp_file(name, df);
2956         argv_array_push(argv, temp->name);
2957         argv_array_push(argv, temp->hex);
2958         argv_array_push(argv, temp->mode);
2959 }
2960
2961 /* An external diff command takes:
2962  *
2963  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2964  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2965  *
2966  */
2967 static void run_external_diff(const char *pgm,
2968                               const char *name,
2969                               const char *other,
2970                               struct diff_filespec *one,
2971                               struct diff_filespec *two,
2972                               const char *xfrm_msg,
2973                               int complete_rewrite,
2974                               struct diff_options *o)
2975 {
2976         struct argv_array argv = ARGV_ARRAY_INIT;
2977         struct argv_array env = ARGV_ARRAY_INIT;
2978         struct diff_queue_struct *q = &diff_queued_diff;
2979
2980         argv_array_push(&argv, pgm);
2981         argv_array_push(&argv, name);
2982
2983         if (one && two) {
2984                 add_external_diff_name(&argv, name, one);
2985                 if (!other)
2986                         add_external_diff_name(&argv, name, two);
2987                 else {
2988                         add_external_diff_name(&argv, other, two);
2989                         argv_array_push(&argv, other);
2990                         argv_array_push(&argv, xfrm_msg);
2991                 }
2992         }
2993
2994         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
2995         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
2996
2997         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
2998                 die(_("external diff died, stopping at %s"), name);
2999
3000         remove_tempfile();
3001         argv_array_clear(&argv);
3002         argv_array_clear(&env);
3003 }
3004
3005 static int similarity_index(struct diff_filepair *p)
3006 {
3007         return p->score * 100 / MAX_SCORE;
3008 }
3009
3010 static void fill_metainfo(struct strbuf *msg,
3011                           const char *name,
3012                           const char *other,
3013                           struct diff_filespec *one,
3014                           struct diff_filespec *two,
3015                           struct diff_options *o,
3016                           struct diff_filepair *p,
3017                           int *must_show_header,
3018                           int use_color)
3019 {
3020         const char *set = diff_get_color(use_color, DIFF_METAINFO);
3021         const char *reset = diff_get_color(use_color, DIFF_RESET);
3022         const char *line_prefix = diff_line_prefix(o);
3023
3024         *must_show_header = 1;
3025         strbuf_init(msg, PATH_MAX * 2 + 300);
3026         switch (p->status) {
3027         case DIFF_STATUS_COPIED:
3028                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3029                             line_prefix, set, similarity_index(p));
3030                 strbuf_addf(msg, "%s\n%s%scopy from ",
3031                             reset,  line_prefix, set);
3032                 quote_c_style(name, msg, NULL, 0);
3033                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
3034                 quote_c_style(other, msg, NULL, 0);
3035                 strbuf_addf(msg, "%s\n", reset);
3036                 break;
3037         case DIFF_STATUS_RENAMED:
3038                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3039                             line_prefix, set, similarity_index(p));
3040                 strbuf_addf(msg, "%s\n%s%srename from ",
3041                             reset, line_prefix, set);
3042                 quote_c_style(name, msg, NULL, 0);
3043                 strbuf_addf(msg, "%s\n%s%srename to ",
3044                             reset, line_prefix, set);
3045                 quote_c_style(other, msg, NULL, 0);
3046                 strbuf_addf(msg, "%s\n", reset);
3047                 break;
3048         case DIFF_STATUS_MODIFIED:
3049                 if (p->score) {
3050                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3051                                     line_prefix,
3052                                     set, similarity_index(p), reset);
3053                         break;
3054                 }
3055                 /* fallthru */
3056         default:
3057                 *must_show_header = 0;
3058         }
3059         if (one && two && hashcmp(one->sha1, two->sha1)) {
3060                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3061
3062                 if (DIFF_OPT_TST(o, BINARY)) {
3063                         mmfile_t mf;
3064                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3065                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3066                                 abbrev = 40;
3067                 }
3068                 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
3069                             find_unique_abbrev(one->sha1, abbrev));
3070                 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
3071                 if (one->mode == two->mode)
3072                         strbuf_addf(msg, " %06o", one->mode);
3073                 strbuf_addf(msg, "%s\n", reset);
3074         }
3075 }
3076
3077 static void run_diff_cmd(const char *pgm,
3078                          const char *name,
3079                          const char *other,
3080                          const char *attr_path,
3081                          struct diff_filespec *one,
3082                          struct diff_filespec *two,
3083                          struct strbuf *msg,
3084                          struct diff_options *o,
3085                          struct diff_filepair *p)
3086 {
3087         const char *xfrm_msg = NULL;
3088         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3089         int must_show_header = 0;
3090
3091
3092         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3093                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3094                 if (drv && drv->external)
3095                         pgm = drv->external;
3096         }
3097
3098         if (msg) {
3099                 /*
3100                  * don't use colors when the header is intended for an
3101                  * external diff driver
3102                  */
3103                 fill_metainfo(msg, name, other, one, two, o, p,
3104                               &must_show_header,
3105                               want_color(o->use_color) && !pgm);
3106                 xfrm_msg = msg->len ? msg->buf : NULL;
3107         }
3108
3109         if (pgm) {
3110                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3111                                   complete_rewrite, o);
3112                 return;
3113         }
3114         if (one && two)
3115                 builtin_diff(name, other ? other : name,
3116                              one, two, xfrm_msg, must_show_header,
3117                              o, complete_rewrite);
3118         else
3119                 fprintf(o->file, "* Unmerged path %s\n", name);
3120 }
3121
3122 static void diff_fill_sha1_info(struct diff_filespec *one)
3123 {
3124         if (DIFF_FILE_VALID(one)) {
3125                 if (!one->sha1_valid) {
3126                         struct stat st;
3127                         if (one->is_stdin) {
3128                                 hashcpy(one->sha1, null_sha1);
3129                                 return;
3130                         }
3131                         if (lstat(one->path, &st) < 0)
3132                                 die_errno("stat '%s'", one->path);
3133                         if (index_path(one->sha1, one->path, &st, 0))
3134                                 die("cannot hash %s", one->path);
3135                 }
3136         }
3137         else
3138                 hashclr(one->sha1);
3139 }
3140
3141 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3142 {
3143         /* Strip the prefix but do not molest /dev/null and absolute paths */
3144         if (*namep && **namep != '/') {
3145                 *namep += prefix_length;
3146                 if (**namep == '/')
3147                         ++*namep;
3148         }
3149         if (*otherp && **otherp != '/') {
3150                 *otherp += prefix_length;
3151                 if (**otherp == '/')
3152                         ++*otherp;
3153         }
3154 }
3155
3156 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3157 {
3158         const char *pgm = external_diff();
3159         struct strbuf msg;
3160         struct diff_filespec *one = p->one;
3161         struct diff_filespec *two = p->two;
3162         const char *name;
3163         const char *other;
3164         const char *attr_path;
3165
3166         name  = p->one->path;
3167         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3168         attr_path = name;
3169         if (o->prefix_length)
3170                 strip_prefix(o->prefix_length, &name, &other);
3171
3172         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3173                 pgm = NULL;
3174
3175         if (DIFF_PAIR_UNMERGED(p)) {
3176                 run_diff_cmd(pgm, name, NULL, attr_path,
3177                              NULL, NULL, NULL, o, p);
3178                 return;
3179         }
3180
3181         diff_fill_sha1_info(one);
3182         diff_fill_sha1_info(two);
3183
3184         if (!pgm &&
3185             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3186             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3187                 /*
3188                  * a filepair that changes between file and symlink
3189                  * needs to be split into deletion and creation.
3190                  */
3191                 struct diff_filespec *null = alloc_filespec(two->path);
3192                 run_diff_cmd(NULL, name, other, attr_path,
3193                              one, null, &msg, o, p);
3194                 free(null);
3195                 strbuf_release(&msg);
3196
3197                 null = alloc_filespec(one->path);
3198                 run_diff_cmd(NULL, name, other, attr_path,
3199                              null, two, &msg, o, p);
3200                 free(null);
3201         }
3202         else
3203                 run_diff_cmd(pgm, name, other, attr_path,
3204                              one, two, &msg, o, p);
3205
3206         strbuf_release(&msg);
3207 }
3208
3209 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3210                          struct diffstat_t *diffstat)
3211 {
3212         const char *name;
3213         const char *other;
3214
3215         if (DIFF_PAIR_UNMERGED(p)) {
3216                 /* unmerged */
3217                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3218                 return;
3219         }
3220
3221         name = p->one->path;
3222         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3223
3224         if (o->prefix_length)
3225                 strip_prefix(o->prefix_length, &name, &other);
3226
3227         diff_fill_sha1_info(p->one);
3228         diff_fill_sha1_info(p->two);
3229
3230         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3231 }
3232
3233 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3234 {
3235         const char *name;
3236         const char *other;
3237         const char *attr_path;
3238
3239         if (DIFF_PAIR_UNMERGED(p)) {
3240                 /* unmerged */
3241                 return;
3242         }
3243
3244         name = p->one->path;
3245         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3246         attr_path = other ? other : name;
3247
3248         if (o->prefix_length)
3249                 strip_prefix(o->prefix_length, &name, &other);
3250
3251         diff_fill_sha1_info(p->one);
3252         diff_fill_sha1_info(p->two);
3253
3254         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3255 }
3256
3257 void diff_setup(struct diff_options *options)
3258 {
3259         memcpy(options, &default_diff_options, sizeof(*options));
3260
3261         options->file = stdout;
3262
3263         options->line_termination = '\n';
3264         options->break_opt = -1;
3265         options->rename_limit = -1;
3266         options->dirstat_permille = diff_dirstat_permille_default;
3267         options->context = diff_context_default;
3268         options->ws_error_highlight = WSEH_NEW;
3269         DIFF_OPT_SET(options, RENAME_EMPTY);
3270
3271         /* pathchange left =NULL by default */
3272         options->change = diff_change;
3273         options->add_remove = diff_addremove;
3274         options->use_color = diff_use_color_default;
3275         options->detect_rename = diff_detect_rename_default;
3276         options->xdl_opts |= diff_algorithm;
3277
3278         options->orderfile = diff_order_file_cfg;
3279
3280         if (diff_no_prefix) {
3281                 options->a_prefix = options->b_prefix = "";
3282         } else if (!diff_mnemonic_prefix) {
3283                 options->a_prefix = "a/";
3284                 options->b_prefix = "b/";
3285         }
3286 }
3287
3288 void diff_setup_done(struct diff_options *options)
3289 {
3290         int count = 0;
3291
3292         if (options->set_default)
3293                 options->set_default(options);
3294
3295         if (options->output_format & DIFF_FORMAT_NAME)
3296                 count++;
3297         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3298                 count++;
3299         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3300                 count++;
3301         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3302                 count++;
3303         if (count > 1)
3304                 die("--name-only, --name-status, --check and -s are mutually exclusive");
3305
3306         /*
3307          * Most of the time we can say "there are changes"
3308          * only by checking if there are changed paths, but
3309          * --ignore-whitespace* options force us to look
3310          * inside contents.
3311          */
3312
3313         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3314             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3315             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3316                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3317         else
3318                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3319
3320         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3321                 options->detect_rename = DIFF_DETECT_COPY;
3322
3323         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3324                 options->prefix = NULL;
3325         if (options->prefix)
3326                 options->prefix_length = strlen(options->prefix);
3327         else
3328                 options->prefix_length = 0;
3329
3330         if (options->output_format & (DIFF_FORMAT_NAME |
3331                                       DIFF_FORMAT_NAME_STATUS |
3332                                       DIFF_FORMAT_CHECKDIFF |
3333                                       DIFF_FORMAT_NO_OUTPUT))
3334                 options->output_format &= ~(DIFF_FORMAT_RAW |
3335                                             DIFF_FORMAT_NUMSTAT |
3336                                             DIFF_FORMAT_DIFFSTAT |
3337                                             DIFF_FORMAT_SHORTSTAT |
3338                                             DIFF_FORMAT_DIRSTAT |
3339                                             DIFF_FORMAT_SUMMARY |
3340                                             DIFF_FORMAT_PATCH);
3341
3342         /*
3343          * These cases always need recursive; we do not drop caller-supplied
3344          * recursive bits for other formats here.
3345          */
3346         if (options->output_format & (DIFF_FORMAT_PATCH |
3347                                       DIFF_FORMAT_NUMSTAT |
3348                                       DIFF_FORMAT_DIFFSTAT |
3349                                       DIFF_FORMAT_SHORTSTAT |
3350                                       DIFF_FORMAT_DIRSTAT |
3351                                       DIFF_FORMAT_SUMMARY |
3352                                       DIFF_FORMAT_CHECKDIFF))
3353                 DIFF_OPT_SET(options, RECURSIVE);
3354         /*
3355          * Also pickaxe would not work very well if you do not say recursive
3356          */
3357         if (options->pickaxe)
3358                 DIFF_OPT_SET(options, RECURSIVE);
3359         /*
3360          * When patches are generated, submodules diffed against the work tree
3361          * must be checked for dirtiness too so it can be shown in the output
3362          */
3363         if (options->output_format & DIFF_FORMAT_PATCH)
3364                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3365
3366         if (options->detect_rename && options->rename_limit < 0)
3367                 options->rename_limit = diff_rename_limit_default;
3368         if (options->setup & DIFF_SETUP_USE_CACHE) {
3369                 if (!active_cache)
3370                         /* read-cache does not die even when it fails
3371                          * so it is safe for us to do this here.  Also
3372                          * it does not smudge active_cache or active_nr
3373                          * when it fails, so we do not have to worry about
3374                          * cleaning it up ourselves either.
3375                          */
3376                         read_cache();
3377         }
3378         if (options->abbrev <= 0 || 40 < options->abbrev)
3379                 options->abbrev = 40; /* full */
3380
3381         /*
3382          * It does not make sense to show the first hit we happened
3383          * to have found.  It does not make sense not to return with
3384          * exit code in such a case either.
3385          */
3386         if (DIFF_OPT_TST(options, QUICK)) {
3387                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3388                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3389         }
3390
3391         options->diff_path_counter = 0;
3392
3393         if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
3394                 die(_("--follow requires exactly one pathspec"));
3395 }
3396
3397 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3398 {
3399         char c, *eq;
3400         int len;
3401
3402         if (*arg != '-')
3403                 return 0;
3404         c = *++arg;
3405         if (!c)
3406                 return 0;
3407         if (c == arg_short) {
3408                 c = *++arg;
3409                 if (!c)
3410                         return 1;
3411                 if (val && isdigit(c)) {
3412                         char *end;
3413                         int n = strtoul(arg, &end, 10);
3414                         if (*end)
3415                                 return 0;
3416                         *val = n;
3417                         return 1;
3418                 }
3419                 return 0;
3420         }
3421         if (c != '-')
3422                 return 0;
3423         arg++;
3424         eq = strchrnul(arg, '=');
3425         len = eq - arg;
3426         if (!len || strncmp(arg, arg_long, len))
3427                 return 0;
3428         if (*eq) {
3429                 int n;
3430                 char *end;
3431                 if (!isdigit(*++eq))
3432                         return 0;
3433                 n = strtoul(eq, &end, 10);
3434                 if (*end)
3435                         return 0;
3436                 *val = n;
3437         }
3438         return 1;
3439 }
3440
3441 static int diff_scoreopt_parse(const char *opt);
3442
3443 static inline int short_opt(char opt, const char **argv,
3444                             const char **optarg)
3445 {
3446         const char *arg = argv[0];
3447         if (arg[0] != '-' || arg[1] != opt)
3448                 return 0;
3449         if (arg[2] != '\0') {
3450                 *optarg = arg + 2;
3451                 return 1;
3452         }
3453         if (!argv[1])
3454                 die("Option '%c' requires a value", opt);
3455         *optarg = argv[1];
3456         return 2;
3457 }
3458
3459 int parse_long_opt(const char *opt, const char **argv,
3460                    const char **optarg)
3461 {
3462         const char *arg = argv[0];
3463         if (!skip_prefix(arg, "--", &arg))
3464                 return 0;
3465         if (!skip_prefix(arg, opt, &arg))
3466                 return 0;
3467         if (*arg == '=') { /* stuck form: --option=value */
3468                 *optarg = arg + 1;
3469                 return 1;
3470         }
3471         if (*arg != '\0')
3472                 return 0;
3473         /* separate form: --option value */
3474         if (!argv[1])
3475                 die("Option '--%s' requires a value", opt);
3476         *optarg = argv[1];
3477         return 2;
3478 }
3479
3480 static int stat_opt(struct diff_options *options, const char **av)
3481 {
3482         const char *arg = av[0];
3483         char *end;
3484         int width = options->stat_width;
3485         int name_width = options->stat_name_width;
3486         int graph_width = options->stat_graph_width;
3487         int count = options->stat_count;
3488         int argcount = 1;
3489
3490         if (!skip_prefix(arg, "--stat", &arg))
3491                 die("BUG: stat option does not begin with --stat: %s", arg);
3492         end = (char *)arg;
3493
3494         switch (*arg) {
3495         case '-':
3496                 if (skip_prefix(arg, "-width", &arg)) {
3497                         if (*arg == '=')
3498                                 width = strtoul(arg + 1, &end, 10);
3499                         else if (!*arg && !av[1])
3500                                 die("Option '--stat-width' requires a value");
3501                         else if (!*arg) {
3502                                 width = strtoul(av[1], &end, 10);
3503                                 argcount = 2;
3504                         }
3505                 } else if (skip_prefix(arg, "-name-width", &arg)) {
3506                         if (*arg == '=')
3507                                 name_width = strtoul(arg + 1, &end, 10);
3508                         else if (!*arg && !av[1])
3509                                 die("Option '--stat-name-width' requires a value");
3510                         else if (!*arg) {
3511                                 name_width = strtoul(av[1], &end, 10);
3512                                 argcount = 2;
3513                         }
3514                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
3515                         if (*arg == '=')
3516                                 graph_width = strtoul(arg + 1, &end, 10);
3517                         else if (!*arg && !av[1])
3518                                 die("Option '--stat-graph-width' requires a value");
3519                         else if (!*arg) {
3520                                 graph_width = strtoul(av[1], &end, 10);
3521                                 argcount = 2;
3522                         }
3523                 } else if (skip_prefix(arg, "-count", &arg)) {
3524                         if (*arg == '=')
3525                                 count = strtoul(arg + 1, &end, 10);
3526                         else if (!*arg && !av[1])
3527                                 die("Option '--stat-count' requires a value");
3528                         else if (!*arg) {
3529                                 count = strtoul(av[1], &end, 10);
3530                                 argcount = 2;
3531                         }
3532                 }
3533                 break;
3534         case '=':
3535                 width = strtoul(arg+1, &end, 10);
3536                 if (*end == ',')
3537                         name_width = strtoul(end+1, &end, 10);
3538                 if (*end == ',')
3539                         count = strtoul(end+1, &end, 10);
3540         }
3541
3542         /* Important! This checks all the error cases! */
3543         if (*end)
3544                 return 0;
3545         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3546         options->stat_name_width = name_width;
3547         options->stat_graph_width = graph_width;
3548         options->stat_width = width;
3549         options->stat_count = count;
3550         return argcount;
3551 }
3552
3553 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3554 {
3555         struct strbuf errmsg = STRBUF_INIT;
3556         if (parse_dirstat_params(options, params, &errmsg))
3557                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3558                     errmsg.buf);
3559         strbuf_release(&errmsg);
3560         /*
3561          * The caller knows a dirstat-related option is given from the command
3562          * line; allow it to say "return this_function();"
3563          */
3564         options->output_format |= DIFF_FORMAT_DIRSTAT;
3565         return 1;
3566 }
3567
3568 static int parse_submodule_opt(struct diff_options *options, const char *value)
3569 {
3570         if (parse_submodule_params(options, value))
3571                 die(_("Failed to parse --submodule option parameter: '%s'"),
3572                         value);
3573         return 1;
3574 }
3575
3576 static const char diff_status_letters[] = {
3577         DIFF_STATUS_ADDED,
3578         DIFF_STATUS_COPIED,
3579         DIFF_STATUS_DELETED,
3580         DIFF_STATUS_MODIFIED,
3581         DIFF_STATUS_RENAMED,
3582         DIFF_STATUS_TYPE_CHANGED,
3583         DIFF_STATUS_UNKNOWN,
3584         DIFF_STATUS_UNMERGED,
3585         DIFF_STATUS_FILTER_AON,
3586         DIFF_STATUS_FILTER_BROKEN,
3587         '\0',
3588 };
3589
3590 static unsigned int filter_bit['Z' + 1];
3591
3592 static void prepare_filter_bits(void)
3593 {
3594         int i;
3595
3596         if (!filter_bit[DIFF_STATUS_ADDED]) {
3597                 for (i = 0; diff_status_letters[i]; i++)
3598                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
3599         }
3600 }
3601
3602 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
3603 {
3604         return opt->filter & filter_bit[(int) status];
3605 }
3606
3607 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
3608 {
3609         int i, optch;
3610
3611         prepare_filter_bits();
3612
3613         /*
3614          * If there is a negation e.g. 'd' in the input, and we haven't
3615          * initialized the filter field with another --diff-filter, start
3616          * from full set of bits, except for AON.
3617          */
3618         if (!opt->filter) {
3619                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3620                         if (optch < 'a' || 'z' < optch)
3621                                 continue;
3622                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
3623                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
3624                         break;
3625                 }
3626         }
3627
3628         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3629                 unsigned int bit;
3630                 int negate;
3631
3632                 if ('a' <= optch && optch <= 'z') {
3633                         negate = 1;
3634                         optch = toupper(optch);
3635                 } else {
3636                         negate = 0;
3637                 }
3638
3639                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
3640                 if (!bit)
3641                         return optarg[i];
3642                 if (negate)
3643                         opt->filter &= ~bit;
3644                 else
3645                         opt->filter |= bit;
3646         }
3647         return 0;
3648 }
3649
3650 static void enable_patch_output(int *fmt) {
3651         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
3652         *fmt |= DIFF_FORMAT_PATCH;
3653 }
3654
3655 static int parse_one_token(const char **arg, const char *token)
3656 {
3657         const char *rest;
3658         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
3659                 *arg = rest;
3660                 return 1;
3661         }
3662         return 0;
3663 }
3664
3665 static int parse_ws_error_highlight(struct diff_options *opt, const char *arg)
3666 {
3667         const char *orig_arg = arg;
3668         unsigned val = 0;
3669         while (*arg) {
3670                 if (parse_one_token(&arg, "none"))
3671                         val = 0;
3672                 else if (parse_one_token(&arg, "default"))
3673                         val = WSEH_NEW;
3674                 else if (parse_one_token(&arg, "all"))
3675                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
3676                 else if (parse_one_token(&arg, "new"))
3677                         val |= WSEH_NEW;
3678                 else if (parse_one_token(&arg, "old"))
3679                         val |= WSEH_OLD;
3680                 else if (parse_one_token(&arg, "context"))
3681                         val |= WSEH_CONTEXT;
3682                 else {
3683                         error("unknown value after ws-error-highlight=%.*s",
3684                               (int)(arg - orig_arg), orig_arg);
3685                         return 0;
3686                 }
3687                 if (*arg)
3688                         arg++;
3689         }
3690         opt->ws_error_highlight = val;
3691         return 1;
3692 }
3693
3694 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3695 {
3696         const char *arg = av[0];
3697         const char *optarg;
3698         int argcount;
3699
3700         /* Output format options */
3701         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
3702             || opt_arg(arg, 'U', "unified", &options->context))
3703                 enable_patch_output(&options->output_format);
3704         else if (!strcmp(arg, "--raw"))
3705                 options->output_format |= DIFF_FORMAT_RAW;
3706         else if (!strcmp(arg, "--patch-with-raw")) {
3707                 enable_patch_output(&options->output_format);
3708                 options->output_format |= DIFF_FORMAT_RAW;
3709         } else if (!strcmp(arg, "--numstat"))
3710                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3711         else if (!strcmp(arg, "--shortstat"))
3712                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3713         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3714                 return parse_dirstat_opt(options, "");
3715         else if (skip_prefix(arg, "-X", &arg))
3716                 return parse_dirstat_opt(options, arg);
3717         else if (skip_prefix(arg, "--dirstat=", &arg))
3718                 return parse_dirstat_opt(options, arg);
3719         else if (!strcmp(arg, "--cumulative"))
3720                 return parse_dirstat_opt(options, "cumulative");
3721         else if (!strcmp(arg, "--dirstat-by-file"))
3722                 return parse_dirstat_opt(options, "files");
3723         else if (skip_prefix(arg, "--dirstat-by-file=", &arg)) {
3724                 parse_dirstat_opt(options, "files");
3725                 return parse_dirstat_opt(options, arg);
3726         }
3727         else if (!strcmp(arg, "--check"))
3728                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3729         else if (!strcmp(arg, "--summary"))
3730                 options->output_format |= DIFF_FORMAT_SUMMARY;
3731         else if (!strcmp(arg, "--patch-with-stat")) {
3732                 enable_patch_output(&options->output_format);
3733                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3734         } else if (!strcmp(arg, "--name-only"))
3735                 options->output_format |= DIFF_FORMAT_NAME;
3736         else if (!strcmp(arg, "--name-status"))
3737                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3738         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
3739                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3740         else if (starts_with(arg, "--stat"))
3741                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3742                 return stat_opt(options, av);
3743
3744         /* renames options */
3745         else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
3746                  !strcmp(arg, "--break-rewrites")) {
3747                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3748                         return error("invalid argument to -B: %s", arg+2);
3749         }
3750         else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
3751                  !strcmp(arg, "--find-renames")) {
3752                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3753                         return error("invalid argument to -M: %s", arg+2);
3754                 options->detect_rename = DIFF_DETECT_RENAME;
3755         }
3756         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3757                 options->irreversible_delete = 1;
3758         }
3759         else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
3760                  !strcmp(arg, "--find-copies")) {
3761                 if (options->detect_rename == DIFF_DETECT_COPY)
3762                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3763                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3764                         return error("invalid argument to -C: %s", arg+2);
3765                 options->detect_rename = DIFF_DETECT_COPY;
3766         }
3767         else if (!strcmp(arg, "--no-renames"))
3768                 options->detect_rename = 0;
3769         else if (!strcmp(arg, "--rename-empty"))
3770                 DIFF_OPT_SET(options, RENAME_EMPTY);
3771         else if (!strcmp(arg, "--no-rename-empty"))
3772                 DIFF_OPT_CLR(options, RENAME_EMPTY);
3773         else if (!strcmp(arg, "--relative"))
3774                 DIFF_OPT_SET(options, RELATIVE_NAME);
3775         else if (skip_prefix(arg, "--relative=", &arg)) {
3776                 DIFF_OPT_SET(options, RELATIVE_NAME);
3777                 options->prefix = arg;
3778         }
3779
3780         /* xdiff options */
3781         else if (!strcmp(arg, "--minimal"))
3782                 DIFF_XDL_SET(options, NEED_MINIMAL);
3783         else if (!strcmp(arg, "--no-minimal"))
3784                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3785         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3786                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3787         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3788                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3789         else if (!strcmp(arg, "--ignore-space-at-eol"))
3790                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3791         else if (!strcmp(arg, "--ignore-blank-lines"))
3792                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
3793         else if (!strcmp(arg, "--patience"))
3794                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3795         else if (!strcmp(arg, "--histogram"))
3796                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3797         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
3798                 long value = parse_algorithm_value(optarg);
3799                 if (value < 0)
3800                         return error("option diff-algorithm accepts \"myers\", "
3801                                      "\"minimal\", \"patience\" and \"histogram\"");
3802                 /* clear out previous settings */
3803                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3804                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
3805                 options->xdl_opts |= value;
3806                 return argcount;
3807         }
3808
3809         /* flags options */
3810         else if (!strcmp(arg, "--binary")) {
3811                 enable_patch_output(&options->output_format);
3812                 DIFF_OPT_SET(options, BINARY);
3813         }
3814         else if (!strcmp(arg, "--full-index"))
3815                 DIFF_OPT_SET(options, FULL_INDEX);
3816         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3817                 DIFF_OPT_SET(options, TEXT);
3818         else if (!strcmp(arg, "-R"))
3819                 DIFF_OPT_SET(options, REVERSE_DIFF);
3820         else if (!strcmp(arg, "--find-copies-harder"))
3821                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3822         else if (!strcmp(arg, "--follow"))
3823                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3824         else if (!strcmp(arg, "--no-follow")) {
3825                 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
3826                 DIFF_OPT_CLR(options, DEFAULT_FOLLOW_RENAMES);
3827         } else if (!strcmp(arg, "--color"))
3828                 options->use_color = 1;
3829         else if (skip_prefix(arg, "--color=", &arg)) {
3830                 int value = git_config_colorbool(NULL, arg);
3831                 if (value < 0)
3832                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3833                 options->use_color = value;
3834         }
3835         else if (!strcmp(arg, "--no-color"))
3836                 options->use_color = 0;
3837         else if (!strcmp(arg, "--color-words")) {
3838                 options->use_color = 1;
3839                 options->word_diff = DIFF_WORDS_COLOR;
3840         }
3841         else if (skip_prefix(arg, "--color-words=", &arg)) {
3842                 options->use_color = 1;
3843                 options->word_diff = DIFF_WORDS_COLOR;
3844                 options->word_regex = arg;
3845         }
3846         else if (!strcmp(arg, "--word-diff")) {
3847                 if (options->word_diff == DIFF_WORDS_NONE)
3848                         options->word_diff = DIFF_WORDS_PLAIN;
3849         }
3850         else if (skip_prefix(arg, "--word-diff=", &arg)) {
3851                 if (!strcmp(arg, "plain"))
3852                         options->word_diff = DIFF_WORDS_PLAIN;
3853                 else if (!strcmp(arg, "color")) {
3854                         options->use_color = 1;
3855                         options->word_diff = DIFF_WORDS_COLOR;
3856                 }
3857                 else if (!strcmp(arg, "porcelain"))
3858                         options->word_diff = DIFF_WORDS_PORCELAIN;
3859                 else if (!strcmp(arg, "none"))
3860                         options->word_diff = DIFF_WORDS_NONE;
3861                 else
3862                         die("bad --word-diff argument: %s", arg);
3863         }
3864         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3865                 if (options->word_diff == DIFF_WORDS_NONE)
3866                         options->word_diff = DIFF_WORDS_PLAIN;
3867                 options->word_regex = optarg;
3868                 return argcount;
3869         }
3870         else if (!strcmp(arg, "--exit-code"))
3871                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3872         else if (!strcmp(arg, "--quiet"))
3873                 DIFF_OPT_SET(options, QUICK);
3874         else if (!strcmp(arg, "--ext-diff"))
3875                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3876         else if (!strcmp(arg, "--no-ext-diff"))
3877                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3878         else if (!strcmp(arg, "--textconv"))
3879                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3880         else if (!strcmp(arg, "--no-textconv"))
3881                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3882         else if (!strcmp(arg, "--ignore-submodules")) {
3883                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3884                 handle_ignore_submodules_arg(options, "all");
3885         } else if (skip_prefix(arg, "--ignore-submodules=", &arg)) {
3886                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3887                 handle_ignore_submodules_arg(options, arg);
3888         } else if (!strcmp(arg, "--submodule"))
3889                 DIFF_OPT_SET(options, SUBMODULE_LOG);
3890         else if (skip_prefix(arg, "--submodule=", &arg))
3891                 return parse_submodule_opt(options, arg);
3892         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
3893                 return parse_ws_error_highlight(options, arg);
3894
3895         /* misc options */
3896         else if (!strcmp(arg, "-z"))
3897                 options->line_termination = 0;
3898         else if ((argcount = short_opt('l', av, &optarg))) {
3899                 options->rename_limit = strtoul(optarg, NULL, 10);
3900                 return argcount;
3901         }
3902         else if ((argcount = short_opt('S', av, &optarg))) {
3903                 options->pickaxe = optarg;
3904                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3905                 return argcount;
3906         } else if ((argcount = short_opt('G', av, &optarg))) {
3907                 options->pickaxe = optarg;
3908                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3909                 return argcount;
3910         }
3911         else if (!strcmp(arg, "--pickaxe-all"))
3912                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3913         else if (!strcmp(arg, "--pickaxe-regex"))
3914                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3915         else if ((argcount = short_opt('O', av, &optarg))) {
3916                 options->orderfile = optarg;
3917                 return argcount;
3918         }
3919         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3920                 int offending = parse_diff_filter_opt(optarg, options);
3921                 if (offending)
3922                         die("unknown change class '%c' in --diff-filter=%s",
3923                             offending, optarg);
3924                 return argcount;
3925         }
3926         else if (!strcmp(arg, "--abbrev"))
3927                 options->abbrev = DEFAULT_ABBREV;
3928         else if (skip_prefix(arg, "--abbrev=", &arg)) {
3929                 options->abbrev = strtoul(arg, NULL, 10);
3930                 if (options->abbrev < MINIMUM_ABBREV)
3931                         options->abbrev = MINIMUM_ABBREV;
3932                 else if (40 < options->abbrev)
3933                         options->abbrev = 40;
3934         }
3935         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3936                 options->a_prefix = optarg;
3937                 return argcount;
3938         }
3939         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3940                 options->b_prefix = optarg;
3941                 return argcount;
3942         }
3943         else if (!strcmp(arg, "--no-prefix"))
3944                 options->a_prefix = options->b_prefix = "";
3945         else if (opt_arg(arg, '\0', "inter-hunk-context",
3946                          &options->interhunkcontext))
3947                 ;
3948         else if (!strcmp(arg, "-W"))
3949                 DIFF_OPT_SET(options, FUNCCONTEXT);
3950         else if (!strcmp(arg, "--function-context"))
3951                 DIFF_OPT_SET(options, FUNCCONTEXT);
3952         else if (!strcmp(arg, "--no-function-context"))
3953                 DIFF_OPT_CLR(options, FUNCCONTEXT);
3954         else if ((argcount = parse_long_opt("output", av, &optarg))) {
3955                 options->file = fopen(optarg, "w");
3956                 if (!options->file)
3957                         die_errno("Could not open '%s'", optarg);
3958                 options->close_file = 1;
3959                 return argcount;
3960         } else
3961                 return 0;
3962         return 1;
3963 }
3964
3965 int parse_rename_score(const char **cp_p)
3966 {
3967         unsigned long num, scale;
3968         int ch, dot;
3969         const char *cp = *cp_p;
3970
3971         num = 0;
3972         scale = 1;
3973         dot = 0;
3974         for (;;) {
3975                 ch = *cp;
3976                 if ( !dot && ch == '.' ) {
3977                         scale = 1;
3978                         dot = 1;
3979                 } else if ( ch == '%' ) {
3980                         scale = dot ? scale*100 : 100;
3981                         cp++;   /* % is always at the end */
3982                         break;
3983                 } else if ( ch >= '0' && ch <= '9' ) {
3984                         if ( scale < 100000 ) {
3985                                 scale *= 10;
3986                                 num = (num*10) + (ch-'0');
3987                         }
3988                 } else {
3989                         break;
3990                 }
3991                 cp++;
3992         }
3993         *cp_p = cp;
3994
3995         /* user says num divided by scale and we say internally that
3996          * is MAX_SCORE * num / scale.
3997          */
3998         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
3999 }
4000
4001 static int diff_scoreopt_parse(const char *opt)
4002 {
4003         int opt1, opt2, cmd;
4004
4005         if (*opt++ != '-')
4006                 return -1;
4007         cmd = *opt++;
4008         if (cmd == '-') {
4009                 /* convert the long-form arguments into short-form versions */
4010                 if (skip_prefix(opt, "break-rewrites", &opt)) {
4011                         if (*opt == 0 || *opt++ == '=')
4012                                 cmd = 'B';
4013                 } else if (skip_prefix(opt, "find-copies", &opt)) {
4014                         if (*opt == 0 || *opt++ == '=')
4015                                 cmd = 'C';
4016                 } else if (skip_prefix(opt, "find-renames", &opt)) {
4017                         if (*opt == 0 || *opt++ == '=')
4018                                 cmd = 'M';
4019                 }
4020         }
4021         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
4022                 return -1; /* that is not a -M, -C, or -B option */
4023
4024         opt1 = parse_rename_score(&opt);
4025         if (cmd != 'B')
4026                 opt2 = 0;
4027         else {
4028                 if (*opt == 0)
4029                         opt2 = 0;
4030                 else if (*opt != '/')
4031                         return -1; /* we expect -B80/99 or -B80 */
4032                 else {
4033                         opt++;
4034                         opt2 = parse_rename_score(&opt);
4035                 }
4036         }
4037         if (*opt != 0)
4038                 return -1;
4039         return opt1 | (opt2 << 16);
4040 }
4041
4042 struct diff_queue_struct diff_queued_diff;
4043
4044 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
4045 {
4046         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
4047         queue->queue[queue->nr++] = dp;
4048 }
4049
4050 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
4051                                  struct diff_filespec *one,
4052                                  struct diff_filespec *two)
4053 {
4054         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
4055         dp->one = one;
4056         dp->two = two;
4057         if (queue)
4058                 diff_q(queue, dp);
4059         return dp;
4060 }
4061
4062 void diff_free_filepair(struct diff_filepair *p)
4063 {
4064         free_filespec(p->one);
4065         free_filespec(p->two);
4066         free(p);
4067 }
4068
4069 /* This is different from find_unique_abbrev() in that
4070  * it stuffs the result with dots for alignment.
4071  */
4072 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
4073 {
4074         int abblen;
4075         const char *abbrev;
4076         if (len == 40)
4077                 return sha1_to_hex(sha1);
4078
4079         abbrev = find_unique_abbrev(sha1, len);
4080         abblen = strlen(abbrev);
4081         if (abblen < 37) {
4082                 static char hex[41];
4083                 if (len < abblen && abblen <= len + 2)
4084                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
4085                 else
4086                         sprintf(hex, "%s...", abbrev);
4087                 return hex;
4088         }
4089         return sha1_to_hex(sha1);
4090 }
4091
4092 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4093 {
4094         int line_termination = opt->line_termination;
4095         int inter_name_termination = line_termination ? '\t' : '\0';
4096
4097         fprintf(opt->file, "%s", diff_line_prefix(opt));
4098         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4099                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4100                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
4101                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
4102         }
4103         if (p->score) {
4104                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4105                         inter_name_termination);
4106         } else {
4107                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4108         }
4109
4110         if (p->status == DIFF_STATUS_COPIED ||
4111             p->status == DIFF_STATUS_RENAMED) {
4112                 const char *name_a, *name_b;
4113                 name_a = p->one->path;
4114                 name_b = p->two->path;
4115                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4116                 write_name_quoted(name_a, opt->file, inter_name_termination);
4117                 write_name_quoted(name_b, opt->file, line_termination);
4118         } else {
4119                 const char *name_a, *name_b;
4120                 name_a = p->one->mode ? p->one->path : p->two->path;
4121                 name_b = NULL;
4122                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4123                 write_name_quoted(name_a, opt->file, line_termination);
4124         }
4125 }
4126
4127 int diff_unmodified_pair(struct diff_filepair *p)
4128 {
4129         /* This function is written stricter than necessary to support
4130          * the currently implemented transformers, but the idea is to
4131          * let transformers to produce diff_filepairs any way they want,
4132          * and filter and clean them up here before producing the output.
4133          */
4134         struct diff_filespec *one = p->one, *two = p->two;
4135
4136         if (DIFF_PAIR_UNMERGED(p))
4137                 return 0; /* unmerged is interesting */
4138
4139         /* deletion, addition, mode or type change
4140          * and rename are all interesting.
4141          */
4142         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4143             DIFF_PAIR_MODE_CHANGED(p) ||
4144             strcmp(one->path, two->path))
4145                 return 0;
4146
4147         /* both are valid and point at the same path.  that is, we are
4148          * dealing with a change.
4149          */
4150         if (one->sha1_valid && two->sha1_valid &&
4151             !hashcmp(one->sha1, two->sha1) &&
4152             !one->dirty_submodule && !two->dirty_submodule)
4153                 return 1; /* no change */
4154         if (!one->sha1_valid && !two->sha1_valid)
4155                 return 1; /* both look at the same file on the filesystem. */
4156         return 0;
4157 }
4158
4159 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4160 {
4161         if (diff_unmodified_pair(p))
4162                 return;
4163
4164         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4165             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4166                 return; /* no tree diffs in patch format */
4167
4168         run_diff(p, o);
4169 }
4170
4171 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4172                             struct diffstat_t *diffstat)
4173 {
4174         if (diff_unmodified_pair(p))
4175                 return;
4176
4177         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4178             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4179                 return; /* no useful stat for tree diffs */
4180
4181         run_diffstat(p, o, diffstat);
4182 }
4183
4184 static void diff_flush_checkdiff(struct diff_filepair *p,
4185                 struct diff_options *o)
4186 {
4187         if (diff_unmodified_pair(p))
4188                 return;
4189
4190         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4191             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4192                 return; /* nothing to check in tree diffs */
4193
4194         run_checkdiff(p, o);
4195 }
4196
4197 int diff_queue_is_empty(void)
4198 {
4199         struct diff_queue_struct *q = &diff_queued_diff;
4200         int i;
4201         for (i = 0; i < q->nr; i++)
4202                 if (!diff_unmodified_pair(q->queue[i]))
4203                         return 0;
4204         return 1;
4205 }
4206
4207 #if DIFF_DEBUG
4208 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4209 {
4210         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4211                 x, one ? one : "",
4212                 s->path,
4213                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
4214                 s->mode,
4215                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
4216         fprintf(stderr, "queue[%d] %s size %lu\n",
4217                 x, one ? one : "",
4218                 s->size);
4219 }
4220
4221 void diff_debug_filepair(const struct diff_filepair *p, int i)
4222 {
4223         diff_debug_filespec(p->one, i, "one");
4224         diff_debug_filespec(p->two, i, "two");
4225         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4226                 p->score, p->status ? p->status : '?',
4227                 p->one->rename_used, p->broken_pair);
4228 }
4229
4230 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4231 {
4232         int i;
4233         if (msg)
4234                 fprintf(stderr, "%s\n", msg);
4235         fprintf(stderr, "q->nr = %d\n", q->nr);
4236         for (i = 0; i < q->nr; i++) {
4237                 struct diff_filepair *p = q->queue[i];
4238                 diff_debug_filepair(p, i);
4239         }
4240 }
4241 #endif
4242
4243 static void diff_resolve_rename_copy(void)
4244 {
4245         int i;
4246         struct diff_filepair *p;
4247         struct diff_queue_struct *q = &diff_queued_diff;
4248
4249         diff_debug_queue("resolve-rename-copy", q);
4250
4251         for (i = 0; i < q->nr; i++) {
4252                 p = q->queue[i];
4253                 p->status = 0; /* undecided */
4254                 if (DIFF_PAIR_UNMERGED(p))
4255                         p->status = DIFF_STATUS_UNMERGED;
4256                 else if (!DIFF_FILE_VALID(p->one))
4257                         p->status = DIFF_STATUS_ADDED;
4258                 else if (!DIFF_FILE_VALID(p->two))
4259                         p->status = DIFF_STATUS_DELETED;
4260                 else if (DIFF_PAIR_TYPE_CHANGED(p))
4261                         p->status = DIFF_STATUS_TYPE_CHANGED;
4262
4263                 /* from this point on, we are dealing with a pair
4264                  * whose both sides are valid and of the same type, i.e.
4265                  * either in-place edit or rename/copy edit.
4266                  */
4267                 else if (DIFF_PAIR_RENAME(p)) {
4268                         /*
4269                          * A rename might have re-connected a broken
4270                          * pair up, causing the pathnames to be the
4271                          * same again. If so, that's not a rename at
4272                          * all, just a modification..
4273                          *
4274                          * Otherwise, see if this source was used for
4275                          * multiple renames, in which case we decrement
4276                          * the count, and call it a copy.
4277                          */
4278                         if (!strcmp(p->one->path, p->two->path))
4279                                 p->status = DIFF_STATUS_MODIFIED;
4280                         else if (--p->one->rename_used > 0)
4281                                 p->status = DIFF_STATUS_COPIED;
4282                         else
4283                                 p->status = DIFF_STATUS_RENAMED;
4284                 }
4285                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
4286                          p->one->mode != p->two->mode ||
4287                          p->one->dirty_submodule ||
4288                          p->two->dirty_submodule ||
4289                          is_null_sha1(p->one->sha1))
4290                         p->status = DIFF_STATUS_MODIFIED;
4291                 else {
4292                         /* This is a "no-change" entry and should not
4293                          * happen anymore, but prepare for broken callers.
4294                          */
4295                         error("feeding unmodified %s to diffcore",
4296                               p->one->path);
4297                         p->status = DIFF_STATUS_UNKNOWN;
4298                 }
4299         }
4300         diff_debug_queue("resolve-rename-copy done", q);
4301 }
4302
4303 static int check_pair_status(struct diff_filepair *p)
4304 {
4305         switch (p->status) {
4306         case DIFF_STATUS_UNKNOWN:
4307                 return 0;
4308         case 0:
4309                 die("internal error in diff-resolve-rename-copy");
4310         default:
4311                 return 1;
4312         }
4313 }
4314
4315 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4316 {
4317         int fmt = opt->output_format;
4318
4319         if (fmt & DIFF_FORMAT_CHECKDIFF)
4320                 diff_flush_checkdiff(p, opt);
4321         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4322                 diff_flush_raw(p, opt);
4323         else if (fmt & DIFF_FORMAT_NAME) {
4324                 const char *name_a, *name_b;
4325                 name_a = p->two->path;
4326                 name_b = NULL;
4327                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4328                 write_name_quoted(name_a, opt->file, opt->line_termination);
4329         }
4330 }
4331
4332 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4333 {
4334         if (fs->mode)
4335                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4336         else
4337                 fprintf(file, " %s ", newdelete);
4338         write_name_quoted(fs->path, file, '\n');
4339 }
4340
4341
4342 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4343                 const char *line_prefix)
4344 {
4345         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4346                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4347                         p->two->mode, show_name ? ' ' : '\n');
4348                 if (show_name) {
4349                         write_name_quoted(p->two->path, file, '\n');
4350                 }
4351         }
4352 }
4353
4354 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4355                         const char *line_prefix)
4356 {
4357         char *names = pprint_rename(p->one->path, p->two->path);
4358
4359         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4360         free(names);
4361         show_mode_change(file, p, 0, line_prefix);
4362 }
4363
4364 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4365 {
4366         FILE *file = opt->file;
4367         const char *line_prefix = diff_line_prefix(opt);
4368
4369         switch(p->status) {
4370         case DIFF_STATUS_DELETED:
4371                 fputs(line_prefix, file);
4372                 show_file_mode_name(file, "delete", p->one);
4373                 break;
4374         case DIFF_STATUS_ADDED:
4375                 fputs(line_prefix, file);
4376                 show_file_mode_name(file, "create", p->two);
4377                 break;
4378         case DIFF_STATUS_COPIED:
4379                 fputs(line_prefix, file);
4380                 show_rename_copy(file, "copy", p, line_prefix);
4381                 break;
4382         case DIFF_STATUS_RENAMED:
4383                 fputs(line_prefix, file);
4384                 show_rename_copy(file, "rename", p, line_prefix);
4385                 break;
4386         default:
4387                 if (p->score) {
4388                         fprintf(file, "%s rewrite ", line_prefix);
4389                         write_name_quoted(p->two->path, file, ' ');
4390                         fprintf(file, "(%d%%)\n", similarity_index(p));
4391                 }
4392                 show_mode_change(file, p, !p->score, line_prefix);
4393                 break;
4394         }
4395 }
4396
4397 struct patch_id_t {
4398         git_SHA_CTX *ctx;
4399         int patchlen;
4400 };
4401
4402 static int remove_space(char *line, int len)
4403 {
4404         int i;
4405         char *dst = line;
4406         unsigned char c;
4407
4408         for (i = 0; i < len; i++)
4409                 if (!isspace((c = line[i])))
4410                         *dst++ = c;
4411
4412         return dst - line;
4413 }
4414
4415 static void patch_id_consume(void *priv, char *line, unsigned long len)
4416 {
4417         struct patch_id_t *data = priv;
4418         int new_len;
4419
4420         /* Ignore line numbers when computing the SHA1 of the patch */
4421         if (starts_with(line, "@@ -"))
4422                 return;
4423
4424         new_len = remove_space(line, len);
4425
4426         git_SHA1_Update(data->ctx, line, new_len);
4427         data->patchlen += new_len;
4428 }
4429
4430 /* returns 0 upon success, and writes result into sha1 */
4431 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
4432 {
4433         struct diff_queue_struct *q = &diff_queued_diff;
4434         int i;
4435         git_SHA_CTX ctx;
4436         struct patch_id_t data;
4437         char buffer[PATH_MAX * 4 + 20];
4438
4439         git_SHA1_Init(&ctx);
4440         memset(&data, 0, sizeof(struct patch_id_t));
4441         data.ctx = &ctx;
4442
4443         for (i = 0; i < q->nr; i++) {
4444                 xpparam_t xpp;
4445                 xdemitconf_t xecfg;
4446                 mmfile_t mf1, mf2;
4447                 struct diff_filepair *p = q->queue[i];
4448                 int len1, len2;
4449
4450                 memset(&xpp, 0, sizeof(xpp));
4451                 memset(&xecfg, 0, sizeof(xecfg));
4452                 if (p->status == 0)
4453                         return error("internal diff status error");
4454                 if (p->status == DIFF_STATUS_UNKNOWN)
4455                         continue;
4456                 if (diff_unmodified_pair(p))
4457                         continue;
4458                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4459                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4460                         continue;
4461                 if (DIFF_PAIR_UNMERGED(p))
4462                         continue;
4463
4464                 diff_fill_sha1_info(p->one);
4465                 diff_fill_sha1_info(p->two);
4466                 if (fill_mmfile(&mf1, p->one) < 0 ||
4467                                 fill_mmfile(&mf2, p->two) < 0)
4468                         return error("unable to read files to diff");
4469
4470                 len1 = remove_space(p->one->path, strlen(p->one->path));
4471                 len2 = remove_space(p->two->path, strlen(p->two->path));
4472                 if (p->one->mode == 0)
4473                         len1 = snprintf(buffer, sizeof(buffer),
4474                                         "diff--gita/%.*sb/%.*s"
4475                                         "newfilemode%06o"
4476                                         "---/dev/null"
4477                                         "+++b/%.*s",
4478                                         len1, p->one->path,
4479                                         len2, p->two->path,
4480                                         p->two->mode,
4481                                         len2, p->two->path);
4482                 else if (p->two->mode == 0)
4483                         len1 = snprintf(buffer, sizeof(buffer),
4484                                         "diff--gita/%.*sb/%.*s"
4485                                         "deletedfilemode%06o"
4486                                         "---a/%.*s"
4487                                         "+++/dev/null",
4488                                         len1, p->one->path,
4489                                         len2, p->two->path,
4490                                         p->one->mode,
4491                                         len1, p->one->path);
4492                 else
4493                         len1 = snprintf(buffer, sizeof(buffer),
4494                                         "diff--gita/%.*sb/%.*s"
4495                                         "---a/%.*s"
4496                                         "+++b/%.*s",
4497                                         len1, p->one->path,
4498                                         len2, p->two->path,
4499                                         len1, p->one->path,
4500                                         len2, p->two->path);
4501                 git_SHA1_Update(&ctx, buffer, len1);
4502
4503                 if (diff_filespec_is_binary(p->one) ||
4504                     diff_filespec_is_binary(p->two)) {
4505                         git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
4506                         git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
4507                         continue;
4508                 }
4509
4510                 xpp.flags = 0;
4511                 xecfg.ctxlen = 3;
4512                 xecfg.flags = 0;
4513                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4514                               &xpp, &xecfg);
4515         }
4516
4517         git_SHA1_Final(sha1, &ctx);
4518         return 0;
4519 }
4520
4521 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
4522 {
4523         struct diff_queue_struct *q = &diff_queued_diff;
4524         int i;
4525         int result = diff_get_patch_id(options, sha1);
4526
4527         for (i = 0; i < q->nr; i++)
4528                 diff_free_filepair(q->queue[i]);
4529
4530         free(q->queue);
4531         DIFF_QUEUE_CLEAR(q);
4532
4533         return result;
4534 }
4535
4536 static int is_summary_empty(const struct diff_queue_struct *q)
4537 {
4538         int i;
4539
4540         for (i = 0; i < q->nr; i++) {
4541                 const struct diff_filepair *p = q->queue[i];
4542
4543                 switch (p->status) {
4544                 case DIFF_STATUS_DELETED:
4545                 case DIFF_STATUS_ADDED:
4546                 case DIFF_STATUS_COPIED:
4547                 case DIFF_STATUS_RENAMED:
4548                         return 0;
4549                 default:
4550                         if (p->score)
4551                                 return 0;
4552                         if (p->one->mode && p->two->mode &&
4553                             p->one->mode != p->two->mode)
4554                                 return 0;
4555                         break;
4556                 }
4557         }
4558         return 1;
4559 }
4560
4561 static const char rename_limit_warning[] =
4562 "inexact rename detection was skipped due to too many files.";
4563
4564 static const char degrade_cc_to_c_warning[] =
4565 "only found copies from modified paths due to too many files.";
4566
4567 static const char rename_limit_advice[] =
4568 "you may want to set your %s variable to at least "
4569 "%d and retry the command.";
4570
4571 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4572 {
4573         if (degraded_cc)
4574                 warning(degrade_cc_to_c_warning);
4575         else if (needed)
4576                 warning(rename_limit_warning);
4577         else
4578                 return;
4579         if (0 < needed && needed < 32767)
4580                 warning(rename_limit_advice, varname, needed);
4581 }
4582
4583 void diff_flush(struct diff_options *options)
4584 {
4585         struct diff_queue_struct *q = &diff_queued_diff;
4586         int i, output_format = options->output_format;
4587         int separator = 0;
4588         int dirstat_by_line = 0;
4589
4590         /*
4591          * Order: raw, stat, summary, patch
4592          * or:    name/name-status/checkdiff (other bits clear)
4593          */
4594         if (!q->nr)
4595                 goto free_queue;
4596
4597         if (output_format & (DIFF_FORMAT_RAW |
4598                              DIFF_FORMAT_NAME |
4599                              DIFF_FORMAT_NAME_STATUS |
4600                              DIFF_FORMAT_CHECKDIFF)) {
4601                 for (i = 0; i < q->nr; i++) {
4602                         struct diff_filepair *p = q->queue[i];
4603                         if (check_pair_status(p))
4604                                 flush_one_pair(p, options);
4605                 }
4606                 separator++;
4607         }
4608
4609         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4610                 dirstat_by_line = 1;
4611
4612         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4613             dirstat_by_line) {
4614                 struct diffstat_t diffstat;
4615
4616                 memset(&diffstat, 0, sizeof(struct diffstat_t));
4617                 for (i = 0; i < q->nr; i++) {
4618                         struct diff_filepair *p = q->queue[i];
4619                         if (check_pair_status(p))
4620                                 diff_flush_stat(p, options, &diffstat);
4621                 }
4622                 if (output_format & DIFF_FORMAT_NUMSTAT)
4623                         show_numstat(&diffstat, options);
4624                 if (output_format & DIFF_FORMAT_DIFFSTAT)
4625                         show_stats(&diffstat, options);
4626                 if (output_format & DIFF_FORMAT_SHORTSTAT)
4627                         show_shortstats(&diffstat, options);
4628                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
4629                         show_dirstat_by_line(&diffstat, options);
4630                 free_diffstat_info(&diffstat);
4631                 separator++;
4632         }
4633         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4634                 show_dirstat(options);
4635
4636         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4637                 for (i = 0; i < q->nr; i++) {
4638                         diff_summary(options, q->queue[i]);
4639                 }
4640                 separator++;
4641         }
4642
4643         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4644             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4645             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4646                 /*
4647                  * run diff_flush_patch for the exit status. setting
4648                  * options->file to /dev/null should be safe, because we
4649                  * aren't supposed to produce any output anyway.
4650                  */
4651                 if (options->close_file)
4652                         fclose(options->file);
4653                 options->file = fopen("/dev/null", "w");
4654                 if (!options->file)
4655                         die_errno("Could not open /dev/null");
4656                 options->close_file = 1;
4657                 for (i = 0; i < q->nr; i++) {
4658                         struct diff_filepair *p = q->queue[i];
4659                         if (check_pair_status(p))
4660                                 diff_flush_patch(p, options);
4661                         if (options->found_changes)
4662                                 break;
4663                 }
4664         }
4665
4666         if (output_format & DIFF_FORMAT_PATCH) {
4667                 if (separator) {
4668                         fprintf(options->file, "%s%c",
4669                                 diff_line_prefix(options),
4670                                 options->line_termination);
4671                         if (options->stat_sep) {
4672                                 /* attach patch instead of inline */
4673                                 fputs(options->stat_sep, options->file);
4674                         }
4675                 }
4676
4677                 for (i = 0; i < q->nr; i++) {
4678                         struct diff_filepair *p = q->queue[i];
4679                         if (check_pair_status(p))
4680                                 diff_flush_patch(p, options);
4681                 }
4682         }
4683
4684         if (output_format & DIFF_FORMAT_CALLBACK)
4685                 options->format_callback(q, options, options->format_callback_data);
4686
4687         for (i = 0; i < q->nr; i++)
4688                 diff_free_filepair(q->queue[i]);
4689 free_queue:
4690         free(q->queue);
4691         DIFF_QUEUE_CLEAR(q);
4692         if (options->close_file)
4693                 fclose(options->file);
4694
4695         /*
4696          * Report the content-level differences with HAS_CHANGES;
4697          * diff_addremove/diff_change does not set the bit when
4698          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4699          */
4700         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4701                 if (options->found_changes)
4702                         DIFF_OPT_SET(options, HAS_CHANGES);
4703                 else
4704                         DIFF_OPT_CLR(options, HAS_CHANGES);
4705         }
4706 }
4707
4708 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
4709 {
4710         return (((p->status == DIFF_STATUS_MODIFIED) &&
4711                  ((p->score &&
4712                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
4713                   (!p->score &&
4714                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
4715                 ((p->status != DIFF_STATUS_MODIFIED) &&
4716                  filter_bit_tst(p->status, options)));
4717 }
4718
4719 static void diffcore_apply_filter(struct diff_options *options)
4720 {
4721         int i;
4722         struct diff_queue_struct *q = &diff_queued_diff;
4723         struct diff_queue_struct outq;
4724
4725         DIFF_QUEUE_CLEAR(&outq);
4726
4727         if (!options->filter)
4728                 return;
4729
4730         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
4731                 int found;
4732                 for (i = found = 0; !found && i < q->nr; i++) {
4733                         if (match_filter(options, q->queue[i]))
4734                                 found++;
4735                 }
4736                 if (found)
4737                         return;
4738
4739                 /* otherwise we will clear the whole queue
4740                  * by copying the empty outq at the end of this
4741                  * function, but first clear the current entries
4742                  * in the queue.
4743                  */
4744                 for (i = 0; i < q->nr; i++)
4745                         diff_free_filepair(q->queue[i]);
4746         }
4747         else {
4748                 /* Only the matching ones */
4749                 for (i = 0; i < q->nr; i++) {
4750                         struct diff_filepair *p = q->queue[i];
4751                         if (match_filter(options, p))
4752                                 diff_q(&outq, p);
4753                         else
4754                                 diff_free_filepair(p);
4755                 }
4756         }
4757         free(q->queue);
4758         *q = outq;
4759 }
4760
4761 /* Check whether two filespecs with the same mode and size are identical */
4762 static int diff_filespec_is_identical(struct diff_filespec *one,
4763                                       struct diff_filespec *two)
4764 {
4765         if (S_ISGITLINK(one->mode))
4766                 return 0;
4767         if (diff_populate_filespec(one, 0))
4768                 return 0;
4769         if (diff_populate_filespec(two, 0))
4770                 return 0;
4771         return !memcmp(one->data, two->data, one->size);
4772 }
4773
4774 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
4775 {
4776         if (p->done_skip_stat_unmatch)
4777                 return p->skip_stat_unmatch_result;
4778
4779         p->done_skip_stat_unmatch = 1;
4780         p->skip_stat_unmatch_result = 0;
4781         /*
4782          * 1. Entries that come from stat info dirtiness
4783          *    always have both sides (iow, not create/delete),
4784          *    one side of the object name is unknown, with
4785          *    the same mode and size.  Keep the ones that
4786          *    do not match these criteria.  They have real
4787          *    differences.
4788          *
4789          * 2. At this point, the file is known to be modified,
4790          *    with the same mode and size, and the object
4791          *    name of one side is unknown.  Need to inspect
4792          *    the identical contents.
4793          */
4794         if (!DIFF_FILE_VALID(p->one) || /* (1) */
4795             !DIFF_FILE_VALID(p->two) ||
4796             (p->one->sha1_valid && p->two->sha1_valid) ||
4797             (p->one->mode != p->two->mode) ||
4798             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
4799             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
4800             (p->one->size != p->two->size) ||
4801             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4802                 p->skip_stat_unmatch_result = 1;
4803         return p->skip_stat_unmatch_result;
4804 }
4805
4806 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4807 {
4808         int i;
4809         struct diff_queue_struct *q = &diff_queued_diff;
4810         struct diff_queue_struct outq;
4811         DIFF_QUEUE_CLEAR(&outq);
4812
4813         for (i = 0; i < q->nr; i++) {
4814                 struct diff_filepair *p = q->queue[i];
4815
4816                 if (diff_filespec_check_stat_unmatch(p))
4817                         diff_q(&outq, p);
4818                 else {
4819                         /*
4820                          * The caller can subtract 1 from skip_stat_unmatch
4821                          * to determine how many paths were dirty only
4822                          * due to stat info mismatch.
4823                          */
4824                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4825                                 diffopt->skip_stat_unmatch++;
4826                         diff_free_filepair(p);
4827                 }
4828         }
4829         free(q->queue);
4830         *q = outq;
4831 }
4832
4833 static int diffnamecmp(const void *a_, const void *b_)
4834 {
4835         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4836         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4837         const char *name_a, *name_b;
4838
4839         name_a = a->one ? a->one->path : a->two->path;
4840         name_b = b->one ? b->one->path : b->two->path;
4841         return strcmp(name_a, name_b);
4842 }
4843
4844 void diffcore_fix_diff_index(struct diff_options *options)
4845 {
4846         struct diff_queue_struct *q = &diff_queued_diff;
4847         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4848 }
4849
4850 void diffcore_std(struct diff_options *options)
4851 {
4852         /* NOTE please keep the following in sync with diff_tree_combined() */
4853         if (options->skip_stat_unmatch)
4854                 diffcore_skip_stat_unmatch(options);
4855         if (!options->found_follow) {
4856                 /* See try_to_follow_renames() in tree-diff.c */
4857                 if (options->break_opt != -1)
4858                         diffcore_break(options->break_opt);
4859                 if (options->detect_rename)
4860                         diffcore_rename(options);
4861                 if (options->break_opt != -1)
4862                         diffcore_merge_broken();
4863         }
4864         if (options->pickaxe)
4865                 diffcore_pickaxe(options);
4866         if (options->orderfile)
4867                 diffcore_order(options->orderfile);
4868         if (!options->found_follow)
4869                 /* See try_to_follow_renames() in tree-diff.c */
4870                 diff_resolve_rename_copy();
4871         diffcore_apply_filter(options);
4872
4873         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4874                 DIFF_OPT_SET(options, HAS_CHANGES);
4875         else
4876                 DIFF_OPT_CLR(options, HAS_CHANGES);
4877
4878         options->found_follow = 0;
4879 }
4880
4881 int diff_result_code(struct diff_options *opt, int status)
4882 {
4883         int result = 0;
4884
4885         diff_warn_rename_limit("diff.renameLimit",
4886                                opt->needed_rename_limit,
4887                                opt->degraded_cc_to_c);
4888         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4889             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4890                 return status;
4891         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4892             DIFF_OPT_TST(opt, HAS_CHANGES))
4893                 result |= 01;
4894         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4895             DIFF_OPT_TST(opt, CHECK_FAILED))
4896                 result |= 02;
4897         return result;
4898 }
4899
4900 int diff_can_quit_early(struct diff_options *opt)
4901 {
4902         return (DIFF_OPT_TST(opt, QUICK) &&
4903                 !opt->filter &&
4904                 DIFF_OPT_TST(opt, HAS_CHANGES));
4905 }
4906
4907 /*
4908  * Shall changes to this submodule be ignored?
4909  *
4910  * Submodule changes can be configured to be ignored separately for each path,
4911  * but that configuration can be overridden from the command line.
4912  */
4913 static int is_submodule_ignored(const char *path, struct diff_options *options)
4914 {
4915         int ignored = 0;
4916         unsigned orig_flags = options->flags;
4917         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4918                 set_diffopt_flags_from_submodule_config(options, path);
4919         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4920                 ignored = 1;
4921         options->flags = orig_flags;
4922         return ignored;
4923 }
4924
4925 void diff_addremove(struct diff_options *options,
4926                     int addremove, unsigned mode,
4927                     const unsigned char *sha1,
4928                     int sha1_valid,
4929                     const char *concatpath, unsigned dirty_submodule)
4930 {
4931         struct diff_filespec *one, *two;
4932
4933         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4934                 return;
4935
4936         /* This may look odd, but it is a preparation for
4937          * feeding "there are unchanged files which should
4938          * not produce diffs, but when you are doing copy
4939          * detection you would need them, so here they are"
4940          * entries to the diff-core.  They will be prefixed
4941          * with something like '=' or '*' (I haven't decided
4942          * which but should not make any difference).
4943          * Feeding the same new and old to diff_change()
4944          * also has the same effect.
4945          * Before the final output happens, they are pruned after
4946          * merged into rename/copy pairs as appropriate.
4947          */
4948         if (DIFF_OPT_TST(options, REVERSE_DIFF))
4949                 addremove = (addremove == '+' ? '-' :
4950                              addremove == '-' ? '+' : addremove);
4951
4952         if (options->prefix &&
4953             strncmp(concatpath, options->prefix, options->prefix_length))
4954                 return;
4955
4956         one = alloc_filespec(concatpath);
4957         two = alloc_filespec(concatpath);
4958
4959         if (addremove != '+')
4960                 fill_filespec(one, sha1, sha1_valid, mode);
4961         if (addremove != '-') {
4962                 fill_filespec(two, sha1, sha1_valid, mode);
4963                 two->dirty_submodule = dirty_submodule;
4964         }
4965
4966         diff_queue(&diff_queued_diff, one, two);
4967         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4968                 DIFF_OPT_SET(options, HAS_CHANGES);
4969 }
4970
4971 void diff_change(struct diff_options *options,
4972                  unsigned old_mode, unsigned new_mode,
4973                  const unsigned char *old_sha1,
4974                  const unsigned char *new_sha1,
4975                  int old_sha1_valid, int new_sha1_valid,
4976                  const char *concatpath,
4977                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4978 {
4979         struct diff_filespec *one, *two;
4980         struct diff_filepair *p;
4981
4982         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4983             is_submodule_ignored(concatpath, options))
4984                 return;
4985
4986         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4987                 unsigned tmp;
4988                 const unsigned char *tmp_c;
4989                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4990                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4991                 tmp = old_sha1_valid; old_sha1_valid = new_sha1_valid;
4992                         new_sha1_valid = tmp;
4993                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
4994                         new_dirty_submodule = tmp;
4995         }
4996
4997         if (options->prefix &&
4998             strncmp(concatpath, options->prefix, options->prefix_length))
4999                 return;
5000
5001         one = alloc_filespec(concatpath);
5002         two = alloc_filespec(concatpath);
5003         fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
5004         fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
5005         one->dirty_submodule = old_dirty_submodule;
5006         two->dirty_submodule = new_dirty_submodule;
5007         p = diff_queue(&diff_queued_diff, one, two);
5008
5009         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5010                 return;
5011
5012         if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
5013             !diff_filespec_check_stat_unmatch(p))
5014                 return;
5015
5016         DIFF_OPT_SET(options, HAS_CHANGES);
5017 }
5018
5019 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
5020 {
5021         struct diff_filepair *pair;
5022         struct diff_filespec *one, *two;
5023
5024         if (options->prefix &&
5025             strncmp(path, options->prefix, options->prefix_length))
5026                 return NULL;
5027
5028         one = alloc_filespec(path);
5029         two = alloc_filespec(path);
5030         pair = diff_queue(&diff_queued_diff, one, two);
5031         pair->is_unmerged = 1;
5032         return pair;
5033 }
5034
5035 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
5036                 size_t *outsize)
5037 {
5038         struct diff_tempfile *temp;
5039         const char *argv[3];
5040         const char **arg = argv;
5041         struct child_process child = CHILD_PROCESS_INIT;
5042         struct strbuf buf = STRBUF_INIT;
5043         int err = 0;
5044
5045         temp = prepare_temp_file(spec->path, spec);
5046         *arg++ = pgm;
5047         *arg++ = temp->name;
5048         *arg = NULL;
5049
5050         child.use_shell = 1;
5051         child.argv = argv;
5052         child.out = -1;
5053         if (start_command(&child)) {
5054                 remove_tempfile();
5055                 return NULL;
5056         }
5057
5058         if (strbuf_read(&buf, child.out, 0) < 0)
5059                 err = error("error reading from textconv command '%s'", pgm);
5060         close(child.out);
5061
5062         if (finish_command(&child) || err) {
5063                 strbuf_release(&buf);
5064                 remove_tempfile();
5065                 return NULL;
5066         }
5067         remove_tempfile();
5068
5069         return strbuf_detach(&buf, outsize);
5070 }
5071
5072 size_t fill_textconv(struct userdiff_driver *driver,
5073                      struct diff_filespec *df,
5074                      char **outbuf)
5075 {
5076         size_t size;
5077
5078         if (!driver || !driver->textconv) {
5079                 if (!DIFF_FILE_VALID(df)) {
5080                         *outbuf = "";
5081                         return 0;
5082                 }
5083                 if (diff_populate_filespec(df, 0))
5084                         die("unable to read files to diff");
5085                 *outbuf = df->data;
5086                 return df->size;
5087         }
5088
5089         if (driver->textconv_cache && df->sha1_valid) {
5090                 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
5091                                           &size);
5092                 if (*outbuf)
5093                         return size;
5094         }
5095
5096         *outbuf = run_textconv(driver->textconv, df, &size);
5097         if (!*outbuf)
5098                 die("unable to read files to diff");
5099
5100         if (driver->textconv_cache && df->sha1_valid) {
5101                 /* ignore errors, as we might be in a readonly repository */
5102                 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
5103                                 size);
5104                 /*
5105                  * we could save up changes and flush them all at the end,
5106                  * but we would need an extra call after all diffing is done.
5107                  * Since generating a cache entry is the slow path anyway,
5108                  * this extra overhead probably isn't a big deal.
5109                  */
5110                 notes_cache_write(driver->textconv_cache);
5111         }
5112
5113         return size;
5114 }
5115
5116 void setup_diff_pager(struct diff_options *opt)
5117 {
5118         /*
5119          * If the user asked for our exit code, then either they want --quiet
5120          * or --exit-code. We should definitely not bother with a pager in the
5121          * former case, as we will generate no output. Since we still properly
5122          * report our exit code even when a pager is run, we _could_ run a
5123          * pager with --exit-code. But since we have not done so historically,
5124          * and because it is easy to find people oneline advising "git diff
5125          * --exit-code" in hooks and other scripts, we do not do so.
5126          */
5127         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5128             check_pager_config("diff") != 0)
5129                 setup_pager();
5130 }