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