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

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