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