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                         /*
1121                          * Skip the prefix character, if any.  With
1122                          * diff_suppress_blank_empty, there may be
1123                          * none.
1124                          */
1125                         if (line[0] != '\n') {
1126                               line++;
1127                               len--;
1128                         }
1129                         emit_line(ecbdata->opt, plain, reset, line, len);
1130                 }
1131                 return;
1132         }
1133
1134         if (line[0] != '+') {
1135                 const char *color =
1136                         diff_get_color(ecbdata->color_diff,
1137                                        line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
1138                 ecbdata->lno_in_preimage++;
1139                 if (line[0] == ' ')
1140                         ecbdata->lno_in_postimage++;
1141                 emit_line(ecbdata->opt, color, reset, line, len);
1142         } else {
1143                 ecbdata->lno_in_postimage++;
1144                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1145         }
1146 }
1147
1148 static char *pprint_rename(const char *a, const char *b)
1149 {
1150         const char *old = a;
1151         const char *new = b;
1152         struct strbuf name = STRBUF_INIT;
1153         int pfx_length, sfx_length;
1154         int len_a = strlen(a);
1155         int len_b = strlen(b);
1156         int a_midlen, b_midlen;
1157         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1158         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1159
1160         if (qlen_a || qlen_b) {
1161                 quote_c_style(a, &name, NULL, 0);
1162                 strbuf_addstr(&name, " => ");
1163                 quote_c_style(b, &name, NULL, 0);
1164                 return strbuf_detach(&name, NULL);
1165         }
1166
1167         /* Find common prefix */
1168         pfx_length = 0;
1169         while (*old && *new && *old == *new) {
1170                 if (*old == '/')
1171                         pfx_length = old - a + 1;
1172                 old++;
1173                 new++;
1174         }
1175
1176         /* Find common suffix */
1177         old = a + len_a;
1178         new = b + len_b;
1179         sfx_length = 0;
1180         while (a <= old && b <= new && *old == *new) {
1181                 if (*old == '/')
1182                         sfx_length = len_a - (old - a);
1183                 old--;
1184                 new--;
1185         }
1186
1187         /*
1188          * pfx{mid-a => mid-b}sfx
1189          * {pfx-a => pfx-b}sfx
1190          * pfx{sfx-a => sfx-b}
1191          * name-a => name-b
1192          */
1193         a_midlen = len_a - pfx_length - sfx_length;
1194         b_midlen = len_b - pfx_length - sfx_length;
1195         if (a_midlen < 0)
1196                 a_midlen = 0;
1197         if (b_midlen < 0)
1198                 b_midlen = 0;
1199
1200         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1201         if (pfx_length + sfx_length) {
1202                 strbuf_add(&name, a, pfx_length);
1203                 strbuf_addch(&name, '{');
1204         }
1205         strbuf_add(&name, a + pfx_length, a_midlen);
1206         strbuf_addstr(&name, " => ");
1207         strbuf_add(&name, b + pfx_length, b_midlen);
1208         if (pfx_length + sfx_length) {
1209                 strbuf_addch(&name, '}');
1210                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1211         }
1212         return strbuf_detach(&name, NULL);
1213 }
1214
1215 struct diffstat_t {
1216         int nr;
1217         int alloc;
1218         struct diffstat_file {
1219                 char *from_name;
1220                 char *name;
1221                 char *print_name;
1222                 unsigned is_unmerged:1;
1223                 unsigned is_binary:1;
1224                 unsigned is_renamed:1;
1225                 uintmax_t added, deleted;
1226         } **files;
1227 };
1228
1229 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1230                                           const char *name_a,
1231                                           const char *name_b)
1232 {
1233         struct diffstat_file *x;
1234         x = xcalloc(sizeof (*x), 1);
1235         if (diffstat->nr == diffstat->alloc) {
1236                 diffstat->alloc = alloc_nr(diffstat->alloc);
1237                 diffstat->files = xrealloc(diffstat->files,
1238                                 diffstat->alloc * sizeof(x));
1239         }
1240         diffstat->files[diffstat->nr++] = x;
1241         if (name_b) {
1242                 x->from_name = xstrdup(name_a);
1243                 x->name = xstrdup(name_b);
1244                 x->is_renamed = 1;
1245         }
1246         else {
1247                 x->from_name = NULL;
1248                 x->name = xstrdup(name_a);
1249         }
1250         return x;
1251 }
1252
1253 static void diffstat_consume(void *priv, char *line, unsigned long len)
1254 {
1255         struct diffstat_t *diffstat = priv;
1256         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1257
1258         if (line[0] == '+')
1259                 x->added++;
1260         else if (line[0] == '-')
1261                 x->deleted++;
1262 }
1263
1264 const char mime_boundary_leader[] = "------------";
1265
1266 static int scale_linear(int it, int width, int max_change)
1267 {
1268         /*
1269          * make sure that at least one '-' is printed if there were deletions,
1270          * and likewise for '+'.
1271          */
1272         if (max_change < 2)
1273                 return it;
1274         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
1275 }
1276
1277 static void show_name(FILE *file,
1278                       const char *prefix, const char *name, int len)
1279 {
1280         fprintf(file, " %s%-*s |", prefix, len, name);
1281 }
1282
1283 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1284 {
1285         if (cnt <= 0)
1286                 return;
1287         fprintf(file, "%s", set);
1288         while (cnt--)
1289                 putc(ch, file);
1290         fprintf(file, "%s", reset);
1291 }
1292
1293 static void fill_print_name(struct diffstat_file *file)
1294 {
1295         char *pname;
1296
1297         if (file->print_name)
1298                 return;
1299
1300         if (!file->is_renamed) {
1301                 struct strbuf buf = STRBUF_INIT;
1302                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1303                         pname = strbuf_detach(&buf, NULL);
1304                 } else {
1305                         pname = file->name;
1306                         strbuf_release(&buf);
1307                 }
1308         } else {
1309                 pname = pprint_rename(file->from_name, file->name);
1310         }
1311         file->print_name = pname;
1312 }
1313
1314 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1315 {
1316         int i, len, add, del, adds = 0, dels = 0;
1317         uintmax_t max_change = 0, max_len = 0;
1318         int total_files = data->nr;
1319         int width, name_width;
1320         const char *reset, *add_c, *del_c;
1321         const char *line_prefix = "";
1322         struct strbuf *msg = NULL;
1323
1324         if (data->nr == 0)
1325                 return;
1326
1327         if (options->output_prefix) {
1328                 msg = options->output_prefix(options, options->output_prefix_data);
1329                 line_prefix = msg->buf;
1330         }
1331
1332         width = options->stat_width ? options->stat_width : 80;
1333         name_width = options->stat_name_width ? options->stat_name_width : 50;
1334
1335         /* Sanity: give at least 5 columns to the graph,
1336          * but leave at least 10 columns for the name.
1337          */
1338         if (width < 25)
1339                 width = 25;
1340         if (name_width < 10)
1341                 name_width = 10;
1342         else if (width < name_width + 15)
1343                 name_width = width - 15;
1344
1345         /* Find the longest filename and max number of changes */
1346         reset = diff_get_color_opt(options, DIFF_RESET);
1347         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1348         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1349
1350         for (i = 0; i < data->nr; i++) {
1351                 struct diffstat_file *file = data->files[i];
1352                 uintmax_t change = file->added + file->deleted;
1353                 fill_print_name(file);
1354                 len = strlen(file->print_name);
1355                 if (max_len < len)
1356                         max_len = len;
1357
1358                 if (file->is_binary || file->is_unmerged)
1359                         continue;
1360                 if (max_change < change)
1361                         max_change = change;
1362         }
1363
1364         /* Compute the width of the graph part;
1365          * 10 is for one blank at the beginning of the line plus
1366          * " | count " between the name and the graph.
1367          *
1368          * From here on, name_width is the width of the name area,
1369          * and width is the width of the graph area.
1370          */
1371         name_width = (name_width < max_len) ? name_width : max_len;
1372         if (width < (name_width + 10) + max_change)
1373                 width = width - (name_width + 10);
1374         else
1375                 width = max_change;
1376
1377         for (i = 0; i < data->nr; i++) {
1378                 const char *prefix = "";
1379                 char *name = data->files[i]->print_name;
1380                 uintmax_t added = data->files[i]->added;
1381                 uintmax_t deleted = data->files[i]->deleted;
1382                 int name_len;
1383
1384                 /*
1385                  * "scale" the filename
1386                  */
1387                 len = name_width;
1388                 name_len = strlen(name);
1389                 if (name_width < name_len) {
1390                         char *slash;
1391                         prefix = "...";
1392                         len -= 3;
1393                         name += name_len - len;
1394                         slash = strchr(name, '/');
1395                         if (slash)
1396                                 name = slash;
1397                 }
1398
1399                 if (data->files[i]->is_binary) {
1400                         fprintf(options->file, "%s", line_prefix);
1401                         show_name(options->file, prefix, name, len);
1402                         fprintf(options->file, "  Bin ");
1403                         fprintf(options->file, "%s%"PRIuMAX"%s",
1404                                 del_c, deleted, reset);
1405                         fprintf(options->file, " -> ");
1406                         fprintf(options->file, "%s%"PRIuMAX"%s",
1407                                 add_c, added, reset);
1408                         fprintf(options->file, " bytes");
1409                         fprintf(options->file, "\n");
1410                         continue;
1411                 }
1412                 else if (data->files[i]->is_unmerged) {
1413                         fprintf(options->file, "%s", line_prefix);
1414                         show_name(options->file, prefix, name, len);
1415                         fprintf(options->file, "  Unmerged\n");
1416                         continue;
1417                 }
1418                 else if (!data->files[i]->is_renamed &&
1419                          (added + deleted == 0)) {
1420                         total_files--;
1421                         continue;
1422                 }
1423
1424                 /*
1425                  * scale the add/delete
1426                  */
1427                 add = added;
1428                 del = deleted;
1429                 adds += add;
1430                 dels += del;
1431
1432                 if (width <= max_change) {
1433                         add = scale_linear(add, width, max_change);
1434                         del = scale_linear(del, width, max_change);
1435                 }
1436                 fprintf(options->file, "%s", line_prefix);
1437                 show_name(options->file, prefix, name, len);
1438                 fprintf(options->file, "%5"PRIuMAX"%s", added + deleted,
1439                                 added + deleted ? " " : "");
1440                 show_graph(options->file, '+', add, add_c, reset);
1441                 show_graph(options->file, '-', del, del_c, reset);
1442                 fprintf(options->file, "\n");
1443         }
1444         fprintf(options->file, "%s", line_prefix);
1445         fprintf(options->file,
1446                " %d files changed, %d insertions(+), %d deletions(-)\n",
1447                total_files, adds, dels);
1448 }
1449
1450 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1451 {
1452         int i, adds = 0, dels = 0, total_files = data->nr;
1453
1454         if (data->nr == 0)
1455                 return;
1456
1457         for (i = 0; i < data->nr; i++) {
1458                 if (!data->files[i]->is_binary &&
1459                     !data->files[i]->is_unmerged) {
1460                         int added = data->files[i]->added;
1461                         int deleted= data->files[i]->deleted;
1462                         if (!data->files[i]->is_renamed &&
1463                             (added + deleted == 0)) {
1464                                 total_files--;
1465                         } else {
1466                                 adds += added;
1467                                 dels += deleted;
1468                         }
1469                 }
1470         }
1471         if (options->output_prefix) {
1472                 struct strbuf *msg = NULL;
1473                 msg = options->output_prefix(options,
1474                                 options->output_prefix_data);
1475                 fprintf(options->file, "%s", msg->buf);
1476         }
1477         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
1478                total_files, adds, dels);
1479 }
1480
1481 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1482 {
1483         int i;
1484
1485         if (data->nr == 0)
1486                 return;
1487
1488         for (i = 0; i < data->nr; i++) {
1489                 struct diffstat_file *file = data->files[i];
1490
1491                 if (options->output_prefix) {
1492                         struct strbuf *msg = NULL;
1493                         msg = options->output_prefix(options,
1494                                         options->output_prefix_data);
1495                         fprintf(options->file, "%s", msg->buf);
1496                 }
1497
1498                 if (file->is_binary)
1499                         fprintf(options->file, "-\t-\t");
1500                 else
1501                         fprintf(options->file,
1502                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1503                                 file->added, file->deleted);
1504                 if (options->line_termination) {
1505                         fill_print_name(file);
1506                         if (!file->is_renamed)
1507                                 write_name_quoted(file->name, options->file,
1508                                                   options->line_termination);
1509                         else {
1510                                 fputs(file->print_name, options->file);
1511                                 putc(options->line_termination, options->file);
1512                         }
1513                 } else {
1514                         if (file->is_renamed) {
1515                                 putc('\0', options->file);
1516                                 write_name_quoted(file->from_name, options->file, '\0');
1517                         }
1518                         write_name_quoted(file->name, options->file, '\0');
1519                 }
1520         }
1521 }
1522
1523 struct dirstat_file {
1524         const char *name;
1525         unsigned long changed;
1526 };
1527
1528 struct dirstat_dir {
1529         struct dirstat_file *files;
1530         int alloc, nr, permille, cumulative;
1531 };
1532
1533 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1534                 unsigned long changed, const char *base, int baselen)
1535 {
1536         unsigned long this_dir = 0;
1537         unsigned int sources = 0;
1538         const char *line_prefix = "";
1539         struct strbuf *msg = NULL;
1540
1541         if (opt->output_prefix) {
1542                 msg = opt->output_prefix(opt, opt->output_prefix_data);
1543                 line_prefix = msg->buf;
1544         }
1545
1546         while (dir->nr) {
1547                 struct dirstat_file *f = dir->files;
1548                 int namelen = strlen(f->name);
1549                 unsigned long this;
1550                 char *slash;
1551
1552                 if (namelen < baselen)
1553                         break;
1554                 if (memcmp(f->name, base, baselen))
1555                         break;
1556                 slash = strchr(f->name + baselen, '/');
1557                 if (slash) {
1558                         int newbaselen = slash + 1 - f->name;
1559                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1560                         sources++;
1561                 } else {
1562                         this = f->changed;
1563                         dir->files++;
1564                         dir->nr--;
1565                         sources += 2;
1566                 }
1567                 this_dir += this;
1568         }
1569
1570         /*
1571          * We don't report dirstat's for
1572          *  - the top level
1573          *  - or cases where everything came from a single directory
1574          *    under this directory (sources == 1).
1575          */
1576         if (baselen && sources != 1) {
1577                 if (this_dir) {
1578                         int permille = this_dir * 1000 / changed;
1579                         if (permille >= dir->permille) {
1580                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1581                                         permille / 10, permille % 10, baselen, base);
1582                                 if (!dir->cumulative)
1583                                         return 0;
1584                         }
1585                 }
1586         }
1587         return this_dir;
1588 }
1589
1590 static int dirstat_compare(const void *_a, const void *_b)
1591 {
1592         const struct dirstat_file *a = _a;
1593         const struct dirstat_file *b = _b;
1594         return strcmp(a->name, b->name);
1595 }
1596
1597 static void show_dirstat(struct diff_options *options)
1598 {
1599         int i;
1600         unsigned long changed;
1601         struct dirstat_dir dir;
1602         struct diff_queue_struct *q = &diff_queued_diff;
1603
1604         dir.files = NULL;
1605         dir.alloc = 0;
1606         dir.nr = 0;
1607         dir.permille = options->dirstat_permille;
1608         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1609
1610         changed = 0;
1611         for (i = 0; i < q->nr; i++) {
1612                 struct diff_filepair *p = q->queue[i];
1613                 const char *name;
1614                 unsigned long copied, added, damage;
1615                 int content_changed;
1616
1617                 name = p->two->path ? p->two->path : p->one->path;
1618
1619                 if (p->one->sha1_valid && p->two->sha1_valid)
1620                         content_changed = hashcmp(p->one->sha1, p->two->sha1);
1621                 else
1622                         content_changed = 1;
1623
1624                 if (!content_changed) {
1625                         /*
1626                          * The SHA1 has not changed, so pre-/post-content is
1627                          * identical. We can therefore skip looking at the
1628                          * file contents altogether.
1629                          */
1630                         damage = 0;
1631                         goto found_damage;
1632                 }
1633
1634                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1635                         /*
1636                          * In --dirstat-by-file mode, we don't really need to
1637                          * look at the actual file contents at all.
1638                          * The fact that the SHA1 changed is enough for us to
1639                          * add this file to the list of results
1640                          * (with each file contributing equal damage).
1641                          */
1642                         damage = 1;
1643                         goto found_damage;
1644                 }
1645
1646                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1647                         diff_populate_filespec(p->one, 0);
1648                         diff_populate_filespec(p->two, 0);
1649                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1650                                                &copied, &added);
1651                         diff_free_filespec_data(p->one);
1652                         diff_free_filespec_data(p->two);
1653                 } else if (DIFF_FILE_VALID(p->one)) {
1654                         diff_populate_filespec(p->one, 1);
1655                         copied = added = 0;
1656                         diff_free_filespec_data(p->one);
1657                 } else if (DIFF_FILE_VALID(p->two)) {
1658                         diff_populate_filespec(p->two, 1);
1659                         copied = 0;
1660                         added = p->two->size;
1661                         diff_free_filespec_data(p->two);
1662                 } else
1663                         continue;
1664
1665                 /*
1666                  * Original minus copied is the removed material,
1667                  * added is the new material.  They are both damages
1668                  * made to the preimage.
1669                  * If the resulting damage is zero, we know that
1670                  * diffcore_count_changes() considers the two entries to
1671                  * be identical, but since content_changed is true, we
1672                  * know that there must have been _some_ kind of change,
1673                  * so we force all entries to have damage > 0.
1674                  */
1675                 damage = (p->one->size - copied) + added;
1676                 if (!damage)
1677                         damage = 1;
1678
1679 found_damage:
1680                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1681                 dir.files[dir.nr].name = name;
1682                 dir.files[dir.nr].changed = damage;
1683                 changed += damage;
1684                 dir.nr++;
1685         }
1686
1687         /* This can happen even with many files, if everything was renames */
1688         if (!changed)
1689                 return;
1690
1691         /* Show all directories with more than x% of the changes */
1692         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1693         gather_dirstat(options, &dir, changed, "", 0);
1694 }
1695
1696 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
1697 {
1698         int i;
1699         unsigned long changed;
1700         struct dirstat_dir dir;
1701
1702         if (data->nr == 0)
1703                 return;
1704
1705         dir.files = NULL;
1706         dir.alloc = 0;
1707         dir.nr = 0;
1708         dir.permille = options->dirstat_permille;
1709         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1710
1711         changed = 0;
1712         for (i = 0; i < data->nr; i++) {
1713                 struct diffstat_file *file = data->files[i];
1714                 unsigned long damage = file->added + file->deleted;
1715                 if (file->is_binary)
1716                         /*
1717                          * binary files counts bytes, not lines. Must find some
1718                          * way to normalize binary bytes vs. textual lines.
1719                          * The following heuristic assumes that there are 64
1720                          * bytes per "line".
1721                          * This is stupid and ugly, but very cheap...
1722                          */
1723                         damage = (damage + 63) / 64;
1724                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1725                 dir.files[dir.nr].name = file->name;
1726                 dir.files[dir.nr].changed = damage;
1727                 changed += damage;
1728                 dir.nr++;
1729         }
1730
1731         /* This can happen even with many files, if everything was renames */
1732         if (!changed)
1733                 return;
1734
1735         /* Show all directories with more than x% of the changes */
1736         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1737         gather_dirstat(options, &dir, changed, "", 0);
1738 }
1739
1740 static void free_diffstat_info(struct diffstat_t *diffstat)
1741 {
1742         int i;
1743         for (i = 0; i < diffstat->nr; i++) {
1744                 struct diffstat_file *f = diffstat->files[i];
1745                 if (f->name != f->print_name)
1746                         free(f->print_name);
1747                 free(f->name);
1748                 free(f->from_name);
1749                 free(f);
1750         }
1751         free(diffstat->files);
1752 }
1753
1754 struct checkdiff_t {
1755         const char *filename;
1756         int lineno;
1757         int conflict_marker_size;
1758         struct diff_options *o;
1759         unsigned ws_rule;
1760         unsigned status;
1761 };
1762
1763 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
1764 {
1765         char firstchar;
1766         int cnt;
1767
1768         if (len < marker_size + 1)
1769                 return 0;
1770         firstchar = line[0];
1771         switch (firstchar) {
1772         case '=': case '>': case '<': case '|':
1773                 break;
1774         default:
1775                 return 0;
1776         }
1777         for (cnt = 1; cnt < marker_size; cnt++)
1778                 if (line[cnt] != firstchar)
1779                         return 0;
1780         /* line[1] thru line[marker_size-1] are same as firstchar */
1781         if (len < marker_size + 1 || !isspace(line[marker_size]))
1782                 return 0;
1783         return 1;
1784 }
1785
1786 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1787 {
1788         struct checkdiff_t *data = priv;
1789         int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1790         int marker_size = data->conflict_marker_size;
1791         const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1792         const char *reset = diff_get_color(color_diff, DIFF_RESET);
1793         const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1794         char *err;
1795         char *line_prefix = "";
1796         struct strbuf *msgbuf;
1797
1798         assert(data->o);
1799         if (data->o->output_prefix) {
1800                 msgbuf = data->o->output_prefix(data->o,
1801                         data->o->output_prefix_data);
1802                 line_prefix = msgbuf->buf;
1803         }
1804
1805         if (line[0] == '+') {
1806                 unsigned bad;
1807                 data->lineno++;
1808                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
1809                         data->status |= 1;
1810                         fprintf(data->o->file,
1811                                 "%s%s:%d: leftover conflict marker\n",
1812                                 line_prefix, data->filename, data->lineno);
1813                 }
1814                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1815                 if (!bad)
1816                         return;
1817                 data->status |= bad;
1818                 err = whitespace_error_string(bad);
1819                 fprintf(data->o->file, "%s%s:%d: %s.\n",
1820                         line_prefix, data->filename, data->lineno, err);
1821                 free(err);
1822                 emit_line(data->o, set, reset, line, 1);
1823                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1824                               data->o->file, set, reset, ws);
1825         } else if (line[0] == ' ') {
1826                 data->lineno++;
1827         } else if (line[0] == '@') {
1828                 char *plus = strchr(line, '+');
1829                 if (plus)
1830                         data->lineno = strtol(plus, NULL, 10) - 1;
1831                 else
1832                         die("invalid diff");
1833         }
1834 }
1835
1836 static unsigned char *deflate_it(char *data,
1837                                  unsigned long size,
1838                                  unsigned long *result_size)
1839 {
1840         int bound;
1841         unsigned char *deflated;
1842         z_stream stream;
1843
1844         memset(&stream, 0, sizeof(stream));
1845         deflateInit(&stream, zlib_compression_level);
1846         bound = deflateBound(&stream, size);
1847         deflated = xmalloc(bound);
1848         stream.next_out = deflated;
1849         stream.avail_out = bound;
1850
1851         stream.next_in = (unsigned char *)data;
1852         stream.avail_in = size;
1853         while (deflate(&stream, Z_FINISH) == Z_OK)
1854                 ; /* nothing */
1855         deflateEnd(&stream);
1856         *result_size = stream.total_out;
1857         return deflated;
1858 }
1859
1860 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
1861 {
1862         void *cp;
1863         void *delta;
1864         void *deflated;
1865         void *data;
1866         unsigned long orig_size;
1867         unsigned long delta_size;
1868         unsigned long deflate_size;
1869         unsigned long data_size;
1870
1871         /* We could do deflated delta, or we could do just deflated two,
1872          * whichever is smaller.
1873          */
1874         delta = NULL;
1875         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1876         if (one->size && two->size) {
1877                 delta = diff_delta(one->ptr, one->size,
1878                                    two->ptr, two->size,
1879                                    &delta_size, deflate_size);
1880                 if (delta) {
1881                         void *to_free = delta;
1882                         orig_size = delta_size;
1883                         delta = deflate_it(delta, delta_size, &delta_size);
1884                         free(to_free);
1885                 }
1886         }
1887
1888         if (delta && delta_size < deflate_size) {
1889                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
1890                 free(deflated);
1891                 data = delta;
1892                 data_size = delta_size;
1893         }
1894         else {
1895                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
1896                 free(delta);
1897                 data = deflated;
1898                 data_size = deflate_size;
1899         }
1900
1901         /* emit data encoded in base85 */
1902         cp = data;
1903         while (data_size) {
1904                 int bytes = (52 < data_size) ? 52 : data_size;
1905                 char line[70];
1906                 data_size -= bytes;
1907                 if (bytes <= 26)
1908                         line[0] = bytes + 'A' - 1;
1909                 else
1910                         line[0] = bytes - 26 + 'a' - 1;
1911                 encode_85(line + 1, cp, bytes);
1912                 cp = (char *) cp + bytes;
1913                 fprintf(file, "%s", prefix);
1914                 fputs(line, file);
1915                 fputc('\n', file);
1916         }
1917         fprintf(file, "%s\n", prefix);
1918         free(data);
1919 }
1920
1921 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
1922 {
1923         fprintf(file, "%sGIT binary patch\n", prefix);
1924         emit_binary_diff_body(file, one, two, prefix);
1925         emit_binary_diff_body(file, two, one, prefix);
1926 }
1927
1928 static void diff_filespec_load_driver(struct diff_filespec *one)
1929 {
1930         /* Use already-loaded driver */
1931         if (one->driver)
1932                 return;
1933
1934         if (S_ISREG(one->mode))
1935                 one->driver = userdiff_find_by_path(one->path);
1936
1937         /* Fallback to default settings */
1938         if (!one->driver)
1939                 one->driver = userdiff_find_by_name("default");
1940 }
1941
1942 int diff_filespec_is_binary(struct diff_filespec *one)
1943 {
1944         if (one->is_binary == -1) {
1945                 diff_filespec_load_driver(one);
1946                 if (one->driver->binary != -1)
1947                         one->is_binary = one->driver->binary;
1948                 else {
1949                         if (!one->data && DIFF_FILE_VALID(one))
1950                                 diff_populate_filespec(one, 0);
1951                         if (one->data)
1952                                 one->is_binary = buffer_is_binary(one->data,
1953                                                 one->size);
1954                         if (one->is_binary == -1)
1955                                 one->is_binary = 0;
1956                 }
1957         }
1958         return one->is_binary;
1959 }
1960
1961 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1962 {
1963         diff_filespec_load_driver(one);
1964         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1965 }
1966
1967 static const char *userdiff_word_regex(struct diff_filespec *one)
1968 {
1969         diff_filespec_load_driver(one);
1970         return one->driver->word_regex;
1971 }
1972
1973 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1974 {
1975         if (!options->a_prefix)
1976                 options->a_prefix = a;
1977         if (!options->b_prefix)
1978                 options->b_prefix = b;
1979 }
1980
1981 struct userdiff_driver *get_textconv(struct diff_filespec *one)
1982 {
1983         if (!DIFF_FILE_VALID(one))
1984                 return NULL;
1985
1986         diff_filespec_load_driver(one);
1987         if (!one->driver->textconv)
1988                 return NULL;
1989
1990         if (one->driver->textconv_want_cache && !one->driver->textconv_cache) {
1991                 struct notes_cache *c = xmalloc(sizeof(*c));
1992                 struct strbuf name = STRBUF_INIT;
1993
1994                 strbuf_addf(&name, "textconv/%s", one->driver->name);
1995                 notes_cache_init(c, name.buf, one->driver->textconv);
1996                 one->driver->textconv_cache = c;
1997         }
1998
1999         return one->driver;
2000 }
2001
2002 static void builtin_diff(const char *name_a,
2003                          const char *name_b,
2004                          struct diff_filespec *one,
2005                          struct diff_filespec *two,
2006                          const char *xfrm_msg,
2007                          int must_show_header,
2008                          struct diff_options *o,
2009                          int complete_rewrite)
2010 {
2011         mmfile_t mf1, mf2;
2012         const char *lbl[2];
2013         char *a_one, *b_two;
2014         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
2015         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2016         const char *a_prefix, *b_prefix;
2017         struct userdiff_driver *textconv_one = NULL;
2018         struct userdiff_driver *textconv_two = NULL;
2019         struct strbuf header = STRBUF_INIT;
2020         struct strbuf *msgbuf;
2021         char *line_prefix = "";
2022
2023         if (o->output_prefix) {
2024                 msgbuf = o->output_prefix(o, o->output_prefix_data);
2025                 line_prefix = msgbuf->buf;
2026         }
2027
2028         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2029                         (!one->mode || S_ISGITLINK(one->mode)) &&
2030                         (!two->mode || S_ISGITLINK(two->mode))) {
2031                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2032                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2033                 show_submodule_summary(o->file, one ? one->path : two->path,
2034                                 one->sha1, two->sha1, two->dirty_submodule,
2035                                 del, add, reset);
2036                 return;
2037         }
2038
2039         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2040                 textconv_one = get_textconv(one);
2041                 textconv_two = get_textconv(two);
2042         }
2043
2044         diff_set_mnemonic_prefix(o, "a/", "b/");
2045         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2046                 a_prefix = o->b_prefix;
2047                 b_prefix = o->a_prefix;
2048         } else {
2049                 a_prefix = o->a_prefix;
2050                 b_prefix = o->b_prefix;
2051         }
2052
2053         /* Never use a non-valid filename anywhere if at all possible */
2054         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2055         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2056
2057         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2058         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2059         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2060         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2061         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, set, a_one, b_two, reset);
2062         if (lbl[0][0] == '/') {
2063                 /* /dev/null */
2064                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, set, two->mode, reset);
2065                 if (xfrm_msg)
2066                         strbuf_addstr(&header, xfrm_msg);
2067                 must_show_header = 1;
2068         }
2069         else if (lbl[1][0] == '/') {
2070                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, set, one->mode, reset);
2071                 if (xfrm_msg)
2072                         strbuf_addstr(&header, xfrm_msg);
2073                 must_show_header = 1;
2074         }
2075         else {
2076                 if (one->mode != two->mode) {
2077                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, set, one->mode, reset);
2078                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, set, two->mode, reset);
2079                         must_show_header = 1;
2080                 }
2081                 if (xfrm_msg)
2082                         strbuf_addstr(&header, xfrm_msg);
2083
2084                 /*
2085                  * we do not run diff between different kind
2086                  * of objects.
2087                  */
2088                 if ((one->mode ^ two->mode) & S_IFMT)
2089                         goto free_ab_and_return;
2090                 if (complete_rewrite &&
2091                     (textconv_one || !diff_filespec_is_binary(one)) &&
2092                     (textconv_two || !diff_filespec_is_binary(two))) {
2093                         fprintf(o->file, "%s", header.buf);
2094                         strbuf_reset(&header);
2095                         emit_rewrite_diff(name_a, name_b, one, two,
2096                                                 textconv_one, textconv_two, o);
2097                         o->found_changes = 1;
2098                         goto free_ab_and_return;
2099                 }
2100         }
2101
2102         if (o->irreversible_delete && lbl[1][0] == '/') {
2103                 fprintf(o->file, "%s", header.buf);
2104                 strbuf_reset(&header);
2105                 goto free_ab_and_return;
2106         } else if (!DIFF_OPT_TST(o, TEXT) &&
2107             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2108               (!textconv_two && diff_filespec_is_binary(two)) )) {
2109                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2110                         die("unable to read files to diff");
2111                 /* Quite common confusing case */
2112                 if (mf1.size == mf2.size &&
2113                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2114                         if (must_show_header)
2115                                 fprintf(o->file, "%s", header.buf);
2116                         goto free_ab_and_return;
2117                 }
2118                 fprintf(o->file, "%s", header.buf);
2119                 strbuf_reset(&header);
2120                 if (DIFF_OPT_TST(o, BINARY))
2121                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2122                 else
2123                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2124                                 line_prefix, lbl[0], lbl[1]);
2125                 o->found_changes = 1;
2126         } else {
2127                 /* Crazy xdl interfaces.. */
2128                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2129                 xpparam_t xpp;
2130                 xdemitconf_t xecfg;
2131                 struct emit_callback ecbdata;
2132                 const struct userdiff_funcname *pe;
2133
2134                 if (!DIFF_XDL_TST(o, WHITESPACE_FLAGS) || must_show_header) {
2135                         fprintf(o->file, "%s", header.buf);
2136                         strbuf_reset(&header);
2137                 }
2138
2139                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2140                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2141
2142                 pe = diff_funcname_pattern(one);
2143                 if (!pe)
2144                         pe = diff_funcname_pattern(two);
2145
2146                 memset(&xpp, 0, sizeof(xpp));
2147                 memset(&xecfg, 0, sizeof(xecfg));
2148                 memset(&ecbdata, 0, sizeof(ecbdata));
2149                 ecbdata.label_path = lbl;
2150                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
2151                 ecbdata.found_changesp = &o->found_changes;
2152                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
2153                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2154                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2155                 ecbdata.opt = o;
2156                 ecbdata.header = header.len ? &header : NULL;
2157                 xpp.flags = o->xdl_opts;
2158                 xecfg.ctxlen = o->context;
2159                 xecfg.interhunkctxlen = o->interhunkcontext;
2160                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2161                 if (pe)
2162                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2163                 if (!diffopts)
2164                         ;
2165                 else if (!prefixcmp(diffopts, "--unified="))
2166                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
2167                 else if (!prefixcmp(diffopts, "-u"))
2168                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
2169                 if (o->word_diff) {
2170                         int i;
2171
2172                         ecbdata.diff_words =
2173                                 xcalloc(1, sizeof(struct diff_words_data));
2174                         ecbdata.diff_words->type = o->word_diff;
2175                         ecbdata.diff_words->opt = o;
2176                         if (!o->word_regex)
2177                                 o->word_regex = userdiff_word_regex(one);
2178                         if (!o->word_regex)
2179                                 o->word_regex = userdiff_word_regex(two);
2180                         if (!o->word_regex)
2181                                 o->word_regex = diff_word_regex_cfg;
2182                         if (o->word_regex) {
2183                                 ecbdata.diff_words->word_regex = (regex_t *)
2184                                         xmalloc(sizeof(regex_t));
2185                                 if (regcomp(ecbdata.diff_words->word_regex,
2186                                                 o->word_regex,
2187                                                 REG_EXTENDED | REG_NEWLINE))
2188                                         die ("Invalid regular expression: %s",
2189                                                         o->word_regex);
2190                         }
2191                         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2192                                 if (o->word_diff == diff_words_styles[i].type) {
2193                                         ecbdata.diff_words->style =
2194                                                 &diff_words_styles[i];
2195                                         break;
2196                                 }
2197                         }
2198                         if (DIFF_OPT_TST(o, COLOR_DIFF)) {
2199                                 struct diff_words_style *st = ecbdata.diff_words->style;
2200                                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2201                                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2202                                 st->ctx.color = diff_get_color_opt(o, DIFF_PLAIN);
2203                         }
2204                 }
2205                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2206                               &xpp, &xecfg);
2207                 if (o->word_diff)
2208                         free_diff_words_data(&ecbdata);
2209                 if (textconv_one)
2210                         free(mf1.ptr);
2211                 if (textconv_two)
2212                         free(mf2.ptr);
2213                 xdiff_clear_find_func(&xecfg);
2214         }
2215
2216  free_ab_and_return:
2217         strbuf_release(&header);
2218         diff_free_filespec_data(one);
2219         diff_free_filespec_data(two);
2220         free(a_one);
2221         free(b_two);
2222         return;
2223 }
2224
2225 static void builtin_diffstat(const char *name_a, const char *name_b,
2226                              struct diff_filespec *one,
2227                              struct diff_filespec *two,
2228                              struct diffstat_t *diffstat,
2229                              struct diff_options *o,
2230                              int complete_rewrite)
2231 {
2232         mmfile_t mf1, mf2;
2233         struct diffstat_file *data;
2234
2235         data = diffstat_add(diffstat, name_a, name_b);
2236
2237         if (!one || !two) {
2238                 data->is_unmerged = 1;
2239                 return;
2240         }
2241
2242         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2243                 data->is_binary = 1;
2244                 data->added = diff_filespec_size(two);
2245                 data->deleted = diff_filespec_size(one);
2246         }
2247
2248         else if (complete_rewrite) {
2249                 diff_populate_filespec(one, 0);
2250                 diff_populate_filespec(two, 0);
2251                 data->deleted = count_lines(one->data, one->size);
2252                 data->added = count_lines(two->data, two->size);
2253         }
2254
2255         else {
2256                 /* Crazy xdl interfaces.. */
2257                 xpparam_t xpp;
2258                 xdemitconf_t xecfg;
2259
2260                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2261                         die("unable to read files to diff");
2262
2263                 memset(&xpp, 0, sizeof(xpp));
2264                 memset(&xecfg, 0, sizeof(xecfg));
2265                 xpp.flags = o->xdl_opts;
2266                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2267                               &xpp, &xecfg);
2268         }
2269
2270         diff_free_filespec_data(one);
2271         diff_free_filespec_data(two);
2272 }
2273
2274 static void builtin_checkdiff(const char *name_a, const char *name_b,
2275                               const char *attr_path,
2276                               struct diff_filespec *one,
2277                               struct diff_filespec *two,
2278                               struct diff_options *o)
2279 {
2280         mmfile_t mf1, mf2;
2281         struct checkdiff_t data;
2282
2283         if (!two)
2284                 return;
2285
2286         memset(&data, 0, sizeof(data));
2287         data.filename = name_b ? name_b : name_a;
2288         data.lineno = 0;
2289         data.o = o;
2290         data.ws_rule = whitespace_rule(attr_path);
2291         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2292
2293         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2294                 die("unable to read files to diff");
2295
2296         /*
2297          * All the other codepaths check both sides, but not checking
2298          * the "old" side here is deliberate.  We are checking the newly
2299          * introduced changes, and as long as the "new" side is text, we
2300          * can and should check what it introduces.
2301          */
2302         if (diff_filespec_is_binary(two))
2303                 goto free_and_return;
2304         else {
2305                 /* Crazy xdl interfaces.. */
2306                 xpparam_t xpp;
2307                 xdemitconf_t xecfg;
2308
2309                 memset(&xpp, 0, sizeof(xpp));
2310                 memset(&xecfg, 0, sizeof(xecfg));
2311                 xecfg.ctxlen = 1; /* at least one context line */
2312                 xpp.flags = 0;
2313                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2314                               &xpp, &xecfg);
2315
2316                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2317                         struct emit_callback ecbdata;
2318                         int blank_at_eof;
2319
2320                         ecbdata.ws_rule = data.ws_rule;
2321                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2322                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2323
2324                         if (blank_at_eof) {
2325                                 static char *err;
2326                                 if (!err)
2327                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2328                                 fprintf(o->file, "%s:%d: %s.\n",
2329                                         data.filename, blank_at_eof, err);
2330                                 data.status = 1; /* report errors */
2331                         }
2332                 }
2333         }
2334  free_and_return:
2335         diff_free_filespec_data(one);
2336         diff_free_filespec_data(two);
2337         if (data.status)
2338                 DIFF_OPT_SET(o, CHECK_FAILED);
2339 }
2340
2341 struct diff_filespec *alloc_filespec(const char *path)
2342 {
2343         int namelen = strlen(path);
2344         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2345
2346         memset(spec, 0, sizeof(*spec));
2347         spec->path = (char *)(spec + 1);
2348         memcpy(spec->path, path, namelen+1);
2349         spec->count = 1;
2350         spec->is_binary = -1;
2351         return spec;
2352 }
2353
2354 void free_filespec(struct diff_filespec *spec)
2355 {
2356         if (!--spec->count) {
2357                 diff_free_filespec_data(spec);
2358                 free(spec);
2359         }
2360 }
2361
2362 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2363                    unsigned short mode)
2364 {
2365         if (mode) {
2366                 spec->mode = canon_mode(mode);
2367                 hashcpy(spec->sha1, sha1);
2368                 spec->sha1_valid = !is_null_sha1(sha1);
2369         }
2370 }
2371
2372 /*
2373  * Given a name and sha1 pair, if the index tells us the file in
2374  * the work tree has that object contents, return true, so that
2375  * prepare_temp_file() does not have to inflate and extract.
2376  */
2377 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2378 {
2379         struct cache_entry *ce;
2380         struct stat st;
2381         int pos, len;
2382
2383         /*
2384          * We do not read the cache ourselves here, because the
2385          * benchmark with my previous version that always reads cache
2386          * shows that it makes things worse for diff-tree comparing
2387          * two linux-2.6 kernel trees in an already checked out work
2388          * tree.  This is because most diff-tree comparisons deal with
2389          * only a small number of files, while reading the cache is
2390          * expensive for a large project, and its cost outweighs the
2391          * savings we get by not inflating the object to a temporary
2392          * file.  Practically, this code only helps when we are used
2393          * by diff-cache --cached, which does read the cache before
2394          * calling us.
2395          */
2396         if (!active_cache)
2397                 return 0;
2398
2399         /* We want to avoid the working directory if our caller
2400          * doesn't need the data in a normal file, this system
2401          * is rather slow with its stat/open/mmap/close syscalls,
2402          * and the object is contained in a pack file.  The pack
2403          * is probably already open and will be faster to obtain
2404          * the data through than the working directory.  Loose
2405          * objects however would tend to be slower as they need
2406          * to be individually opened and inflated.
2407          */
2408         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2409                 return 0;
2410
2411         len = strlen(name);
2412         pos = cache_name_pos(name, len);
2413         if (pos < 0)
2414                 return 0;
2415         ce = active_cache[pos];
2416
2417         /*
2418          * This is not the sha1 we are looking for, or
2419          * unreusable because it is not a regular file.
2420          */
2421         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2422                 return 0;
2423
2424         /*
2425          * If ce is marked as "assume unchanged", there is no
2426          * guarantee that work tree matches what we are looking for.
2427          */
2428         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2429                 return 0;
2430
2431         /*
2432          * If ce matches the file in the work tree, we can reuse it.
2433          */
2434         if (ce_uptodate(ce) ||
2435             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2436                 return 1;
2437
2438         return 0;
2439 }
2440
2441 static int populate_from_stdin(struct diff_filespec *s)
2442 {
2443         struct strbuf buf = STRBUF_INIT;
2444         size_t size = 0;
2445
2446         if (strbuf_read(&buf, 0, 0) < 0)
2447                 return error("error while reading from stdin %s",
2448                                      strerror(errno));
2449
2450         s->should_munmap = 0;
2451         s->data = strbuf_detach(&buf, &size);
2452         s->size = size;
2453         s->should_free = 1;
2454         return 0;
2455 }
2456
2457 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2458 {
2459         int len;
2460         char *data = xmalloc(100), *dirty = "";
2461
2462         /* Are we looking at the work tree? */
2463         if (s->dirty_submodule)
2464                 dirty = "-dirty";
2465
2466         len = snprintf(data, 100,
2467                        "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2468         s->data = data;
2469         s->size = len;
2470         s->should_free = 1;
2471         if (size_only) {
2472                 s->data = NULL;
2473                 free(data);
2474         }
2475         return 0;
2476 }
2477
2478 /*
2479  * While doing rename detection and pickaxe operation, we may need to
2480  * grab the data for the blob (or file) for our own in-core comparison.
2481  * diff_filespec has data and size fields for this purpose.
2482  */
2483 int diff_populate_filespec(struct diff_filespec *s, int size_only)
2484 {
2485         int err = 0;
2486         if (!DIFF_FILE_VALID(s))
2487                 die("internal error: asking to populate invalid file.");
2488         if (S_ISDIR(s->mode))
2489                 return -1;
2490
2491         if (s->data)
2492                 return 0;
2493
2494         if (size_only && 0 < s->size)
2495                 return 0;
2496
2497         if (S_ISGITLINK(s->mode))
2498                 return diff_populate_gitlink(s, size_only);
2499
2500         if (!s->sha1_valid ||
2501             reuse_worktree_file(s->path, s->sha1, 0)) {
2502                 struct strbuf buf = STRBUF_INIT;
2503                 struct stat st;
2504                 int fd;
2505
2506                 if (!strcmp(s->path, "-"))
2507                         return populate_from_stdin(s);
2508
2509                 if (lstat(s->path, &st) < 0) {
2510                         if (errno == ENOENT) {
2511                         err_empty:
2512                                 err = -1;
2513                         empty:
2514                                 s->data = (char *)"";
2515                                 s->size = 0;
2516                                 return err;
2517                         }
2518                 }
2519                 s->size = xsize_t(st.st_size);
2520                 if (!s->size)
2521                         goto empty;
2522                 if (S_ISLNK(st.st_mode)) {
2523                         struct strbuf sb = STRBUF_INIT;
2524
2525                         if (strbuf_readlink(&sb, s->path, s->size))
2526                                 goto err_empty;
2527                         s->size = sb.len;
2528                         s->data = strbuf_detach(&sb, NULL);
2529                         s->should_free = 1;
2530                         return 0;
2531                 }
2532                 if (size_only)
2533                         return 0;
2534                 fd = open(s->path, O_RDONLY);
2535                 if (fd < 0)
2536                         goto err_empty;
2537                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2538                 close(fd);
2539                 s->should_munmap = 1;
2540
2541                 /*
2542                  * Convert from working tree format to canonical git format
2543                  */
2544                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2545                         size_t size = 0;
2546                         munmap(s->data, s->size);
2547                         s->should_munmap = 0;
2548                         s->data = strbuf_detach(&buf, &size);
2549                         s->size = size;
2550                         s->should_free = 1;
2551                 }
2552         }
2553         else {
2554                 enum object_type type;
2555                 if (size_only) {
2556                         type = sha1_object_info(s->sha1, &s->size);
2557                         if (type < 0)
2558                                 die("unable to read %s", sha1_to_hex(s->sha1));
2559                 } else {
2560                         s->data = read_sha1_file(s->sha1, &type, &s->size);
2561                         if (!s->data)
2562                                 die("unable to read %s", sha1_to_hex(s->sha1));
2563                         s->should_free = 1;
2564                 }
2565         }
2566         return 0;
2567 }
2568
2569 void diff_free_filespec_blob(struct diff_filespec *s)
2570 {
2571         if (s->should_free)
2572                 free(s->data);
2573         else if (s->should_munmap)
2574                 munmap(s->data, s->size);
2575
2576         if (s->should_free || s->should_munmap) {
2577                 s->should_free = s->should_munmap = 0;
2578                 s->data = NULL;
2579         }
2580 }
2581
2582 void diff_free_filespec_data(struct diff_filespec *s)
2583 {
2584         diff_free_filespec_blob(s);
2585         free(s->cnt_data);
2586         s->cnt_data = NULL;
2587 }
2588
2589 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2590                            void *blob,
2591                            unsigned long size,
2592                            const unsigned char *sha1,
2593                            int mode)
2594 {
2595         int fd;
2596         struct strbuf buf = STRBUF_INIT;
2597         struct strbuf template = STRBUF_INIT;
2598         char *path_dup = xstrdup(path);
2599         const char *base = basename(path_dup);
2600
2601         /* Generate "XXXXXX_basename.ext" */
2602         strbuf_addstr(&template, "XXXXXX_");
2603         strbuf_addstr(&template, base);
2604
2605         fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2606                         strlen(base) + 1);
2607         if (fd < 0)
2608                 die_errno("unable to create temp-file");
2609         if (convert_to_working_tree(path,
2610                         (const char *)blob, (size_t)size, &buf)) {
2611                 blob = buf.buf;
2612                 size = buf.len;
2613         }
2614         if (write_in_full(fd, blob, size) != size)
2615                 die_errno("unable to write temp-file");
2616         close(fd);
2617         temp->name = temp->tmp_path;
2618         strcpy(temp->hex, sha1_to_hex(sha1));
2619         temp->hex[40] = 0;
2620         sprintf(temp->mode, "%06o", mode);
2621         strbuf_release(&buf);
2622         strbuf_release(&template);
2623         free(path_dup);
2624 }
2625
2626 static struct diff_tempfile *prepare_temp_file(const char *name,
2627                 struct diff_filespec *one)
2628 {
2629         struct diff_tempfile *temp = claim_diff_tempfile();
2630
2631         if (!DIFF_FILE_VALID(one)) {
2632         not_a_valid_file:
2633                 /* A '-' entry produces this for file-2, and
2634                  * a '+' entry produces this for file-1.
2635                  */
2636                 temp->name = "/dev/null";
2637                 strcpy(temp->hex, ".");
2638                 strcpy(temp->mode, ".");
2639                 return temp;
2640         }
2641
2642         if (!remove_tempfile_installed) {
2643                 atexit(remove_tempfile);
2644                 sigchain_push_common(remove_tempfile_on_signal);
2645                 remove_tempfile_installed = 1;
2646         }
2647
2648         if (!one->sha1_valid ||
2649             reuse_worktree_file(name, one->sha1, 1)) {
2650                 struct stat st;
2651                 if (lstat(name, &st) < 0) {
2652                         if (errno == ENOENT)
2653                                 goto not_a_valid_file;
2654                         die_errno("stat(%s)", name);
2655                 }
2656                 if (S_ISLNK(st.st_mode)) {
2657                         struct strbuf sb = STRBUF_INIT;
2658                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2659                                 die_errno("readlink(%s)", name);
2660                         prep_temp_blob(name, temp, sb.buf, sb.len,
2661                                        (one->sha1_valid ?
2662                                         one->sha1 : null_sha1),
2663                                        (one->sha1_valid ?
2664                                         one->mode : S_IFLNK));
2665                         strbuf_release(&sb);
2666                 }
2667                 else {
2668                         /* we can borrow from the file in the work tree */
2669                         temp->name = name;
2670                         if (!one->sha1_valid)
2671                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2672                         else
2673                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2674                         /* Even though we may sometimes borrow the
2675                          * contents from the work tree, we always want
2676                          * one->mode.  mode is trustworthy even when
2677                          * !(one->sha1_valid), as long as
2678                          * DIFF_FILE_VALID(one).
2679                          */
2680                         sprintf(temp->mode, "%06o", one->mode);
2681                 }
2682                 return temp;
2683         }
2684         else {
2685                 if (diff_populate_filespec(one, 0))
2686                         die("cannot read data blob for %s", one->path);
2687                 prep_temp_blob(name, temp, one->data, one->size,
2688                                one->sha1, one->mode);
2689         }
2690         return temp;
2691 }
2692
2693 /* An external diff command takes:
2694  *
2695  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2696  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2697  *
2698  */
2699 static void run_external_diff(const char *pgm,
2700                               const char *name,
2701                               const char *other,
2702                               struct diff_filespec *one,
2703                               struct diff_filespec *two,
2704                               const char *xfrm_msg,
2705                               int complete_rewrite)
2706 {
2707         const char *spawn_arg[10];
2708         int retval;
2709         const char **arg = &spawn_arg[0];
2710
2711         if (one && two) {
2712                 struct diff_tempfile *temp_one, *temp_two;
2713                 const char *othername = (other ? other : name);
2714                 temp_one = prepare_temp_file(name, one);
2715                 temp_two = prepare_temp_file(othername, two);
2716                 *arg++ = pgm;
2717                 *arg++ = name;
2718                 *arg++ = temp_one->name;
2719                 *arg++ = temp_one->hex;
2720                 *arg++ = temp_one->mode;
2721                 *arg++ = temp_two->name;
2722                 *arg++ = temp_two->hex;
2723                 *arg++ = temp_two->mode;
2724                 if (other) {
2725                         *arg++ = other;
2726                         *arg++ = xfrm_msg;
2727                 }
2728         } else {
2729                 *arg++ = pgm;
2730                 *arg++ = name;
2731         }
2732         *arg = NULL;
2733         fflush(NULL);
2734         retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2735         remove_tempfile();
2736         if (retval) {
2737                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2738                 exit(1);
2739         }
2740 }
2741
2742 static int similarity_index(struct diff_filepair *p)
2743 {
2744         return p->score * 100 / MAX_SCORE;
2745 }
2746
2747 static void fill_metainfo(struct strbuf *msg,
2748                           const char *name,
2749                           const char *other,
2750                           struct diff_filespec *one,
2751                           struct diff_filespec *two,
2752                           struct diff_options *o,
2753                           struct diff_filepair *p,
2754                           int *must_show_header,
2755                           int use_color)
2756 {
2757         const char *set = diff_get_color(use_color, DIFF_METAINFO);
2758         const char *reset = diff_get_color(use_color, DIFF_RESET);
2759         struct strbuf *msgbuf;
2760         char *line_prefix = "";
2761
2762         *must_show_header = 1;
2763         if (o->output_prefix) {
2764                 msgbuf = o->output_prefix(o, o->output_prefix_data);
2765                 line_prefix = msgbuf->buf;
2766         }
2767         strbuf_init(msg, PATH_MAX * 2 + 300);
2768         switch (p->status) {
2769         case DIFF_STATUS_COPIED:
2770                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2771                             line_prefix, set, similarity_index(p));
2772                 strbuf_addf(msg, "%s\n%s%scopy from ",
2773                             reset,  line_prefix, set);
2774                 quote_c_style(name, msg, NULL, 0);
2775                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
2776                 quote_c_style(other, msg, NULL, 0);
2777                 strbuf_addf(msg, "%s\n", reset);
2778                 break;
2779         case DIFF_STATUS_RENAMED:
2780                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2781                             line_prefix, set, similarity_index(p));
2782                 strbuf_addf(msg, "%s\n%s%srename from ",
2783                             reset, line_prefix, set);
2784                 quote_c_style(name, msg, NULL, 0);
2785                 strbuf_addf(msg, "%s\n%s%srename to ",
2786                             reset, line_prefix, set);
2787                 quote_c_style(other, msg, NULL, 0);
2788                 strbuf_addf(msg, "%s\n", reset);
2789                 break;
2790         case DIFF_STATUS_MODIFIED:
2791                 if (p->score) {
2792                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
2793                                     line_prefix,
2794                                     set, similarity_index(p), reset);
2795                         break;
2796                 }
2797                 /* fallthru */
2798         default:
2799                 *must_show_header = 0;
2800         }
2801         if (one && two && hashcmp(one->sha1, two->sha1)) {
2802                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2803
2804                 if (DIFF_OPT_TST(o, BINARY)) {
2805                         mmfile_t mf;
2806                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2807                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2808                                 abbrev = 40;
2809                 }
2810                 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
2811                             find_unique_abbrev(one->sha1, abbrev));
2812                 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
2813                 if (one->mode == two->mode)
2814                         strbuf_addf(msg, " %06o", one->mode);
2815                 strbuf_addf(msg, "%s\n", reset);
2816         }
2817 }
2818
2819 static void run_diff_cmd(const char *pgm,
2820                          const char *name,
2821                          const char *other,
2822                          const char *attr_path,
2823                          struct diff_filespec *one,
2824                          struct diff_filespec *two,
2825                          struct strbuf *msg,
2826                          struct diff_options *o,
2827                          struct diff_filepair *p)
2828 {
2829         const char *xfrm_msg = NULL;
2830         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2831         int must_show_header = 0;
2832
2833         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2834                 pgm = NULL;
2835         else {
2836                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2837                 if (drv && drv->external)
2838                         pgm = drv->external;
2839         }
2840
2841         if (msg) {
2842                 /*
2843                  * don't use colors when the header is intended for an
2844                  * external diff driver
2845                  */
2846                 fill_metainfo(msg, name, other, one, two, o, p,
2847                               &must_show_header,
2848                               DIFF_OPT_TST(o, COLOR_DIFF) && !pgm);
2849                 xfrm_msg = msg->len ? msg->buf : NULL;
2850         }
2851
2852         if (pgm) {
2853                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2854                                   complete_rewrite);
2855                 return;
2856         }
2857         if (one && two)
2858                 builtin_diff(name, other ? other : name,
2859                              one, two, xfrm_msg, must_show_header,
2860                              o, complete_rewrite);
2861         else
2862                 fprintf(o->file, "* Unmerged path %s\n", name);
2863 }
2864
2865 static void diff_fill_sha1_info(struct diff_filespec *one)
2866 {
2867         if (DIFF_FILE_VALID(one)) {
2868                 if (!one->sha1_valid) {
2869                         struct stat st;
2870                         if (!strcmp(one->path, "-")) {
2871                                 hashcpy(one->sha1, null_sha1);
2872                                 return;
2873                         }
2874                         if (lstat(one->path, &st) < 0)
2875                                 die_errno("stat '%s'", one->path);
2876                         if (index_path(one->sha1, one->path, &st, 0))
2877                                 die("cannot hash %s", one->path);
2878                 }
2879         }
2880         else
2881                 hashclr(one->sha1);
2882 }
2883
2884 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2885 {
2886         /* Strip the prefix but do not molest /dev/null and absolute paths */
2887         if (*namep && **namep != '/') {
2888                 *namep += prefix_length;
2889                 if (**namep == '/')
2890                         ++*namep;
2891         }
2892         if (*otherp && **otherp != '/') {
2893                 *otherp += prefix_length;
2894                 if (**otherp == '/')
2895                         ++*otherp;
2896         }
2897 }
2898
2899 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2900 {
2901         const char *pgm = external_diff();
2902         struct strbuf msg;
2903         struct diff_filespec *one = p->one;
2904         struct diff_filespec *two = p->two;
2905         const char *name;
2906         const char *other;
2907         const char *attr_path;
2908
2909         name  = p->one->path;
2910         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2911         attr_path = name;
2912         if (o->prefix_length)
2913                 strip_prefix(o->prefix_length, &name, &other);
2914
2915         if (DIFF_PAIR_UNMERGED(p)) {
2916                 run_diff_cmd(pgm, name, NULL, attr_path,
2917                              NULL, NULL, NULL, o, p);
2918                 return;
2919         }
2920
2921         diff_fill_sha1_info(one);
2922         diff_fill_sha1_info(two);
2923
2924         if (!pgm &&
2925             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2926             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2927                 /*
2928                  * a filepair that changes between file and symlink
2929                  * needs to be split into deletion and creation.
2930                  */
2931                 struct diff_filespec *null = alloc_filespec(two->path);
2932                 run_diff_cmd(NULL, name, other, attr_path,
2933                              one, null, &msg, o, p);
2934                 free(null);
2935                 strbuf_release(&msg);
2936
2937                 null = alloc_filespec(one->path);
2938                 run_diff_cmd(NULL, name, other, attr_path,
2939                              null, two, &msg, o, p);
2940                 free(null);
2941         }
2942         else
2943                 run_diff_cmd(pgm, name, other, attr_path,
2944                              one, two, &msg, o, p);
2945
2946         strbuf_release(&msg);
2947 }
2948
2949 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2950                          struct diffstat_t *diffstat)
2951 {
2952         const char *name;
2953         const char *other;
2954         int complete_rewrite = 0;
2955
2956         if (DIFF_PAIR_UNMERGED(p)) {
2957                 /* unmerged */
2958                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2959                 return;
2960         }
2961
2962         name = p->one->path;
2963         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2964
2965         if (o->prefix_length)
2966                 strip_prefix(o->prefix_length, &name, &other);
2967
2968         diff_fill_sha1_info(p->one);
2969         diff_fill_sha1_info(p->two);
2970
2971         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2972                 complete_rewrite = 1;
2973         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2974 }
2975
2976 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2977 {
2978         const char *name;
2979         const char *other;
2980         const char *attr_path;
2981
2982         if (DIFF_PAIR_UNMERGED(p)) {
2983                 /* unmerged */
2984                 return;
2985         }
2986
2987         name = p->one->path;
2988         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2989         attr_path = other ? other : name;
2990
2991         if (o->prefix_length)
2992                 strip_prefix(o->prefix_length, &name, &other);
2993
2994         diff_fill_sha1_info(p->one);
2995         diff_fill_sha1_info(p->two);
2996
2997         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2998 }
2999
3000 void diff_setup(struct diff_options *options)
3001 {
3002         memcpy(options, &default_diff_options, sizeof(*options));
3003
3004         options->file = stdout;
3005
3006         options->line_termination = '\n';
3007         options->break_opt = -1;
3008         options->rename_limit = -1;
3009         options->dirstat_permille = diff_dirstat_permille_default;
3010         options->context = 3;
3011
3012         options->change = diff_change;
3013         options->add_remove = diff_addremove;
3014         if (diff_use_color_default > 0)
3015                 DIFF_OPT_SET(options, COLOR_DIFF);
3016         options->detect_rename = diff_detect_rename_default;
3017
3018         if (diff_no_prefix) {
3019                 options->a_prefix = options->b_prefix = "";
3020         } else if (!diff_mnemonic_prefix) {
3021                 options->a_prefix = "a/";
3022                 options->b_prefix = "b/";
3023         }
3024 }
3025
3026 int diff_setup_done(struct diff_options *options)
3027 {
3028         int count = 0;
3029
3030         if (options->output_format & DIFF_FORMAT_NAME)
3031                 count++;
3032         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3033                 count++;
3034         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3035                 count++;
3036         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3037                 count++;
3038         if (count > 1)
3039                 die("--name-only, --name-status, --check and -s are mutually exclusive");
3040
3041         /*
3042          * Most of the time we can say "there are changes"
3043          * only by checking if there are changed paths, but
3044          * --ignore-whitespace* options force us to look
3045          * inside contents.
3046          */
3047
3048         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3049             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3050             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3051                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3052         else
3053                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3054
3055         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3056                 options->detect_rename = DIFF_DETECT_COPY;
3057
3058         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3059                 options->prefix = NULL;
3060         if (options->prefix)
3061                 options->prefix_length = strlen(options->prefix);
3062         else
3063                 options->prefix_length = 0;
3064
3065         if (options->output_format & (DIFF_FORMAT_NAME |
3066                                       DIFF_FORMAT_NAME_STATUS |
3067                                       DIFF_FORMAT_CHECKDIFF |
3068                                       DIFF_FORMAT_NO_OUTPUT))
3069                 options->output_format &= ~(DIFF_FORMAT_RAW |
3070                                             DIFF_FORMAT_NUMSTAT |
3071                                             DIFF_FORMAT_DIFFSTAT |
3072                                             DIFF_FORMAT_SHORTSTAT |
3073                                             DIFF_FORMAT_DIRSTAT |
3074                                             DIFF_FORMAT_SUMMARY |
3075                                             DIFF_FORMAT_PATCH);
3076
3077         /*
3078          * These cases always need recursive; we do not drop caller-supplied
3079          * recursive bits for other formats here.
3080          */
3081         if (options->output_format & (DIFF_FORMAT_PATCH |
3082                                       DIFF_FORMAT_NUMSTAT |
3083                                       DIFF_FORMAT_DIFFSTAT |
3084                                       DIFF_FORMAT_SHORTSTAT |
3085                                       DIFF_FORMAT_DIRSTAT |
3086                                       DIFF_FORMAT_SUMMARY |
3087                                       DIFF_FORMAT_CHECKDIFF))
3088                 DIFF_OPT_SET(options, RECURSIVE);
3089         /*
3090          * Also pickaxe would not work very well if you do not say recursive
3091          */
3092         if (options->pickaxe)
3093                 DIFF_OPT_SET(options, RECURSIVE);
3094         /*
3095          * When patches are generated, submodules diffed against the work tree
3096          * must be checked for dirtiness too so it can be shown in the output
3097          */
3098         if (options->output_format & DIFF_FORMAT_PATCH)
3099                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3100
3101         if (options->detect_rename && options->rename_limit < 0)
3102                 options->rename_limit = diff_rename_limit_default;
3103         if (options->setup & DIFF_SETUP_USE_CACHE) {
3104                 if (!active_cache)
3105                         /* read-cache does not die even when it fails
3106                          * so it is safe for us to do this here.  Also
3107                          * it does not smudge active_cache or active_nr
3108                          * when it fails, so we do not have to worry about
3109                          * cleaning it up ourselves either.
3110                          */
3111                         read_cache();
3112         }
3113         if (options->abbrev <= 0 || 40 < options->abbrev)
3114                 options->abbrev = 40; /* full */
3115
3116         /*
3117          * It does not make sense to show the first hit we happened
3118          * to have found.  It does not make sense not to return with
3119          * exit code in such a case either.
3120          */
3121         if (DIFF_OPT_TST(options, QUICK)) {
3122                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3123                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3124         }
3125
3126         return 0;
3127 }
3128
3129 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3130 {
3131         char c, *eq;
3132         int len;
3133
3134         if (*arg != '-')
3135                 return 0;
3136         c = *++arg;
3137         if (!c)
3138                 return 0;
3139         if (c == arg_short) {
3140                 c = *++arg;
3141                 if (!c)
3142                         return 1;
3143                 if (val && isdigit(c)) {
3144                         char *end;
3145                         int n = strtoul(arg, &end, 10);
3146                         if (*end)
3147                                 return 0;
3148                         *val = n;
3149                         return 1;
3150                 }
3151                 return 0;
3152         }
3153         if (c != '-')
3154                 return 0;
3155         arg++;
3156         eq = strchr(arg, '=');
3157         if (eq)
3158                 len = eq - arg;
3159         else
3160                 len = strlen(arg);
3161         if (!len || strncmp(arg, arg_long, len))
3162                 return 0;
3163         if (eq) {
3164                 int n;
3165                 char *end;
3166                 if (!isdigit(*++eq))
3167                         return 0;
3168                 n = strtoul(eq, &end, 10);
3169                 if (*end)
3170                         return 0;
3171                 *val = n;
3172         }
3173         return 1;
3174 }
3175
3176 static int diff_scoreopt_parse(const char *opt);
3177
3178 static inline int short_opt(char opt, const char **argv,
3179                             const char **optarg)
3180 {
3181         const char *arg = argv[0];
3182         if (arg[0] != '-' || arg[1] != opt)
3183                 return 0;
3184         if (arg[2] != '\0') {
3185                 *optarg = arg + 2;
3186                 return 1;
3187         }
3188         if (!argv[1])
3189                 die("Option '%c' requires a value", opt);
3190         *optarg = argv[1];
3191         return 2;
3192 }
3193
3194 int parse_long_opt(const char *opt, const char **argv,
3195                    const char **optarg)
3196 {
3197         const char *arg = argv[0];
3198         if (arg[0] != '-' || arg[1] != '-')
3199                 return 0;
3200         arg += strlen("--");
3201         if (prefixcmp(arg, opt))
3202                 return 0;
3203         arg += strlen(opt);
3204         if (*arg == '=') { /* sticked form: --option=value */
3205                 *optarg = arg + 1;
3206                 return 1;
3207         }
3208         if (*arg != '\0')
3209                 return 0;
3210         /* separate form: --option value */
3211         if (!argv[1])
3212                 die("Option '--%s' requires a value", opt);
3213         *optarg = argv[1];
3214         return 2;
3215 }
3216
3217 static int stat_opt(struct diff_options *options, const char **av)
3218 {
3219         const char *arg = av[0];
3220         char *end;
3221         int width = options->stat_width;
3222         int name_width = options->stat_name_width;
3223         int argcount = 1;
3224
3225         arg += strlen("--stat");
3226         end = (char *)arg;
3227
3228         switch (*arg) {
3229         case '-':
3230                 if (!prefixcmp(arg, "-width")) {
3231                         arg += strlen("-width");
3232                         if (*arg == '=')
3233                                 width = strtoul(arg + 1, &end, 10);
3234                         else if (!*arg && !av[1])
3235                                 die("Option '--stat-width' requires a value");
3236                         else if (!*arg) {
3237                                 width = strtoul(av[1], &end, 10);
3238                                 argcount = 2;
3239                         }
3240                 } else if (!prefixcmp(arg, "-name-width")) {
3241                         arg += strlen("-name-width");
3242                         if (*arg == '=')
3243                                 name_width = strtoul(arg + 1, &end, 10);
3244                         else if (!*arg && !av[1])
3245                                 die("Option '--stat-name-width' requires a value");
3246                         else if (!*arg) {
3247                                 name_width = strtoul(av[1], &end, 10);
3248                                 argcount = 2;
3249                         }
3250                 }
3251                 break;
3252         case '=':
3253                 width = strtoul(arg+1, &end, 10);
3254                 if (*end == ',')
3255                         name_width = strtoul(end+1, &end, 10);
3256         }
3257
3258         /* Important! This checks all the error cases! */
3259         if (*end)
3260                 return 0;
3261         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3262         options->stat_name_width = name_width;
3263         options->stat_width = width;
3264         return argcount;
3265 }
3266
3267 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3268 {
3269         struct strbuf errmsg = STRBUF_INIT;
3270         if (parse_dirstat_params(options, params, &errmsg))
3271                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3272                     errmsg.buf);
3273         strbuf_release(&errmsg);
3274         /*
3275          * The caller knows a dirstat-related option is given from the command
3276          * line; allow it to say "return this_function();"
3277          */
3278         options->output_format |= DIFF_FORMAT_DIRSTAT;
3279         return 1;
3280 }
3281
3282 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3283 {
3284         const char *arg = av[0];
3285         const char *optarg;
3286         int argcount;
3287
3288         /* Output format options */
3289         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch"))
3290                 options->output_format |= DIFF_FORMAT_PATCH;
3291         else if (opt_arg(arg, 'U', "unified", &options->context))
3292                 options->output_format |= DIFF_FORMAT_PATCH;
3293         else if (!strcmp(arg, "--raw"))
3294                 options->output_format |= DIFF_FORMAT_RAW;
3295         else if (!strcmp(arg, "--patch-with-raw"))
3296                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
3297         else if (!strcmp(arg, "--numstat"))
3298                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3299         else if (!strcmp(arg, "--shortstat"))
3300                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3301         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3302                 return parse_dirstat_opt(options, "");
3303         else if (!prefixcmp(arg, "-X"))
3304                 return parse_dirstat_opt(options, arg + 2);
3305         else if (!prefixcmp(arg, "--dirstat="))
3306                 return parse_dirstat_opt(options, arg + 10);
3307         else if (!strcmp(arg, "--cumulative"))
3308                 return parse_dirstat_opt(options, "cumulative");
3309         else if (!strcmp(arg, "--dirstat-by-file"))
3310                 return parse_dirstat_opt(options, "files");
3311         else if (!prefixcmp(arg, "--dirstat-by-file=")) {
3312                 parse_dirstat_opt(options, "files");
3313                 return parse_dirstat_opt(options, arg + 18);
3314         }
3315         else if (!strcmp(arg, "--check"))
3316                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3317         else if (!strcmp(arg, "--summary"))
3318                 options->output_format |= DIFF_FORMAT_SUMMARY;
3319         else if (!strcmp(arg, "--patch-with-stat"))
3320                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
3321         else if (!strcmp(arg, "--name-only"))
3322                 options->output_format |= DIFF_FORMAT_NAME;
3323         else if (!strcmp(arg, "--name-status"))
3324                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3325         else if (!strcmp(arg, "-s"))
3326                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3327         else if (!prefixcmp(arg, "--stat"))
3328                 /* --stat, --stat-width, or --stat-name-width */
3329                 return stat_opt(options, av);
3330
3331         /* renames options */
3332         else if (!prefixcmp(arg, "-B") || !prefixcmp(arg, "--break-rewrites=") ||
3333                  !strcmp(arg, "--break-rewrites")) {
3334                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3335                         return error("invalid argument to -B: %s", arg+2);
3336         }
3337         else if (!prefixcmp(arg, "-M") || !prefixcmp(arg, "--find-renames=") ||
3338                  !strcmp(arg, "--find-renames")) {
3339                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3340                         return error("invalid argument to -M: %s", arg+2);
3341                 options->detect_rename = DIFF_DETECT_RENAME;
3342         }
3343         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3344                 options->irreversible_delete = 1;
3345         }
3346         else if (!prefixcmp(arg, "-C") || !prefixcmp(arg, "--find-copies=") ||
3347                  !strcmp(arg, "--find-copies")) {
3348                 if (options->detect_rename == DIFF_DETECT_COPY)
3349                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3350                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3351                         return error("invalid argument to -C: %s", arg+2);
3352                 options->detect_rename = DIFF_DETECT_COPY;
3353         }
3354         else if (!strcmp(arg, "--no-renames"))
3355                 options->detect_rename = 0;
3356         else if (!strcmp(arg, "--relative"))
3357                 DIFF_OPT_SET(options, RELATIVE_NAME);
3358         else if (!prefixcmp(arg, "--relative=")) {
3359                 DIFF_OPT_SET(options, RELATIVE_NAME);
3360                 options->prefix = arg + 11;
3361         }
3362
3363         /* xdiff options */
3364         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3365                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3366         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3367                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3368         else if (!strcmp(arg, "--ignore-space-at-eol"))
3369                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3370         else if (!strcmp(arg, "--patience"))
3371                 DIFF_XDL_SET(options, PATIENCE_DIFF);
3372
3373         /* flags options */
3374         else if (!strcmp(arg, "--binary")) {
3375                 options->output_format |= DIFF_FORMAT_PATCH;
3376                 DIFF_OPT_SET(options, BINARY);
3377         }
3378         else if (!strcmp(arg, "--full-index"))
3379                 DIFF_OPT_SET(options, FULL_INDEX);
3380         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3381                 DIFF_OPT_SET(options, TEXT);
3382         else if (!strcmp(arg, "-R"))
3383                 DIFF_OPT_SET(options, REVERSE_DIFF);
3384         else if (!strcmp(arg, "--find-copies-harder"))
3385                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3386         else if (!strcmp(arg, "--follow"))
3387                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3388         else if (!strcmp(arg, "--color"))
3389                 DIFF_OPT_SET(options, COLOR_DIFF);
3390         else if (!prefixcmp(arg, "--color=")) {
3391                 int value = git_config_colorbool(NULL, arg+8, -1);
3392                 if (value == 0)
3393                         DIFF_OPT_CLR(options, COLOR_DIFF);
3394                 else if (value > 0)
3395                         DIFF_OPT_SET(options, COLOR_DIFF);
3396                 else
3397                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3398         }
3399         else if (!strcmp(arg, "--no-color"))
3400                 DIFF_OPT_CLR(options, COLOR_DIFF);
3401         else if (!strcmp(arg, "--color-words")) {
3402                 DIFF_OPT_SET(options, COLOR_DIFF);
3403                 options->word_diff = DIFF_WORDS_COLOR;
3404         }
3405         else if (!prefixcmp(arg, "--color-words=")) {
3406                 DIFF_OPT_SET(options, COLOR_DIFF);
3407                 options->word_diff = DIFF_WORDS_COLOR;
3408                 options->word_regex = arg + 14;
3409         }
3410         else if (!strcmp(arg, "--word-diff")) {
3411                 if (options->word_diff == DIFF_WORDS_NONE)
3412                         options->word_diff = DIFF_WORDS_PLAIN;
3413         }
3414         else if (!prefixcmp(arg, "--word-diff=")) {
3415                 const char *type = arg + 12;
3416                 if (!strcmp(type, "plain"))
3417                         options->word_diff = DIFF_WORDS_PLAIN;
3418                 else if (!strcmp(type, "color")) {
3419                         DIFF_OPT_SET(options, COLOR_DIFF);
3420                         options->word_diff = DIFF_WORDS_COLOR;
3421                 }
3422                 else if (!strcmp(type, "porcelain"))
3423                         options->word_diff = DIFF_WORDS_PORCELAIN;
3424                 else if (!strcmp(type, "none"))
3425                         options->word_diff = DIFF_WORDS_NONE;
3426                 else
3427                         die("bad --word-diff argument: %s", type);
3428         }
3429         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3430                 if (options->word_diff == DIFF_WORDS_NONE)
3431                         options->word_diff = DIFF_WORDS_PLAIN;
3432                 options->word_regex = optarg;
3433                 return argcount;
3434         }
3435         else if (!strcmp(arg, "--exit-code"))
3436                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3437         else if (!strcmp(arg, "--quiet"))
3438                 DIFF_OPT_SET(options, QUICK);
3439         else if (!strcmp(arg, "--ext-diff"))
3440                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3441         else if (!strcmp(arg, "--no-ext-diff"))
3442                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3443         else if (!strcmp(arg, "--textconv"))
3444                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3445         else if (!strcmp(arg, "--no-textconv"))
3446                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3447         else if (!strcmp(arg, "--ignore-submodules")) {
3448                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3449                 handle_ignore_submodules_arg(options, "all");
3450         } else if (!prefixcmp(arg, "--ignore-submodules=")) {
3451                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3452                 handle_ignore_submodules_arg(options, arg + 20);
3453         } else if (!strcmp(arg, "--submodule"))
3454                 DIFF_OPT_SET(options, SUBMODULE_LOG);
3455         else if (!prefixcmp(arg, "--submodule=")) {
3456                 if (!strcmp(arg + 12, "log"))
3457                         DIFF_OPT_SET(options, SUBMODULE_LOG);
3458         }
3459
3460         /* misc options */
3461         else if (!strcmp(arg, "-z"))
3462                 options->line_termination = 0;
3463         else if ((argcount = short_opt('l', av, &optarg))) {
3464                 options->rename_limit = strtoul(optarg, NULL, 10);
3465                 return argcount;
3466         }
3467         else if ((argcount = short_opt('S', av, &optarg))) {
3468                 options->pickaxe = optarg;
3469                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3470                 return argcount;
3471         } else if ((argcount = short_opt('G', av, &optarg))) {
3472                 options->pickaxe = optarg;
3473                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3474                 return argcount;
3475         }
3476         else if (!strcmp(arg, "--pickaxe-all"))
3477                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3478         else if (!strcmp(arg, "--pickaxe-regex"))
3479                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3480         else if ((argcount = short_opt('O', av, &optarg))) {
3481                 options->orderfile = optarg;
3482                 return argcount;
3483         }
3484         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3485                 options->filter = optarg;
3486                 return argcount;
3487         }
3488         else if (!strcmp(arg, "--abbrev"))
3489                 options->abbrev = DEFAULT_ABBREV;
3490         else if (!prefixcmp(arg, "--abbrev=")) {
3491                 options->abbrev = strtoul(arg + 9, NULL, 10);
3492                 if (options->abbrev < MINIMUM_ABBREV)
3493                         options->abbrev = MINIMUM_ABBREV;
3494                 else if (40 < options->abbrev)
3495                         options->abbrev = 40;
3496         }
3497         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3498                 options->a_prefix = optarg;
3499                 return argcount;
3500         }
3501         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3502                 options->b_prefix = optarg;
3503                 return argcount;
3504         }
3505         else if (!strcmp(arg, "--no-prefix"))
3506                 options->a_prefix = options->b_prefix = "";
3507         else if (opt_arg(arg, '\0', "inter-hunk-context",
3508                          &options->interhunkcontext))
3509                 ;
3510         else if ((argcount = parse_long_opt("output", av, &optarg))) {
3511                 options->file = fopen(optarg, "w");
3512                 if (!options->file)
3513                         die_errno("Could not open '%s'", optarg);
3514                 options->close_file = 1;
3515                 return argcount;
3516         } else
3517                 return 0;
3518         return 1;
3519 }
3520
3521 int parse_rename_score(const char **cp_p)
3522 {
3523         unsigned long num, scale;
3524         int ch, dot;
3525         const char *cp = *cp_p;
3526
3527         num = 0;
3528         scale = 1;
3529         dot = 0;
3530         for (;;) {
3531                 ch = *cp;
3532                 if ( !dot && ch == '.' ) {
3533                         scale = 1;
3534                         dot = 1;
3535                 } else if ( ch == '%' ) {
3536                         scale = dot ? scale*100 : 100;
3537                         cp++;   /* % is always at the end */
3538                         break;
3539                 } else if ( ch >= '0' && ch <= '9' ) {
3540                         if ( scale < 100000 ) {
3541                                 scale *= 10;
3542                                 num = (num*10) + (ch-'0');
3543                         }
3544                 } else {
3545                         break;
3546                 }
3547                 cp++;
3548         }
3549         *cp_p = cp;
3550
3551         /* user says num divided by scale and we say internally that
3552          * is MAX_SCORE * num / scale.
3553          */
3554         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
3555 }
3556
3557 static int diff_scoreopt_parse(const char *opt)
3558 {
3559         int opt1, opt2, cmd;
3560
3561         if (*opt++ != '-')
3562                 return -1;
3563         cmd = *opt++;
3564         if (cmd == '-') {
3565                 /* convert the long-form arguments into short-form versions */
3566                 if (!prefixcmp(opt, "break-rewrites")) {
3567                         opt += strlen("break-rewrites");
3568                         if (*opt == 0 || *opt++ == '=')
3569                                 cmd = 'B';
3570                 } else if (!prefixcmp(opt, "find-copies")) {
3571                         opt += strlen("find-copies");
3572                         if (*opt == 0 || *opt++ == '=')
3573                                 cmd = 'C';
3574                 } else if (!prefixcmp(opt, "find-renames")) {
3575                         opt += strlen("find-renames");
3576                         if (*opt == 0 || *opt++ == '=')
3577                                 cmd = 'M';
3578                 }
3579         }
3580         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
3581                 return -1; /* that is not a -M, -C nor -B option */
3582
3583         opt1 = parse_rename_score(&opt);
3584         if (cmd != 'B')
3585                 opt2 = 0;
3586         else {
3587                 if (*opt == 0)
3588                         opt2 = 0;
3589                 else if (*opt != '/')
3590                         return -1; /* we expect -B80/99 or -B80 */
3591                 else {
3592                         opt++;
3593                         opt2 = parse_rename_score(&opt);
3594                 }
3595         }
3596         if (*opt != 0)
3597                 return -1;
3598         return opt1 | (opt2 << 16);
3599 }
3600
3601 struct diff_queue_struct diff_queued_diff;
3602
3603 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
3604 {
3605         if (queue->alloc <= queue->nr) {
3606                 queue->alloc = alloc_nr(queue->alloc);
3607                 queue->queue = xrealloc(queue->queue,
3608                                         sizeof(dp) * queue->alloc);
3609         }
3610         queue->queue[queue->nr++] = dp;
3611 }
3612
3613 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
3614                                  struct diff_filespec *one,
3615                                  struct diff_filespec *two)
3616 {
3617         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3618         dp->one = one;
3619         dp->two = two;
3620         if (queue)
3621                 diff_q(queue, dp);
3622         return dp;
3623 }
3624
3625 void diff_free_filepair(struct diff_filepair *p)
3626 {
3627         free_filespec(p->one);
3628         free_filespec(p->two);
3629         free(p);
3630 }
3631
3632 /* This is different from find_unique_abbrev() in that
3633  * it stuffs the result with dots for alignment.
3634  */
3635 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3636 {
3637         int abblen;
3638         const char *abbrev;
3639         if (len == 40)
3640                 return sha1_to_hex(sha1);
3641
3642         abbrev = find_unique_abbrev(sha1, len);
3643         abblen = strlen(abbrev);
3644         if (abblen < 37) {
3645                 static char hex[41];
3646                 if (len < abblen && abblen <= len + 2)
3647                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3648                 else
3649                         sprintf(hex, "%s...", abbrev);
3650                 return hex;
3651         }
3652         return sha1_to_hex(sha1);
3653 }
3654
3655 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3656 {
3657         int line_termination = opt->line_termination;
3658         int inter_name_termination = line_termination ? '\t' : '\0';
3659         if (opt->output_prefix) {
3660                 struct strbuf *msg = NULL;
3661                 msg = opt->output_prefix(opt, opt->output_prefix_data);
3662                 fprintf(opt->file, "%s", msg->buf);
3663         }
3664
3665         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3666                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3667                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
3668                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3669         }
3670         if (p->score) {
3671                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3672                         inter_name_termination);
3673         } else {
3674                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
3675         }
3676
3677         if (p->status == DIFF_STATUS_COPIED ||
3678             p->status == DIFF_STATUS_RENAMED) {
3679                 const char *name_a, *name_b;
3680                 name_a = p->one->path;
3681                 name_b = p->two->path;
3682                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3683                 write_name_quoted(name_a, opt->file, inter_name_termination);
3684                 write_name_quoted(name_b, opt->file, line_termination);
3685         } else {
3686                 const char *name_a, *name_b;
3687                 name_a = p->one->mode ? p->one->path : p->two->path;
3688                 name_b = NULL;
3689                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3690                 write_name_quoted(name_a, opt->file, line_termination);
3691         }
3692 }
3693
3694 int diff_unmodified_pair(struct diff_filepair *p)
3695 {
3696         /* This function is written stricter than necessary to support
3697          * the currently implemented transformers, but the idea is to
3698          * let transformers to produce diff_filepairs any way they want,
3699          * and filter and clean them up here before producing the output.
3700          */
3701         struct diff_filespec *one = p->one, *two = p->two;
3702
3703         if (DIFF_PAIR_UNMERGED(p))
3704                 return 0; /* unmerged is interesting */
3705
3706         /* deletion, addition, mode or type change
3707          * and rename are all interesting.
3708          */
3709         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
3710             DIFF_PAIR_MODE_CHANGED(p) ||
3711             strcmp(one->path, two->path))
3712                 return 0;
3713
3714         /* both are valid and point at the same path.  that is, we are
3715          * dealing with a change.
3716          */
3717         if (one->sha1_valid && two->sha1_valid &&
3718             !hashcmp(one->sha1, two->sha1) &&
3719             !one->dirty_submodule && !two->dirty_submodule)
3720                 return 1; /* no change */
3721         if (!one->sha1_valid && !two->sha1_valid)
3722                 return 1; /* both look at the same file on the filesystem. */
3723         return 0;
3724 }
3725
3726 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
3727 {
3728         if (diff_unmodified_pair(p))
3729                 return;
3730
3731         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3732             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3733                 return; /* no tree diffs in patch format */
3734
3735         run_diff(p, o);
3736 }
3737
3738 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
3739                             struct diffstat_t *diffstat)
3740 {
3741         if (diff_unmodified_pair(p))
3742                 return;
3743
3744         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3745             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3746                 return; /* no useful stat for tree diffs */
3747
3748         run_diffstat(p, o, diffstat);
3749 }
3750
3751 static void diff_flush_checkdiff(struct diff_filepair *p,
3752                 struct diff_options *o)
3753 {
3754         if (diff_unmodified_pair(p))
3755                 return;
3756
3757         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3758             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3759                 return; /* nothing to check in tree diffs */
3760
3761         run_checkdiff(p, o);
3762 }
3763
3764 int diff_queue_is_empty(void)
3765 {
3766         struct diff_queue_struct *q = &diff_queued_diff;
3767         int i;
3768         for (i = 0; i < q->nr; i++)
3769                 if (!diff_unmodified_pair(q->queue[i]))
3770                         return 0;
3771         return 1;
3772 }
3773
3774 #if DIFF_DEBUG
3775 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
3776 {
3777         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
3778                 x, one ? one : "",
3779                 s->path,
3780                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
3781                 s->mode,
3782                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
3783         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
3784                 x, one ? one : "",
3785                 s->size, s->xfrm_flags);
3786 }
3787
3788 void diff_debug_filepair(const struct diff_filepair *p, int i)
3789 {
3790         diff_debug_filespec(p->one, i, "one");
3791         diff_debug_filespec(p->two, i, "two");
3792         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
3793                 p->score, p->status ? p->status : '?',
3794                 p->one->rename_used, p->broken_pair);
3795 }
3796
3797 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
3798 {
3799         int i;
3800         if (msg)
3801                 fprintf(stderr, "%s\n", msg);
3802         fprintf(stderr, "q->nr = %d\n", q->nr);
3803         for (i = 0; i < q->nr; i++) {
3804                 struct diff_filepair *p = q->queue[i];
3805                 diff_debug_filepair(p, i);
3806         }
3807 }
3808 #endif
3809
3810 static void diff_resolve_rename_copy(void)
3811 {
3812         int i;
3813         struct diff_filepair *p;
3814         struct diff_queue_struct *q = &diff_queued_diff;
3815
3816         diff_debug_queue("resolve-rename-copy", q);
3817
3818         for (i = 0; i < q->nr; i++) {
3819                 p = q->queue[i];
3820                 p->status = 0; /* undecided */
3821                 if (DIFF_PAIR_UNMERGED(p))
3822                         p->status = DIFF_STATUS_UNMERGED;
3823                 else if (!DIFF_FILE_VALID(p->one))
3824                         p->status = DIFF_STATUS_ADDED;
3825                 else if (!DIFF_FILE_VALID(p->two))
3826                         p->status = DIFF_STATUS_DELETED;
3827                 else if (DIFF_PAIR_TYPE_CHANGED(p))
3828                         p->status = DIFF_STATUS_TYPE_CHANGED;
3829
3830                 /* from this point on, we are dealing with a pair
3831                  * whose both sides are valid and of the same type, i.e.
3832                  * either in-place edit or rename/copy edit.
3833                  */
3834                 else if (DIFF_PAIR_RENAME(p)) {
3835                         /*
3836                          * A rename might have re-connected a broken
3837                          * pair up, causing the pathnames to be the
3838                          * same again. If so, that's not a rename at
3839                          * all, just a modification..
3840                          *
3841                          * Otherwise, see if this source was used for
3842                          * multiple renames, in which case we decrement
3843                          * the count, and call it a copy.
3844                          */
3845                         if (!strcmp(p->one->path, p->two->path))
3846                                 p->status = DIFF_STATUS_MODIFIED;
3847                         else if (--p->one->rename_used > 0)
3848                                 p->status = DIFF_STATUS_COPIED;
3849                         else
3850                                 p->status = DIFF_STATUS_RENAMED;
3851                 }
3852                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
3853                          p->one->mode != p->two->mode ||
3854                          p->one->dirty_submodule ||
3855                          p->two->dirty_submodule ||
3856                          is_null_sha1(p->one->sha1))
3857                         p->status = DIFF_STATUS_MODIFIED;
3858                 else {
3859                         /* This is a "no-change" entry and should not
3860                          * happen anymore, but prepare for broken callers.
3861                          */
3862                         error("feeding unmodified %s to diffcore",
3863                               p->one->path);
3864                         p->status = DIFF_STATUS_UNKNOWN;
3865                 }
3866         }
3867         diff_debug_queue("resolve-rename-copy done", q);
3868 }
3869
3870 static int check_pair_status(struct diff_filepair *p)
3871 {
3872         switch (p->status) {
3873         case DIFF_STATUS_UNKNOWN:
3874                 return 0;
3875         case 0:
3876                 die("internal error in diff-resolve-rename-copy");
3877         default:
3878                 return 1;
3879         }
3880 }
3881
3882 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3883 {
3884         int fmt = opt->output_format;
3885
3886         if (fmt & DIFF_FORMAT_CHECKDIFF)
3887                 diff_flush_checkdiff(p, opt);
3888         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3889                 diff_flush_raw(p, opt);
3890         else if (fmt & DIFF_FORMAT_NAME) {
3891                 const char *name_a, *name_b;
3892                 name_a = p->two->path;
3893                 name_b = NULL;
3894                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3895                 write_name_quoted(name_a, opt->file, opt->line_termination);
3896         }
3897 }
3898
3899 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3900 {
3901         if (fs->mode)
3902                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3903         else
3904                 fprintf(file, " %s ", newdelete);
3905         write_name_quoted(fs->path, file, '\n');
3906 }
3907
3908
3909 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
3910                 const char *line_prefix)
3911 {
3912         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3913                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
3914                         p->two->mode, show_name ? ' ' : '\n');
3915                 if (show_name) {
3916                         write_name_quoted(p->two->path, file, '\n');
3917                 }
3918         }
3919 }
3920
3921 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
3922                         const char *line_prefix)
3923 {
3924         char *names = pprint_rename(p->one->path, p->two->path);
3925
3926         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3927         free(names);
3928         show_mode_change(file, p, 0, line_prefix);
3929 }
3930
3931 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
3932 {
3933         FILE *file = opt->file;
3934         char *line_prefix = "";
3935
3936         if (opt->output_prefix) {
3937                 struct strbuf *buf = opt->output_prefix(opt, opt->output_prefix_data);
3938                 line_prefix = buf->buf;
3939         }
3940
3941         switch(p->status) {
3942         case DIFF_STATUS_DELETED:
3943                 fputs(line_prefix, file);
3944                 show_file_mode_name(file, "delete", p->one);
3945                 break;
3946         case DIFF_STATUS_ADDED:
3947                 fputs(line_prefix, file);
3948                 show_file_mode_name(file, "create", p->two);
3949                 break;
3950         case DIFF_STATUS_COPIED:
3951                 fputs(line_prefix, file);
3952                 show_rename_copy(file, "copy", p, line_prefix);
3953                 break;
3954         case DIFF_STATUS_RENAMED:
3955                 fputs(line_prefix, file);
3956                 show_rename_copy(file, "rename", p, line_prefix);
3957                 break;
3958         default:
3959                 if (p->score) {
3960                         fprintf(file, "%s rewrite ", line_prefix);
3961                         write_name_quoted(p->two->path, file, ' ');
3962                         fprintf(file, "(%d%%)\n", similarity_index(p));
3963                 }
3964                 show_mode_change(file, p, !p->score, line_prefix);
3965                 break;
3966         }
3967 }
3968
3969 struct patch_id_t {
3970         git_SHA_CTX *ctx;
3971         int patchlen;
3972 };
3973
3974 static int remove_space(char *line, int len)
3975 {
3976         int i;
3977         char *dst = line;
3978         unsigned char c;
3979
3980         for (i = 0; i < len; i++)
3981                 if (!isspace((c = line[i])))
3982                         *dst++ = c;
3983
3984         return dst - line;
3985 }
3986
3987 static void patch_id_consume(void *priv, char *line, unsigned long len)
3988 {
3989         struct patch_id_t *data = priv;
3990         int new_len;
3991
3992         /* Ignore line numbers when computing the SHA1 of the patch */
3993         if (!prefixcmp(line, "@@ -"))
3994                 return;
3995
3996         new_len = remove_space(line, len);
3997
3998         git_SHA1_Update(data->ctx, line, new_len);
3999         data->patchlen += new_len;
4000 }
4001
4002 /* returns 0 upon success, and writes result into sha1 */
4003 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
4004 {
4005         struct diff_queue_struct *q = &diff_queued_diff;
4006         int i;
4007         git_SHA_CTX ctx;
4008         struct patch_id_t data;
4009         char buffer[PATH_MAX * 4 + 20];
4010
4011         git_SHA1_Init(&ctx);
4012         memset(&data, 0, sizeof(struct patch_id_t));
4013         data.ctx = &ctx;
4014
4015         for (i = 0; i < q->nr; i++) {
4016                 xpparam_t xpp;
4017                 xdemitconf_t xecfg;
4018                 mmfile_t mf1, mf2;
4019                 struct diff_filepair *p = q->queue[i];
4020                 int len1, len2;
4021
4022                 memset(&xpp, 0, sizeof(xpp));
4023                 memset(&xecfg, 0, sizeof(xecfg));
4024                 if (p->status == 0)
4025                         return error("internal diff status error");
4026                 if (p->status == DIFF_STATUS_UNKNOWN)
4027                         continue;
4028                 if (diff_unmodified_pair(p))
4029                         continue;
4030                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4031                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4032                         continue;
4033                 if (DIFF_PAIR_UNMERGED(p))
4034                         continue;
4035
4036                 diff_fill_sha1_info(p->one);
4037                 diff_fill_sha1_info(p->two);
4038                 if (fill_mmfile(&mf1, p->one) < 0 ||
4039                                 fill_mmfile(&mf2, p->two) < 0)
4040                         return error("unable to read files to diff");
4041
4042                 len1 = remove_space(p->one->path, strlen(p->one->path));
4043                 len2 = remove_space(p->two->path, strlen(p->two->path));
4044                 if (p->one->mode == 0)
4045                         len1 = snprintf(buffer, sizeof(buffer),
4046                                         "diff--gita/%.*sb/%.*s"
4047                                         "newfilemode%06o"
4048                                         "---/dev/null"
4049                                         "+++b/%.*s",
4050                                         len1, p->one->path,
4051                                         len2, p->two->path,
4052                                         p->two->mode,
4053                                         len2, p->two->path);
4054                 else if (p->two->mode == 0)
4055                         len1 = snprintf(buffer, sizeof(buffer),
4056                                         "diff--gita/%.*sb/%.*s"
4057                                         "deletedfilemode%06o"
4058                                         "---a/%.*s"
4059                                         "+++/dev/null",
4060                                         len1, p->one->path,
4061                                         len2, p->two->path,
4062                                         p->one->mode,
4063                                         len1, p->one->path);
4064                 else
4065                         len1 = snprintf(buffer, sizeof(buffer),
4066                                         "diff--gita/%.*sb/%.*s"
4067                                         "---a/%.*s"
4068                                         "+++b/%.*s",
4069                                         len1, p->one->path,
4070                                         len2, p->two->path,
4071                                         len1, p->one->path,
4072                                         len2, p->two->path);
4073                 git_SHA1_Update(&ctx, buffer, len1);
4074
4075                 if (diff_filespec_is_binary(p->one) ||
4076                     diff_filespec_is_binary(p->two)) {
4077                         git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
4078                         git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
4079                         continue;
4080                 }
4081
4082                 xpp.flags = 0;
4083                 xecfg.ctxlen = 3;
4084                 xecfg.flags = 0;
4085                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4086                               &xpp, &xecfg);
4087         }
4088
4089         git_SHA1_Final(sha1, &ctx);
4090         return 0;
4091 }
4092
4093 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
4094 {
4095         struct diff_queue_struct *q = &diff_queued_diff;
4096         int i;
4097         int result = diff_get_patch_id(options, sha1);
4098
4099         for (i = 0; i < q->nr; i++)
4100                 diff_free_filepair(q->queue[i]);
4101
4102         free(q->queue);
4103         DIFF_QUEUE_CLEAR(q);
4104
4105         return result;
4106 }
4107
4108 static int is_summary_empty(const struct diff_queue_struct *q)
4109 {
4110         int i;
4111
4112         for (i = 0; i < q->nr; i++) {
4113                 const struct diff_filepair *p = q->queue[i];
4114
4115                 switch (p->status) {
4116                 case DIFF_STATUS_DELETED:
4117                 case DIFF_STATUS_ADDED:
4118                 case DIFF_STATUS_COPIED:
4119                 case DIFF_STATUS_RENAMED:
4120                         return 0;
4121                 default:
4122                         if (p->score)
4123                                 return 0;
4124                         if (p->one->mode && p->two->mode &&
4125                             p->one->mode != p->two->mode)
4126                                 return 0;
4127                         break;
4128                 }
4129         }
4130         return 1;
4131 }
4132
4133 static const char rename_limit_warning[] =
4134 "inexact rename detection was skipped due to too many files.";
4135
4136 static const char degrade_cc_to_c_warning[] =
4137 "only found copies from modified paths due to too many files.";
4138
4139 static const char rename_limit_advice[] =
4140 "you may want to set your %s variable to at least "
4141 "%d and retry the command.";
4142
4143 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4144 {
4145         if (degraded_cc)
4146                 warning(degrade_cc_to_c_warning);
4147         else if (needed)
4148                 warning(rename_limit_warning);
4149         else
4150                 return;
4151         if (0 < needed && needed < 32767)
4152                 warning(rename_limit_advice, varname, needed);
4153 }
4154
4155 void diff_flush(struct diff_options *options)
4156 {
4157         struct diff_queue_struct *q = &diff_queued_diff;
4158         int i, output_format = options->output_format;
4159         int separator = 0;
4160         int dirstat_by_line = 0;
4161
4162         /*
4163          * Order: raw, stat, summary, patch
4164          * or:    name/name-status/checkdiff (other bits clear)
4165          */
4166         if (!q->nr)
4167                 goto free_queue;
4168
4169         if (output_format & (DIFF_FORMAT_RAW |
4170                              DIFF_FORMAT_NAME |
4171                              DIFF_FORMAT_NAME_STATUS |
4172                              DIFF_FORMAT_CHECKDIFF)) {
4173                 for (i = 0; i < q->nr; i++) {
4174                         struct diff_filepair *p = q->queue[i];
4175                         if (check_pair_status(p))
4176                                 flush_one_pair(p, options);
4177                 }
4178                 separator++;
4179         }
4180
4181         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4182                 dirstat_by_line = 1;
4183
4184         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4185             dirstat_by_line) {
4186                 struct diffstat_t diffstat;
4187
4188                 memset(&diffstat, 0, sizeof(struct diffstat_t));
4189                 for (i = 0; i < q->nr; i++) {
4190                         struct diff_filepair *p = q->queue[i];
4191                         if (check_pair_status(p))
4192                                 diff_flush_stat(p, options, &diffstat);
4193                 }
4194                 if (output_format & DIFF_FORMAT_NUMSTAT)
4195                         show_numstat(&diffstat, options);
4196                 if (output_format & DIFF_FORMAT_DIFFSTAT)
4197                         show_stats(&diffstat, options);
4198                 if (output_format & DIFF_FORMAT_SHORTSTAT)
4199                         show_shortstats(&diffstat, options);
4200                 if (output_format & DIFF_FORMAT_DIRSTAT)
4201                         show_dirstat_by_line(&diffstat, options);
4202                 free_diffstat_info(&diffstat);
4203                 separator++;
4204         }
4205         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4206                 show_dirstat(options);
4207
4208         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4209                 for (i = 0; i < q->nr; i++) {
4210                         diff_summary(options, q->queue[i]);
4211                 }
4212                 separator++;
4213         }
4214
4215         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4216             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4217             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4218                 /*
4219                  * run diff_flush_patch for the exit status. setting
4220                  * options->file to /dev/null should be safe, becaue we
4221                  * aren't supposed to produce any output anyway.
4222                  */
4223                 if (options->close_file)
4224                         fclose(options->file);
4225                 options->file = fopen("/dev/null", "w");
4226                 if (!options->file)
4227                         die_errno("Could not open /dev/null");
4228                 options->close_file = 1;
4229                 for (i = 0; i < q->nr; i++) {
4230                         struct diff_filepair *p = q->queue[i];
4231                         if (check_pair_status(p))
4232                                 diff_flush_patch(p, options);
4233                         if (options->found_changes)
4234                                 break;
4235                 }
4236         }
4237
4238         if (output_format & DIFF_FORMAT_PATCH) {
4239                 if (separator) {
4240                         putc(options->line_termination, options->file);
4241                         if (options->stat_sep) {
4242                                 /* attach patch instead of inline */
4243                                 fputs(options->stat_sep, options->file);
4244                         }
4245                 }
4246
4247                 for (i = 0; i < q->nr; i++) {
4248                         struct diff_filepair *p = q->queue[i];
4249                         if (check_pair_status(p))
4250                                 diff_flush_patch(p, options);
4251                 }
4252         }
4253
4254         if (output_format & DIFF_FORMAT_CALLBACK)
4255                 options->format_callback(q, options, options->format_callback_data);
4256
4257         for (i = 0; i < q->nr; i++)
4258                 diff_free_filepair(q->queue[i]);
4259 free_queue:
4260         free(q->queue);
4261         DIFF_QUEUE_CLEAR(q);
4262         if (options->close_file)
4263                 fclose(options->file);
4264
4265         /*
4266          * Report the content-level differences with HAS_CHANGES;
4267          * diff_addremove/diff_change does not set the bit when
4268          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4269          */
4270         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4271                 if (options->found_changes)
4272                         DIFF_OPT_SET(options, HAS_CHANGES);
4273                 else
4274                         DIFF_OPT_CLR(options, HAS_CHANGES);
4275         }
4276 }
4277
4278 static void diffcore_apply_filter(const char *filter)
4279 {
4280         int i;
4281         struct diff_queue_struct *q = &diff_queued_diff;
4282         struct diff_queue_struct outq;
4283         DIFF_QUEUE_CLEAR(&outq);
4284
4285         if (!filter)
4286                 return;
4287
4288         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
4289                 int found;
4290                 for (i = found = 0; !found && i < q->nr; i++) {
4291                         struct diff_filepair *p = q->queue[i];
4292                         if (((p->status == DIFF_STATUS_MODIFIED) &&
4293                              ((p->score &&
4294                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
4295                               (!p->score &&
4296                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
4297                             ((p->status != DIFF_STATUS_MODIFIED) &&
4298                              strchr(filter, p->status)))
4299                                 found++;
4300                 }
4301                 if (found)
4302                         return;
4303
4304                 /* otherwise we will clear the whole queue
4305                  * by copying the empty outq at the end of this
4306                  * function, but first clear the current entries
4307                  * in the queue.
4308                  */
4309                 for (i = 0; i < q->nr; i++)
4310                         diff_free_filepair(q->queue[i]);
4311         }
4312         else {
4313                 /* Only the matching ones */
4314                 for (i = 0; i < q->nr; i++) {
4315                         struct diff_filepair *p = q->queue[i];
4316
4317                         if (((p->status == DIFF_STATUS_MODIFIED) &&
4318                              ((p->score &&
4319                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
4320                               (!p->score &&
4321                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
4322                             ((p->status != DIFF_STATUS_MODIFIED) &&
4323                              strchr(filter, p->status)))
4324                                 diff_q(&outq, p);
4325                         else
4326                                 diff_free_filepair(p);
4327                 }
4328         }
4329         free(q->queue);
4330         *q = outq;
4331 }
4332
4333 /* Check whether two filespecs with the same mode and size are identical */
4334 static int diff_filespec_is_identical(struct diff_filespec *one,
4335                                       struct diff_filespec *two)
4336 {
4337         if (S_ISGITLINK(one->mode))
4338                 return 0;
4339         if (diff_populate_filespec(one, 0))
4340                 return 0;
4341         if (diff_populate_filespec(two, 0))
4342                 return 0;
4343         return !memcmp(one->data, two->data, one->size);
4344 }
4345
4346 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4347 {
4348         int i;
4349         struct diff_queue_struct *q = &diff_queued_diff;
4350         struct diff_queue_struct outq;
4351         DIFF_QUEUE_CLEAR(&outq);
4352
4353         for (i = 0; i < q->nr; i++) {
4354                 struct diff_filepair *p = q->queue[i];
4355
4356                 /*
4357                  * 1. Entries that come from stat info dirtiness
4358                  *    always have both sides (iow, not create/delete),
4359                  *    one side of the object name is unknown, with
4360                  *    the same mode and size.  Keep the ones that
4361                  *    do not match these criteria.  They have real
4362                  *    differences.
4363                  *
4364                  * 2. At this point, the file is known to be modified,
4365                  *    with the same mode and size, and the object
4366                  *    name of one side is unknown.  Need to inspect
4367                  *    the identical contents.
4368                  */
4369                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
4370                     !DIFF_FILE_VALID(p->two) ||
4371                     (p->one->sha1_valid && p->two->sha1_valid) ||
4372                     (p->one->mode != p->two->mode) ||
4373                     diff_populate_filespec(p->one, 1) ||
4374                     diff_populate_filespec(p->two, 1) ||
4375                     (p->one->size != p->two->size) ||
4376                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4377                         diff_q(&outq, p);
4378                 else {
4379                         /*
4380                          * The caller can subtract 1 from skip_stat_unmatch
4381                          * to determine how many paths were dirty only
4382                          * due to stat info mismatch.
4383                          */
4384                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4385                                 diffopt->skip_stat_unmatch++;
4386                         diff_free_filepair(p);
4387                 }
4388         }
4389         free(q->queue);
4390         *q = outq;
4391 }
4392
4393 static int diffnamecmp(const void *a_, const void *b_)
4394 {
4395         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4396         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4397         const char *name_a, *name_b;
4398
4399         name_a = a->one ? a->one->path : a->two->path;
4400         name_b = b->one ? b->one->path : b->two->path;
4401         return strcmp(name_a, name_b);
4402 }
4403
4404 void diffcore_fix_diff_index(struct diff_options *options)
4405 {
4406         struct diff_queue_struct *q = &diff_queued_diff;
4407         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4408 }
4409
4410 void diffcore_std(struct diff_options *options)
4411 {
4412         if (options->skip_stat_unmatch)
4413                 diffcore_skip_stat_unmatch(options);
4414         if (!options->found_follow) {
4415                 /* See try_to_follow_renames() in tree-diff.c */
4416                 if (options->break_opt != -1)
4417                         diffcore_break(options->break_opt);
4418                 if (options->detect_rename)
4419                         diffcore_rename(options);
4420                 if (options->break_opt != -1)
4421                         diffcore_merge_broken();
4422         }
4423         if (options->pickaxe)
4424                 diffcore_pickaxe(options);
4425         if (options->orderfile)
4426                 diffcore_order(options->orderfile);
4427         if (!options->found_follow)
4428                 /* See try_to_follow_renames() in tree-diff.c */
4429                 diff_resolve_rename_copy();
4430         diffcore_apply_filter(options->filter);
4431
4432         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4433                 DIFF_OPT_SET(options, HAS_CHANGES);
4434         else
4435                 DIFF_OPT_CLR(options, HAS_CHANGES);
4436
4437         options->found_follow = 0;
4438 }
4439
4440 int diff_result_code(struct diff_options *opt, int status)
4441 {
4442         int result = 0;
4443
4444         diff_warn_rename_limit("diff.renamelimit",
4445                                opt->needed_rename_limit,
4446                                opt->degraded_cc_to_c);
4447         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4448             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4449                 return status;
4450         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4451             DIFF_OPT_TST(opt, HAS_CHANGES))
4452                 result |= 01;
4453         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4454             DIFF_OPT_TST(opt, CHECK_FAILED))
4455                 result |= 02;
4456         return result;
4457 }
4458
4459 int diff_can_quit_early(struct diff_options *opt)
4460 {
4461         return (DIFF_OPT_TST(opt, QUICK) &&
4462                 !opt->filter &&
4463                 DIFF_OPT_TST(opt, HAS_CHANGES));
4464 }
4465
4466 /*
4467  * Shall changes to this submodule be ignored?
4468  *
4469  * Submodule changes can be configured to be ignored separately for each path,
4470  * but that configuration can be overridden from the command line.
4471  */
4472 static int is_submodule_ignored(const char *path, struct diff_options *options)
4473 {
4474         int ignored = 0;
4475         unsigned orig_flags = options->flags;
4476         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4477                 set_diffopt_flags_from_submodule_config(options, path);
4478         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4479                 ignored = 1;
4480         options->flags = orig_flags;
4481         return ignored;
4482 }
4483
4484 void diff_addremove(struct diff_options *options,
4485                     int addremove, unsigned mode,
4486                     const unsigned char *sha1,
4487                     const char *concatpath, unsigned dirty_submodule)
4488 {
4489         struct diff_filespec *one, *two;
4490
4491         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4492                 return;
4493
4494         /* This may look odd, but it is a preparation for
4495          * feeding "there are unchanged files which should
4496          * not produce diffs, but when you are doing copy
4497          * detection you would need them, so here they are"
4498          * entries to the diff-core.  They will be prefixed
4499          * with something like '=' or '*' (I haven't decided
4500          * which but should not make any difference).
4501          * Feeding the same new and old to diff_change()
4502          * also has the same effect.
4503          * Before the final output happens, they are pruned after
4504          * merged into rename/copy pairs as appropriate.
4505          */
4506         if (DIFF_OPT_TST(options, REVERSE_DIFF))
4507                 addremove = (addremove == '+' ? '-' :
4508                              addremove == '-' ? '+' : addremove);
4509
4510         if (options->prefix &&
4511             strncmp(concatpath, options->prefix, options->prefix_length))
4512                 return;
4513
4514         one = alloc_filespec(concatpath);
4515         two = alloc_filespec(concatpath);
4516
4517         if (addremove != '+')
4518                 fill_filespec(one, sha1, mode);
4519         if (addremove != '-') {
4520                 fill_filespec(two, sha1, mode);
4521                 two->dirty_submodule = dirty_submodule;
4522         }
4523
4524         diff_queue(&diff_queued_diff, one, two);
4525         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4526                 DIFF_OPT_SET(options, HAS_CHANGES);
4527 }
4528
4529 void diff_change(struct diff_options *options,
4530                  unsigned old_mode, unsigned new_mode,
4531                  const unsigned char *old_sha1,
4532                  const unsigned char *new_sha1,
4533                  const char *concatpath,
4534                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4535 {
4536         struct diff_filespec *one, *two;
4537
4538         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4539             is_submodule_ignored(concatpath, options))
4540                 return;
4541
4542         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4543                 unsigned tmp;
4544                 const unsigned char *tmp_c;
4545                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4546                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4547                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
4548                         new_dirty_submodule = tmp;
4549         }
4550
4551         if (options->prefix &&
4552             strncmp(concatpath, options->prefix, options->prefix_length))
4553                 return;
4554
4555         one = alloc_filespec(concatpath);
4556         two = alloc_filespec(concatpath);
4557         fill_filespec(one, old_sha1, old_mode);
4558         fill_filespec(two, new_sha1, new_mode);
4559         one->dirty_submodule = old_dirty_submodule;
4560         two->dirty_submodule = new_dirty_submodule;
4561
4562         diff_queue(&diff_queued_diff, one, two);
4563         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4564                 DIFF_OPT_SET(options, HAS_CHANGES);
4565 }
4566
4567 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
4568 {
4569         struct diff_filepair *pair;
4570         struct diff_filespec *one, *two;
4571
4572         if (options->prefix &&
4573             strncmp(path, options->prefix, options->prefix_length))
4574                 return NULL;
4575
4576         one = alloc_filespec(path);
4577         two = alloc_filespec(path);
4578         pair = diff_queue(&diff_queued_diff, one, two);
4579         pair->is_unmerged = 1;
4580         return pair;
4581 }
4582
4583 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
4584                 size_t *outsize)
4585 {
4586         struct diff_tempfile *temp;
4587         const char *argv[3];
4588         const char **arg = argv;
4589         struct child_process child;
4590         struct strbuf buf = STRBUF_INIT;
4591         int err = 0;
4592
4593         temp = prepare_temp_file(spec->path, spec);
4594         *arg++ = pgm;
4595         *arg++ = temp->name;
4596         *arg = NULL;
4597
4598         memset(&child, 0, sizeof(child));
4599         child.use_shell = 1;
4600         child.argv = argv;
4601         child.out = -1;
4602         if (start_command(&child)) {
4603                 remove_tempfile();
4604                 return NULL;
4605         }
4606
4607         if (strbuf_read(&buf, child.out, 0) < 0)
4608                 err = error("error reading from textconv command '%s'", pgm);
4609         close(child.out);
4610
4611         if (finish_command(&child) || err) {
4612                 strbuf_release(&buf);
4613                 remove_tempfile();
4614                 return NULL;
4615         }
4616         remove_tempfile();
4617
4618         return strbuf_detach(&buf, outsize);
4619 }
4620
4621 size_t fill_textconv(struct userdiff_driver *driver,
4622                      struct diff_filespec *df,
4623                      char **outbuf)
4624 {
4625         size_t size;
4626
4627         if (!driver || !driver->textconv) {
4628                 if (!DIFF_FILE_VALID(df)) {
4629                         *outbuf = "";
4630                         return 0;
4631                 }
4632                 if (diff_populate_filespec(df, 0))
4633                         die("unable to read files to diff");
4634                 *outbuf = df->data;
4635                 return df->size;
4636         }
4637
4638         if (driver->textconv_cache && df->sha1_valid) {
4639                 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
4640                                           &size);
4641                 if (*outbuf)
4642                         return size;
4643         }
4644
4645         *outbuf = run_textconv(driver->textconv, df, &size);
4646         if (!*outbuf)
4647                 die("unable to read files to diff");
4648
4649         if (driver->textconv_cache && df->sha1_valid) {
4650                 /* ignore errors, as we might be in a readonly repository */
4651                 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
4652                                 size);
4653                 /*
4654                  * we could save up changes and flush them all at the end,
4655                  * but we would need an extra call after all diffing is done.
4656                  * Since generating a cache entry is the slow path anyway,
4657                  * this extra overhead probably isn't a big deal.
4658                  */
4659                 notes_cache_write(driver->textconv_cache);
4660         }
4661
4662         return size;
4663 }