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