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