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

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