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 static 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_stat_graph_width;
  35 static int diff_dirstat_permille_default = 30;
  36 static struct diff_options default_diff_options;
  37
  38 static char diff_colors[][COLOR_MAXLEN] = {
  39         GIT_COLOR_RESET,
  40         GIT_COLOR_NORMAL,       /* PLAIN */
  41         GIT_COLOR_BOLD,         /* METAINFO */
  42         GIT_COLOR_CYAN,         /* FRAGINFO */
  43         GIT_COLOR_RED,          /* OLD */
  44         GIT_COLOR_GREEN,        /* NEW */
  45         GIT_COLOR_YELLOW,       /* COMMIT */
  46         GIT_COLOR_BG_RED,       /* WHITESPACE */
  47         GIT_COLOR_NORMAL,       /* FUNCINFO */
  48 };
  49
  50 static int parse_diff_color_slot(const char *var, int ofs)
  51 {
  52         if (!strcasecmp(var+ofs, "plain"))
  53                 return DIFF_PLAIN;
  54         if (!strcasecmp(var+ofs, "meta"))
  55                 return DIFF_METAINFO;
  56         if (!strcasecmp(var+ofs, "frag"))
  57                 return DIFF_FRAGINFO;
  58         if (!strcasecmp(var+ofs, "old"))
  59                 return DIFF_FILE_OLD;
  60         if (!strcasecmp(var+ofs, "new"))
  61                 return DIFF_FILE_NEW;
  62         if (!strcasecmp(var+ofs, "commit"))
  63                 return DIFF_COMMIT;
  64         if (!strcasecmp(var+ofs, "whitespace"))
  65                 return DIFF_WHITESPACE;
  66         if (!strcasecmp(var+ofs, "func"))
  67                 return DIFF_FUNCINFO;
  68         return -1;
  69 }
  70
  71 static int parse_dirstat_params(struct diff_options *options, const char *params,
  72                                 struct strbuf *errmsg)
  73 {
  74         const char *p = params;
  75         int p_len, ret = 0;
  76
  77         while (*p) {
  78                 p_len = strchrnul(p, ',') - p;
  79                 if (!memcmp(p, "changes", p_len)) {
  80                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  81                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
  82                 } else if (!memcmp(p, "lines", p_len)) {
  83                         DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
  84                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
  85                 } else if (!memcmp(p, "files", p_len)) {
  86                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  87                         DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
  88                 } else if (!memcmp(p, "noncumulative", p_len)) {
  89                         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
  90                 } else if (!memcmp(p, "cumulative", p_len)) {
  91                         DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
  92                 } else if (isdigit(*p)) {
  93                         char *end;
  94                         int permille = strtoul(p, &end, 10) * 10;
  95                         if (*end == '.' && isdigit(*++end)) {
  96                                 /* only use first digit */
  97                                 permille += *end - '0';
  98                                 /* .. and ignore any further digits */
  99                                 while (isdigit(*++end))
 100                                         ; /* nothing */
 101                         }
 102                         if (end - p == p_len)
 103                                 options->dirstat_permille = permille;
 104                         else {
 105                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%.*s'\n"),
 106                                             p_len, p);
 107                                 ret++;
 108                         }
 109                 } else {
 110                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%.*s'\n"),
 111                                     p_len, p);
 112                         ret++;
 113                 }
 114
 115                 p += p_len;
 116
 117                 if (*p)
 118                         p++; /* more parameters, swallow separator */
 119         }
 120         return ret;
 121 }
 122
 123 static int git_config_rename(const char *var, const char *value)
 124 {
 125         if (!value)
 126                 return DIFF_DETECT_RENAME;
 127         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 128                 return  DIFF_DETECT_COPY;
 129         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 130 }
 131
 132 /*
 133  * These are to give UI layer defaults.
 134  * The core-level commands such as git-diff-files should
 135  * never be affected by the setting of diff.renames
 136  * the user happens to have in the configuration file.
 137  */
 138 int git_diff_ui_config(const char *var, const char *value, void *cb)
 139 {
 140         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 141                 diff_use_color_default = git_config_colorbool(var, value);
 142                 return 0;
 143         }
 144         if (!strcmp(var, "diff.renames")) {
 145                 diff_detect_rename_default = git_config_rename(var, value);
 146                 return 0;
 147         }
 148         if (!strcmp(var, "diff.autorefreshindex")) {
 149                 diff_auto_refresh_index = git_config_bool(var, value);
 150                 return 0;
 151         }
 152         if (!strcmp(var, "diff.mnemonicprefix")) {
 153                 diff_mnemonic_prefix = git_config_bool(var, value);
 154                 return 0;
 155         }
 156         if (!strcmp(var, "diff.noprefix")) {
 157                 diff_no_prefix = git_config_bool(var, value);
 158                 return 0;
 159         }
 160         if (!strcmp(var, "diff.statgraphwidth")) {
 161                 diff_stat_graph_width = git_config_int(var, value);
 162                 return 0;
 163         }
 164         if (!strcmp(var, "diff.external"))
 165                 return git_config_string(&external_diff_cmd_cfg, var, value);
 166         if (!strcmp(var, "diff.wordregex"))
 167                 return git_config_string(&diff_word_regex_cfg, var, value);
 168
 169         if (!strcmp(var, "diff.ignoresubmodules"))
 170                 handle_ignore_submodules_arg(&default_diff_options, value);
 171
 172         if (git_color_config(var, value, cb) < 0)
 173                 return -1;
 174
 175         return git_diff_basic_config(var, value, cb);
 176 }
 177
 178 int git_diff_basic_config(const char *var, const char *value, void *cb)
 179 {
 180         if (!strcmp(var, "diff.renamelimit")) {
 181                 diff_rename_limit_default = git_config_int(var, value);
 182                 return 0;
 183         }
 184
 185         if (userdiff_config(var, value) < 0)
 186                 return -1;
 187
 188         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 189                 int slot = parse_diff_color_slot(var, 11);
 190                 if (slot < 0)
 191                         return 0;
 192                 if (!value)
 193                         return config_error_nonbool(var);
 194                 color_parse(value, var, diff_colors[slot]);
 195                 return 0;
 196         }
 197
 198         /* like GNU diff's --suppress-blank-empty option  */
 199         if (!strcmp(var, "diff.suppressblankempty") ||
 200                         /* for backwards compatibility */
 201                         !strcmp(var, "diff.suppress-blank-empty")) {
 202                 diff_suppress_blank_empty = git_config_bool(var, value);
 203                 return 0;
 204         }
 205
 206         if (!strcmp(var, "diff.dirstat")) {
 207                 struct strbuf errmsg = STRBUF_INIT;
 208                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 209                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 210                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 211                                 errmsg.buf);
 212                 strbuf_release(&errmsg);
 213                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 214                 return 0;
 215         }
 216
 217         if (!prefixcmp(var, "submodule."))
 218                 return parse_submodule_config_option(var, value);
 219
 220         return git_default_config(var, value, cb);
 221 }
 222
 223 static char *quote_two(const char *one, const char *two)
 224 {
 225         int need_one = quote_c_style(one, NULL, NULL, 1);
 226         int need_two = quote_c_style(two, NULL, NULL, 1);
 227         struct strbuf res = STRBUF_INIT;
 228
 229         if (need_one + need_two) {
 230                 strbuf_addch(&res, '"');
 231                 quote_c_style(one, &res, NULL, 1);
 232                 quote_c_style(two, &res, NULL, 1);
 233                 strbuf_addch(&res, '"');
 234         } else {
 235                 strbuf_addstr(&res, one);
 236                 strbuf_addstr(&res, two);
 237         }
 238         return strbuf_detach(&res, NULL);
 239 }
 240
 241 static const char *external_diff(void)
 242 {
 243         static const char *external_diff_cmd = NULL;
 244         static int done_preparing = 0;
 245
 246         if (done_preparing)
 247                 return external_diff_cmd;
 248         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 249         if (!external_diff_cmd)
 250                 external_diff_cmd = external_diff_cmd_cfg;
 251         done_preparing = 1;
 252         return external_diff_cmd;
 253 }
 254
 255 static struct diff_tempfile {
 256         const char *name; /* filename external diff should read from */
 257         char hex[41];
 258         char mode[10];
 259         char tmp_path[PATH_MAX];
 260 } diff_temp[2];
 261
 262 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 263
 264 struct emit_callback {
 265         int color_diff;
 266         unsigned ws_rule;
 267         int blank_at_eof_in_preimage;
 268         int blank_at_eof_in_postimage;
 269         int lno_in_preimage;
 270         int lno_in_postimage;
 271         sane_truncate_fn truncate;
 272         const char **label_path;
 273         struct diff_words_data *diff_words;
 274         struct diff_options *opt;
 275         int *found_changesp;
 276         struct strbuf *header;
 277 };
 278
 279 static int count_lines(const char *data, int size)
 280 {
 281         int count, ch, completely_empty = 1, nl_just_seen = 0;
 282         count = 0;
 283         while (0 < size--) {
 284                 ch = *data++;
 285                 if (ch == '\n') {
 286                         count++;
 287                         nl_just_seen = 1;
 288                         completely_empty = 0;
 289                 }
 290                 else {
 291                         nl_just_seen = 0;
 292                         completely_empty = 0;
 293                 }
 294         }
 295         if (completely_empty)
 296                 return 0;
 297         if (!nl_just_seen)
 298                 count++; /* no trailing newline */
 299         return count;
 300 }
 301
 302 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 303 {
 304         if (!DIFF_FILE_VALID(one)) {
 305                 mf->ptr = (char *)""; /* does not matter */
 306                 mf->size = 0;
 307                 return 0;
 308         }
 309         else if (diff_populate_filespec(one, 0))
 310                 return -1;
 311
 312         mf->ptr = one->data;
 313         mf->size = one->size;
 314         return 0;
 315 }
 316
 317 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 318 static unsigned long diff_filespec_size(struct diff_filespec *one)
 319 {
 320         if (!DIFF_FILE_VALID(one))
 321                 return 0;
 322         diff_populate_filespec(one, 1);
 323         return one->size;
 324 }
 325
 326 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 327 {
 328         char *ptr = mf->ptr;
 329         long size = mf->size;
 330         int cnt = 0;
 331
 332         if (!size)
 333                 return cnt;
 334         ptr += size - 1; /* pointing at the very end */
 335         if (*ptr != '\n')
 336                 ; /* incomplete line */
 337         else
 338                 ptr--; /* skip the last LF */
 339         while (mf->ptr < ptr) {
 340                 char *prev_eol;
 341                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 342                         if (*prev_eol == '\n')
 343                                 break;
 344                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 345                         break;
 346                 cnt++;
 347                 ptr = prev_eol - 1;
 348         }
 349         return cnt;
 350 }
 351
 352 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 353                                struct emit_callback *ecbdata)
 354 {
 355         int l1, l2, at;
 356         unsigned ws_rule = ecbdata->ws_rule;
 357         l1 = count_trailing_blank(mf1, ws_rule);
 358         l2 = count_trailing_blank(mf2, ws_rule);
 359         if (l2 <= l1) {
 360                 ecbdata->blank_at_eof_in_preimage = 0;
 361                 ecbdata->blank_at_eof_in_postimage = 0;
 362                 return;
 363         }
 364         at = count_lines(mf1->ptr, mf1->size);
 365         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 366
 367         at = count_lines(mf2->ptr, mf2->size);
 368         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 369 }
 370
 371 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 372                         int first, const char *line, int len)
 373 {
 374         int has_trailing_newline, has_trailing_carriage_return;
 375         int nofirst;
 376         FILE *file = o->file;
 377
 378         if (o->output_prefix) {
 379                 struct strbuf *msg = NULL;
 380                 msg = o->output_prefix(o, o->output_prefix_data);
 381                 assert(msg);
 382                 fwrite(msg->buf, msg->len, 1, file);
 383         }
 384
 385         if (len == 0) {
 386                 has_trailing_newline = (first == '\n');
 387                 has_trailing_carriage_return = (!has_trailing_newline &&
 388                                                 (first == '\r'));
 389                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 390         } else {
 391                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 392                 if (has_trailing_newline)
 393                         len--;
 394                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 395                 if (has_trailing_carriage_return)
 396                         len--;
 397                 nofirst = 0;
 398         }
 399
 400         if (len || !nofirst) {
 401                 fputs(set, file);
 402                 if (!nofirst)
 403                         fputc(first, file);
 404                 fwrite(line, len, 1, file);
 405                 fputs(reset, file);
 406         }
 407         if (has_trailing_carriage_return)
 408                 fputc('\r', file);
 409         if (has_trailing_newline)
 410                 fputc('\n', file);
 411 }
 412
 413 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 414                       const char *line, int len)
 415 {
 416         emit_line_0(o, set, reset, line[0], line+1, len-1);
 417 }
 418
 419 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
 420 {
 421         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
 422               ecbdata->blank_at_eof_in_preimage &&
 423               ecbdata->blank_at_eof_in_postimage &&
 424               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
 425               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
 426                 return 0;
 427         return ws_blank_line(line, len, ecbdata->ws_rule);
 428 }
 429
 430 static void emit_add_line(const char *reset,
 431                           struct emit_callback *ecbdata,
 432                           const char *line, int len)
 433 {
 434         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 435         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 436
 437         if (!*ws)
 438                 emit_line_0(ecbdata->opt, set, reset, '+', line, len);
 439         else if (new_blank_line_at_eof(ecbdata, line, len))
 440                 /* Blank line at EOF - paint '+' as well */
 441                 emit_line_0(ecbdata->opt, ws, reset, '+', line, len);
 442         else {
 443                 /* Emit just the prefix, then the rest. */
 444                 emit_line_0(ecbdata->opt, set, reset, '+', "", 0);
 445                 ws_check_emit(line, len, ecbdata->ws_rule,
 446                               ecbdata->opt->file, set, reset, ws);
 447         }
 448 }
 449
 450 static void emit_hunk_header(struct emit_callback *ecbdata,
 451                              const char *line, int len)
 452 {
 453         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 454         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
 455         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
 456         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 457         static const char atat[2] = { '@', '@' };
 458         const char *cp, *ep;
 459         struct strbuf msgbuf = STRBUF_INIT;
 460         int org_len = len;
 461         int i = 1;
 462
 463         /*
 464          * As a hunk header must begin with "@@ -<old>, +<new> @@",
 465          * it always is at least 10 bytes long.
 466          */
 467         if (len < 10 ||
 468             memcmp(line, atat, 2) ||
 469             !(ep = memmem(line + 2, len - 2, atat, 2))) {
 470                 emit_line(ecbdata->opt, plain, reset, line, len);
 471                 return;
 472         }
 473         ep += 2; /* skip over @@ */
 474
 475         /* The hunk header in fraginfo color */
 476         strbuf_add(&msgbuf, frag, strlen(frag));
 477         strbuf_add(&msgbuf, line, ep - line);
 478         strbuf_add(&msgbuf, reset, strlen(reset));
 479
 480         /*
 481          * trailing "\r\n"
 482          */
 483         for ( ; i < 3; i++)
 484                 if (line[len - i] == '\r' || line[len - i] == '\n')
 485                         len--;
 486
 487         /* blank before the func header */
 488         for (cp = ep; ep - line < len; ep++)
 489                 if (*ep != ' ' && *ep != '\t')
 490                         break;
 491         if (ep != cp) {
 492                 strbuf_add(&msgbuf, plain, strlen(plain));
 493                 strbuf_add(&msgbuf, cp, ep - cp);
 494                 strbuf_add(&msgbuf, reset, strlen(reset));
 495         }
 496
 497         if (ep < line + len) {
 498                 strbuf_add(&msgbuf, func, strlen(func));
 499                 strbuf_add(&msgbuf, ep, line + len - ep);
 500                 strbuf_add(&msgbuf, reset, strlen(reset));
 501         }
 502
 503         strbuf_add(&msgbuf, line + len, org_len - len);
 504         emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
 505         strbuf_release(&msgbuf);
 506 }
 507
 508 static struct diff_tempfile *claim_diff_tempfile(void) {
 509         int i;
 510         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
 511                 if (!diff_temp[i].name)
 512                         return diff_temp + i;
 513         die("BUG: diff is failing to clean up its tempfiles");
 514 }
 515
 516 static int remove_tempfile_installed;
 517
 518 static void remove_tempfile(void)
 519 {
 520         int i;
 521         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
 522                 if (diff_temp[i].name == diff_temp[i].tmp_path)
 523                         unlink_or_warn(diff_temp[i].name);
 524                 diff_temp[i].name = NULL;
 525         }
 526 }
 527
 528 static void remove_tempfile_on_signal(int signo)
 529 {
 530         remove_tempfile();
 531         sigchain_pop(signo);
 532         raise(signo);
 533 }
 534
 535 static void print_line_count(FILE *file, int count)
 536 {
 537         switch (count) {
 538         case 0:
 539                 fprintf(file, "0,0");
 540                 break;
 541         case 1:
 542                 fprintf(file, "1");
 543                 break;
 544         default:
 545                 fprintf(file, "1,%d", count);
 546                 break;
 547         }
 548 }
 549
 550 static void emit_rewrite_lines(struct emit_callback *ecb,
 551                                int prefix, const char *data, int size)
 552 {
 553         const char *endp = NULL;
 554         static const char *nneof = " No newline at end of file\n";
 555         const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
 556         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
 557
 558         while (0 < size) {
 559                 int len;
 560
 561                 endp = memchr(data, '\n', size);
 562                 len = endp ? (endp - data + 1) : size;
 563                 if (prefix != '+') {
 564                         ecb->lno_in_preimage++;
 565                         emit_line_0(ecb->opt, old, reset, '-',
 566                                     data, len);
 567                 } else {
 568                         ecb->lno_in_postimage++;
 569                         emit_add_line(reset, ecb, data, len);
 570                 }
 571                 size -= len;
 572                 data += len;
 573         }
 574         if (!endp) {
 575                 const char *plain = diff_get_color(ecb->color_diff,
 576                                                    DIFF_PLAIN);
 577                 putc('\n', ecb->opt->file);
 578                 emit_line_0(ecb->opt, plain, reset, '\\',
 579                             nneof, strlen(nneof));
 580         }
 581 }
 582
 583 static void emit_rewrite_diff(const char *name_a,
 584                               const char *name_b,
 585                               struct diff_filespec *one,
 586                               struct diff_filespec *two,
 587                               struct userdiff_driver *textconv_one,
 588                               struct userdiff_driver *textconv_two,
 589                               struct diff_options *o)
 590 {
 591         int lc_a, lc_b;
 592         const char *name_a_tab, *name_b_tab;
 593         const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
 594         const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
 595         const char *reset = diff_get_color(o->use_color, DIFF_RESET);
 596         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 597         const char *a_prefix, *b_prefix;
 598         char *data_one, *data_two;
 599         size_t size_one, size_two;
 600         struct emit_callback ecbdata;
 601         char *line_prefix = "";
 602         struct strbuf *msgbuf;
 603
 604         if (o && o->output_prefix) {
 605                 msgbuf = o->output_prefix(o, o->output_prefix_data);
 606                 line_prefix = msgbuf->buf;
 607         }
 608
 609         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 610                 a_prefix = o->b_prefix;
 611                 b_prefix = o->a_prefix;
 612         } else {
 613                 a_prefix = o->a_prefix;
 614                 b_prefix = o->b_prefix;
 615         }
 616
 617         name_a += (*name_a == '/');
 618         name_b += (*name_b == '/');
 619         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 620         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 621
 622         strbuf_reset(&a_name);
 623         strbuf_reset(&b_name);
 624         quote_two_c_style(&a_name, a_prefix, name_a, 0);
 625         quote_two_c_style(&b_name, b_prefix, name_b, 0);
 626
 627         size_one = fill_textconv(textconv_one, one, &data_one);
 628         size_two = fill_textconv(textconv_two, two, &data_two);
 629
 630         memset(&ecbdata, 0, sizeof(ecbdata));
 631         ecbdata.color_diff = want_color(o->use_color);
 632         ecbdata.found_changesp = &o->found_changes;
 633         ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
 634         ecbdata.opt = o;
 635         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
 636                 mmfile_t mf1, mf2;
 637                 mf1.ptr = (char *)data_one;
 638                 mf2.ptr = (char *)data_two;
 639                 mf1.size = size_one;
 640                 mf2.size = size_two;
 641                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
 642         }
 643         ecbdata.lno_in_preimage = 1;
 644         ecbdata.lno_in_postimage = 1;
 645
 646         lc_a = count_lines(data_one, size_one);
 647         lc_b = count_lines(data_two, size_two);
 648         fprintf(o->file,
 649                 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
 650                 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
 651                 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
 652                 line_prefix, fraginfo);
 653         if (!o->irreversible_delete)
 654                 print_line_count(o->file, lc_a);
 655         else
 656                 fprintf(o->file, "?,?");
 657         fprintf(o->file, " +");
 658         print_line_count(o->file, lc_b);
 659         fprintf(o->file, " @@%s\n", reset);
 660         if (lc_a && !o->irreversible_delete)
 661                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
 662         if (lc_b)
 663                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
 664         if (textconv_one)
 665                 free((char *)data_one);
 666         if (textconv_two)
 667                 free((char *)data_two);
 668 }
 669
 670 struct diff_words_buffer {
 671         mmfile_t text;
 672         long alloc;
 673         struct diff_words_orig {
 674                 const char *begin, *end;
 675         } *orig;
 676         int orig_nr, orig_alloc;
 677 };
 678
 679 static void diff_words_append(char *line, unsigned long len,
 680                 struct diff_words_buffer *buffer)
 681 {
 682         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
 683         line++;
 684         len--;
 685         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 686         buffer->text.size += len;
 687         buffer->text.ptr[buffer->text.size] = '\0';
 688 }
 689
 690 struct diff_words_style_elem {
 691         const char *prefix;
 692         const char *suffix;
 693         const char *color; /* NULL; filled in by the setup code if
 694                             * color is enabled */
 695 };
 696
 697 struct diff_words_style {
 698         enum diff_words_type type;
 699         struct diff_words_style_elem new, old, ctx;
 700         const char *newline;
 701 };
 702
 703 static struct diff_words_style diff_words_styles[] = {
 704         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
 705         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
 706         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
 707 };
 708
 709 struct diff_words_data {
 710         struct diff_words_buffer minus, plus;
 711         const char *current_plus;
 712         int last_minus;
 713         struct diff_options *opt;
 714         regex_t *word_regex;
 715         enum diff_words_type type;
 716         struct diff_words_style *style;
 717 };
 718
 719 static int fn_out_diff_words_write_helper(FILE *fp,
 720                                           struct diff_words_style_elem *st_el,
 721                                           const char *newline,
 722                                           size_t count, const char *buf,
 723                                           const char *line_prefix)
 724 {
 725         int print = 0;
 726
 727         while (count) {
 728                 char *p = memchr(buf, '\n', count);
 729                 if (print)
 730                         fputs(line_prefix, fp);
 731                 if (p != buf) {
 732                         if (st_el->color && fputs(st_el->color, fp) < 0)
 733                                 return -1;
 734                         if (fputs(st_el->prefix, fp) < 0 ||
 735                             fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
 736                             fputs(st_el->suffix, fp) < 0)
 737                                 return -1;
 738                         if (st_el->color && *st_el->color
 739                             && fputs(GIT_COLOR_RESET, fp) < 0)
 740                                 return -1;
 741                 }
 742                 if (!p)
 743                         return 0;
 744                 if (fputs(newline, fp) < 0)
 745                         return -1;
 746                 count -= p + 1 - buf;
 747                 buf = p + 1;
 748                 print = 1;
 749         }
 750         return 0;
 751 }
 752
 753 /*
 754  * '--color-words' algorithm can be described as:
 755  *
 756  *   1. collect a the minus/plus lines of a diff hunk, divided into
 757  *      minus-lines and plus-lines;
 758  *
 759  *   2. break both minus-lines and plus-lines into words and
 760  *      place them into two mmfile_t with one word for each line;
 761  *
 762  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
 763  *
 764  * And for the common parts of the both file, we output the plus side text.
 765  * diff_words->current_plus is used to trace the current position of the plus file
 766  * which printed. diff_words->last_minus is used to trace the last minus word
 767  * printed.
 768  *
 769  * For '--graph' to work with '--color-words', we need to output the graph prefix
 770  * on each line of color words output. Generally, there are two conditions on
 771  * which we should output the prefix.
 772  *
 773  *   1. diff_words->last_minus == 0 &&
 774  *      diff_words->current_plus == diff_words->plus.text.ptr
 775  *
 776  *      that is: the plus text must start as a new line, and if there is no minus
 777  *      word printed, a graph prefix must be printed.
 778  *
 779  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
 780  *      *(diff_words->current_plus - 1) == '\n'
 781  *
 782  *      that is: a graph prefix must be printed following a '\n'
 783  */
 784 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
 785 {
 786         if ((diff_words->last_minus == 0 &&
 787                 diff_words->current_plus == diff_words->plus.text.ptr) ||
 788                 (diff_words->current_plus > diff_words->plus.text.ptr &&
 789                 *(diff_words->current_plus - 1) == '\n')) {
 790                 return 1;
 791         } else {
 792                 return 0;
 793         }
 794 }
 795
 796 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 797 {
 798         struct diff_words_data *diff_words = priv;
 799         struct diff_words_style *style = diff_words->style;
 800         int minus_first, minus_len, plus_first, plus_len;
 801         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 802         struct diff_options *opt = diff_words->opt;
 803         struct strbuf *msgbuf;
 804         char *line_prefix = "";
 805
 806         if (line[0] != '@' || parse_hunk_header(line, len,
 807                         &minus_first, &minus_len, &plus_first, &plus_len))
 808                 return;
 809
 810         assert(opt);
 811         if (opt->output_prefix) {
 812                 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
 813                 line_prefix = msgbuf->buf;
 814         }
 815
 816         /* POSIX requires that first be decremented by one if len == 0... */
 817         if (minus_len) {
 818                 minus_begin = diff_words->minus.orig[minus_first].begin;
 819                 minus_end =
 820                         diff_words->minus.orig[minus_first + minus_len - 1].end;
 821         } else
 822                 minus_begin = minus_end =
 823                         diff_words->minus.orig[minus_first].end;
 824
 825         if (plus_len) {
 826                 plus_begin = diff_words->plus.orig[plus_first].begin;
 827                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
 828         } else
 829                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
 830
 831         if (color_words_output_graph_prefix(diff_words)) {
 832                 fputs(line_prefix, diff_words->opt->file);
 833         }
 834         if (diff_words->current_plus != plus_begin) {
 835                 fn_out_diff_words_write_helper(diff_words->opt->file,
 836                                 &style->ctx, style->newline,
 837                                 plus_begin - diff_words->current_plus,
 838                                 diff_words->current_plus, line_prefix);
 839                 if (*(plus_begin - 1) == '\n')
 840                         fputs(line_prefix, diff_words->opt->file);
 841         }
 842         if (minus_begin != minus_end) {
 843                 fn_out_diff_words_write_helper(diff_words->opt->file,
 844                                 &style->old, style->newline,
 845                                 minus_end - minus_begin, minus_begin,
 846                                 line_prefix);
 847         }
 848         if (plus_begin != plus_end) {
 849                 fn_out_diff_words_write_helper(diff_words->opt->file,
 850                                 &style->new, style->newline,
 851                                 plus_end - plus_begin, plus_begin,
 852                                 line_prefix);
 853         }
 854
 855         diff_words->current_plus = plus_end;
 856         diff_words->last_minus = minus_first;
 857 }
 858
 859 /* This function starts looking at *begin, and returns 0 iff a word was found. */
 860 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
 861                 int *begin, int *end)
 862 {
 863         if (word_regex && *begin < buffer->size) {
 864                 regmatch_t match[1];
 865                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
 866                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
 867                                         '\n', match[0].rm_eo - match[0].rm_so);
 868                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
 869                         *begin += match[0].rm_so;
 870                         return *begin >= *end;
 871                 }
 872                 return -1;
 873         }
 874
 875         /* find the next word */
 876         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
 877                 (*begin)++;
 878         if (*begin >= buffer->size)
 879                 return -1;
 880
 881         /* find the end of the word */
 882         *end = *begin + 1;
 883         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
 884                 (*end)++;
 885
 886         return 0;
 887 }
 888
 889 /*
 890  * This function splits the words in buffer->text, stores the list with
 891  * newline separator into out, and saves the offsets of the original words
 892  * in buffer->orig.
 893  */
 894 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
 895                 regex_t *word_regex)
 896 {
 897         int i, j;
 898         long alloc = 0;
 899
 900         out->size = 0;
 901         out->ptr = NULL;
 902
 903         /* fake an empty "0th" word */
 904         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
 905         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
 906         buffer->orig_nr = 1;
 907
 908         for (i = 0; i < buffer->text.size; i++) {
 909                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
 910                         return;
 911
 912                 /* store original boundaries */
 913                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
 914                                 buffer->orig_alloc);
 915                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
 916                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
 917                 buffer->orig_nr++;
 918
 919                 /* store one word */
 920                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
 921                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
 922                 out->ptr[out->size + j - i] = '\n';
 923                 out->size += j - i + 1;
 924
 925                 i = j - 1;
 926         }
 927 }
 928
 929 /* this executes the word diff on the accumulated buffers */
 930 static void diff_words_show(struct diff_words_data *diff_words)
 931 {
 932         xpparam_t xpp;
 933         xdemitconf_t xecfg;
 934         mmfile_t minus, plus;
 935         struct diff_words_style *style = diff_words->style;
 936
 937         struct diff_options *opt = diff_words->opt;
 938         struct strbuf *msgbuf;
 939         char *line_prefix = "";
 940
 941         assert(opt);
 942         if (opt->output_prefix) {
 943                 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
 944                 line_prefix = msgbuf->buf;
 945         }
 946
 947         /* special case: only removal */
 948         if (!diff_words->plus.text.size) {
 949                 fputs(line_prefix, diff_words->opt->file);
 950                 fn_out_diff_words_write_helper(diff_words->opt->file,
 951                         &style->old, style->newline,
 952                         diff_words->minus.text.size,
 953                         diff_words->minus.text.ptr, line_prefix);
 954                 diff_words->minus.text.size = 0;
 955                 return;
 956         }
 957
 958         diff_words->current_plus = diff_words->plus.text.ptr;
 959         diff_words->last_minus = 0;
 960
 961         memset(&xpp, 0, sizeof(xpp));
 962         memset(&xecfg, 0, sizeof(xecfg));
 963         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
 964         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
 965         xpp.flags = 0;
 966         /* as only the hunk header will be parsed, we need a 0-context */
 967         xecfg.ctxlen = 0;
 968         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
 969                       &xpp, &xecfg);
 970         free(minus.ptr);
 971         free(plus.ptr);
 972         if (diff_words->current_plus != diff_words->plus.text.ptr +
 973                         diff_words->plus.text.size) {
 974                 if (color_words_output_graph_prefix(diff_words))
 975                         fputs(line_prefix, diff_words->opt->file);
 976                 fn_out_diff_words_write_helper(diff_words->opt->file,
 977                         &style->ctx, style->newline,
 978                         diff_words->plus.text.ptr + diff_words->plus.text.size
 979                         - diff_words->current_plus, diff_words->current_plus,
 980                         line_prefix);
 981         }
 982         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 983 }
 984
 985 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
 986 static void diff_words_flush(struct emit_callback *ecbdata)
 987 {
 988         if (ecbdata->diff_words->minus.text.size ||
 989             ecbdata->diff_words->plus.text.size)
 990                 diff_words_show(ecbdata->diff_words);
 991 }
 992
 993 static void diff_filespec_load_driver(struct diff_filespec *one)
 994 {
 995         /* Use already-loaded driver */
 996         if (one->driver)
 997                 return;
 998
 999         if (S_ISREG(one->mode))
1000                 one->driver = userdiff_find_by_path(one->path);
1001
1002         /* Fallback to default settings */
1003         if (!one->driver)
1004                 one->driver = userdiff_find_by_name("default");
1005 }
1006
1007 static const char *userdiff_word_regex(struct diff_filespec *one)
1008 {
1009         diff_filespec_load_driver(one);
1010         return one->driver->word_regex;
1011 }
1012
1013 static void init_diff_words_data(struct emit_callback *ecbdata,
1014                                  struct diff_options *orig_opts,
1015                                  struct diff_filespec *one,
1016                                  struct diff_filespec *two)
1017 {
1018         int i;
1019         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1020         memcpy(o, orig_opts, sizeof(struct diff_options));
1021
1022         ecbdata->diff_words =
1023                 xcalloc(1, sizeof(struct diff_words_data));
1024         ecbdata->diff_words->type = o->word_diff;
1025         ecbdata->diff_words->opt = o;
1026         if (!o->word_regex)
1027                 o->word_regex = userdiff_word_regex(one);
1028         if (!o->word_regex)
1029                 o->word_regex = userdiff_word_regex(two);
1030         if (!o->word_regex)
1031                 o->word_regex = diff_word_regex_cfg;
1032         if (o->word_regex) {
1033                 ecbdata->diff_words->word_regex = (regex_t *)
1034                         xmalloc(sizeof(regex_t));
1035                 if (regcomp(ecbdata->diff_words->word_regex,
1036                             o->word_regex,
1037                             REG_EXTENDED | REG_NEWLINE))
1038                         die ("Invalid regular expression: %s",
1039                              o->word_regex);
1040         }
1041         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1042                 if (o->word_diff == diff_words_styles[i].type) {
1043                         ecbdata->diff_words->style =
1044                                 &diff_words_styles[i];
1045                         break;
1046                 }
1047         }
1048         if (want_color(o->use_color)) {
1049                 struct diff_words_style *st = ecbdata->diff_words->style;
1050                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1051                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1052                 st->ctx.color = diff_get_color_opt(o, DIFF_PLAIN);
1053         }
1054 }
1055
1056 static void free_diff_words_data(struct emit_callback *ecbdata)
1057 {
1058         if (ecbdata->diff_words) {
1059                 diff_words_flush(ecbdata);
1060                 free (ecbdata->diff_words->opt);
1061                 free (ecbdata->diff_words->minus.text.ptr);
1062                 free (ecbdata->diff_words->minus.orig);
1063                 free (ecbdata->diff_words->plus.text.ptr);
1064                 free (ecbdata->diff_words->plus.orig);
1065                 if (ecbdata->diff_words->word_regex) {
1066                         regfree(ecbdata->diff_words->word_regex);
1067                         free(ecbdata->diff_words->word_regex);
1068                 }
1069                 free(ecbdata->diff_words);
1070                 ecbdata->diff_words = NULL;
1071         }
1072 }
1073
1074 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1075 {
1076         if (want_color(diff_use_color))
1077                 return diff_colors[ix];
1078         return "";
1079 }
1080
1081 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1082 {
1083         const char *cp;
1084         unsigned long allot;
1085         size_t l = len;
1086
1087         if (ecb->truncate)
1088                 return ecb->truncate(line, len);
1089         cp = line;
1090         allot = l;
1091         while (0 < l) {
1092                 (void) utf8_width(&cp, &l);
1093                 if (!cp)
1094                         break; /* truncated in the middle? */
1095         }
1096         return allot - l;
1097 }
1098
1099 static void find_lno(const char *line, struct emit_callback *ecbdata)
1100 {
1101         const char *p;
1102         ecbdata->lno_in_preimage = 0;
1103         ecbdata->lno_in_postimage = 0;
1104         p = strchr(line, '-');
1105         if (!p)
1106                 return; /* cannot happen */
1107         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1108         p = strchr(p, '+');
1109         if (!p)
1110                 return; /* cannot happen */
1111         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1112 }
1113
1114 static void fn_out_consume(void *priv, char *line, unsigned long len)
1115 {
1116         struct emit_callback *ecbdata = priv;
1117         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1118         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
1119         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1120         struct diff_options *o = ecbdata->opt;
1121         char *line_prefix = "";
1122         struct strbuf *msgbuf;
1123
1124         if (o && o->output_prefix) {
1125                 msgbuf = o->output_prefix(o, o->output_prefix_data);
1126                 line_prefix = msgbuf->buf;
1127         }
1128
1129         if (ecbdata->header) {
1130                 fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
1131                 strbuf_reset(ecbdata->header);
1132                 ecbdata->header = NULL;
1133         }
1134         *(ecbdata->found_changesp) = 1;
1135
1136         if (ecbdata->label_path[0]) {
1137                 const char *name_a_tab, *name_b_tab;
1138
1139                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1140                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1141
1142                 fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1143                         line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1144                 fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1145                         line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1146                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1147         }
1148
1149         if (diff_suppress_blank_empty
1150             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1151                 line[0] = '\n';
1152                 len = 1;
1153         }
1154
1155         if (line[0] == '@') {
1156                 if (ecbdata->diff_words)
1157                         diff_words_flush(ecbdata);
1158                 len = sane_truncate_line(ecbdata, line, len);
1159                 find_lno(line, ecbdata);
1160                 emit_hunk_header(ecbdata, line, len);
1161                 if (line[len-1] != '\n')
1162                         putc('\n', ecbdata->opt->file);
1163                 return;
1164         }
1165
1166         if (len < 1) {
1167                 emit_line(ecbdata->opt, reset, reset, line, len);
1168                 if (ecbdata->diff_words
1169                     && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1170                         fputs("~\n", ecbdata->opt->file);
1171                 return;
1172         }
1173
1174         if (ecbdata->diff_words) {
1175                 if (line[0] == '-') {
1176                         diff_words_append(line, len,
1177                                           &ecbdata->diff_words->minus);
1178                         return;
1179                 } else if (line[0] == '+') {
1180                         diff_words_append(line, len,
1181                                           &ecbdata->diff_words->plus);
1182                         return;
1183                 } else if (!prefixcmp(line, "\\ ")) {
1184                         /*
1185                          * Eat the "no newline at eof" marker as if we
1186                          * saw a "+" or "-" line with nothing on it,
1187                          * and return without diff_words_flush() to
1188                          * defer processing. If this is the end of
1189                          * preimage, more "+" lines may come after it.
1190                          */
1191                         return;
1192                 }
1193                 diff_words_flush(ecbdata);
1194                 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1195                         emit_line(ecbdata->opt, plain, reset, line, len);
1196                         fputs("~\n", ecbdata->opt->file);
1197                 } else {
1198                         /*
1199                          * Skip the prefix character, if any.  With
1200                          * diff_suppress_blank_empty, there may be
1201                          * none.
1202                          */
1203                         if (line[0] != '\n') {
1204                               line++;
1205                               len--;
1206                         }
1207                         emit_line(ecbdata->opt, plain, reset, line, len);
1208                 }
1209                 return;
1210         }
1211
1212         if (line[0] != '+') {
1213                 const char *color =
1214                         diff_get_color(ecbdata->color_diff,
1215                                        line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
1216                 ecbdata->lno_in_preimage++;
1217                 if (line[0] == ' ')
1218                         ecbdata->lno_in_postimage++;
1219                 emit_line(ecbdata->opt, color, reset, line, len);
1220         } else {
1221                 ecbdata->lno_in_postimage++;
1222                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1223         }
1224 }
1225
1226 static char *pprint_rename(const char *a, const char *b)
1227 {
1228         const char *old = a;
1229         const char *new = b;
1230         struct strbuf name = STRBUF_INIT;
1231         int pfx_length, sfx_length;
1232         int len_a = strlen(a);
1233         int len_b = strlen(b);
1234         int a_midlen, b_midlen;
1235         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1236         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1237
1238         if (qlen_a || qlen_b) {
1239                 quote_c_style(a, &name, NULL, 0);
1240                 strbuf_addstr(&name, " => ");
1241                 quote_c_style(b, &name, NULL, 0);
1242                 return strbuf_detach(&name, NULL);
1243         }
1244
1245         /* Find common prefix */
1246         pfx_length = 0;
1247         while (*old && *new && *old == *new) {
1248                 if (*old == '/')
1249                         pfx_length = old - a + 1;
1250                 old++;
1251                 new++;
1252         }
1253
1254         /* Find common suffix */
1255         old = a + len_a;
1256         new = b + len_b;
1257         sfx_length = 0;
1258         while (a <= old && b <= new && *old == *new) {
1259                 if (*old == '/')
1260                         sfx_length = len_a - (old - a);
1261                 old--;
1262                 new--;
1263         }
1264
1265         /*
1266          * pfx{mid-a => mid-b}sfx
1267          * {pfx-a => pfx-b}sfx
1268          * pfx{sfx-a => sfx-b}
1269          * name-a => name-b
1270          */
1271         a_midlen = len_a - pfx_length - sfx_length;
1272         b_midlen = len_b - pfx_length - sfx_length;
1273         if (a_midlen < 0)
1274                 a_midlen = 0;
1275         if (b_midlen < 0)
1276                 b_midlen = 0;
1277
1278         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1279         if (pfx_length + sfx_length) {
1280                 strbuf_add(&name, a, pfx_length);
1281                 strbuf_addch(&name, '{');
1282         }
1283         strbuf_add(&name, a + pfx_length, a_midlen);
1284         strbuf_addstr(&name, " => ");
1285         strbuf_add(&name, b + pfx_length, b_midlen);
1286         if (pfx_length + sfx_length) {
1287                 strbuf_addch(&name, '}');
1288                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1289         }
1290         return strbuf_detach(&name, NULL);
1291 }
1292
1293 struct diffstat_t {
1294         int nr;
1295         int alloc;
1296         struct diffstat_file {
1297                 char *from_name;
1298                 char *name;
1299                 char *print_name;
1300                 unsigned is_unmerged:1;
1301                 unsigned is_binary:1;
1302                 unsigned is_renamed:1;
1303                 unsigned is_interesting:1;
1304                 uintmax_t added, deleted;
1305         } **files;
1306 };
1307
1308 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1309                                           const char *name_a,
1310                                           const char *name_b)
1311 {
1312         struct diffstat_file *x;
1313         x = xcalloc(sizeof (*x), 1);
1314         if (diffstat->nr == diffstat->alloc) {
1315                 diffstat->alloc = alloc_nr(diffstat->alloc);
1316                 diffstat->files = xrealloc(diffstat->files,
1317                                 diffstat->alloc * sizeof(x));
1318         }
1319         diffstat->files[diffstat->nr++] = x;
1320         if (name_b) {
1321                 x->from_name = xstrdup(name_a);
1322                 x->name = xstrdup(name_b);
1323                 x->is_renamed = 1;
1324         }
1325         else {
1326                 x->from_name = NULL;
1327                 x->name = xstrdup(name_a);
1328         }
1329         return x;
1330 }
1331
1332 static void diffstat_consume(void *priv, char *line, unsigned long len)
1333 {
1334         struct diffstat_t *diffstat = priv;
1335         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1336
1337         if (line[0] == '+')
1338                 x->added++;
1339         else if (line[0] == '-')
1340                 x->deleted++;
1341 }
1342
1343 const char mime_boundary_leader[] = "------------";
1344
1345 static int scale_linear(int it, int width, int max_change)
1346 {
1347         if (!it)
1348                 return 0;
1349         /*
1350          * make sure that at least one '-' or '+' is printed if
1351          * there is any change to this path. The easiest way is to
1352          * scale linearly as if the alloted width is one column shorter
1353          * than it is, and then add 1 to the result.
1354          */
1355         return 1 + (it * (width - 1) / max_change);
1356 }
1357
1358 static void show_name(FILE *file,
1359                       const char *prefix, const char *name, int len)
1360 {
1361         fprintf(file, " %s%-*s |", prefix, len, name);
1362 }
1363
1364 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1365 {
1366         if (cnt <= 0)
1367                 return;
1368         fprintf(file, "%s", set);
1369         while (cnt--)
1370                 putc(ch, file);
1371         fprintf(file, "%s", reset);
1372 }
1373
1374 static void fill_print_name(struct diffstat_file *file)
1375 {
1376         char *pname;
1377
1378         if (file->print_name)
1379                 return;
1380
1381         if (!file->is_renamed) {
1382                 struct strbuf buf = STRBUF_INIT;
1383                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1384                         pname = strbuf_detach(&buf, NULL);
1385                 } else {
1386                         pname = file->name;
1387                         strbuf_release(&buf);
1388                 }
1389         } else {
1390                 pname = pprint_rename(file->from_name, file->name);
1391         }
1392         file->print_name = pname;
1393 }
1394
1395 int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1396 {
1397         struct strbuf sb = STRBUF_INIT;
1398         int ret;
1399
1400         if (!files) {
1401                 assert(insertions == 0 && deletions == 0);
1402                 return fprintf(fp, "%s\n", " 0 files changed");
1403         }
1404
1405         strbuf_addf(&sb,
1406                     (files == 1) ? " %d file changed" : " %d files changed",
1407                     files);
1408
1409         /*
1410          * For binary diff, the caller may want to print "x files
1411          * changed" with insertions == 0 && deletions == 0.
1412          *
1413          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1414          * is probably less confusing (i.e skip over "2 files changed
1415          * but nothing about added/removed lines? Is this a bug in Git?").
1416          */
1417         if (insertions || deletions == 0) {
1418                 /*
1419                  * TRANSLATORS: "+" in (+) is a line addition marker;
1420                  * do not translate it.
1421                  */
1422                 strbuf_addf(&sb,
1423                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1424                             insertions);
1425         }
1426
1427         if (deletions || insertions == 0) {
1428                 /*
1429                  * TRANSLATORS: "-" in (-) is a line removal marker;
1430                  * do not translate it.
1431                  */
1432                 strbuf_addf(&sb,
1433                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1434                             deletions);
1435         }
1436         strbuf_addch(&sb, '\n');
1437         ret = fputs(sb.buf, fp);
1438         strbuf_release(&sb);
1439         return ret;
1440 }
1441
1442 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1443 {
1444         int i, len, add, del, adds = 0, dels = 0;
1445         uintmax_t max_change = 0, max_len = 0;
1446         int total_files = data->nr, count;
1447         int width, name_width, graph_width, number_width = 0, bin_width = 0;
1448         const char *reset, *add_c, *del_c;
1449         const char *line_prefix = "";
1450         int extra_shown = 0;
1451         struct strbuf *msg = NULL;
1452
1453         if (data->nr == 0)
1454                 return;
1455
1456         if (options->output_prefix) {
1457                 msg = options->output_prefix(options, options->output_prefix_data);
1458                 line_prefix = msg->buf;
1459         }
1460
1461         count = options->stat_count ? options->stat_count : data->nr;
1462
1463         reset = diff_get_color_opt(options, DIFF_RESET);
1464         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1465         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1466
1467         /*
1468          * Find the longest filename and max number of changes
1469          */
1470         for (i = 0; (i < count) && (i < data->nr); i++) {
1471                 struct diffstat_file *file = data->files[i];
1472                 uintmax_t change = file->added + file->deleted;
1473
1474                 if (!file->is_interesting && (change == 0)) {
1475                         count++; /* not shown == room for one more */
1476                         continue;
1477                 }
1478                 fill_print_name(file);
1479                 len = strlen(file->print_name);
1480                 if (max_len < len)
1481                         max_len = len;
1482
1483                 if (file->is_unmerged) {
1484                         /* "Unmerged" is 8 characters */
1485                         bin_width = bin_width < 8 ? 8 : bin_width;
1486                         continue;
1487                 }
1488                 if (file->is_binary) {
1489                         /* "Bin XXX -> YYY bytes" */
1490                         int w = 14 + decimal_width(file->added)
1491                                 + decimal_width(file->deleted);
1492                         bin_width = bin_width < w ? w : bin_width;
1493                         /* Display change counts aligned with "Bin" */
1494                         number_width = 3;
1495                         continue;
1496                 }
1497
1498                 if (max_change < change)
1499                         max_change = change;
1500         }
1501         count = i; /* where we can stop scanning in data->files[] */
1502
1503         /*
1504          * We have width = stat_width or term_columns() columns total.
1505          * We want a maximum of min(max_len, stat_name_width) for the name part.
1506          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1507          * We also need 1 for " " and 4 + decimal_width(max_change)
1508          * for " | NNNN " and one the empty column at the end, altogether
1509          * 6 + decimal_width(max_change).
1510          *
1511          * If there's not enough space, we will use the smaller of
1512          * stat_name_width (if set) and 5/8*width for the filename,
1513          * and the rest for constant elements + graph part, but no more
1514          * than stat_graph_width for the graph part.
1515          * (5/8 gives 50 for filename and 30 for the constant parts + graph
1516          * for the standard terminal size).
1517          *
1518          * In other words: stat_width limits the maximum width, and
1519          * stat_name_width fixes the maximum width of the filename,
1520          * and is also used to divide available columns if there
1521          * aren't enough.
1522          *
1523          * Binary files are displayed with "Bin XXX -> YYY bytes"
1524          * instead of the change count and graph. This part is treated
1525          * similarly to the graph part, except that it is not
1526          * "scaled". If total width is too small to accomodate the
1527          * guaranteed minimum width of the filename part and the
1528          * separators and this message, this message will "overflow"
1529          * making the line longer than the maximum width.
1530          */
1531
1532         if (options->stat_width == -1)
1533                 width = term_columns() - options->output_prefix_length;
1534         else
1535                 width = options->stat_width ? options->stat_width : 80;
1536         number_width = decimal_width(max_change) > number_width ?
1537                 decimal_width(max_change) : number_width;
1538
1539         if (options->stat_graph_width == -1)
1540                 options->stat_graph_width = diff_stat_graph_width;
1541
1542         /*
1543          * Guarantee 3/8*16==6 for the graph part
1544          * and 5/8*16==10 for the filename part
1545          */
1546         if (width < 16 + 6 + number_width)
1547                 width = 16 + 6 + number_width;
1548
1549         /*
1550          * First assign sizes that are wanted, ignoring available width.
1551          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1552          * starting from "XXX" should fit in graph_width.
1553          */
1554         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1555         if (options->stat_graph_width &&
1556             options->stat_graph_width < graph_width)
1557                 graph_width = options->stat_graph_width;
1558
1559         name_width = (options->stat_name_width > 0 &&
1560                       options->stat_name_width < max_len) ?
1561                 options->stat_name_width : max_len;
1562
1563         /*
1564          * Adjust adjustable widths not to exceed maximum width
1565          */
1566         if (name_width + number_width + 6 + graph_width > width) {
1567                 if (graph_width > width * 3/8 - number_width - 6) {
1568                         graph_width = width * 3/8 - number_width - 6;
1569                         if (graph_width < 6)
1570                                 graph_width = 6;
1571                 }
1572
1573                 if (options->stat_graph_width &&
1574                     graph_width > options->stat_graph_width)
1575                         graph_width = options->stat_graph_width;
1576                 if (name_width > width - number_width - 6 - graph_width)
1577                         name_width = width - number_width - 6 - graph_width;
1578                 else
1579                         graph_width = width - number_width - 6 - name_width;
1580         }
1581
1582         /*
1583          * From here name_width is the width of the name area,
1584          * and graph_width is the width of the graph area.
1585          * max_change is used to scale graph properly.
1586          */
1587         for (i = 0; i < count; i++) {
1588                 const char *prefix = "";
1589                 struct diffstat_file *file = data->files[i];
1590                 char *name = file->print_name;
1591                 uintmax_t added = file->added;
1592                 uintmax_t deleted = file->deleted;
1593                 int name_len;
1594
1595                 if (!file->is_interesting && (added + deleted == 0))
1596                         continue;
1597
1598                 /*
1599                  * "scale" the filename
1600                  */
1601                 len = name_width;
1602                 name_len = strlen(name);
1603                 if (name_width < name_len) {
1604                         char *slash;
1605                         prefix = "...";
1606                         len -= 3;
1607                         name += name_len - len;
1608                         slash = strchr(name, '/');
1609                         if (slash)
1610                                 name = slash;
1611                 }
1612
1613                 if (file->is_binary) {
1614                         fprintf(options->file, "%s", line_prefix);
1615                         show_name(options->file, prefix, name, len);
1616                         fprintf(options->file, " %*s", number_width, "Bin");
1617                         if (!added && !deleted) {
1618                                 putc('\n', options->file);
1619                                 continue;
1620                         }
1621                         fprintf(options->file, " %s%"PRIuMAX"%s",
1622                                 del_c, deleted, reset);
1623                         fprintf(options->file, " -> ");
1624                         fprintf(options->file, "%s%"PRIuMAX"%s",
1625                                 add_c, added, reset);
1626                         fprintf(options->file, " bytes");
1627                         fprintf(options->file, "\n");
1628                         continue;
1629                 }
1630                 else if (file->is_unmerged) {
1631                         fprintf(options->file, "%s", line_prefix);
1632                         show_name(options->file, prefix, name, len);
1633                         fprintf(options->file, " Unmerged\n");
1634                         continue;
1635                 }
1636
1637                 /*
1638                  * scale the add/delete
1639                  */
1640                 add = added;
1641                 del = deleted;
1642
1643                 if (graph_width <= max_change) {
1644                         int total = add + del;
1645
1646                         total = scale_linear(add + del, graph_width, max_change);
1647                         if (total < 2 && add && del)
1648                                 /* width >= 2 due to the sanity check */
1649                                 total = 2;
1650                         if (add < del) {
1651                                 add = scale_linear(add, graph_width, max_change);
1652                                 del = total - add;
1653                         } else {
1654                                 del = scale_linear(del, graph_width, max_change);
1655                                 add = total - del;
1656                         }
1657                 }
1658                 fprintf(options->file, "%s", line_prefix);
1659                 show_name(options->file, prefix, name, len);
1660                 fprintf(options->file, " %*"PRIuMAX"%s",
1661                         number_width, added + deleted,
1662                         added + deleted ? " " : "");
1663                 show_graph(options->file, '+', add, add_c, reset);
1664                 show_graph(options->file, '-', del, del_c, reset);
1665                 fprintf(options->file, "\n");
1666         }
1667
1668         for (i = 0; i < data->nr; i++) {
1669                 struct diffstat_file *file = data->files[i];
1670                 uintmax_t added = file->added;
1671                 uintmax_t deleted = file->deleted;
1672
1673                 if (file->is_unmerged ||
1674                     (!file->is_interesting && (added + deleted == 0))) {
1675                         total_files--;
1676                         continue;
1677                 }
1678
1679                 if (!file->is_binary) {
1680                         adds += added;
1681                         dels += deleted;
1682                 }
1683                 if (i < count)
1684                         continue;
1685                 if (!extra_shown)
1686                         fprintf(options->file, "%s ...\n", line_prefix);
1687                 extra_shown = 1;
1688         }
1689         fprintf(options->file, "%s", line_prefix);
1690         print_stat_summary(options->file, total_files, adds, dels);
1691 }
1692
1693 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1694 {
1695         int i, adds = 0, dels = 0, total_files = data->nr;
1696
1697         if (data->nr == 0)
1698                 return;
1699
1700         for (i = 0; i < data->nr; i++) {
1701                 int added = data->files[i]->added;
1702                 int deleted= data->files[i]->deleted;
1703
1704                 if (data->files[i]->is_unmerged ||
1705                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
1706                         total_files--;
1707                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
1708                         adds += added;
1709                         dels += deleted;
1710                 }
1711         }
1712         if (options->output_prefix) {
1713                 struct strbuf *msg = NULL;
1714                 msg = options->output_prefix(options,
1715                                 options->output_prefix_data);
1716                 fprintf(options->file, "%s", msg->buf);
1717         }
1718         print_stat_summary(options->file, total_files, adds, dels);
1719 }
1720
1721 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1722 {
1723         int i;
1724
1725         if (data->nr == 0)
1726                 return;
1727
1728         for (i = 0; i < data->nr; i++) {
1729                 struct diffstat_file *file = data->files[i];
1730
1731                 if (options->output_prefix) {
1732                         struct strbuf *msg = NULL;
1733                         msg = options->output_prefix(options,
1734                                         options->output_prefix_data);
1735                         fprintf(options->file, "%s", msg->buf);
1736                 }
1737
1738                 if (file->is_binary)
1739                         fprintf(options->file, "-\t-\t");
1740                 else
1741                         fprintf(options->file,
1742                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1743                                 file->added, file->deleted);
1744                 if (options->line_termination) {
1745                         fill_print_name(file);
1746                         if (!file->is_renamed)
1747                                 write_name_quoted(file->name, options->file,
1748                                                   options->line_termination);
1749                         else {
1750                                 fputs(file->print_name, options->file);
1751                                 putc(options->line_termination, options->file);
1752                         }
1753                 } else {
1754                         if (file->is_renamed) {
1755                                 putc('\0', options->file);
1756                                 write_name_quoted(file->from_name, options->file, '\0');
1757                         }
1758                         write_name_quoted(file->name, options->file, '\0');
1759                 }
1760         }
1761 }
1762
1763 struct dirstat_file {
1764         const char *name;
1765         unsigned long changed;
1766 };
1767
1768 struct dirstat_dir {
1769         struct dirstat_file *files;
1770         int alloc, nr, permille, cumulative;
1771 };
1772
1773 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1774                 unsigned long changed, const char *base, int baselen)
1775 {
1776         unsigned long this_dir = 0;
1777         unsigned int sources = 0;
1778         const char *line_prefix = "";
1779         struct strbuf *msg = NULL;
1780
1781         if (opt->output_prefix) {
1782                 msg = opt->output_prefix(opt, opt->output_prefix_data);
1783                 line_prefix = msg->buf;
1784         }
1785
1786         while (dir->nr) {
1787                 struct dirstat_file *f = dir->files;
1788                 int namelen = strlen(f->name);
1789                 unsigned long this;
1790                 char *slash;
1791
1792                 if (namelen < baselen)
1793                         break;
1794                 if (memcmp(f->name, base, baselen))
1795                         break;
1796                 slash = strchr(f->name + baselen, '/');
1797                 if (slash) {
1798                         int newbaselen = slash + 1 - f->name;
1799                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1800                         sources++;
1801                 } else {
1802                         this = f->changed;
1803                         dir->files++;
1804                         dir->nr--;
1805                         sources += 2;
1806                 }
1807                 this_dir += this;
1808         }
1809
1810         /*
1811          * We don't report dirstat's for
1812          *  - the top level
1813          *  - or cases where everything came from a single directory
1814          *    under this directory (sources == 1).
1815          */
1816         if (baselen && sources != 1) {
1817                 if (this_dir) {
1818                         int permille = this_dir * 1000 / changed;
1819                         if (permille >= dir->permille) {
1820                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1821                                         permille / 10, permille % 10, baselen, base);
1822                                 if (!dir->cumulative)
1823                                         return 0;
1824                         }
1825                 }
1826         }
1827         return this_dir;
1828 }
1829
1830 static int dirstat_compare(const void *_a, const void *_b)
1831 {
1832         const struct dirstat_file *a = _a;
1833         const struct dirstat_file *b = _b;
1834         return strcmp(a->name, b->name);
1835 }
1836
1837 static void show_dirstat(struct diff_options *options)
1838 {
1839         int i;
1840         unsigned long changed;
1841         struct dirstat_dir dir;
1842         struct diff_queue_struct *q = &diff_queued_diff;
1843
1844         dir.files = NULL;
1845         dir.alloc = 0;
1846         dir.nr = 0;
1847         dir.permille = options->dirstat_permille;
1848         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1849
1850         changed = 0;
1851         for (i = 0; i < q->nr; i++) {
1852                 struct diff_filepair *p = q->queue[i];
1853                 const char *name;
1854                 unsigned long copied, added, damage;
1855                 int content_changed;
1856
1857                 name = p->two->path ? p->two->path : p->one->path;
1858
1859                 if (p->one->sha1_valid && p->two->sha1_valid)
1860                         content_changed = hashcmp(p->one->sha1, p->two->sha1);
1861                 else
1862                         content_changed = 1;
1863
1864                 if (!content_changed) {
1865                         /*
1866                          * The SHA1 has not changed, so pre-/post-content is
1867                          * identical. We can therefore skip looking at the
1868                          * file contents altogether.
1869                          */
1870                         damage = 0;
1871                         goto found_damage;
1872                 }
1873
1874                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1875                         /*
1876                          * In --dirstat-by-file mode, we don't really need to
1877                          * look at the actual file contents at all.
1878                          * The fact that the SHA1 changed is enough for us to
1879                          * add this file to the list of results
1880                          * (with each file contributing equal damage).
1881                          */
1882                         damage = 1;
1883                         goto found_damage;
1884                 }
1885
1886                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1887                         diff_populate_filespec(p->one, 0);
1888                         diff_populate_filespec(p->two, 0);
1889                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1890                                                &copied, &added);
1891                         diff_free_filespec_data(p->one);
1892                         diff_free_filespec_data(p->two);
1893                 } else if (DIFF_FILE_VALID(p->one)) {
1894                         diff_populate_filespec(p->one, 1);
1895                         copied = added = 0;
1896                         diff_free_filespec_data(p->one);
1897                 } else if (DIFF_FILE_VALID(p->two)) {
1898                         diff_populate_filespec(p->two, 1);
1899                         copied = 0;
1900                         added = p->two->size;
1901                         diff_free_filespec_data(p->two);
1902                 } else
1903                         continue;
1904
1905                 /*
1906                  * Original minus copied is the removed material,
1907                  * added is the new material.  They are both damages
1908                  * made to the preimage.
1909                  * If the resulting damage is zero, we know that
1910                  * diffcore_count_changes() considers the two entries to
1911                  * be identical, but since content_changed is true, we
1912                  * know that there must have been _some_ kind of change,
1913                  * so we force all entries to have damage > 0.
1914                  */
1915                 damage = (p->one->size - copied) + added;
1916                 if (!damage)
1917                         damage = 1;
1918
1919 found_damage:
1920                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1921                 dir.files[dir.nr].name = name;
1922                 dir.files[dir.nr].changed = damage;
1923                 changed += damage;
1924                 dir.nr++;
1925         }
1926
1927         /* This can happen even with many files, if everything was renames */
1928         if (!changed)
1929                 return;
1930
1931         /* Show all directories with more than x% of the changes */
1932         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1933         gather_dirstat(options, &dir, changed, "", 0);
1934 }
1935
1936 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
1937 {
1938         int i;
1939         unsigned long changed;
1940         struct dirstat_dir dir;
1941
1942         if (data->nr == 0)
1943                 return;
1944
1945         dir.files = NULL;
1946         dir.alloc = 0;
1947         dir.nr = 0;
1948         dir.permille = options->dirstat_permille;
1949         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1950
1951         changed = 0;
1952         for (i = 0; i < data->nr; i++) {
1953                 struct diffstat_file *file = data->files[i];
1954                 unsigned long damage = file->added + file->deleted;
1955                 if (file->is_binary)
1956                         /*
1957                          * binary files counts bytes, not lines. Must find some
1958                          * way to normalize binary bytes vs. textual lines.
1959                          * The following heuristic assumes that there are 64
1960                          * bytes per "line".
1961                          * This is stupid and ugly, but very cheap...
1962                          */
1963                         damage = (damage + 63) / 64;
1964                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1965                 dir.files[dir.nr].name = file->name;
1966                 dir.files[dir.nr].changed = damage;
1967                 changed += damage;
1968                 dir.nr++;
1969         }
1970
1971         /* This can happen even with many files, if everything was renames */
1972         if (!changed)
1973                 return;
1974
1975         /* Show all directories with more than x% of the changes */
1976         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1977         gather_dirstat(options, &dir, changed, "", 0);
1978 }
1979
1980 static void free_diffstat_info(struct diffstat_t *diffstat)
1981 {
1982         int i;
1983         for (i = 0; i < diffstat->nr; i++) {
1984                 struct diffstat_file *f = diffstat->files[i];
1985                 if (f->name != f->print_name)
1986                         free(f->print_name);
1987                 free(f->name);
1988                 free(f->from_name);
1989                 free(f);
1990         }
1991         free(diffstat->files);
1992 }
1993
1994 struct checkdiff_t {
1995         const char *filename;
1996         int lineno;
1997         int conflict_marker_size;
1998         struct diff_options *o;
1999         unsigned ws_rule;
2000         unsigned status;
2001 };
2002
2003 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2004 {
2005         char firstchar;
2006         int cnt;
2007
2008         if (len < marker_size + 1)
2009                 return 0;
2010         firstchar = line[0];
2011         switch (firstchar) {
2012         case '=': case '>': case '<': case '|':
2013                 break;
2014         default:
2015                 return 0;
2016         }
2017         for (cnt = 1; cnt < marker_size; cnt++)
2018                 if (line[cnt] != firstchar)
2019                         return 0;
2020         /* line[1] thru line[marker_size-1] are same as firstchar */
2021         if (len < marker_size + 1 || !isspace(line[marker_size]))
2022                 return 0;
2023         return 1;
2024 }
2025
2026 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2027 {
2028         struct checkdiff_t *data = priv;
2029         int marker_size = data->conflict_marker_size;
2030         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2031         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2032         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2033         char *err;
2034         char *line_prefix = "";
2035         struct strbuf *msgbuf;
2036
2037         assert(data->o);
2038         if (data->o->output_prefix) {
2039                 msgbuf = data->o->output_prefix(data->o,
2040                         data->o->output_prefix_data);
2041                 line_prefix = msgbuf->buf;
2042         }
2043
2044         if (line[0] == '+') {
2045                 unsigned bad;
2046                 data->lineno++;
2047                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2048                         data->status |= 1;
2049                         fprintf(data->o->file,
2050                                 "%s%s:%d: leftover conflict marker\n",
2051                                 line_prefix, data->filename, data->lineno);
2052                 }
2053                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2054                 if (!bad)
2055                         return;
2056                 data->status |= bad;
2057                 err = whitespace_error_string(bad);
2058                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2059                         line_prefix, data->filename, data->lineno, err);
2060                 free(err);
2061                 emit_line(data->o, set, reset, line, 1);
2062                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2063                               data->o->file, set, reset, ws);
2064         } else if (line[0] == ' ') {
2065                 data->lineno++;
2066         } else if (line[0] == '@') {
2067                 char *plus = strchr(line, '+');
2068                 if (plus)
2069                         data->lineno = strtol(plus, NULL, 10) - 1;
2070                 else
2071                         die("invalid diff");
2072         }
2073 }
2074
2075 static unsigned char *deflate_it(char *data,
2076                                  unsigned long size,
2077                                  unsigned long *result_size)
2078 {
2079         int bound;
2080         unsigned char *deflated;
2081         git_zstream stream;
2082
2083         memset(&stream, 0, sizeof(stream));
2084         git_deflate_init(&stream, zlib_compression_level);
2085         bound = git_deflate_bound(&stream, size);
2086         deflated = xmalloc(bound);
2087         stream.next_out = deflated;
2088         stream.avail_out = bound;
2089
2090         stream.next_in = (unsigned char *)data;
2091         stream.avail_in = size;
2092         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2093                 ; /* nothing */
2094         git_deflate_end(&stream);
2095         *result_size = stream.total_out;
2096         return deflated;
2097 }
2098
2099 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
2100 {
2101         void *cp;
2102         void *delta;
2103         void *deflated;
2104         void *data;
2105         unsigned long orig_size;
2106         unsigned long delta_size;
2107         unsigned long deflate_size;
2108         unsigned long data_size;
2109
2110         /* We could do deflated delta, or we could do just deflated two,
2111          * whichever is smaller.
2112          */
2113         delta = NULL;
2114         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2115         if (one->size && two->size) {
2116                 delta = diff_delta(one->ptr, one->size,
2117                                    two->ptr, two->size,
2118                                    &delta_size, deflate_size);
2119                 if (delta) {
2120                         void *to_free = delta;
2121                         orig_size = delta_size;
2122                         delta = deflate_it(delta, delta_size, &delta_size);
2123                         free(to_free);
2124                 }
2125         }
2126
2127         if (delta && delta_size < deflate_size) {
2128                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2129                 free(deflated);
2130                 data = delta;
2131                 data_size = delta_size;
2132         }
2133         else {
2134                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
2135                 free(delta);
2136                 data = deflated;
2137                 data_size = deflate_size;
2138         }
2139
2140         /* emit data encoded in base85 */
2141         cp = data;
2142         while (data_size) {
2143                 int bytes = (52 < data_size) ? 52 : data_size;
2144                 char line[70];
2145                 data_size -= bytes;
2146                 if (bytes <= 26)
2147                         line[0] = bytes + 'A' - 1;
2148                 else
2149                         line[0] = bytes - 26 + 'a' - 1;
2150                 encode_85(line + 1, cp, bytes);
2151                 cp = (char *) cp + bytes;
2152                 fprintf(file, "%s", prefix);
2153                 fputs(line, file);
2154                 fputc('\n', file);
2155         }
2156         fprintf(file, "%s\n", prefix);
2157         free(data);
2158 }
2159
2160 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
2161 {
2162         fprintf(file, "%sGIT binary patch\n", prefix);
2163         emit_binary_diff_body(file, one, two, prefix);
2164         emit_binary_diff_body(file, two, one, prefix);
2165 }
2166
2167 int diff_filespec_is_binary(struct diff_filespec *one)
2168 {
2169         if (one->is_binary == -1) {
2170                 diff_filespec_load_driver(one);
2171                 if (one->driver->binary != -1)
2172                         one->is_binary = one->driver->binary;
2173                 else {
2174                         if (!one->data && DIFF_FILE_VALID(one))
2175                                 diff_populate_filespec(one, 0);
2176                         if (one->data)
2177                                 one->is_binary = buffer_is_binary(one->data,
2178                                                 one->size);
2179                         if (one->is_binary == -1)
2180                                 one->is_binary = 0;
2181                 }
2182         }
2183         return one->is_binary;
2184 }
2185
2186 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2187 {
2188         diff_filespec_load_driver(one);
2189         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2190 }
2191
2192 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2193 {
2194         if (!options->a_prefix)
2195                 options->a_prefix = a;
2196         if (!options->b_prefix)
2197                 options->b_prefix = b;
2198 }
2199
2200 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2201 {
2202         if (!DIFF_FILE_VALID(one))
2203                 return NULL;
2204
2205         diff_filespec_load_driver(one);
2206         return userdiff_get_textconv(one->driver);
2207 }
2208
2209 static void builtin_diff(const char *name_a,
2210                          const char *name_b,
2211                          struct diff_filespec *one,
2212                          struct diff_filespec *two,
2213                          const char *xfrm_msg,
2214                          int must_show_header,
2215                          struct diff_options *o,
2216                          int complete_rewrite)
2217 {
2218         mmfile_t mf1, mf2;
2219         const char *lbl[2];
2220         char *a_one, *b_two;
2221         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
2222         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2223         const char *a_prefix, *b_prefix;
2224         struct userdiff_driver *textconv_one = NULL;
2225         struct userdiff_driver *textconv_two = NULL;
2226         struct strbuf header = STRBUF_INIT;
2227         struct strbuf *msgbuf;
2228         char *line_prefix = "";
2229
2230         if (o->output_prefix) {
2231                 msgbuf = o->output_prefix(o, o->output_prefix_data);
2232                 line_prefix = msgbuf->buf;
2233         }
2234
2235         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2236                         (!one->mode || S_ISGITLINK(one->mode)) &&
2237                         (!two->mode || S_ISGITLINK(two->mode))) {
2238                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2239                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2240                 show_submodule_summary(o->file, one ? one->path : two->path,
2241                                 one->sha1, two->sha1, two->dirty_submodule,
2242                                 del, add, reset);
2243                 return;
2244         }
2245
2246         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2247                 textconv_one = get_textconv(one);
2248                 textconv_two = get_textconv(two);
2249         }
2250
2251         diff_set_mnemonic_prefix(o, "a/", "b/");
2252         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2253                 a_prefix = o->b_prefix;
2254                 b_prefix = o->a_prefix;
2255         } else {
2256                 a_prefix = o->a_prefix;
2257                 b_prefix = o->b_prefix;
2258         }
2259
2260         /* Never use a non-valid filename anywhere if at all possible */
2261         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2262         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2263
2264         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2265         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2266         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2267         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2268         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, set, a_one, b_two, reset);
2269         if (lbl[0][0] == '/') {
2270                 /* /dev/null */
2271                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, set, two->mode, reset);
2272                 if (xfrm_msg)
2273                         strbuf_addstr(&header, xfrm_msg);
2274                 must_show_header = 1;
2275         }
2276         else if (lbl[1][0] == '/') {
2277                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, set, one->mode, reset);
2278                 if (xfrm_msg)
2279                         strbuf_addstr(&header, xfrm_msg);
2280                 must_show_header = 1;
2281         }
2282         else {
2283                 if (one->mode != two->mode) {
2284                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, set, one->mode, reset);
2285                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, set, two->mode, reset);
2286                         must_show_header = 1;
2287                 }
2288                 if (xfrm_msg)
2289                         strbuf_addstr(&header, xfrm_msg);
2290
2291                 /*
2292                  * we do not run diff between different kind
2293                  * of objects.
2294                  */
2295                 if ((one->mode ^ two->mode) & S_IFMT)
2296                         goto free_ab_and_return;
2297                 if (complete_rewrite &&
2298                     (textconv_one || !diff_filespec_is_binary(one)) &&
2299                     (textconv_two || !diff_filespec_is_binary(two))) {
2300                         fprintf(o->file, "%s", header.buf);
2301                         strbuf_reset(&header);
2302                         emit_rewrite_diff(name_a, name_b, one, two,
2303                                                 textconv_one, textconv_two, o);
2304                         o->found_changes = 1;
2305                         goto free_ab_and_return;
2306                 }
2307         }
2308
2309         if (o->irreversible_delete && lbl[1][0] == '/') {
2310                 fprintf(o->file, "%s", header.buf);
2311                 strbuf_reset(&header);
2312                 goto free_ab_and_return;
2313         } else if (!DIFF_OPT_TST(o, TEXT) &&
2314             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2315               (!textconv_two && diff_filespec_is_binary(two)) )) {
2316                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2317                         die("unable to read files to diff");
2318                 /* Quite common confusing case */
2319                 if (mf1.size == mf2.size &&
2320                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2321                         if (must_show_header)
2322                                 fprintf(o->file, "%s", header.buf);
2323                         goto free_ab_and_return;
2324                 }
2325                 fprintf(o->file, "%s", header.buf);
2326                 strbuf_reset(&header);
2327                 if (DIFF_OPT_TST(o, BINARY))
2328                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2329                 else
2330                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2331                                 line_prefix, lbl[0], lbl[1]);
2332                 o->found_changes = 1;
2333         } else {
2334                 /* Crazy xdl interfaces.. */
2335                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2336                 xpparam_t xpp;
2337                 xdemitconf_t xecfg;
2338                 struct emit_callback ecbdata;
2339                 const struct userdiff_funcname *pe;
2340
2341                 if (must_show_header) {
2342                         fprintf(o->file, "%s", header.buf);
2343                         strbuf_reset(&header);
2344                 }
2345
2346                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2347                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2348
2349                 pe = diff_funcname_pattern(one);
2350                 if (!pe)
2351                         pe = diff_funcname_pattern(two);
2352
2353                 memset(&xpp, 0, sizeof(xpp));
2354                 memset(&xecfg, 0, sizeof(xecfg));
2355                 memset(&ecbdata, 0, sizeof(ecbdata));
2356                 ecbdata.label_path = lbl;
2357                 ecbdata.color_diff = want_color(o->use_color);
2358                 ecbdata.found_changesp = &o->found_changes;
2359                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
2360                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2361                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2362                 ecbdata.opt = o;
2363                 ecbdata.header = header.len ? &header : NULL;
2364                 xpp.flags = o->xdl_opts;
2365                 xecfg.ctxlen = o->context;
2366                 xecfg.interhunkctxlen = o->interhunkcontext;
2367                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2368                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2369                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2370                 if (pe)
2371                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2372                 if (!diffopts)
2373                         ;
2374                 else if (!prefixcmp(diffopts, "--unified="))
2375                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
2376                 else if (!prefixcmp(diffopts, "-u"))
2377                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
2378                 if (o->word_diff)
2379                         init_diff_words_data(&ecbdata, o, one, two);
2380                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2381                               &xpp, &xecfg);
2382                 if (o->word_diff)
2383                         free_diff_words_data(&ecbdata);
2384                 if (textconv_one)
2385                         free(mf1.ptr);
2386                 if (textconv_two)
2387                         free(mf2.ptr);
2388                 xdiff_clear_find_func(&xecfg);
2389         }
2390
2391  free_ab_and_return:
2392         strbuf_release(&header);
2393         diff_free_filespec_data(one);
2394         diff_free_filespec_data(two);
2395         free(a_one);
2396         free(b_two);
2397         return;
2398 }
2399
2400 static void builtin_diffstat(const char *name_a, const char *name_b,
2401                              struct diff_filespec *one,
2402                              struct diff_filespec *two,
2403                              struct diffstat_t *diffstat,
2404                              struct diff_options *o,
2405                              struct diff_filepair *p)
2406 {
2407         mmfile_t mf1, mf2;
2408         struct diffstat_file *data;
2409         int same_contents;
2410         int complete_rewrite = 0;
2411
2412         if (!DIFF_PAIR_UNMERGED(p)) {
2413                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2414                         complete_rewrite = 1;
2415         }
2416
2417         data = diffstat_add(diffstat, name_a, name_b);
2418         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2419
2420         if (!one || !two) {
2421                 data->is_unmerged = 1;
2422                 return;
2423         }
2424
2425         same_contents = !hashcmp(one->sha1, two->sha1);
2426
2427         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2428                 data->is_binary = 1;
2429                 if (same_contents) {
2430                         data->added = 0;
2431                         data->deleted = 0;
2432                 } else {
2433                         data->added = diff_filespec_size(two);
2434                         data->deleted = diff_filespec_size(one);
2435                 }
2436         }
2437
2438         else if (complete_rewrite) {
2439                 diff_populate_filespec(one, 0);
2440                 diff_populate_filespec(two, 0);
2441                 data->deleted = count_lines(one->data, one->size);
2442                 data->added = count_lines(two->data, two->size);
2443         }
2444
2445         else if (!same_contents) {
2446                 /* Crazy xdl interfaces.. */
2447                 xpparam_t xpp;
2448                 xdemitconf_t xecfg;
2449
2450                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2451                         die("unable to read files to diff");
2452
2453                 memset(&xpp, 0, sizeof(xpp));
2454                 memset(&xecfg, 0, sizeof(xecfg));
2455                 xpp.flags = o->xdl_opts;
2456                 xecfg.ctxlen = o->context;
2457                 xecfg.interhunkctxlen = o->interhunkcontext;
2458                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2459                               &xpp, &xecfg);
2460         }
2461
2462         diff_free_filespec_data(one);
2463         diff_free_filespec_data(two);
2464 }
2465
2466 static void builtin_checkdiff(const char *name_a, const char *name_b,
2467                               const char *attr_path,
2468                               struct diff_filespec *one,
2469                               struct diff_filespec *two,
2470                               struct diff_options *o)
2471 {
2472         mmfile_t mf1, mf2;
2473         struct checkdiff_t data;
2474
2475         if (!two)
2476                 return;
2477
2478         memset(&data, 0, sizeof(data));
2479         data.filename = name_b ? name_b : name_a;
2480         data.lineno = 0;
2481         data.o = o;
2482         data.ws_rule = whitespace_rule(attr_path);
2483         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2484
2485         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2486                 die("unable to read files to diff");
2487
2488         /*
2489          * All the other codepaths check both sides, but not checking
2490          * the "old" side here is deliberate.  We are checking the newly
2491          * introduced changes, and as long as the "new" side is text, we
2492          * can and should check what it introduces.
2493          */
2494         if (diff_filespec_is_binary(two))
2495                 goto free_and_return;
2496         else {
2497                 /* Crazy xdl interfaces.. */
2498                 xpparam_t xpp;
2499                 xdemitconf_t xecfg;
2500
2501                 memset(&xpp, 0, sizeof(xpp));
2502                 memset(&xecfg, 0, sizeof(xecfg));
2503                 xecfg.ctxlen = 1; /* at least one context line */
2504                 xpp.flags = 0;
2505                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2506                               &xpp, &xecfg);
2507
2508                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2509                         struct emit_callback ecbdata;
2510                         int blank_at_eof;
2511
2512                         ecbdata.ws_rule = data.ws_rule;
2513                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2514                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2515
2516                         if (blank_at_eof) {
2517                                 static char *err;
2518                                 if (!err)
2519                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2520                                 fprintf(o->file, "%s:%d: %s.\n",
2521                                         data.filename, blank_at_eof, err);
2522                                 data.status = 1; /* report errors */
2523                         }
2524                 }
2525         }
2526  free_and_return:
2527         diff_free_filespec_data(one);
2528         diff_free_filespec_data(two);
2529         if (data.status)
2530                 DIFF_OPT_SET(o, CHECK_FAILED);
2531 }
2532
2533 struct diff_filespec *alloc_filespec(const char *path)
2534 {
2535         int namelen = strlen(path);
2536         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2537
2538         memset(spec, 0, sizeof(*spec));
2539         spec->path = (char *)(spec + 1);
2540         memcpy(spec->path, path, namelen+1);
2541         spec->count = 1;
2542         spec->is_binary = -1;
2543         return spec;
2544 }
2545
2546 void free_filespec(struct diff_filespec *spec)
2547 {
2548         if (!--spec->count) {
2549                 diff_free_filespec_data(spec);
2550                 free(spec);
2551         }
2552 }
2553
2554 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2555                    int sha1_valid, unsigned short mode)
2556 {
2557         if (mode) {
2558                 spec->mode = canon_mode(mode);
2559                 hashcpy(spec->sha1, sha1);
2560                 spec->sha1_valid = sha1_valid;
2561         }
2562 }
2563
2564 /*
2565  * Given a name and sha1 pair, if the index tells us the file in
2566  * the work tree has that object contents, return true, so that
2567  * prepare_temp_file() does not have to inflate and extract.
2568  */
2569 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2570 {
2571         struct cache_entry *ce;
2572         struct stat st;
2573         int pos, len;
2574
2575         /*
2576          * We do not read the cache ourselves here, because the
2577          * benchmark with my previous version that always reads cache
2578          * shows that it makes things worse for diff-tree comparing
2579          * two linux-2.6 kernel trees in an already checked out work
2580          * tree.  This is because most diff-tree comparisons deal with
2581          * only a small number of files, while reading the cache is
2582          * expensive for a large project, and its cost outweighs the
2583          * savings we get by not inflating the object to a temporary
2584          * file.  Practically, this code only helps when we are used
2585          * by diff-cache --cached, which does read the cache before
2586          * calling us.
2587          */
2588         if (!active_cache)
2589                 return 0;
2590
2591         /* We want to avoid the working directory if our caller
2592          * doesn't need the data in a normal file, this system
2593          * is rather slow with its stat/open/mmap/close syscalls,
2594          * and the object is contained in a pack file.  The pack
2595          * is probably already open and will be faster to obtain
2596          * the data through than the working directory.  Loose
2597          * objects however would tend to be slower as they need
2598          * to be individually opened and inflated.
2599          */
2600         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2601                 return 0;
2602
2603         len = strlen(name);
2604         pos = cache_name_pos(name, len);
2605         if (pos < 0)
2606                 return 0;
2607         ce = active_cache[pos];
2608
2609         /*
2610          * This is not the sha1 we are looking for, or
2611          * unreusable because it is not a regular file.
2612          */
2613         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2614                 return 0;
2615
2616         /*
2617          * If ce is marked as "assume unchanged", there is no
2618          * guarantee that work tree matches what we are looking for.
2619          */
2620         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2621                 return 0;
2622
2623         /*
2624          * If ce matches the file in the work tree, we can reuse it.
2625          */
2626         if (ce_uptodate(ce) ||
2627             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2628                 return 1;
2629
2630         return 0;
2631 }
2632
2633 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2634 {
2635         int len;
2636         char *data = xmalloc(100), *dirty = "";
2637
2638         /* Are we looking at the work tree? */
2639         if (s->dirty_submodule)
2640                 dirty = "-dirty";
2641
2642         len = snprintf(data, 100,
2643                        "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2644         s->data = data;
2645         s->size = len;
2646         s->should_free = 1;
2647         if (size_only) {
2648                 s->data = NULL;
2649                 free(data);
2650         }
2651         return 0;
2652 }
2653
2654 /*
2655  * While doing rename detection and pickaxe operation, we may need to
2656  * grab the data for the blob (or file) for our own in-core comparison.
2657  * diff_filespec has data and size fields for this purpose.
2658  */
2659 int diff_populate_filespec(struct diff_filespec *s, int size_only)
2660 {
2661         int err = 0;
2662         if (!DIFF_FILE_VALID(s))
2663                 die("internal error: asking to populate invalid file.");
2664         if (S_ISDIR(s->mode))
2665                 return -1;
2666
2667         if (s->data)
2668                 return 0;
2669
2670         if (size_only && 0 < s->size)
2671                 return 0;
2672
2673         if (S_ISGITLINK(s->mode))
2674                 return diff_populate_gitlink(s, size_only);
2675
2676         if (!s->sha1_valid ||
2677             reuse_worktree_file(s->path, s->sha1, 0)) {
2678                 struct strbuf buf = STRBUF_INIT;
2679                 struct stat st;
2680                 int fd;
2681
2682                 if (lstat(s->path, &st) < 0) {
2683                         if (errno == ENOENT) {
2684                         err_empty:
2685                                 err = -1;
2686                         empty:
2687                                 s->data = (char *)"";
2688                                 s->size = 0;
2689                                 return err;
2690                         }
2691                 }
2692                 s->size = xsize_t(st.st_size);
2693                 if (!s->size)
2694                         goto empty;
2695                 if (S_ISLNK(st.st_mode)) {
2696                         struct strbuf sb = STRBUF_INIT;
2697
2698                         if (strbuf_readlink(&sb, s->path, s->size))
2699                                 goto err_empty;
2700                         s->size = sb.len;
2701                         s->data = strbuf_detach(&sb, NULL);
2702                         s->should_free = 1;
2703                         return 0;
2704                 }
2705                 if (size_only)
2706                         return 0;
2707                 fd = open(s->path, O_RDONLY);
2708                 if (fd < 0)
2709                         goto err_empty;
2710                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2711                 close(fd);
2712                 s->should_munmap = 1;
2713
2714                 /*
2715                  * Convert from working tree format to canonical git format
2716                  */
2717                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2718                         size_t size = 0;
2719                         munmap(s->data, s->size);
2720                         s->should_munmap = 0;
2721                         s->data = strbuf_detach(&buf, &size);
2722                         s->size = size;
2723                         s->should_free = 1;
2724                 }
2725         }
2726         else {
2727                 enum object_type type;
2728                 if (size_only) {
2729                         type = sha1_object_info(s->sha1, &s->size);
2730                         if (type < 0)
2731                                 die("unable to read %s", sha1_to_hex(s->sha1));
2732                 } else {
2733                         s->data = read_sha1_file(s->sha1, &type, &s->size);
2734                         if (!s->data)
2735                                 die("unable to read %s", sha1_to_hex(s->sha1));
2736                         s->should_free = 1;
2737                 }
2738         }
2739         return 0;
2740 }
2741
2742 void diff_free_filespec_blob(struct diff_filespec *s)
2743 {
2744         if (s->should_free)
2745                 free(s->data);
2746         else if (s->should_munmap)
2747                 munmap(s->data, s->size);
2748
2749         if (s->should_free || s->should_munmap) {
2750                 s->should_free = s->should_munmap = 0;
2751                 s->data = NULL;
2752         }
2753 }
2754
2755 void diff_free_filespec_data(struct diff_filespec *s)
2756 {
2757         diff_free_filespec_blob(s);
2758         free(s->cnt_data);
2759         s->cnt_data = NULL;
2760 }
2761
2762 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2763                            void *blob,
2764                            unsigned long size,
2765                            const unsigned char *sha1,
2766                            int mode)
2767 {
2768         int fd;
2769         struct strbuf buf = STRBUF_INIT;
2770         struct strbuf template = STRBUF_INIT;
2771         char *path_dup = xstrdup(path);
2772         const char *base = basename(path_dup);
2773
2774         /* Generate "XXXXXX_basename.ext" */
2775         strbuf_addstr(&template, "XXXXXX_");
2776         strbuf_addstr(&template, base);
2777
2778         fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2779                         strlen(base) + 1);
2780         if (fd < 0)
2781                 die_errno("unable to create temp-file");
2782         if (convert_to_working_tree(path,
2783                         (const char *)blob, (size_t)size, &buf)) {
2784                 blob = buf.buf;
2785                 size = buf.len;
2786         }
2787         if (write_in_full(fd, blob, size) != size)
2788                 die_errno("unable to write temp-file");
2789         close(fd);
2790         temp->name = temp->tmp_path;
2791         strcpy(temp->hex, sha1_to_hex(sha1));
2792         temp->hex[40] = 0;
2793         sprintf(temp->mode, "%06o", mode);
2794         strbuf_release(&buf);
2795         strbuf_release(&template);
2796         free(path_dup);
2797 }
2798
2799 static struct diff_tempfile *prepare_temp_file(const char *name,
2800                 struct diff_filespec *one)
2801 {
2802         struct diff_tempfile *temp = claim_diff_tempfile();
2803
2804         if (!DIFF_FILE_VALID(one)) {
2805         not_a_valid_file:
2806                 /* A '-' entry produces this for file-2, and
2807                  * a '+' entry produces this for file-1.
2808                  */
2809                 temp->name = "/dev/null";
2810                 strcpy(temp->hex, ".");
2811                 strcpy(temp->mode, ".");
2812                 return temp;
2813         }
2814
2815         if (!remove_tempfile_installed) {
2816                 atexit(remove_tempfile);
2817                 sigchain_push_common(remove_tempfile_on_signal);
2818                 remove_tempfile_installed = 1;
2819         }
2820
2821         if (!one->sha1_valid ||
2822             reuse_worktree_file(name, one->sha1, 1)) {
2823                 struct stat st;
2824                 if (lstat(name, &st) < 0) {
2825                         if (errno == ENOENT)
2826                                 goto not_a_valid_file;
2827                         die_errno("stat(%s)", name);
2828                 }
2829                 if (S_ISLNK(st.st_mode)) {
2830                         struct strbuf sb = STRBUF_INIT;
2831                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2832                                 die_errno("readlink(%s)", name);
2833                         prep_temp_blob(name, temp, sb.buf, sb.len,
2834                                        (one->sha1_valid ?
2835                                         one->sha1 : null_sha1),
2836                                        (one->sha1_valid ?
2837                                         one->mode : S_IFLNK));
2838                         strbuf_release(&sb);
2839                 }
2840                 else {
2841                         /* we can borrow from the file in the work tree */
2842                         temp->name = name;
2843                         if (!one->sha1_valid)
2844                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2845                         else
2846                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2847                         /* Even though we may sometimes borrow the
2848                          * contents from the work tree, we always want
2849                          * one->mode.  mode is trustworthy even when
2850                          * !(one->sha1_valid), as long as
2851                          * DIFF_FILE_VALID(one).
2852                          */
2853                         sprintf(temp->mode, "%06o", one->mode);
2854                 }
2855                 return temp;
2856         }
2857         else {
2858                 if (diff_populate_filespec(one, 0))
2859                         die("cannot read data blob for %s", one->path);
2860                 prep_temp_blob(name, temp, one->data, one->size,
2861                                one->sha1, one->mode);
2862         }
2863         return temp;
2864 }
2865
2866 /* An external diff command takes:
2867  *
2868  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2869  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2870  *
2871  */
2872 static void run_external_diff(const char *pgm,
2873                               const char *name,
2874                               const char *other,
2875                               struct diff_filespec *one,
2876                               struct diff_filespec *two,
2877                               const char *xfrm_msg,
2878                               int complete_rewrite)
2879 {
2880         const char *spawn_arg[10];
2881         int retval;
2882         const char **arg = &spawn_arg[0];
2883
2884         if (one && two) {
2885                 struct diff_tempfile *temp_one, *temp_two;
2886                 const char *othername = (other ? other : name);
2887                 temp_one = prepare_temp_file(name, one);
2888                 temp_two = prepare_temp_file(othername, two);
2889                 *arg++ = pgm;
2890                 *arg++ = name;
2891                 *arg++ = temp_one->name;
2892                 *arg++ = temp_one->hex;
2893                 *arg++ = temp_one->mode;
2894                 *arg++ = temp_two->name;
2895                 *arg++ = temp_two->hex;
2896                 *arg++ = temp_two->mode;
2897                 if (other) {
2898                         *arg++ = other;
2899                         *arg++ = xfrm_msg;
2900                 }
2901         } else {
2902                 *arg++ = pgm;
2903                 *arg++ = name;
2904         }
2905         *arg = NULL;
2906         fflush(NULL);
2907         retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2908         remove_tempfile();
2909         if (retval) {
2910                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2911                 exit(1);
2912         }
2913 }
2914
2915 static int similarity_index(struct diff_filepair *p)
2916 {
2917         return p->score * 100 / MAX_SCORE;
2918 }
2919
2920 static void fill_metainfo(struct strbuf *msg,
2921                           const char *name,
2922                           const char *other,
2923                           struct diff_filespec *one,
2924                           struct diff_filespec *two,
2925                           struct diff_options *o,
2926                           struct diff_filepair *p,
2927                           int *must_show_header,
2928                           int use_color)
2929 {
2930         const char *set = diff_get_color(use_color, DIFF_METAINFO);
2931         const char *reset = diff_get_color(use_color, DIFF_RESET);
2932         struct strbuf *msgbuf;
2933         char *line_prefix = "";
2934
2935         *must_show_header = 1;
2936         if (o->output_prefix) {
2937                 msgbuf = o->output_prefix(o, o->output_prefix_data);
2938                 line_prefix = msgbuf->buf;
2939         }
2940         strbuf_init(msg, PATH_MAX * 2 + 300);
2941         switch (p->status) {
2942         case DIFF_STATUS_COPIED:
2943                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2944                             line_prefix, set, similarity_index(p));
2945                 strbuf_addf(msg, "%s\n%s%scopy from ",
2946                             reset,  line_prefix, set);
2947                 quote_c_style(name, msg, NULL, 0);
2948                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
2949                 quote_c_style(other, msg, NULL, 0);
2950                 strbuf_addf(msg, "%s\n", reset);
2951                 break;
2952         case DIFF_STATUS_RENAMED:
2953                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2954                             line_prefix, set, similarity_index(p));
2955                 strbuf_addf(msg, "%s\n%s%srename from ",
2956                             reset, line_prefix, set);
2957                 quote_c_style(name, msg, NULL, 0);
2958                 strbuf_addf(msg, "%s\n%s%srename to ",
2959                             reset, line_prefix, set);
2960                 quote_c_style(other, msg, NULL, 0);
2961                 strbuf_addf(msg, "%s\n", reset);
2962                 break;
2963         case DIFF_STATUS_MODIFIED:
2964                 if (p->score) {
2965                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
2966                                     line_prefix,
2967                                     set, similarity_index(p), reset);
2968                         break;
2969                 }
2970                 /* fallthru */
2971         default:
2972                 *must_show_header = 0;
2973         }
2974         if (one && two && hashcmp(one->sha1, two->sha1)) {
2975                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2976
2977                 if (DIFF_OPT_TST(o, BINARY)) {
2978                         mmfile_t mf;
2979                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2980                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2981                                 abbrev = 40;
2982                 }
2983                 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
2984                             find_unique_abbrev(one->sha1, abbrev));
2985                 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
2986                 if (one->mode == two->mode)
2987                         strbuf_addf(msg, " %06o", one->mode);
2988                 strbuf_addf(msg, "%s\n", reset);
2989         }
2990 }
2991
2992 static void run_diff_cmd(const char *pgm,
2993                          const char *name,
2994                          const char *other,
2995                          const char *attr_path,
2996                          struct diff_filespec *one,
2997                          struct diff_filespec *two,
2998                          struct strbuf *msg,
2999                          struct diff_options *o,
3000                          struct diff_filepair *p)
3001 {
3002         const char *xfrm_msg = NULL;
3003         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3004         int must_show_header = 0;
3005
3006
3007         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3008                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3009                 if (drv && drv->external)
3010                         pgm = drv->external;
3011         }
3012
3013         if (msg) {
3014                 /*
3015                  * don't use colors when the header is intended for an
3016                  * external diff driver
3017                  */
3018                 fill_metainfo(msg, name, other, one, two, o, p,
3019                               &must_show_header,
3020                               want_color(o->use_color) && !pgm);
3021                 xfrm_msg = msg->len ? msg->buf : NULL;
3022         }
3023
3024         if (pgm) {
3025                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3026                                   complete_rewrite);
3027                 return;
3028         }
3029         if (one && two)
3030                 builtin_diff(name, other ? other : name,
3031                              one, two, xfrm_msg, must_show_header,
3032                              o, complete_rewrite);
3033         else
3034                 fprintf(o->file, "* Unmerged path %s\n", name);
3035 }
3036
3037 static void diff_fill_sha1_info(struct diff_filespec *one)
3038 {
3039         if (DIFF_FILE_VALID(one)) {
3040                 if (!one->sha1_valid) {
3041                         struct stat st;
3042                         if (one->is_stdin) {
3043                                 hashcpy(one->sha1, null_sha1);
3044                                 return;
3045                         }
3046                         if (lstat(one->path, &st) < 0)
3047                                 die_errno("stat '%s'", one->path);
3048                         if (index_path(one->sha1, one->path, &st, 0))
3049                                 die("cannot hash %s", one->path);
3050                 }
3051         }
3052         else
3053                 hashclr(one->sha1);
3054 }
3055
3056 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3057 {
3058         /* Strip the prefix but do not molest /dev/null and absolute paths */
3059         if (*namep && **namep != '/') {
3060                 *namep += prefix_length;
3061                 if (**namep == '/')
3062                         ++*namep;
3063         }
3064         if (*otherp && **otherp != '/') {
3065                 *otherp += prefix_length;
3066                 if (**otherp == '/')
3067                         ++*otherp;
3068         }
3069 }
3070
3071 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3072 {
3073         const char *pgm = external_diff();
3074         struct strbuf msg;
3075         struct diff_filespec *one = p->one;
3076         struct diff_filespec *two = p->two;
3077         const char *name;
3078         const char *other;
3079         const char *attr_path;
3080
3081         name  = p->one->path;
3082         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3083         attr_path = name;
3084         if (o->prefix_length)
3085                 strip_prefix(o->prefix_length, &name, &other);
3086
3087         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3088                 pgm = NULL;
3089
3090         if (DIFF_PAIR_UNMERGED(p)) {
3091                 run_diff_cmd(pgm, name, NULL, attr_path,
3092                              NULL, NULL, NULL, o, p);
3093                 return;
3094         }
3095
3096         diff_fill_sha1_info(one);
3097         diff_fill_sha1_info(two);
3098
3099         if (!pgm &&
3100             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3101             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3102                 /*
3103                  * a filepair that changes between file and symlink
3104                  * needs to be split into deletion and creation.
3105                  */
3106                 struct diff_filespec *null = alloc_filespec(two->path);
3107                 run_diff_cmd(NULL, name, other, attr_path,
3108                              one, null, &msg, o, p);
3109                 free(null);
3110                 strbuf_release(&msg);
3111
3112                 null = alloc_filespec(one->path);
3113                 run_diff_cmd(NULL, name, other, attr_path,
3114                              null, two, &msg, o, p);
3115                 free(null);
3116         }
3117         else
3118                 run_diff_cmd(pgm, name, other, attr_path,
3119                              one, two, &msg, o, p);
3120
3121         strbuf_release(&msg);
3122 }
3123
3124 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3125                          struct diffstat_t *diffstat)
3126 {
3127         const char *name;
3128         const char *other;
3129
3130         if (DIFF_PAIR_UNMERGED(p)) {
3131                 /* unmerged */
3132                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3133                 return;
3134         }
3135
3136         name = p->one->path;
3137         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3138
3139         if (o->prefix_length)
3140                 strip_prefix(o->prefix_length, &name, &other);
3141
3142         diff_fill_sha1_info(p->one);
3143         diff_fill_sha1_info(p->two);
3144
3145         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3146 }
3147
3148 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3149 {
3150         const char *name;
3151         const char *other;
3152         const char *attr_path;
3153
3154         if (DIFF_PAIR_UNMERGED(p)) {
3155                 /* unmerged */
3156                 return;
3157         }
3158
3159         name = p->one->path;
3160         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3161         attr_path = other ? other : name;
3162
3163         if (o->prefix_length)
3164                 strip_prefix(o->prefix_length, &name, &other);
3165
3166         diff_fill_sha1_info(p->one);
3167         diff_fill_sha1_info(p->two);
3168
3169         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3170 }
3171
3172 void diff_setup(struct diff_options *options)
3173 {
3174         memcpy(options, &default_diff_options, sizeof(*options));
3175
3176         options->file = stdout;
3177
3178         options->line_termination = '\n';
3179         options->break_opt = -1;
3180         options->rename_limit = -1;
3181         options->dirstat_permille = diff_dirstat_permille_default;
3182         options->context = 3;
3183         DIFF_OPT_SET(options, RENAME_EMPTY);
3184
3185         options->change = diff_change;
3186         options->add_remove = diff_addremove;
3187         options->use_color = diff_use_color_default;
3188         options->detect_rename = diff_detect_rename_default;
3189
3190         if (diff_no_prefix) {
3191                 options->a_prefix = options->b_prefix = "";
3192         } else if (!diff_mnemonic_prefix) {
3193                 options->a_prefix = "a/";
3194                 options->b_prefix = "b/";
3195         }
3196 }
3197
3198 void diff_setup_done(struct diff_options *options)
3199 {
3200         int count = 0;
3201
3202         if (options->output_format & DIFF_FORMAT_NAME)
3203                 count++;
3204         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3205                 count++;
3206         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3207                 count++;
3208         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3209                 count++;
3210         if (count > 1)
3211                 die("--name-only, --name-status, --check and -s are mutually exclusive");
3212
3213         /*
3214          * Most of the time we can say "there are changes"
3215          * only by checking if there are changed paths, but
3216          * --ignore-whitespace* options force us to look
3217          * inside contents.
3218          */
3219
3220         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3221             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3222             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3223                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3224         else
3225                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3226
3227         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3228                 options->detect_rename = DIFF_DETECT_COPY;
3229
3230         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3231                 options->prefix = NULL;
3232         if (options->prefix)
3233                 options->prefix_length = strlen(options->prefix);
3234         else
3235                 options->prefix_length = 0;
3236
3237         if (options->output_format & (DIFF_FORMAT_NAME |
3238                                       DIFF_FORMAT_NAME_STATUS |
3239                                       DIFF_FORMAT_CHECKDIFF |
3240                                       DIFF_FORMAT_NO_OUTPUT))
3241                 options->output_format &= ~(DIFF_FORMAT_RAW |
3242                                             DIFF_FORMAT_NUMSTAT |
3243                                             DIFF_FORMAT_DIFFSTAT |
3244                                             DIFF_FORMAT_SHORTSTAT |
3245                                             DIFF_FORMAT_DIRSTAT |
3246                                             DIFF_FORMAT_SUMMARY |
3247                                             DIFF_FORMAT_PATCH);
3248
3249         /*
3250          * These cases always need recursive; we do not drop caller-supplied
3251          * recursive bits for other formats here.
3252          */
3253         if (options->output_format & (DIFF_FORMAT_PATCH |
3254                                       DIFF_FORMAT_NUMSTAT |
3255                                       DIFF_FORMAT_DIFFSTAT |
3256                                       DIFF_FORMAT_SHORTSTAT |
3257                                       DIFF_FORMAT_DIRSTAT |
3258                                       DIFF_FORMAT_SUMMARY |
3259                                       DIFF_FORMAT_CHECKDIFF))
3260                 DIFF_OPT_SET(options, RECURSIVE);
3261         /*
3262          * Also pickaxe would not work very well if you do not say recursive
3263          */
3264         if (options->pickaxe)
3265                 DIFF_OPT_SET(options, RECURSIVE);
3266         /*
3267          * When patches are generated, submodules diffed against the work tree
3268          * must be checked for dirtiness too so it can be shown in the output
3269          */
3270         if (options->output_format & DIFF_FORMAT_PATCH)
3271                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3272
3273         if (options->detect_rename && options->rename_limit < 0)
3274                 options->rename_limit = diff_rename_limit_default;
3275         if (options->setup & DIFF_SETUP_USE_CACHE) {
3276                 if (!active_cache)
3277                         /* read-cache does not die even when it fails
3278                          * so it is safe for us to do this here.  Also
3279                          * it does not smudge active_cache or active_nr
3280                          * when it fails, so we do not have to worry about
3281                          * cleaning it up ourselves either.
3282                          */
3283                         read_cache();
3284         }
3285         if (options->abbrev <= 0 || 40 < options->abbrev)
3286                 options->abbrev = 40; /* full */
3287
3288         /*
3289          * It does not make sense to show the first hit we happened
3290          * to have found.  It does not make sense not to return with
3291          * exit code in such a case either.
3292          */
3293         if (DIFF_OPT_TST(options, QUICK)) {
3294                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3295                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3296         }
3297 }
3298
3299 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3300 {
3301         char c, *eq;
3302         int len;
3303
3304         if (*arg != '-')
3305                 return 0;
3306         c = *++arg;
3307         if (!c)
3308                 return 0;
3309         if (c == arg_short) {
3310                 c = *++arg;
3311                 if (!c)
3312                         return 1;
3313                 if (val && isdigit(c)) {
3314                         char *end;
3315                         int n = strtoul(arg, &end, 10);
3316                         if (*end)
3317                                 return 0;
3318                         *val = n;
3319                         return 1;
3320                 }
3321                 return 0;
3322         }
3323         if (c != '-')
3324                 return 0;
3325         arg++;
3326         eq = strchr(arg, '=');
3327         if (eq)
3328                 len = eq - arg;
3329         else
3330                 len = strlen(arg);
3331         if (!len || strncmp(arg, arg_long, len))
3332                 return 0;
3333         if (eq) {
3334                 int n;
3335                 char *end;
3336                 if (!isdigit(*++eq))
3337                         return 0;
3338                 n = strtoul(eq, &end, 10);
3339                 if (*end)
3340                         return 0;
3341                 *val = n;
3342         }
3343         return 1;
3344 }
3345
3346 static int diff_scoreopt_parse(const char *opt);
3347
3348 static inline int short_opt(char opt, const char **argv,
3349                             const char **optarg)
3350 {
3351         const char *arg = argv[0];
3352         if (arg[0] != '-' || arg[1] != opt)
3353                 return 0;
3354         if (arg[2] != '\0') {
3355                 *optarg = arg + 2;
3356                 return 1;
3357         }
3358         if (!argv[1])
3359                 die("Option '%c' requires a value", opt);
3360         *optarg = argv[1];
3361         return 2;
3362 }
3363
3364 int parse_long_opt(const char *opt, const char **argv,
3365                    const char **optarg)
3366 {
3367         const char *arg = argv[0];
3368         if (arg[0] != '-' || arg[1] != '-')
3369                 return 0;
3370         arg += strlen("--");
3371         if (prefixcmp(arg, opt))
3372                 return 0;
3373         arg += strlen(opt);
3374         if (*arg == '=') { /* sticked form: --option=value */
3375                 *optarg = arg + 1;
3376                 return 1;
3377         }
3378         if (*arg != '\0')
3379                 return 0;
3380         /* separate form: --option value */
3381         if (!argv[1])
3382                 die("Option '--%s' requires a value", opt);
3383         *optarg = argv[1];
3384         return 2;
3385 }
3386
3387 static int stat_opt(struct diff_options *options, const char **av)
3388 {
3389         const char *arg = av[0];
3390         char *end;
3391         int width = options->stat_width;
3392         int name_width = options->stat_name_width;
3393         int graph_width = options->stat_graph_width;
3394         int count = options->stat_count;
3395         int argcount = 1;
3396
3397         arg += strlen("--stat");
3398         end = (char *)arg;
3399
3400         switch (*arg) {
3401         case '-':
3402                 if (!prefixcmp(arg, "-width")) {
3403                         arg += strlen("-width");
3404                         if (*arg == '=')
3405                                 width = strtoul(arg + 1, &end, 10);
3406                         else if (!*arg && !av[1])
3407                                 die("Option '--stat-width' requires a value");
3408                         else if (!*arg) {
3409                                 width = strtoul(av[1], &end, 10);
3410                                 argcount = 2;
3411                         }
3412                 } else if (!prefixcmp(arg, "-name-width")) {
3413                         arg += strlen("-name-width");
3414                         if (*arg == '=')
3415                                 name_width = strtoul(arg + 1, &end, 10);
3416                         else if (!*arg && !av[1])
3417                                 die("Option '--stat-name-width' requires a value");
3418                         else if (!*arg) {
3419                                 name_width = strtoul(av[1], &end, 10);
3420                                 argcount = 2;
3421                         }
3422                 } else if (!prefixcmp(arg, "-graph-width")) {
3423                         arg += strlen("-graph-width");
3424                         if (*arg == '=')
3425                                 graph_width = strtoul(arg + 1, &end, 10);
3426                         else if (!*arg && !av[1])
3427                                 die("Option '--stat-graph-width' requires a value");
3428                         else if (!*arg) {
3429                                 graph_width = strtoul(av[1], &end, 10);
3430                                 argcount = 2;
3431                         }
3432                 } else if (!prefixcmp(arg, "-count")) {
3433                         arg += strlen("-count");
3434                         if (*arg == '=')
3435                                 count = strtoul(arg + 1, &end, 10);
3436                         else if (!*arg && !av[1])
3437                                 die("Option '--stat-count' requires a value");
3438                         else if (!*arg) {
3439                                 count = strtoul(av[1], &end, 10);
3440                                 argcount = 2;
3441                         }
3442                 }
3443                 break;
3444         case '=':
3445                 width = strtoul(arg+1, &end, 10);
3446                 if (*end == ',')
3447                         name_width = strtoul(end+1, &end, 10);
3448                 if (*end == ',')
3449                         count = strtoul(end+1, &end, 10);
3450         }
3451
3452         /* Important! This checks all the error cases! */
3453         if (*end)
3454                 return 0;
3455         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3456         options->stat_name_width = name_width;
3457         options->stat_graph_width = graph_width;
3458         options->stat_width = width;
3459         options->stat_count = count;
3460         return argcount;
3461 }
3462
3463 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3464 {
3465         struct strbuf errmsg = STRBUF_INIT;
3466         if (parse_dirstat_params(options, params, &errmsg))
3467                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3468                     errmsg.buf);
3469         strbuf_release(&errmsg);
3470         /*
3471          * The caller knows a dirstat-related option is given from the command
3472          * line; allow it to say "return this_function();"
3473          */
3474         options->output_format |= DIFF_FORMAT_DIRSTAT;
3475         return 1;
3476 }
3477
3478 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3479 {
3480         const char *arg = av[0];
3481         const char *optarg;
3482         int argcount;
3483
3484         /* Output format options */
3485         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch"))
3486                 options->output_format |= DIFF_FORMAT_PATCH;
3487         else if (opt_arg(arg, 'U', "unified", &options->context))
3488                 options->output_format |= DIFF_FORMAT_PATCH;
3489         else if (!strcmp(arg, "--raw"))
3490                 options->output_format |= DIFF_FORMAT_RAW;
3491         else if (!strcmp(arg, "--patch-with-raw"))
3492                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
3493         else if (!strcmp(arg, "--numstat"))
3494                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3495         else if (!strcmp(arg, "--shortstat"))
3496                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3497         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3498                 return parse_dirstat_opt(options, "");
3499         else if (!prefixcmp(arg, "-X"))
3500                 return parse_dirstat_opt(options, arg + 2);
3501         else if (!prefixcmp(arg, "--dirstat="))
3502                 return parse_dirstat_opt(options, arg + 10);
3503         else if (!strcmp(arg, "--cumulative"))
3504                 return parse_dirstat_opt(options, "cumulative");
3505         else if (!strcmp(arg, "--dirstat-by-file"))
3506                 return parse_dirstat_opt(options, "files");
3507         else if (!prefixcmp(arg, "--dirstat-by-file=")) {
3508                 parse_dirstat_opt(options, "files");
3509                 return parse_dirstat_opt(options, arg + 18);
3510         }
3511         else if (!strcmp(arg, "--check"))
3512                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3513         else if (!strcmp(arg, "--summary"))
3514                 options->output_format |= DIFF_FORMAT_SUMMARY;
3515         else if (!strcmp(arg, "--patch-with-stat"))
3516                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
3517         else if (!strcmp(arg, "--name-only"))
3518                 options->output_format |= DIFF_FORMAT_NAME;
3519         else if (!strcmp(arg, "--name-status"))
3520                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3521         else if (!strcmp(arg, "-s"))
3522                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3523         else if (!prefixcmp(arg, "--stat"))
3524                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3525                 return stat_opt(options, av);
3526
3527         /* renames options */
3528         else if (!prefixcmp(arg, "-B") || !prefixcmp(arg, "--break-rewrites=") ||
3529                  !strcmp(arg, "--break-rewrites")) {
3530                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3531                         return error("invalid argument to -B: %s", arg+2);
3532         }
3533         else if (!prefixcmp(arg, "-M") || !prefixcmp(arg, "--find-renames=") ||
3534                  !strcmp(arg, "--find-renames")) {
3535                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3536                         return error("invalid argument to -M: %s", arg+2);
3537                 options->detect_rename = DIFF_DETECT_RENAME;
3538         }
3539         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3540                 options->irreversible_delete = 1;
3541         }
3542         else if (!prefixcmp(arg, "-C") || !prefixcmp(arg, "--find-copies=") ||
3543                  !strcmp(arg, "--find-copies")) {
3544                 if (options->detect_rename == DIFF_DETECT_COPY)
3545                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3546                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3547                         return error("invalid argument to -C: %s", arg+2);
3548                 options->detect_rename = DIFF_DETECT_COPY;
3549         }
3550         else if (!strcmp(arg, "--no-renames"))
3551                 options->detect_rename = 0;
3552         else if (!strcmp(arg, "--rename-empty"))
3553                 DIFF_OPT_SET(options, RENAME_EMPTY);
3554         else if (!strcmp(arg, "--no-rename-empty"))
3555                 DIFF_OPT_CLR(options, RENAME_EMPTY);
3556         else if (!strcmp(arg, "--relative"))
3557                 DIFF_OPT_SET(options, RELATIVE_NAME);
3558         else if (!prefixcmp(arg, "--relative=")) {
3559                 DIFF_OPT_SET(options, RELATIVE_NAME);
3560                 options->prefix = arg + 11;
3561         }
3562
3563         /* xdiff options */
3564         else if (!strcmp(arg, "--minimal"))
3565                 DIFF_XDL_SET(options, NEED_MINIMAL);
3566         else if (!strcmp(arg, "--no-minimal"))
3567                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3568         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3569                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3570         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3571                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3572         else if (!strcmp(arg, "--ignore-space-at-eol"))
3573                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3574         else if (!strcmp(arg, "--patience"))
3575                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3576         else if (!strcmp(arg, "--histogram"))
3577                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3578
3579         /* flags options */
3580         else if (!strcmp(arg, "--binary")) {
3581                 options->output_format |= DIFF_FORMAT_PATCH;
3582                 DIFF_OPT_SET(options, BINARY);
3583         }
3584         else if (!strcmp(arg, "--full-index"))
3585                 DIFF_OPT_SET(options, FULL_INDEX);
3586         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3587                 DIFF_OPT_SET(options, TEXT);
3588         else if (!strcmp(arg, "-R"))
3589                 DIFF_OPT_SET(options, REVERSE_DIFF);
3590         else if (!strcmp(arg, "--find-copies-harder"))
3591                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3592         else if (!strcmp(arg, "--follow"))
3593                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3594         else if (!strcmp(arg, "--color"))
3595                 options->use_color = 1;
3596         else if (!prefixcmp(arg, "--color=")) {
3597                 int value = git_config_colorbool(NULL, arg+8);
3598                 if (value < 0)
3599                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3600                 options->use_color = value;
3601         }
3602         else if (!strcmp(arg, "--no-color"))
3603                 options->use_color = 0;
3604         else if (!strcmp(arg, "--color-words")) {
3605                 options->use_color = 1;
3606                 options->word_diff = DIFF_WORDS_COLOR;
3607         }
3608         else if (!prefixcmp(arg, "--color-words=")) {
3609                 options->use_color = 1;
3610                 options->word_diff = DIFF_WORDS_COLOR;
3611                 options->word_regex = arg + 14;
3612         }
3613         else if (!strcmp(arg, "--word-diff")) {
3614                 if (options->word_diff == DIFF_WORDS_NONE)
3615                         options->word_diff = DIFF_WORDS_PLAIN;
3616         }
3617         else if (!prefixcmp(arg, "--word-diff=")) {
3618                 const char *type = arg + 12;
3619                 if (!strcmp(type, "plain"))
3620                         options->word_diff = DIFF_WORDS_PLAIN;
3621                 else if (!strcmp(type, "color")) {
3622                         options->use_color = 1;
3623                         options->word_diff = DIFF_WORDS_COLOR;
3624                 }
3625                 else if (!strcmp(type, "porcelain"))
3626                         options->word_diff = DIFF_WORDS_PORCELAIN;
3627                 else if (!strcmp(type, "none"))
3628                         options->word_diff = DIFF_WORDS_NONE;
3629                 else
3630                         die("bad --word-diff argument: %s", type);
3631         }
3632         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3633                 if (options->word_diff == DIFF_WORDS_NONE)
3634                         options->word_diff = DIFF_WORDS_PLAIN;
3635                 options->word_regex = optarg;
3636                 return argcount;
3637         }
3638         else if (!strcmp(arg, "--exit-code"))
3639                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3640         else if (!strcmp(arg, "--quiet"))
3641                 DIFF_OPT_SET(options, QUICK);
3642         else if (!strcmp(arg, "--ext-diff"))
3643                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3644         else if (!strcmp(arg, "--no-ext-diff"))
3645                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3646         else if (!strcmp(arg, "--textconv"))
3647                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3648         else if (!strcmp(arg, "--no-textconv"))
3649                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3650         else if (!strcmp(arg, "--ignore-submodules")) {
3651                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3652                 handle_ignore_submodules_arg(options, "all");
3653         } else if (!prefixcmp(arg, "--ignore-submodules=")) {
3654                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3655                 handle_ignore_submodules_arg(options, arg + 20);
3656         } else if (!strcmp(arg, "--submodule"))
3657                 DIFF_OPT_SET(options, SUBMODULE_LOG);
3658         else if (!prefixcmp(arg, "--submodule=")) {
3659                 if (!strcmp(arg + 12, "log"))
3660                         DIFF_OPT_SET(options, SUBMODULE_LOG);
3661         }
3662
3663         /* misc options */
3664         else if (!strcmp(arg, "-z"))
3665                 options->line_termination = 0;
3666         else if ((argcount = short_opt('l', av, &optarg))) {
3667                 options->rename_limit = strtoul(optarg, NULL, 10);
3668                 return argcount;
3669         }
3670         else if ((argcount = short_opt('S', av, &optarg))) {
3671                 options->pickaxe = optarg;
3672                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3673                 return argcount;
3674         } else if ((argcount = short_opt('G', av, &optarg))) {
3675                 options->pickaxe = optarg;
3676                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3677                 return argcount;
3678         }
3679         else if (!strcmp(arg, "--pickaxe-all"))
3680                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3681         else if (!strcmp(arg, "--pickaxe-regex"))
3682                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3683         else if ((argcount = short_opt('O', av, &optarg))) {
3684                 options->orderfile = optarg;
3685                 return argcount;
3686         }
3687         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3688                 options->filter = optarg;
3689                 return argcount;
3690         }
3691         else if (!strcmp(arg, "--abbrev"))
3692                 options->abbrev = DEFAULT_ABBREV;
3693         else if (!prefixcmp(arg, "--abbrev=")) {
3694                 options->abbrev = strtoul(arg + 9, NULL, 10);
3695                 if (options->abbrev < MINIMUM_ABBREV)
3696                         options->abbrev = MINIMUM_ABBREV;
3697                 else if (40 < options->abbrev)
3698                         options->abbrev = 40;
3699         }
3700         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3701                 options->a_prefix = optarg;
3702                 return argcount;
3703         }
3704         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3705                 options->b_prefix = optarg;
3706                 return argcount;
3707         }
3708         else if (!strcmp(arg, "--no-prefix"))
3709                 options->a_prefix = options->b_prefix = "";
3710         else if (opt_arg(arg, '\0', "inter-hunk-context",
3711                          &options->interhunkcontext))
3712                 ;
3713         else if (!strcmp(arg, "-W"))
3714                 DIFF_OPT_SET(options, FUNCCONTEXT);
3715         else if (!strcmp(arg, "--function-context"))
3716                 DIFF_OPT_SET(options, FUNCCONTEXT);
3717         else if (!strcmp(arg, "--no-function-context"))
3718                 DIFF_OPT_CLR(options, FUNCCONTEXT);
3719         else if ((argcount = parse_long_opt("output", av, &optarg))) {
3720                 options->file = fopen(optarg, "w");
3721                 if (!options->file)
3722                         die_errno("Could not open '%s'", optarg);
3723                 options->close_file = 1;
3724                 return argcount;
3725         } else
3726                 return 0;
3727         return 1;
3728 }
3729
3730 int parse_rename_score(const char **cp_p)
3731 {
3732         unsigned long num, scale;
3733         int ch, dot;
3734         const char *cp = *cp_p;
3735
3736         num = 0;
3737         scale = 1;
3738         dot = 0;
3739         for (;;) {
3740                 ch = *cp;
3741                 if ( !dot && ch == '.' ) {
3742                         scale = 1;
3743                         dot = 1;
3744                 } else if ( ch == '%' ) {
3745                         scale = dot ? scale*100 : 100;
3746                         cp++;   /* % is always at the end */
3747                         break;
3748                 } else if ( ch >= '0' && ch <= '9' ) {
3749                         if ( scale < 100000 ) {
3750                                 scale *= 10;
3751                                 num = (num*10) + (ch-'0');
3752                         }
3753                 } else {
3754                         break;
3755                 }
3756                 cp++;
3757         }
3758         *cp_p = cp;
3759
3760         /* user says num divided by scale and we say internally that
3761          * is MAX_SCORE * num / scale.
3762          */
3763         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
3764 }
3765
3766 static int diff_scoreopt_parse(const char *opt)
3767 {
3768         int opt1, opt2, cmd;
3769
3770         if (*opt++ != '-')
3771                 return -1;
3772         cmd = *opt++;
3773         if (cmd == '-') {
3774                 /* convert the long-form arguments into short-form versions */
3775                 if (!prefixcmp(opt, "break-rewrites")) {
3776                         opt += strlen("break-rewrites");
3777                         if (*opt == 0 || *opt++ == '=')
3778                                 cmd = 'B';
3779                 } else if (!prefixcmp(opt, "find-copies")) {
3780                         opt += strlen("find-copies");
3781                         if (*opt == 0 || *opt++ == '=')
3782                                 cmd = 'C';
3783                 } else if (!prefixcmp(opt, "find-renames")) {
3784                         opt += strlen("find-renames");
3785                         if (*opt == 0 || *opt++ == '=')
3786                                 cmd = 'M';
3787                 }
3788         }
3789         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
3790                 return -1; /* that is not a -M, -C nor -B option */
3791
3792         opt1 = parse_rename_score(&opt);
3793         if (cmd != 'B')
3794                 opt2 = 0;
3795         else {
3796                 if (*opt == 0)
3797                         opt2 = 0;
3798                 else if (*opt != '/')
3799                         return -1; /* we expect -B80/99 or -B80 */
3800                 else {
3801                         opt++;
3802                         opt2 = parse_rename_score(&opt);
3803                 }
3804         }
3805         if (*opt != 0)
3806                 return -1;
3807         return opt1 | (opt2 << 16);
3808 }
3809
3810 struct diff_queue_struct diff_queued_diff;
3811
3812 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
3813 {
3814         if (queue->alloc <= queue->nr) {
3815                 queue->alloc = alloc_nr(queue->alloc);
3816                 queue->queue = xrealloc(queue->queue,
3817                                         sizeof(dp) * queue->alloc);
3818         }
3819         queue->queue[queue->nr++] = dp;
3820 }
3821
3822 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
3823                                  struct diff_filespec *one,
3824                                  struct diff_filespec *two)
3825 {
3826         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3827         dp->one = one;
3828         dp->two = two;
3829         if (queue)
3830                 diff_q(queue, dp);
3831         return dp;
3832 }
3833
3834 void diff_free_filepair(struct diff_filepair *p)
3835 {
3836         free_filespec(p->one);
3837         free_filespec(p->two);
3838         free(p);
3839 }
3840
3841 /* This is different from find_unique_abbrev() in that
3842  * it stuffs the result with dots for alignment.
3843  */
3844 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3845 {
3846         int abblen;
3847         const char *abbrev;
3848         if (len == 40)
3849                 return sha1_to_hex(sha1);
3850
3851         abbrev = find_unique_abbrev(sha1, len);
3852         abblen = strlen(abbrev);
3853         if (abblen < 37) {
3854                 static char hex[41];
3855                 if (len < abblen && abblen <= len + 2)
3856                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3857                 else
3858                         sprintf(hex, "%s...", abbrev);
3859                 return hex;
3860         }
3861         return sha1_to_hex(sha1);
3862 }
3863
3864 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3865 {
3866         int line_termination = opt->line_termination;
3867         int inter_name_termination = line_termination ? '\t' : '\0';
3868         if (opt->output_prefix) {
3869                 struct strbuf *msg = NULL;
3870                 msg = opt->output_prefix(opt, opt->output_prefix_data);
3871                 fprintf(opt->file, "%s", msg->buf);
3872         }
3873
3874         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3875                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3876                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
3877                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3878         }
3879         if (p->score) {
3880                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3881                         inter_name_termination);
3882         } else {
3883                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
3884         }
3885
3886         if (p->status == DIFF_STATUS_COPIED ||
3887             p->status == DIFF_STATUS_RENAMED) {
3888                 const char *name_a, *name_b;
3889                 name_a = p->one->path;
3890                 name_b = p->two->path;
3891                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3892                 write_name_quoted(name_a, opt->file, inter_name_termination);
3893                 write_name_quoted(name_b, opt->file, line_termination);
3894         } else {
3895                 const char *name_a, *name_b;
3896                 name_a = p->one->mode ? p->one->path : p->two->path;
3897                 name_b = NULL;
3898                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3899                 write_name_quoted(name_a, opt->file, line_termination);
3900         }
3901 }
3902
3903 int diff_unmodified_pair(struct diff_filepair *p)
3904 {
3905         /* This function is written stricter than necessary to support
3906          * the currently implemented transformers, but the idea is to
3907          * let transformers to produce diff_filepairs any way they want,
3908          * and filter and clean them up here before producing the output.
3909          */
3910         struct diff_filespec *one = p->one, *two = p->two;
3911
3912         if (DIFF_PAIR_UNMERGED(p))
3913                 return 0; /* unmerged is interesting */
3914
3915         /* deletion, addition, mode or type change
3916          * and rename are all interesting.
3917          */
3918         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
3919             DIFF_PAIR_MODE_CHANGED(p) ||
3920             strcmp(one->path, two->path))
3921                 return 0;
3922
3923         /* both are valid and point at the same path.  that is, we are
3924          * dealing with a change.
3925          */
3926         if (one->sha1_valid && two->sha1_valid &&
3927             !hashcmp(one->sha1, two->sha1) &&
3928             !one->dirty_submodule && !two->dirty_submodule)
3929                 return 1; /* no change */
3930         if (!one->sha1_valid && !two->sha1_valid)
3931                 return 1; /* both look at the same file on the filesystem. */
3932         return 0;
3933 }
3934
3935 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
3936 {
3937         if (diff_unmodified_pair(p))
3938                 return;
3939
3940         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3941             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3942                 return; /* no tree diffs in patch format */
3943
3944         run_diff(p, o);
3945 }
3946
3947 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
3948                             struct diffstat_t *diffstat)
3949 {
3950         if (diff_unmodified_pair(p))
3951                 return;
3952
3953         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3954             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3955                 return; /* no useful stat for tree diffs */
3956
3957         run_diffstat(p, o, diffstat);
3958 }
3959
3960 static void diff_flush_checkdiff(struct diff_filepair *p,
3961                 struct diff_options *o)
3962 {
3963         if (diff_unmodified_pair(p))
3964                 return;
3965
3966         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3967             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3968                 return; /* nothing to check in tree diffs */
3969
3970         run_checkdiff(p, o);
3971 }
3972
3973 int diff_queue_is_empty(void)
3974 {
3975         struct diff_queue_struct *q = &diff_queued_diff;
3976         int i;
3977         for (i = 0; i < q->nr; i++)
3978                 if (!diff_unmodified_pair(q->queue[i]))
3979                         return 0;
3980         return 1;
3981 }
3982
3983 #if DIFF_DEBUG
3984 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
3985 {
3986         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
3987                 x, one ? one : "",
3988                 s->path,
3989                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
3990                 s->mode,
3991                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
3992         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
3993                 x, one ? one : "",
3994                 s->size, s->xfrm_flags);
3995 }
3996
3997 void diff_debug_filepair(const struct diff_filepair *p, int i)
3998 {
3999         diff_debug_filespec(p->one, i, "one");
4000         diff_debug_filespec(p->two, i, "two");
4001         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4002                 p->score, p->status ? p->status : '?',
4003                 p->one->rename_used, p->broken_pair);
4004 }
4005
4006 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4007 {
4008         int i;
4009         if (msg)
4010                 fprintf(stderr, "%s\n", msg);
4011         fprintf(stderr, "q->nr = %d\n", q->nr);
4012         for (i = 0; i < q->nr; i++) {
4013                 struct diff_filepair *p = q->queue[i];
4014                 diff_debug_filepair(p, i);
4015         }
4016 }
4017 #endif
4018
4019 static void diff_resolve_rename_copy(void)
4020 {
4021         int i;
4022         struct diff_filepair *p;
4023         struct diff_queue_struct *q = &diff_queued_diff;
4024
4025         diff_debug_queue("resolve-rename-copy", q);
4026
4027         for (i = 0; i < q->nr; i++) {
4028                 p = q->queue[i];
4029                 p->status = 0; /* undecided */
4030                 if (DIFF_PAIR_UNMERGED(p))
4031                         p->status = DIFF_STATUS_UNMERGED;
4032                 else if (!DIFF_FILE_VALID(p->one))
4033                         p->status = DIFF_STATUS_ADDED;
4034                 else if (!DIFF_FILE_VALID(p->two))
4035                         p->status = DIFF_STATUS_DELETED;
4036                 else if (DIFF_PAIR_TYPE_CHANGED(p))
4037                         p->status = DIFF_STATUS_TYPE_CHANGED;
4038
4039                 /* from this point on, we are dealing with a pair
4040                  * whose both sides are valid and of the same type, i.e.
4041                  * either in-place edit or rename/copy edit.
4042                  */
4043                 else if (DIFF_PAIR_RENAME(p)) {
4044                         /*
4045                          * A rename might have re-connected a broken
4046                          * pair up, causing the pathnames to be the
4047                          * same again. If so, that's not a rename at
4048                          * all, just a modification..
4049                          *
4050                          * Otherwise, see if this source was used for
4051                          * multiple renames, in which case we decrement
4052                          * the count, and call it a copy.
4053                          */
4054                         if (!strcmp(p->one->path, p->two->path))
4055                                 p->status = DIFF_STATUS_MODIFIED;
4056                         else if (--p->one->rename_used > 0)
4057                                 p->status = DIFF_STATUS_COPIED;
4058                         else
4059                                 p->status = DIFF_STATUS_RENAMED;
4060                 }
4061                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
4062                          p->one->mode != p->two->mode ||
4063                          p->one->dirty_submodule ||
4064                          p->two->dirty_submodule ||
4065                          is_null_sha1(p->one->sha1))
4066                         p->status = DIFF_STATUS_MODIFIED;
4067                 else {
4068                         /* This is a "no-change" entry and should not
4069                          * happen anymore, but prepare for broken callers.
4070                          */
4071                         error("feeding unmodified %s to diffcore",
4072                               p->one->path);
4073                         p->status = DIFF_STATUS_UNKNOWN;
4074                 }
4075         }
4076         diff_debug_queue("resolve-rename-copy done", q);
4077 }
4078
4079 static int check_pair_status(struct diff_filepair *p)
4080 {
4081         switch (p->status) {
4082         case DIFF_STATUS_UNKNOWN:
4083                 return 0;
4084         case 0:
4085                 die("internal error in diff-resolve-rename-copy");
4086         default:
4087                 return 1;
4088         }
4089 }
4090
4091 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4092 {
4093         int fmt = opt->output_format;
4094
4095         if (fmt & DIFF_FORMAT_CHECKDIFF)
4096                 diff_flush_checkdiff(p, opt);
4097         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4098                 diff_flush_raw(p, opt);
4099         else if (fmt & DIFF_FORMAT_NAME) {
4100                 const char *name_a, *name_b;
4101                 name_a = p->two->path;
4102                 name_b = NULL;
4103                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4104                 write_name_quoted(name_a, opt->file, opt->line_termination);
4105         }
4106 }
4107
4108 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4109 {
4110         if (fs->mode)
4111                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4112         else
4113                 fprintf(file, " %s ", newdelete);
4114         write_name_quoted(fs->path, file, '\n');
4115 }
4116
4117
4118 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4119                 const char *line_prefix)
4120 {
4121         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4122                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4123                         p->two->mode, show_name ? ' ' : '\n');
4124                 if (show_name) {
4125                         write_name_quoted(p->two->path, file, '\n');
4126                 }
4127         }
4128 }
4129
4130 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4131                         const char *line_prefix)
4132 {
4133         char *names = pprint_rename(p->one->path, p->two->path);
4134
4135         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4136         free(names);
4137         show_mode_change(file, p, 0, line_prefix);
4138 }
4139
4140 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4141 {
4142         FILE *file = opt->file;
4143         char *line_prefix = "";
4144
4145         if (opt->output_prefix) {
4146                 struct strbuf *buf = opt->output_prefix(opt, opt->output_prefix_data);
4147                 line_prefix = buf->buf;
4148         }
4149
4150         switch(p->status) {
4151         case DIFF_STATUS_DELETED:
4152                 fputs(line_prefix, file);
4153                 show_file_mode_name(file, "delete", p->one);
4154                 break;
4155         case DIFF_STATUS_ADDED:
4156                 fputs(line_prefix, file);
4157                 show_file_mode_name(file, "create", p->two);
4158                 break;
4159         case DIFF_STATUS_COPIED:
4160                 fputs(line_prefix, file);
4161                 show_rename_copy(file, "copy", p, line_prefix);
4162                 break;
4163         case DIFF_STATUS_RENAMED:
4164                 fputs(line_prefix, file);
4165                 show_rename_copy(file, "rename", p, line_prefix);
4166                 break;
4167         default:
4168                 if (p->score) {
4169                         fprintf(file, "%s rewrite ", line_prefix);
4170                         write_name_quoted(p->two->path, file, ' ');
4171                         fprintf(file, "(%d%%)\n", similarity_index(p));
4172                 }
4173                 show_mode_change(file, p, !p->score, line_prefix);
4174                 break;
4175         }
4176 }
4177
4178 struct patch_id_t {
4179         git_SHA_CTX *ctx;
4180         int patchlen;
4181 };
4182
4183 static int remove_space(char *line, int len)
4184 {
4185         int i;
4186         char *dst = line;
4187         unsigned char c;
4188
4189         for (i = 0; i < len; i++)
4190                 if (!isspace((c = line[i])))
4191                         *dst++ = c;
4192
4193         return dst - line;
4194 }
4195
4196 static void patch_id_consume(void *priv, char *line, unsigned long len)
4197 {
4198         struct patch_id_t *data = priv;
4199         int new_len;
4200
4201         /* Ignore line numbers when computing the SHA1 of the patch */
4202         if (!prefixcmp(line, "@@ -"))
4203                 return;
4204
4205         new_len = remove_space(line, len);
4206
4207         git_SHA1_Update(data->ctx, line, new_len);
4208         data->patchlen += new_len;
4209 }
4210
4211 /* returns 0 upon success, and writes result into sha1 */
4212 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
4213 {
4214         struct diff_queue_struct *q = &diff_queued_diff;
4215         int i;
4216         git_SHA_CTX ctx;
4217         struct patch_id_t data;
4218         char buffer[PATH_MAX * 4 + 20];
4219
4220         git_SHA1_Init(&ctx);
4221         memset(&data, 0, sizeof(struct patch_id_t));
4222         data.ctx = &ctx;
4223
4224         for (i = 0; i < q->nr; i++) {
4225                 xpparam_t xpp;
4226                 xdemitconf_t xecfg;
4227                 mmfile_t mf1, mf2;
4228                 struct diff_filepair *p = q->queue[i];
4229                 int len1, len2;
4230
4231                 memset(&xpp, 0, sizeof(xpp));
4232                 memset(&xecfg, 0, sizeof(xecfg));
4233                 if (p->status == 0)
4234                         return error("internal diff status error");
4235                 if (p->status == DIFF_STATUS_UNKNOWN)
4236                         continue;
4237                 if (diff_unmodified_pair(p))
4238                         continue;
4239                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4240                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4241                         continue;
4242                 if (DIFF_PAIR_UNMERGED(p))
4243                         continue;
4244
4245                 diff_fill_sha1_info(p->one);
4246                 diff_fill_sha1_info(p->two);
4247                 if (fill_mmfile(&mf1, p->one) < 0 ||
4248                                 fill_mmfile(&mf2, p->two) < 0)
4249                         return error("unable to read files to diff");
4250
4251                 len1 = remove_space(p->one->path, strlen(p->one->path));
4252                 len2 = remove_space(p->two->path, strlen(p->two->path));
4253                 if (p->one->mode == 0)
4254                         len1 = snprintf(buffer, sizeof(buffer),
4255                                         "diff--gita/%.*sb/%.*s"
4256                                         "newfilemode%06o"
4257                                         "---/dev/null"
4258                                         "+++b/%.*s",
4259                                         len1, p->one->path,
4260                                         len2, p->two->path,
4261                                         p->two->mode,
4262                                         len2, p->two->path);
4263                 else if (p->two->mode == 0)
4264                         len1 = snprintf(buffer, sizeof(buffer),
4265                                         "diff--gita/%.*sb/%.*s"
4266                                         "deletedfilemode%06o"
4267                                         "---a/%.*s"
4268                                         "+++/dev/null",
4269                                         len1, p->one->path,
4270                                         len2, p->two->path,
4271                                         p->one->mode,
4272                                         len1, p->one->path);
4273                 else
4274                         len1 = snprintf(buffer, sizeof(buffer),
4275                                         "diff--gita/%.*sb/%.*s"
4276                                         "---a/%.*s"
4277                                         "+++b/%.*s",
4278                                         len1, p->one->path,
4279                                         len2, p->two->path,
4280                                         len1, p->one->path,
4281                                         len2, p->two->path);
4282                 git_SHA1_Update(&ctx, buffer, len1);
4283
4284                 if (diff_filespec_is_binary(p->one) ||
4285                     diff_filespec_is_binary(p->two)) {
4286                         git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
4287                         git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
4288                         continue;
4289                 }
4290
4291                 xpp.flags = 0;
4292                 xecfg.ctxlen = 3;
4293                 xecfg.flags = 0;
4294                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4295                               &xpp, &xecfg);
4296         }
4297
4298         git_SHA1_Final(sha1, &ctx);
4299         return 0;
4300 }
4301
4302 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
4303 {
4304         struct diff_queue_struct *q = &diff_queued_diff;
4305         int i;
4306         int result = diff_get_patch_id(options, sha1);
4307
4308         for (i = 0; i < q->nr; i++)
4309                 diff_free_filepair(q->queue[i]);
4310
4311         free(q->queue);
4312         DIFF_QUEUE_CLEAR(q);
4313
4314         return result;
4315 }
4316
4317 static int is_summary_empty(const struct diff_queue_struct *q)
4318 {
4319         int i;
4320
4321         for (i = 0; i < q->nr; i++) {
4322                 const struct diff_filepair *p = q->queue[i];
4323
4324                 switch (p->status) {
4325                 case DIFF_STATUS_DELETED:
4326                 case DIFF_STATUS_ADDED:
4327                 case DIFF_STATUS_COPIED:
4328                 case DIFF_STATUS_RENAMED:
4329                         return 0;
4330                 default:
4331                         if (p->score)
4332                                 return 0;
4333                         if (p->one->mode && p->two->mode &&
4334                             p->one->mode != p->two->mode)
4335                                 return 0;
4336                         break;
4337                 }
4338         }
4339         return 1;
4340 }
4341
4342 static const char rename_limit_warning[] =
4343 "inexact rename detection was skipped due to too many files.";
4344
4345 static const char degrade_cc_to_c_warning[] =
4346 "only found copies from modified paths due to too many files.";
4347
4348 static const char rename_limit_advice[] =
4349 "you may want to set your %s variable to at least "
4350 "%d and retry the command.";
4351
4352 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4353 {
4354         if (degraded_cc)
4355                 warning(degrade_cc_to_c_warning);
4356         else if (needed)
4357                 warning(rename_limit_warning);
4358         else
4359                 return;
4360         if (0 < needed && needed < 32767)
4361                 warning(rename_limit_advice, varname, needed);
4362 }
4363
4364 void diff_flush(struct diff_options *options)
4365 {
4366         struct diff_queue_struct *q = &diff_queued_diff;
4367         int i, output_format = options->output_format;
4368         int separator = 0;
4369         int dirstat_by_line = 0;
4370
4371         /*
4372          * Order: raw, stat, summary, patch
4373          * or:    name/name-status/checkdiff (other bits clear)
4374          */
4375         if (!q->nr)
4376                 goto free_queue;
4377
4378         if (output_format & (DIFF_FORMAT_RAW |
4379                              DIFF_FORMAT_NAME |
4380                              DIFF_FORMAT_NAME_STATUS |
4381                              DIFF_FORMAT_CHECKDIFF)) {
4382                 for (i = 0; i < q->nr; i++) {
4383                         struct diff_filepair *p = q->queue[i];
4384                         if (check_pair_status(p))
4385                                 flush_one_pair(p, options);
4386                 }
4387                 separator++;
4388         }
4389
4390         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4391                 dirstat_by_line = 1;
4392
4393         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4394             dirstat_by_line) {
4395                 struct diffstat_t diffstat;
4396
4397                 memset(&diffstat, 0, sizeof(struct diffstat_t));
4398                 for (i = 0; i < q->nr; i++) {
4399                         struct diff_filepair *p = q->queue[i];
4400                         if (check_pair_status(p))
4401                                 diff_flush_stat(p, options, &diffstat);
4402                 }
4403                 if (output_format & DIFF_FORMAT_NUMSTAT)
4404                         show_numstat(&diffstat, options);
4405                 if (output_format & DIFF_FORMAT_DIFFSTAT)
4406                         show_stats(&diffstat, options);
4407                 if (output_format & DIFF_FORMAT_SHORTSTAT)
4408                         show_shortstats(&diffstat, options);
4409                 if (output_format & DIFF_FORMAT_DIRSTAT)
4410                         show_dirstat_by_line(&diffstat, options);
4411                 free_diffstat_info(&diffstat);
4412                 separator++;
4413         }
4414         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4415                 show_dirstat(options);
4416
4417         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4418                 for (i = 0; i < q->nr; i++) {
4419                         diff_summary(options, q->queue[i]);
4420                 }
4421                 separator++;
4422         }
4423
4424         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4425             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4426             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4427                 /*
4428                  * run diff_flush_patch for the exit status. setting
4429                  * options->file to /dev/null should be safe, becaue we
4430                  * aren't supposed to produce any output anyway.
4431                  */
4432                 if (options->close_file)
4433                         fclose(options->file);
4434                 options->file = fopen("/dev/null", "w");
4435                 if (!options->file)
4436                         die_errno("Could not open /dev/null");
4437                 options->close_file = 1;
4438                 for (i = 0; i < q->nr; i++) {
4439                         struct diff_filepair *p = q->queue[i];
4440                         if (check_pair_status(p))
4441                                 diff_flush_patch(p, options);
4442                         if (options->found_changes)
4443                                 break;
4444                 }
4445         }
4446
4447         if (output_format & DIFF_FORMAT_PATCH) {
4448                 if (separator) {
4449                         if (options->output_prefix) {
4450                                 struct strbuf *msg = NULL;
4451                                 msg = options->output_prefix(options,
4452                                         options->output_prefix_data);
4453                                 fwrite(msg->buf, msg->len, 1, stdout);
4454                         }
4455                         putc(options->line_termination, options->file);
4456                         if (options->stat_sep) {
4457                                 /* attach patch instead of inline */
4458                                 fputs(options->stat_sep, options->file);
4459                         }
4460                 }
4461
4462                 for (i = 0; i < q->nr; i++) {
4463                         struct diff_filepair *p = q->queue[i];
4464                         if (check_pair_status(p))
4465                                 diff_flush_patch(p, options);
4466                 }
4467         }
4468
4469         if (output_format & DIFF_FORMAT_CALLBACK)
4470                 options->format_callback(q, options, options->format_callback_data);
4471
4472         for (i = 0; i < q->nr; i++)
4473                 diff_free_filepair(q->queue[i]);
4474 free_queue:
4475         free(q->queue);
4476         DIFF_QUEUE_CLEAR(q);
4477         if (options->close_file)
4478                 fclose(options->file);
4479
4480         /*
4481          * Report the content-level differences with HAS_CHANGES;
4482          * diff_addremove/diff_change does not set the bit when
4483          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4484          */
4485         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4486                 if (options->found_changes)
4487                         DIFF_OPT_SET(options, HAS_CHANGES);
4488                 else
4489                         DIFF_OPT_CLR(options, HAS_CHANGES);
4490         }
4491 }
4492
4493 static void diffcore_apply_filter(const char *filter)
4494 {
4495         int i;
4496         struct diff_queue_struct *q = &diff_queued_diff;
4497         struct diff_queue_struct outq;
4498         DIFF_QUEUE_CLEAR(&outq);
4499
4500         if (!filter)
4501                 return;
4502
4503         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
4504                 int found;
4505                 for (i = found = 0; !found && i < q->nr; i++) {
4506                         struct diff_filepair *p = q->queue[i];
4507                         if (((p->status == DIFF_STATUS_MODIFIED) &&
4508                              ((p->score &&
4509                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
4510                               (!p->score &&
4511                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
4512                             ((p->status != DIFF_STATUS_MODIFIED) &&
4513                              strchr(filter, p->status)))
4514                                 found++;
4515                 }
4516                 if (found)
4517                         return;
4518
4519                 /* otherwise we will clear the whole queue
4520                  * by copying the empty outq at the end of this
4521                  * function, but first clear the current entries
4522                  * in the queue.
4523                  */
4524                 for (i = 0; i < q->nr; i++)
4525                         diff_free_filepair(q->queue[i]);
4526         }
4527         else {
4528                 /* Only the matching ones */
4529                 for (i = 0; i < q->nr; i++) {
4530                         struct diff_filepair *p = q->queue[i];
4531
4532                         if (((p->status == DIFF_STATUS_MODIFIED) &&
4533                              ((p->score &&
4534                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
4535                               (!p->score &&
4536                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
4537                             ((p->status != DIFF_STATUS_MODIFIED) &&
4538                              strchr(filter, p->status)))
4539                                 diff_q(&outq, p);
4540                         else
4541                                 diff_free_filepair(p);
4542                 }
4543         }
4544         free(q->queue);
4545         *q = outq;
4546 }
4547
4548 /* Check whether two filespecs with the same mode and size are identical */
4549 static int diff_filespec_is_identical(struct diff_filespec *one,
4550                                       struct diff_filespec *two)
4551 {
4552         if (S_ISGITLINK(one->mode))
4553                 return 0;
4554         if (diff_populate_filespec(one, 0))
4555                 return 0;
4556         if (diff_populate_filespec(two, 0))
4557                 return 0;
4558         return !memcmp(one->data, two->data, one->size);
4559 }
4560
4561 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4562 {
4563         int i;
4564         struct diff_queue_struct *q = &diff_queued_diff;
4565         struct diff_queue_struct outq;
4566         DIFF_QUEUE_CLEAR(&outq);
4567
4568         for (i = 0; i < q->nr; i++) {
4569                 struct diff_filepair *p = q->queue[i];
4570
4571                 /*
4572                  * 1. Entries that come from stat info dirtiness
4573                  *    always have both sides (iow, not create/delete),
4574                  *    one side of the object name is unknown, with
4575                  *    the same mode and size.  Keep the ones that
4576                  *    do not match these criteria.  They have real
4577                  *    differences.
4578                  *
4579                  * 2. At this point, the file is known to be modified,
4580                  *    with the same mode and size, and the object
4581                  *    name of one side is unknown.  Need to inspect
4582                  *    the identical contents.
4583                  */
4584                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
4585                     !DIFF_FILE_VALID(p->two) ||
4586                     (p->one->sha1_valid && p->two->sha1_valid) ||
4587                     (p->one->mode != p->two->mode) ||
4588                     diff_populate_filespec(p->one, 1) ||
4589                     diff_populate_filespec(p->two, 1) ||
4590                     (p->one->size != p->two->size) ||
4591                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4592                         diff_q(&outq, p);
4593                 else {
4594                         /*
4595                          * The caller can subtract 1 from skip_stat_unmatch
4596                          * to determine how many paths were dirty only
4597                          * due to stat info mismatch.
4598                          */
4599                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4600                                 diffopt->skip_stat_unmatch++;
4601                         diff_free_filepair(p);
4602                 }
4603         }
4604         free(q->queue);
4605         *q = outq;
4606 }
4607
4608 static int diffnamecmp(const void *a_, const void *b_)
4609 {
4610         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4611         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4612         const char *name_a, *name_b;
4613
4614         name_a = a->one ? a->one->path : a->two->path;
4615         name_b = b->one ? b->one->path : b->two->path;
4616         return strcmp(name_a, name_b);
4617 }
4618
4619 void diffcore_fix_diff_index(struct diff_options *options)
4620 {
4621         struct diff_queue_struct *q = &diff_queued_diff;
4622         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4623 }
4624
4625 void diffcore_std(struct diff_options *options)
4626 {
4627         if (options->skip_stat_unmatch)
4628                 diffcore_skip_stat_unmatch(options);
4629         if (!options->found_follow) {
4630                 /* See try_to_follow_renames() in tree-diff.c */
4631                 if (options->break_opt != -1)
4632                         diffcore_break(options->break_opt);
4633                 if (options->detect_rename)
4634                         diffcore_rename(options);
4635                 if (options->break_opt != -1)
4636                         diffcore_merge_broken();
4637         }
4638         if (options->pickaxe)
4639                 diffcore_pickaxe(options);
4640         if (options->orderfile)
4641                 diffcore_order(options->orderfile);
4642         if (!options->found_follow)
4643                 /* See try_to_follow_renames() in tree-diff.c */
4644                 diff_resolve_rename_copy();
4645         diffcore_apply_filter(options->filter);
4646
4647         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4648                 DIFF_OPT_SET(options, HAS_CHANGES);
4649         else
4650                 DIFF_OPT_CLR(options, HAS_CHANGES);
4651
4652         options->found_follow = 0;
4653 }
4654
4655 int diff_result_code(struct diff_options *opt, int status)
4656 {
4657         int result = 0;
4658
4659         diff_warn_rename_limit("diff.renamelimit",
4660                                opt->needed_rename_limit,
4661                                opt->degraded_cc_to_c);
4662         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4663             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4664                 return status;
4665         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4666             DIFF_OPT_TST(opt, HAS_CHANGES))
4667                 result |= 01;
4668         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4669             DIFF_OPT_TST(opt, CHECK_FAILED))
4670                 result |= 02;
4671         return result;
4672 }
4673
4674 int diff_can_quit_early(struct diff_options *opt)
4675 {
4676         return (DIFF_OPT_TST(opt, QUICK) &&
4677                 !opt->filter &&
4678                 DIFF_OPT_TST(opt, HAS_CHANGES));
4679 }
4680
4681 /*
4682  * Shall changes to this submodule be ignored?
4683  *
4684  * Submodule changes can be configured to be ignored separately for each path,
4685  * but that configuration can be overridden from the command line.
4686  */
4687 static int is_submodule_ignored(const char *path, struct diff_options *options)
4688 {
4689         int ignored = 0;
4690         unsigned orig_flags = options->flags;
4691         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4692                 set_diffopt_flags_from_submodule_config(options, path);
4693         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4694                 ignored = 1;
4695         options->flags = orig_flags;
4696         return ignored;
4697 }
4698
4699 void diff_addremove(struct diff_options *options,
4700                     int addremove, unsigned mode,
4701                     const unsigned char *sha1,
4702                     int sha1_valid,
4703                     const char *concatpath, unsigned dirty_submodule)
4704 {
4705         struct diff_filespec *one, *two;
4706
4707         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4708                 return;
4709
4710         /* This may look odd, but it is a preparation for
4711          * feeding "there are unchanged files which should
4712          * not produce diffs, but when you are doing copy
4713          * detection you would need them, so here they are"
4714          * entries to the diff-core.  They will be prefixed
4715          * with something like '=' or '*' (I haven't decided
4716          * which but should not make any difference).
4717          * Feeding the same new and old to diff_change()
4718          * also has the same effect.
4719          * Before the final output happens, they are pruned after
4720          * merged into rename/copy pairs as appropriate.
4721          */
4722         if (DIFF_OPT_TST(options, REVERSE_DIFF))
4723                 addremove = (addremove == '+' ? '-' :
4724                              addremove == '-' ? '+' : addremove);
4725
4726         if (options->prefix &&
4727             strncmp(concatpath, options->prefix, options->prefix_length))
4728                 return;
4729
4730         one = alloc_filespec(concatpath);
4731         two = alloc_filespec(concatpath);
4732
4733         if (addremove != '+')
4734                 fill_filespec(one, sha1, sha1_valid, mode);
4735         if (addremove != '-') {
4736                 fill_filespec(two, sha1, sha1_valid, mode);
4737                 two->dirty_submodule = dirty_submodule;
4738         }
4739
4740         diff_queue(&diff_queued_diff, one, two);
4741         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4742                 DIFF_OPT_SET(options, HAS_CHANGES);
4743 }
4744
4745 void diff_change(struct diff_options *options,
4746                  unsigned old_mode, unsigned new_mode,
4747                  const unsigned char *old_sha1,
4748                  const unsigned char *new_sha1,
4749                  int old_sha1_valid, int new_sha1_valid,
4750                  const char *concatpath,
4751                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4752 {
4753         struct diff_filespec *one, *two;
4754
4755         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4756             is_submodule_ignored(concatpath, options))
4757                 return;
4758
4759         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4760                 unsigned tmp;
4761                 const unsigned char *tmp_c;
4762                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4763                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4764                 tmp = old_sha1_valid; old_sha1_valid = new_sha1_valid;
4765                         new_sha1_valid = tmp;
4766                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
4767                         new_dirty_submodule = tmp;
4768         }
4769
4770         if (options->prefix &&
4771             strncmp(concatpath, options->prefix, options->prefix_length))
4772                 return;
4773
4774         one = alloc_filespec(concatpath);
4775         two = alloc_filespec(concatpath);
4776         fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
4777         fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
4778         one->dirty_submodule = old_dirty_submodule;
4779         two->dirty_submodule = new_dirty_submodule;
4780
4781         diff_queue(&diff_queued_diff, one, two);
4782         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4783                 DIFF_OPT_SET(options, HAS_CHANGES);
4784 }
4785
4786 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
4787 {
4788         struct diff_filepair *pair;
4789         struct diff_filespec *one, *two;
4790
4791         if (options->prefix &&
4792             strncmp(path, options->prefix, options->prefix_length))
4793                 return NULL;
4794
4795         one = alloc_filespec(path);
4796         two = alloc_filespec(path);
4797         pair = diff_queue(&diff_queued_diff, one, two);
4798         pair->is_unmerged = 1;
4799         return pair;
4800 }
4801
4802 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
4803                 size_t *outsize)
4804 {
4805         struct diff_tempfile *temp;
4806         const char *argv[3];
4807         const char **arg = argv;
4808         struct child_process child;
4809         struct strbuf buf = STRBUF_INIT;
4810         int err = 0;
4811
4812         temp = prepare_temp_file(spec->path, spec);
4813         *arg++ = pgm;
4814         *arg++ = temp->name;
4815         *arg = NULL;
4816
4817         memset(&child, 0, sizeof(child));
4818         child.use_shell = 1;
4819         child.argv = argv;
4820         child.out = -1;
4821         if (start_command(&child)) {
4822                 remove_tempfile();
4823                 return NULL;
4824         }
4825
4826         if (strbuf_read(&buf, child.out, 0) < 0)
4827                 err = error("error reading from textconv command '%s'", pgm);
4828         close(child.out);
4829
4830         if (finish_command(&child) || err) {
4831                 strbuf_release(&buf);
4832                 remove_tempfile();
4833                 return NULL;
4834         }
4835         remove_tempfile();
4836
4837         return strbuf_detach(&buf, outsize);
4838 }
4839
4840 size_t fill_textconv(struct userdiff_driver *driver,
4841                      struct diff_filespec *df,
4842                      char **outbuf)
4843 {
4844         size_t size;
4845
4846         if (!driver || !driver->textconv) {
4847                 if (!DIFF_FILE_VALID(df)) {
4848                         *outbuf = "";
4849                         return 0;
4850                 }
4851                 if (diff_populate_filespec(df, 0))
4852                         die("unable to read files to diff");
4853                 *outbuf = df->data;
4854                 return df->size;
4855         }
4856
4857         if (driver->textconv_cache && df->sha1_valid) {
4858                 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
4859                                           &size);
4860                 if (*outbuf)
4861                         return size;
4862         }
4863
4864         *outbuf = run_textconv(driver->textconv, df, &size);
4865         if (!*outbuf)
4866                 die("unable to read files to diff");
4867
4868         if (driver->textconv_cache && df->sha1_valid) {
4869                 /* ignore errors, as we might be in a readonly repository */
4870                 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
4871                                 size);
4872                 /*
4873                  * we could save up changes and flush them all at the end,
4874                  * but we would need an extra call after all diffing is done.
4875                  * Since generating a cache entry is the slow path anyway,
4876                  * this extra overhead probably isn't a big deal.
4877                  */
4878                 notes_cache_write(driver->textconv_cache);
4879         }
4880
4881         return size;
4882 }