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

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "quote.h"
   6 #include "diff.h"
   7 #include "diffcore.h"
   8 #include "delta.h"
   9 #include "xdiff-interface.h"
  10 #include "color.h"
  11 #include "attr.h"
  12 #include "run-command.h"
  13 #include "utf8.h"
  14 #include "userdiff.h"
  15 #include "sigchain.h"
  16 #include "submodule.h"
  17 #include "ll-merge.h"
  18
  19 #ifdef NO_FAST_WORKING_DIRECTORY
  20 #define FAST_WORKING_DIRECTORY 0
  21 #else
  22 #define FAST_WORKING_DIRECTORY 1
  23 #endif
  24
  25 static int diff_detect_rename_default;
  26 static int diff_rename_limit_default = 400;
  27 static int diff_suppress_blank_empty;
  28 int diff_use_color_default = -1;
  29 static const char *diff_word_regex_cfg;
  30 static const char *external_diff_cmd_cfg;
  31 int diff_auto_refresh_index = 1;
  32 static int diff_mnemonic_prefix;
  33 static int diff_no_prefix;
  34 static int diff_dirstat_permille_default = 30;
  35 static struct diff_options default_diff_options;
  36
  37 static char diff_colors[][COLOR_MAXLEN] = {
  38         GIT_COLOR_RESET,
  39         GIT_COLOR_NORMAL,       /* PLAIN */
  40         GIT_COLOR_BOLD,         /* METAINFO */
  41         GIT_COLOR_CYAN,         /* FRAGINFO */
  42         GIT_COLOR_RED,          /* OLD */
  43         GIT_COLOR_GREEN,        /* NEW */
  44         GIT_COLOR_YELLOW,       /* COMMIT */
  45         GIT_COLOR_BG_RED,       /* WHITESPACE */
  46         GIT_COLOR_NORMAL,       /* FUNCINFO */
  47 };
  48
  49 static int parse_diff_color_slot(const char *var, int ofs)
  50 {
  51         if (!strcasecmp(var+ofs, "plain"))
  52                 return DIFF_PLAIN;
  53         if (!strcasecmp(var+ofs, "meta"))
  54                 return DIFF_METAINFO;
  55         if (!strcasecmp(var+ofs, "frag"))
  56                 return DIFF_FRAGINFO;
  57         if (!strcasecmp(var+ofs, "old"))
  58                 return DIFF_FILE_OLD;
  59         if (!strcasecmp(var+ofs, "new"))
  60                 return DIFF_FILE_NEW;
  61         if (!strcasecmp(var+ofs, "commit"))
  62                 return DIFF_COMMIT;
  63         if (!strcasecmp(var+ofs, "whitespace"))
  64                 return DIFF_WHITESPACE;
  65         if (!strcasecmp(var+ofs, "func"))
  66                 return DIFF_FUNCINFO;
  67         return -1;
  68 }
  69
  70 static int parse_dirstat_params(struct diff_options *options, const char *params,
  71                                 struct strbuf *errmsg)
  72 {
  73         const char *p = params;
  74         int p_len, ret = 0;
  75
  76         while (*p) {
  77                 p_len = strchrnul(p, ',') - p;
  78                 if (!memcmp(p, "changes", p_len)) {
  79                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  80                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
  81                 } else if (!memcmp(p, "lines", p_len)) {
  82                         DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
  83                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
  84                 } else if (!memcmp(p, "files", p_len)) {
  85                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  86                         DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
  87                 } else if (!memcmp(p, "noncumulative", p_len)) {
  88                         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
  89                 } else if (!memcmp(p, "cumulative", p_len)) {
  90                         DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
  91                 } else if (isdigit(*p)) {
  92                         char *end;
  93                         int permille = strtoul(p, &end, 10) * 10;
  94                         if (*end == '.' && isdigit(*++end)) {
  95                                 /* only use first digit */
  96                                 permille += *end - '0';
  97                                 /* .. and ignore any further digits */
  98                                 while (isdigit(*++end))
  99                                         ; /* nothing */
 100                         }
 101                         if (end - p == p_len)
 102                                 options->dirstat_permille = permille;
 103                         else {
 104                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%.*s'\n"),
 105                                             p_len, p);
 106                                 ret++;
 107                         }
 108                 } else {
 109                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%.*s'\n"),
 110                                     p_len, p);
 111                         ret++;
 112                 }
 113
 114                 p += p_len;
 115
 116                 if (*p)
 117                         p++; /* more parameters, swallow separator */
 118         }
 119         return ret;
 120 }
 121
 122 static int git_config_rename(const char *var, const char *value)
 123 {
 124         if (!value)
 125                 return DIFF_DETECT_RENAME;
 126         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 127                 return  DIFF_DETECT_COPY;
 128         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 129 }
 130
 131 /*
 132  * These are to give UI layer defaults.
 133  * The core-level commands such as git-diff-files should
 134  * never be affected by the setting of diff.renames
 135  * the user happens to have in the configuration file.
 136  */
 137 int git_diff_ui_config(const char *var, const char *value, void *cb)
 138 {
 139         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 140                 diff_use_color_default = git_config_colorbool(var, value);
 141                 return 0;
 142         }
 143         if (!strcmp(var, "diff.renames")) {
 144                 diff_detect_rename_default = git_config_rename(var, value);
 145                 return 0;
 146         }
 147         if (!strcmp(var, "diff.autorefreshindex")) {
 148                 diff_auto_refresh_index = git_config_bool(var, value);
 149                 return 0;
 150         }
 151         if (!strcmp(var, "diff.mnemonicprefix")) {
 152                 diff_mnemonic_prefix = git_config_bool(var, value);
 153                 return 0;
 154         }
 155         if (!strcmp(var, "diff.noprefix")) {
 156                 diff_no_prefix = git_config_bool(var, value);
 157                 return 0;
 158         }
 159         if (!strcmp(var, "diff.external"))
 160                 return git_config_string(&external_diff_cmd_cfg, var, value);
 161         if (!strcmp(var, "diff.wordregex"))
 162                 return git_config_string(&diff_word_regex_cfg, var, value);
 163
 164         if (!strcmp(var, "diff.ignoresubmodules"))
 165                 handle_ignore_submodules_arg(&default_diff_options, value);
 166
 167         if (git_color_config(var, value, cb) < 0)
 168                 return -1;
 169
 170         return git_diff_basic_config(var, value, cb);
 171 }
 172
 173 int git_diff_basic_config(const char *var, const char *value, void *cb)
 174 {
 175         if (!strcmp(var, "diff.renamelimit")) {
 176                 diff_rename_limit_default = git_config_int(var, value);
 177                 return 0;
 178         }
 179
 180         switch (userdiff_config(var, value)) {
 181                 case 0: break;
 182                 case -1: return -1;
 183                 default: return 0;
 184         }
 185
 186         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 187                 int slot = parse_diff_color_slot(var, 11);
 188                 if (slot < 0)
 189                         return 0;
 190                 if (!value)
 191                         return config_error_nonbool(var);
 192                 color_parse(value, var, diff_colors[slot]);
 193                 return 0;
 194         }
 195
 196         /* like GNU diff's --suppress-blank-empty option  */
 197         if (!strcmp(var, "diff.suppressblankempty") ||
 198                         /* for backwards compatibility */
 199                         !strcmp(var, "diff.suppress-blank-empty")) {
 200                 diff_suppress_blank_empty = git_config_bool(var, value);
 201                 return 0;
 202         }
 203
 204         if (!strcmp(var, "diff.dirstat")) {
 205                 struct strbuf errmsg = STRBUF_INIT;
 206                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 207                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 208                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 209                                 errmsg.buf);
 210                 strbuf_release(&errmsg);
 211                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 212                 return 0;
 213         }
 214
 215         if (!prefixcmp(var, "submodule."))
 216                 return parse_submodule_config_option(var, value);
 217
 218         return git_default_config(var, value, cb);
 219 }
 220
 221 static char *quote_two(const char *one, const char *two)
 222 {
 223         int need_one = quote_c_style(one, NULL, NULL, 1);
 224         int need_two = quote_c_style(two, NULL, NULL, 1);
 225         struct strbuf res = STRBUF_INIT;
 226
 227         if (need_one + need_two) {
 228                 strbuf_addch(&res, '"');
 229                 quote_c_style(one, &res, NULL, 1);
 230                 quote_c_style(two, &res, NULL, 1);
 231                 strbuf_addch(&res, '"');
 232         } else {
 233                 strbuf_addstr(&res, one);
 234                 strbuf_addstr(&res, two);
 235         }
 236         return strbuf_detach(&res, NULL);
 237 }
 238
 239 static const char *external_diff(void)
 240 {
 241         static const char *external_diff_cmd = NULL;
 242         static int done_preparing = 0;
 243
 244         if (done_preparing)
 245                 return external_diff_cmd;
 246         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 247         if (!external_diff_cmd)
 248                 external_diff_cmd = external_diff_cmd_cfg;
 249         done_preparing = 1;
 250         return external_diff_cmd;
 251 }
 252
 253 static struct diff_tempfile {
 254         const char *name; /* filename external diff should read from */
 255         char hex[41];
 256         char mode[10];
 257         char tmp_path[PATH_MAX];
 258 } diff_temp[2];
 259
 260 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 261
 262 struct emit_callback {
 263         int color_diff;
 264         unsigned ws_rule;
 265         int blank_at_eof_in_preimage;
 266         int blank_at_eof_in_postimage;
 267         int lno_in_preimage;
 268         int lno_in_postimage;
 269         sane_truncate_fn truncate;
 270         const char **label_path;
 271         struct diff_words_data *diff_words;
 272         struct diff_options *opt;
 273         int *found_changesp;
 274         struct strbuf *header;
 275 };
 276
 277 static int count_lines(const char *data, int size)
 278 {
 279         int count, ch, completely_empty = 1, nl_just_seen = 0;
 280         count = 0;
 281         while (0 < size--) {
 282                 ch = *data++;
 283                 if (ch == '\n') {
 284                         count++;
 285                         nl_just_seen = 1;
 286                         completely_empty = 0;
 287                 }
 288                 else {
 289                         nl_just_seen = 0;
 290                         completely_empty = 0;
 291                 }
 292         }
 293         if (completely_empty)
 294                 return 0;
 295         if (!nl_just_seen)
 296                 count++; /* no trailing newline */
 297         return count;
 298 }
 299
 300 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 301 {
 302         if (!DIFF_FILE_VALID(one)) {
 303                 mf->ptr = (char *)""; /* does not matter */
 304                 mf->size = 0;
 305                 return 0;
 306         }
 307         else if (diff_populate_filespec(one, 0))
 308                 return -1;
 309
 310         mf->ptr = one->data;
 311         mf->size = one->size;
 312         return 0;
 313 }
 314
 315 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 316 static unsigned long diff_filespec_size(struct diff_filespec *one)
 317 {
 318         if (!DIFF_FILE_VALID(one))
 319                 return 0;
 320         diff_populate_filespec(one, 1);
 321         return one->size;
 322 }
 323
 324 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 325 {
 326         char *ptr = mf->ptr;
 327         long size = mf->size;
 328         int cnt = 0;
 329
 330         if (!size)
 331                 return cnt;
 332         ptr += size - 1; /* pointing at the very end */
 333         if (*ptr != '\n')
 334                 ; /* incomplete line */
 335         else
 336                 ptr--; /* skip the last LF */
 337         while (mf->ptr < ptr) {
 338                 char *prev_eol;
 339                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 340                         if (*prev_eol == '\n')
 341                                 break;
 342                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 343                         break;
 344                 cnt++;
 345                 ptr = prev_eol - 1;
 346         }
 347         return cnt;
 348 }
 349
 350 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 351                                struct emit_callback *ecbdata)
 352 {
 353         int l1, l2, at;
 354         unsigned ws_rule = ecbdata->ws_rule;
 355         l1 = count_trailing_blank(mf1, ws_rule);
 356         l2 = count_trailing_blank(mf2, ws_rule);
 357         if (l2 <= l1) {
 358                 ecbdata->blank_at_eof_in_preimage = 0;
 359                 ecbdata->blank_at_eof_in_postimage = 0;
 360                 return;
 361         }
 362         at = count_lines(mf1->ptr, mf1->size);
 363         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 364
 365         at = count_lines(mf2->ptr, mf2->size);
 366         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 367 }
 368
 369 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 370                         int first, const char *line, int len)
 371 {
 372         int has_trailing_newline, has_trailing_carriage_return;
 373         int nofirst;
 374         FILE *file = o->file;
 375
 376         if (o->output_prefix) {
 377                 struct strbuf *msg = NULL;
 378                 msg = o->output_prefix(o, o->output_prefix_data);
 379                 assert(msg);
 380                 fwrite(msg->buf, msg->len, 1, file);
 381         }
 382
 383         if (len == 0) {
 384                 has_trailing_newline = (first == '\n');
 385                 has_trailing_carriage_return = (!has_trailing_newline &&
 386                                                 (first == '\r'));
 387                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 388         } else {
 389                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 390                 if (has_trailing_newline)
 391                         len--;
 392                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 393                 if (has_trailing_carriage_return)
 394                         len--;
 395                 nofirst = 0;
 396         }
 397
 398         if (len || !nofirst) {
 399                 fputs(set, file);
 400                 if (!nofirst)
 401                         fputc(first, file);
 402                 fwrite(line, len, 1, file);
 403                 fputs(reset, file);
 404         }
 405         if (has_trailing_carriage_return)
 406                 fputc('\r', file);
 407         if (has_trailing_newline)
 408                 fputc('\n', file);
 409 }
 410
 411 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 412                       const char *line, int len)
 413 {
 414         emit_line_0(o, set, reset, line[0], line+1, len-1);
 415 }
 416
 417 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
 418 {
 419         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
 420               ecbdata->blank_at_eof_in_preimage &&
 421               ecbdata->blank_at_eof_in_postimage &&
 422               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
 423               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
 424                 return 0;
 425         return ws_blank_line(line, len, ecbdata->ws_rule);
 426 }
 427
 428 static void emit_add_line(const char *reset,
 429                           struct emit_callback *ecbdata,
 430                           const char *line, int len)
 431 {
 432         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 433         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 434
 435         if (!*ws)
 436                 emit_line_0(ecbdata->opt, set, reset, '+', line, len);
 437         else if (new_blank_line_at_eof(ecbdata, line, len))
 438                 /* Blank line at EOF - paint '+' as well */
 439                 emit_line_0(ecbdata->opt, ws, reset, '+', line, len);
 440         else {
 441                 /* Emit just the prefix, then the rest. */
 442                 emit_line_0(ecbdata->opt, set, reset, '+', "", 0);
 443                 ws_check_emit(line, len, ecbdata->ws_rule,
 444                               ecbdata->opt->file, set, reset, ws);
 445         }
 446 }
 447
 448 static void emit_hunk_header(struct emit_callback *ecbdata,
 449                              const char *line, int len)
 450 {
 451         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 452         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
 453         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
 454         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 455         static const char atat[2] = { '@', '@' };
 456         const char *cp, *ep;
 457         struct strbuf msgbuf = STRBUF_INIT;
 458         int org_len = len;
 459         int i = 1;
 460
 461         /*
 462          * As a hunk header must begin with "@@ -<old>, +<new> @@",
 463          * it always is at least 10 bytes long.
 464          */
 465         if (len < 10 ||
 466             memcmp(line, atat, 2) ||
 467             !(ep = memmem(line + 2, len - 2, atat, 2))) {
 468                 emit_line(ecbdata->opt, plain, reset, line, len);
 469                 return;
 470         }
 471         ep += 2; /* skip over @@ */
 472
 473         /* The hunk header in fraginfo color */
 474         strbuf_add(&msgbuf, frag, strlen(frag));
 475         strbuf_add(&msgbuf, line, ep - line);
 476         strbuf_add(&msgbuf, reset, strlen(reset));
 477
 478         /*
 479          * trailing "\r\n"
 480          */
 481         for ( ; i < 3; i++)
 482                 if (line[len - i] == '\r' || line[len - i] == '\n')
 483                         len--;
 484
 485         /* blank before the func header */
 486         for (cp = ep; ep - line < len; ep++)
 487                 if (*ep != ' ' && *ep != '\t')
 488                         break;
 489         if (ep != cp) {
 490                 strbuf_add(&msgbuf, plain, strlen(plain));
 491                 strbuf_add(&msgbuf, cp, ep - cp);
 492                 strbuf_add(&msgbuf, reset, strlen(reset));
 493         }
 494
 495         if (ep < line + len) {
 496                 strbuf_add(&msgbuf, func, strlen(func));
 497                 strbuf_add(&msgbuf, ep, line + len - ep);
 498                 strbuf_add(&msgbuf, reset, strlen(reset));
 499         }
 500
 501         strbuf_add(&msgbuf, line + len, org_len - len);
 502         emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
 503         strbuf_release(&msgbuf);
 504 }
 505
 506 static struct diff_tempfile *claim_diff_tempfile(void) {
 507         int i;
 508         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
 509                 if (!diff_temp[i].name)
 510                         return diff_temp + i;
 511         die("BUG: diff is failing to clean up its tempfiles");
 512 }
 513
 514 static int remove_tempfile_installed;
 515
 516 static void remove_tempfile(void)
 517 {
 518         int i;
 519         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
 520                 if (diff_temp[i].name == diff_temp[i].tmp_path)
 521                         unlink_or_warn(diff_temp[i].name);
 522                 diff_temp[i].name = NULL;
 523         }
 524 }
 525
 526 static void remove_tempfile_on_signal(int signo)
 527 {
 528         remove_tempfile();
 529         sigchain_pop(signo);
 530         raise(signo);
 531 }
 532
 533 static void print_line_count(FILE *file, int count)
 534 {
 535         switch (count) {
 536         case 0:
 537                 fprintf(file, "0,0");
 538                 break;
 539         case 1:
 540                 fprintf(file, "1");
 541                 break;
 542         default:
 543                 fprintf(file, "1,%d", count);
 544                 break;
 545         }
 546 }
 547
 548 static void emit_rewrite_lines(struct emit_callback *ecb,
 549                                int prefix, const char *data, int size)
 550 {
 551         const char *endp = NULL;
 552         static const char *nneof = " No newline at end of file\n";
 553         const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
 554         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
 555
 556         while (0 < size) {
 557                 int len;
 558
 559                 endp = memchr(data, '\n', size);
 560                 len = endp ? (endp - data + 1) : size;
 561                 if (prefix != '+') {
 562                         ecb->lno_in_preimage++;
 563                         emit_line_0(ecb->opt, old, reset, '-',
 564                                     data, len);
 565                 } else {
 566                         ecb->lno_in_postimage++;
 567                         emit_add_line(reset, ecb, data, len);
 568                 }
 569                 size -= len;
 570                 data += len;
 571         }
 572         if (!endp) {
 573                 const char *plain = diff_get_color(ecb->color_diff,
 574                                                    DIFF_PLAIN);
 575                 emit_line_0(ecb->opt, plain, reset, '\\',
 576                             nneof, strlen(nneof));
 577         }
 578 }
 579
 580 static void emit_rewrite_diff(const char *name_a,
 581                               const char *name_b,
 582                               struct diff_filespec *one,
 583                               struct diff_filespec *two,
 584                               struct userdiff_driver *textconv_one,
 585                               struct userdiff_driver *textconv_two,
 586                               struct diff_options *o)
 587 {
 588         int lc_a, lc_b;
 589         const char *name_a_tab, *name_b_tab;
 590         const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
 591         const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
 592         const char *reset = diff_get_color(o->use_color, DIFF_RESET);
 593         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 594         const char *a_prefix, *b_prefix;
 595         char *data_one, *data_two;
 596         size_t size_one, size_two;
 597         struct emit_callback ecbdata;
 598         char *line_prefix = "";
 599         struct strbuf *msgbuf;
 600
 601         if (o && o->output_prefix) {
 602                 msgbuf = o->output_prefix(o, o->output_prefix_data);
 603                 line_prefix = msgbuf->buf;
 604         }
 605
 606         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 607                 a_prefix = o->b_prefix;
 608                 b_prefix = o->a_prefix;
 609         } else {
 610                 a_prefix = o->a_prefix;
 611                 b_prefix = o->b_prefix;
 612         }
 613
 614         name_a += (*name_a == '/');
 615         name_b += (*name_b == '/');
 616         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 617         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 618
 619         strbuf_reset(&a_name);
 620         strbuf_reset(&b_name);
 621         quote_two_c_style(&a_name, a_prefix, name_a, 0);
 622         quote_two_c_style(&b_name, b_prefix, name_b, 0);
 623
 624         size_one = fill_textconv(textconv_one, one, &data_one);
 625         size_two = fill_textconv(textconv_two, two, &data_two);
 626
 627         memset(&ecbdata, 0, sizeof(ecbdata));
 628         ecbdata.color_diff = want_color(o->use_color);
 629         ecbdata.found_changesp = &o->found_changes;
 630         ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
 631         ecbdata.opt = o;
 632         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
 633                 mmfile_t mf1, mf2;
 634                 mf1.ptr = (char *)data_one;
 635                 mf2.ptr = (char *)data_two;
 636                 mf1.size = size_one;
 637                 mf2.size = size_two;
 638                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
 639         }
 640         ecbdata.lno_in_preimage = 1;
 641         ecbdata.lno_in_postimage = 1;
 642
 643         lc_a = count_lines(data_one, size_one);
 644         lc_b = count_lines(data_two, size_two);
 645         fprintf(o->file,
 646                 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
 647                 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
 648                 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
 649                 line_prefix, fraginfo);
 650         if (!o->irreversible_delete)
 651                 print_line_count(o->file, lc_a);
 652         else
 653                 fprintf(o->file, "?,?");
 654         fprintf(o->file, " +");
 655         print_line_count(o->file, lc_b);
 656         fprintf(o->file, " @@%s\n", reset);
 657         if (lc_a && !o->irreversible_delete)
 658                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
 659         if (lc_b)
 660                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
 661         if (textconv_one)
 662                 free((char *)data_one);
 663         if (textconv_two)
 664                 free((char *)data_two);
 665 }
 666
 667 struct diff_words_buffer {
 668         mmfile_t text;
 669         long alloc;
 670         struct diff_words_orig {
 671                 const char *begin, *end;
 672         } *orig;
 673         int orig_nr, orig_alloc;
 674 };
 675
 676 static void diff_words_append(char *line, unsigned long len,
 677                 struct diff_words_buffer *buffer)
 678 {
 679         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
 680         line++;
 681         len--;
 682         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 683         buffer->text.size += len;
 684         buffer->text.ptr[buffer->text.size] = '\0';
 685 }
 686
 687 struct diff_words_style_elem {
 688         const char *prefix;
 689         const char *suffix;
 690         const char *color; /* NULL; filled in by the setup code if
 691                             * color is enabled */
 692 };
 693
 694 struct diff_words_style {
 695         enum diff_words_type type;
 696         struct diff_words_style_elem new, old, ctx;
 697         const char *newline;
 698 };
 699
 700 static struct diff_words_style diff_words_styles[] = {
 701         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
 702         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
 703         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
 704 };
 705
 706 struct diff_words_data {
 707         struct diff_words_buffer minus, plus;
 708         const char *current_plus;
 709         int last_minus;
 710         struct diff_options *opt;
 711         regex_t *word_regex;
 712         enum diff_words_type type;
 713         struct diff_words_style *style;
 714 };
 715
 716 static int fn_out_diff_words_write_helper(FILE *fp,
 717                                           struct diff_words_style_elem *st_el,
 718                                           const char *newline,
 719                                           size_t count, const char *buf,
 720                                           const char *line_prefix)
 721 {
 722         int print = 0;
 723
 724         while (count) {
 725                 char *p = memchr(buf, '\n', count);
 726                 if (print)
 727                         fputs(line_prefix, fp);
 728                 if (p != buf) {
 729                         if (st_el->color && fputs(st_el->color, fp) < 0)
 730                                 return -1;
 731                         if (fputs(st_el->prefix, fp) < 0 ||
 732                             fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
 733                             fputs(st_el->suffix, fp) < 0)
 734                                 return -1;
 735                         if (st_el->color && *st_el->color
 736                             && fputs(GIT_COLOR_RESET, fp) < 0)
 737                                 return -1;
 738                 }
 739                 if (!p)
 740                         return 0;
 741                 if (fputs(newline, fp) < 0)
 742                         return -1;
 743                 count -= p + 1 - buf;
 744                 buf = p + 1;
 745                 print = 1;
 746         }
 747         return 0;
 748 }
 749
 750 /*
 751  * '--color-words' algorithm can be described as:
 752  *
 753  *   1. collect a the minus/plus lines of a diff hunk, divided into
 754  *      minus-lines and plus-lines;
 755  *
 756  *   2. break both minus-lines and plus-lines into words and
 757  *      place them into two mmfile_t with one word for each line;
 758  *
 759  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
 760  *
 761  * And for the common parts of the both file, we output the plus side text.
 762  * diff_words->current_plus is used to trace the current position of the plus file
 763  * which printed. diff_words->last_minus is used to trace the last minus word
 764  * printed.
 765  *
 766  * For '--graph' to work with '--color-words', we need to output the graph prefix
 767  * on each line of color words output. Generally, there are two conditions on
 768  * which we should output the prefix.
 769  *
 770  *   1. diff_words->last_minus == 0 &&
 771  *      diff_words->current_plus == diff_words->plus.text.ptr
 772  *
 773  *      that is: the plus text must start as a new line, and if there is no minus
 774  *      word printed, a graph prefix must be printed.
 775  *
 776  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
 777  *      *(diff_words->current_plus - 1) == '\n'
 778  *
 779  *      that is: a graph prefix must be printed following a '\n'
 780  */
 781 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
 782 {
 783         if ((diff_words->last_minus == 0 &&
 784                 diff_words->current_plus == diff_words->plus.text.ptr) ||
 785                 (diff_words->current_plus > diff_words->plus.text.ptr &&
 786                 *(diff_words->current_plus - 1) == '\n')) {
 787                 return 1;
 788         } else {
 789                 return 0;
 790         }
 791 }
 792
 793 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 794 {
 795         struct diff_words_data *diff_words = priv;
 796         struct diff_words_style *style = diff_words->style;
 797         int minus_first, minus_len, plus_first, plus_len;
 798         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 799         struct diff_options *opt = diff_words->opt;
 800         struct strbuf *msgbuf;
 801         char *line_prefix = "";
 802
 803         if (line[0] != '@' || parse_hunk_header(line, len,
 804                         &minus_first, &minus_len, &plus_first, &plus_len))
 805                 return;
 806
 807         assert(opt);
 808         if (opt->output_prefix) {
 809                 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
 810                 line_prefix = msgbuf->buf;
 811         }
 812
 813         /* POSIX requires that first be decremented by one if len == 0... */
 814         if (minus_len) {
 815                 minus_begin = diff_words->minus.orig[minus_first].begin;
 816                 minus_end =
 817                         diff_words->minus.orig[minus_first + minus_len - 1].end;
 818         } else
 819                 minus_begin = minus_end =
 820                         diff_words->minus.orig[minus_first].end;
 821
 822         if (plus_len) {
 823                 plus_begin = diff_words->plus.orig[plus_first].begin;
 824                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
 825         } else
 826                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
 827
 828         if (color_words_output_graph_prefix(diff_words)) {
 829                 fputs(line_prefix, diff_words->opt->file);
 830         }
 831         if (diff_words->current_plus != plus_begin) {
 832                 fn_out_diff_words_write_helper(diff_words->opt->file,
 833                                 &style->ctx, style->newline,
 834                                 plus_begin - diff_words->current_plus,
 835                                 diff_words->current_plus, line_prefix);
 836                 if (*(plus_begin - 1) == '\n')
 837                         fputs(line_prefix, diff_words->opt->file);
 838         }
 839         if (minus_begin != minus_end) {
 840                 fn_out_diff_words_write_helper(diff_words->opt->file,
 841                                 &style->old, style->newline,
 842                                 minus_end - minus_begin, minus_begin,
 843                                 line_prefix);
 844         }
 845         if (plus_begin != plus_end) {
 846                 fn_out_diff_words_write_helper(diff_words->opt->file,
 847                                 &style->new, style->newline,
 848                                 plus_end - plus_begin, plus_begin,
 849                                 line_prefix);
 850         }
 851
 852         diff_words->current_plus = plus_end;
 853         diff_words->last_minus = minus_first;
 854 }
 855
 856 /* This function starts looking at *begin, and returns 0 iff a word was found. */
 857 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
 858                 int *begin, int *end)
 859 {
 860         if (word_regex && *begin < buffer->size) {
 861                 regmatch_t match[1];
 862                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
 863                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
 864                                         '\n', match[0].rm_eo - match[0].rm_so);
 865                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
 866                         *begin += match[0].rm_so;
 867                         return *begin >= *end;
 868                 }
 869                 return -1;
 870         }
 871
 872         /* find the next word */
 873         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
 874                 (*begin)++;
 875         if (*begin >= buffer->size)
 876                 return -1;
 877
 878         /* find the end of the word */
 879         *end = *begin + 1;
 880         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
 881                 (*end)++;
 882
 883         return 0;
 884 }
 885
 886 /*
 887  * This function splits the words in buffer->text, stores the list with
 888  * newline separator into out, and saves the offsets of the original words
 889  * in buffer->orig.
 890  */
 891 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
 892                 regex_t *word_regex)
 893 {
 894         int i, j;
 895         long alloc = 0;
 896
 897         out->size = 0;
 898         out->ptr = NULL;
 899
 900         /* fake an empty "0th" word */
 901         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
 902         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
 903         buffer->orig_nr = 1;
 904
 905         for (i = 0; i < buffer->text.size; i++) {
 906                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
 907                         return;
 908
 909                 /* store original boundaries */
 910                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
 911                                 buffer->orig_alloc);
 912                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
 913                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
 914                 buffer->orig_nr++;
 915
 916                 /* store one word */
 917                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
 918                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
 919                 out->ptr[out->size + j - i] = '\n';
 920                 out->size += j - i + 1;
 921
 922                 i = j - 1;
 923         }
 924 }
 925
 926 /* this executes the word diff on the accumulated buffers */
 927 static void diff_words_show(struct diff_words_data *diff_words)
 928 {
 929         xpparam_t xpp;
 930         xdemitconf_t xecfg;
 931         mmfile_t minus, plus;
 932         struct diff_words_style *style = diff_words->style;
 933
 934         struct diff_options *opt = diff_words->opt;
 935         struct strbuf *msgbuf;
 936         char *line_prefix = "";
 937
 938         assert(opt);
 939         if (opt->output_prefix) {
 940                 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
 941                 line_prefix = msgbuf->buf;
 942         }
 943
 944         /* special case: only removal */
 945         if (!diff_words->plus.text.size) {
 946                 fputs(line_prefix, diff_words->opt->file);
 947                 fn_out_diff_words_write_helper(diff_words->opt->file,
 948                         &style->old, style->newline,
 949                         diff_words->minus.text.size,
 950                         diff_words->minus.text.ptr, line_prefix);
 951                 diff_words->minus.text.size = 0;
 952                 return;
 953         }
 954
 955         diff_words->current_plus = diff_words->plus.text.ptr;
 956         diff_words->last_minus = 0;
 957
 958         memset(&xpp, 0, sizeof(xpp));
 959         memset(&xecfg, 0, sizeof(xecfg));
 960         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
 961         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
 962         xpp.flags = 0;
 963         /* as only the hunk header will be parsed, we need a 0-context */
 964         xecfg.ctxlen = 0;
 965         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
 966                       &xpp, &xecfg);
 967         free(minus.ptr);
 968         free(plus.ptr);
 969         if (diff_words->current_plus != diff_words->plus.text.ptr +
 970                         diff_words->plus.text.size) {
 971                 if (color_words_output_graph_prefix(diff_words))
 972                         fputs(line_prefix, diff_words->opt->file);
 973                 fn_out_diff_words_write_helper(diff_words->opt->file,
 974                         &style->ctx, style->newline,
 975                         diff_words->plus.text.ptr + diff_words->plus.text.size
 976                         - diff_words->current_plus, diff_words->current_plus,
 977                         line_prefix);
 978         }
 979         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 980 }
 981
 982 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
 983 static void diff_words_flush(struct emit_callback *ecbdata)
 984 {
 985         if (ecbdata->diff_words->minus.text.size ||
 986             ecbdata->diff_words->plus.text.size)
 987                 diff_words_show(ecbdata->diff_words);
 988 }
 989
 990 static void free_diff_words_data(struct emit_callback *ecbdata)
 991 {
 992         if (ecbdata->diff_words) {
 993                 diff_words_flush(ecbdata);
 994                 free (ecbdata->diff_words->minus.text.ptr);
 995                 free (ecbdata->diff_words->minus.orig);
 996                 free (ecbdata->diff_words->plus.text.ptr);
 997                 free (ecbdata->diff_words->plus.orig);
 998                 if (ecbdata->diff_words->word_regex) {
 999                         regfree(ecbdata->diff_words->word_regex);
1000                         free(ecbdata->diff_words->word_regex);
1001                 }
1002                 free(ecbdata->diff_words);
1003                 ecbdata->diff_words = NULL;
1004         }
1005 }
1006
1007 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1008 {
1009         if (want_color(diff_use_color))
1010                 return diff_colors[ix];
1011         return "";
1012 }
1013
1014 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1015 {
1016         const char *cp;
1017         unsigned long allot;
1018         size_t l = len;
1019
1020         if (ecb->truncate)
1021                 return ecb->truncate(line, len);
1022         cp = line;
1023         allot = l;
1024         while (0 < l) {
1025                 (void) utf8_width(&cp, &l);
1026                 if (!cp)
1027                         break; /* truncated in the middle? */
1028         }
1029         return allot - l;
1030 }
1031
1032 static void find_lno(const char *line, struct emit_callback *ecbdata)
1033 {
1034         const char *p;
1035         ecbdata->lno_in_preimage = 0;
1036         ecbdata->lno_in_postimage = 0;
1037         p = strchr(line, '-');
1038         if (!p)
1039                 return; /* cannot happen */
1040         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1041         p = strchr(p, '+');
1042         if (!p)
1043                 return; /* cannot happen */
1044         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1045 }
1046
1047 static void fn_out_consume(void *priv, char *line, unsigned long len)
1048 {
1049         struct emit_callback *ecbdata = priv;
1050         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1051         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
1052         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1053         struct diff_options *o = ecbdata->opt;
1054         char *line_prefix = "";
1055         struct strbuf *msgbuf;
1056
1057         if (o && o->output_prefix) {
1058                 msgbuf = o->output_prefix(o, o->output_prefix_data);
1059                 line_prefix = msgbuf->buf;
1060         }
1061
1062         if (ecbdata->header) {
1063                 fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
1064                 strbuf_reset(ecbdata->header);
1065                 ecbdata->header = NULL;
1066         }
1067         *(ecbdata->found_changesp) = 1;
1068
1069         if (ecbdata->label_path[0]) {
1070                 const char *name_a_tab, *name_b_tab;
1071
1072                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1073                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1074
1075                 fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1076                         line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1077                 fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1078                         line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1079                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1080         }
1081
1082         if (diff_suppress_blank_empty
1083             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1084                 line[0] = '\n';
1085                 len = 1;
1086         }
1087
1088         if (line[0] == '@') {
1089                 if (ecbdata->diff_words)
1090                         diff_words_flush(ecbdata);
1091                 len = sane_truncate_line(ecbdata, line, len);
1092                 find_lno(line, ecbdata);
1093                 emit_hunk_header(ecbdata, line, len);
1094                 if (line[len-1] != '\n')
1095                         putc('\n', ecbdata->opt->file);
1096                 return;
1097         }
1098
1099         if (len < 1) {
1100                 emit_line(ecbdata->opt, reset, reset, line, len);
1101                 if (ecbdata->diff_words
1102                     && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1103                         fputs("~\n", ecbdata->opt->file);
1104                 return;
1105         }
1106
1107         if (ecbdata->diff_words) {
1108                 if (line[0] == '-') {
1109                         diff_words_append(line, len,
1110                                           &ecbdata->diff_words->minus);
1111                         return;
1112                 } else if (line[0] == '+') {
1113                         diff_words_append(line, len,
1114                                           &ecbdata->diff_words->plus);
1115                         return;
1116                 }
1117                 diff_words_flush(ecbdata);
1118                 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1119                         emit_line(ecbdata->opt, plain, reset, line, len);
1120                         fputs("~\n", ecbdata->opt->file);
1121                 } else {
1122                         /*
1123                          * Skip the prefix character, if any.  With
1124                          * diff_suppress_blank_empty, there may be
1125                          * none.
1126                          */
1127                         if (line[0] != '\n') {
1128                               line++;
1129                               len--;
1130                         }
1131                         emit_line(ecbdata->opt, plain, reset, line, len);
1132                 }
1133                 return;
1134         }
1135
1136         if (line[0] != '+') {
1137                 const char *color =
1138                         diff_get_color(ecbdata->color_diff,
1139                                        line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
1140                 ecbdata->lno_in_preimage++;
1141                 if (line[0] == ' ')
1142                         ecbdata->lno_in_postimage++;
1143                 emit_line(ecbdata->opt, color, reset, line, len);
1144         } else {
1145                 ecbdata->lno_in_postimage++;
1146                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1147         }
1148 }
1149
1150 static char *pprint_rename(const char *a, const char *b)
1151 {
1152         const char *old = a;
1153         const char *new = b;
1154         struct strbuf name = STRBUF_INIT;
1155         int pfx_length, sfx_length;
1156         int len_a = strlen(a);
1157         int len_b = strlen(b);
1158         int a_midlen, b_midlen;
1159         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1160         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1161
1162         if (qlen_a || qlen_b) {
1163                 quote_c_style(a, &name, NULL, 0);
1164                 strbuf_addstr(&name, " => ");
1165                 quote_c_style(b, &name, NULL, 0);
1166                 return strbuf_detach(&name, NULL);
1167         }
1168
1169         /* Find common prefix */
1170         pfx_length = 0;
1171         while (*old && *new && *old == *new) {
1172                 if (*old == '/')
1173                         pfx_length = old - a + 1;
1174                 old++;
1175                 new++;
1176         }
1177
1178         /* Find common suffix */
1179         old = a + len_a;
1180         new = b + len_b;
1181         sfx_length = 0;
1182         while (a <= old && b <= new && *old == *new) {
1183                 if (*old == '/')
1184                         sfx_length = len_a - (old - a);
1185                 old--;
1186                 new--;
1187         }
1188
1189         /*
1190          * pfx{mid-a => mid-b}sfx
1191          * {pfx-a => pfx-b}sfx
1192          * pfx{sfx-a => sfx-b}
1193          * name-a => name-b
1194          */
1195         a_midlen = len_a - pfx_length - sfx_length;
1196         b_midlen = len_b - pfx_length - sfx_length;
1197         if (a_midlen < 0)
1198                 a_midlen = 0;
1199         if (b_midlen < 0)
1200                 b_midlen = 0;
1201
1202         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1203         if (pfx_length + sfx_length) {
1204                 strbuf_add(&name, a, pfx_length);
1205                 strbuf_addch(&name, '{');
1206         }
1207         strbuf_add(&name, a + pfx_length, a_midlen);
1208         strbuf_addstr(&name, " => ");
1209         strbuf_add(&name, b + pfx_length, b_midlen);
1210         if (pfx_length + sfx_length) {
1211                 strbuf_addch(&name, '}');
1212                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1213         }
1214         return strbuf_detach(&name, NULL);
1215 }
1216
1217 struct diffstat_t {
1218         int nr;
1219         int alloc;
1220         struct diffstat_file {
1221                 char *from_name;
1222                 char *name;
1223                 char *print_name;
1224                 unsigned is_unmerged:1;
1225                 unsigned is_binary:1;
1226                 unsigned is_renamed:1;
1227                 uintmax_t added, deleted;
1228         } **files;
1229 };
1230
1231 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1232                                           const char *name_a,
1233                                           const char *name_b)
1234 {
1235         struct diffstat_file *x;
1236         x = xcalloc(sizeof (*x), 1);
1237         if (diffstat->nr == diffstat->alloc) {
1238                 diffstat->alloc = alloc_nr(diffstat->alloc);
1239                 diffstat->files = xrealloc(diffstat->files,
1240                                 diffstat->alloc * sizeof(x));
1241         }
1242         diffstat->files[diffstat->nr++] = x;
1243         if (name_b) {
1244                 x->from_name = xstrdup(name_a);
1245                 x->name = xstrdup(name_b);
1246                 x->is_renamed = 1;
1247         }
1248         else {
1249                 x->from_name = NULL;
1250                 x->name = xstrdup(name_a);
1251         }
1252         return x;
1253 }
1254
1255 static void diffstat_consume(void *priv, char *line, unsigned long len)
1256 {
1257         struct diffstat_t *diffstat = priv;
1258         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1259
1260         if (line[0] == '+')
1261                 x->added++;
1262         else if (line[0] == '-')
1263                 x->deleted++;
1264 }
1265
1266 const char mime_boundary_leader[] = "------------";
1267
1268 static int scale_linear(int it, int width, int max_change)
1269 {
1270         /*
1271          * make sure that at least one '-' is printed if there were deletions,
1272          * and likewise for '+'.
1273          */
1274         if (max_change < 2)
1275                 return it;
1276         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
1277 }
1278
1279 static void show_name(FILE *file,
1280                       const char *prefix, const char *name, int len)
1281 {
1282         fprintf(file, " %s%-*s |", prefix, len, name);
1283 }
1284
1285 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1286 {
1287         if (cnt <= 0)
1288                 return;
1289         fprintf(file, "%s", set);
1290         while (cnt--)
1291                 putc(ch, file);
1292         fprintf(file, "%s", reset);
1293 }
1294
1295 static void fill_print_name(struct diffstat_file *file)
1296 {
1297         char *pname;
1298
1299         if (file->print_name)
1300                 return;
1301
1302         if (!file->is_renamed) {
1303                 struct strbuf buf = STRBUF_INIT;
1304                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1305                         pname = strbuf_detach(&buf, NULL);
1306                 } else {
1307                         pname = file->name;
1308                         strbuf_release(&buf);
1309                 }
1310         } else {
1311                 pname = pprint_rename(file->from_name, file->name);
1312         }
1313         file->print_name = pname;
1314 }
1315
1316 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1317 {
1318         int i, len, add, del, adds = 0, dels = 0;
1319         uintmax_t max_change = 0, max_len = 0;
1320         int total_files = data->nr;
1321         int width, name_width, count;
1322         const char *reset, *add_c, *del_c;
1323         const char *line_prefix = "";
1324         int extra_shown = 0;
1325         struct strbuf *msg = NULL;
1326
1327         if (data->nr == 0)
1328                 return;
1329
1330         if (options->output_prefix) {
1331                 msg = options->output_prefix(options, options->output_prefix_data);
1332                 line_prefix = msg->buf;
1333         }
1334
1335         width = options->stat_width ? options->stat_width : 80;
1336         name_width = options->stat_name_width ? options->stat_name_width : 50;
1337         count = options->stat_count ? options->stat_count : data->nr;
1338
1339         /* Sanity: give at least 5 columns to the graph,
1340          * but leave at least 10 columns for the name.
1341          */
1342         if (width < 25)
1343                 width = 25;
1344         if (name_width < 10)
1345                 name_width = 10;
1346         else if (width < name_width + 15)
1347                 name_width = width - 15;
1348
1349         /* Find the longest filename and max number of changes */
1350         reset = diff_get_color_opt(options, DIFF_RESET);
1351         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1352         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1353
1354         for (i = 0; (i < count) && (i < data->nr); i++) {
1355                 struct diffstat_file *file = data->files[i];
1356                 uintmax_t change = file->added + file->deleted;
1357                 if (!data->files[i]->is_renamed &&
1358                          (change == 0)) {
1359                         count++; /* not shown == room for one more */
1360                         continue;
1361                 }
1362                 fill_print_name(file);
1363                 len = strlen(file->print_name);
1364                 if (max_len < len)
1365                         max_len = len;
1366
1367                 if (file->is_binary || file->is_unmerged)
1368                         continue;
1369                 if (max_change < change)
1370                         max_change = change;
1371         }
1372         count = i; /* min(count, data->nr) */
1373
1374         /* Compute the width of the graph part;
1375          * 10 is for one blank at the beginning of the line plus
1376          * " | count " between the name and the graph.
1377          *
1378          * From here on, name_width is the width of the name area,
1379          * and width is the width of the graph area.
1380          */
1381         name_width = (name_width < max_len) ? name_width : max_len;
1382         if (width < (name_width + 10) + max_change)
1383                 width = width - (name_width + 10);
1384         else
1385                 width = max_change;
1386
1387         for (i = 0; i < count; i++) {
1388                 const char *prefix = "";
1389                 char *name = data->files[i]->print_name;
1390                 uintmax_t added = data->files[i]->added;
1391                 uintmax_t deleted = data->files[i]->deleted;
1392                 int name_len;
1393
1394                 if (!data->files[i]->is_renamed &&
1395                          (added + deleted == 0)) {
1396                         total_files--;
1397                         continue;
1398                 }
1399                 /*
1400                  * "scale" the filename
1401                  */
1402                 len = name_width;
1403                 name_len = strlen(name);
1404                 if (name_width < name_len) {
1405                         char *slash;
1406                         prefix = "...";
1407                         len -= 3;
1408                         name += name_len - len;
1409                         slash = strchr(name, '/');
1410                         if (slash)
1411                                 name = slash;
1412                 }
1413
1414                 if (data->files[i]->is_binary) {
1415                         fprintf(options->file, "%s", line_prefix);
1416                         show_name(options->file, prefix, name, len);
1417                         fprintf(options->file, "  Bin ");
1418                         fprintf(options->file, "%s%"PRIuMAX"%s",
1419                                 del_c, deleted, reset);
1420                         fprintf(options->file, " -> ");
1421                         fprintf(options->file, "%s%"PRIuMAX"%s",
1422                                 add_c, added, reset);
1423                         fprintf(options->file, " bytes");
1424                         fprintf(options->file, "\n");
1425                         continue;
1426                 }
1427                 else if (data->files[i]->is_unmerged) {
1428                         fprintf(options->file, "%s", line_prefix);
1429                         show_name(options->file, prefix, name, len);
1430                         fprintf(options->file, "  Unmerged\n");
1431                         continue;
1432                 }
1433
1434                 /*
1435                  * scale the add/delete
1436                  */
1437                 add = added;
1438                 del = deleted;
1439                 adds += add;
1440                 dels += del;
1441
1442                 if (width <= max_change) {
1443                         add = scale_linear(add, width, max_change);
1444                         del = scale_linear(del, width, max_change);
1445                 }
1446                 fprintf(options->file, "%s", line_prefix);
1447                 show_name(options->file, prefix, name, len);
1448                 fprintf(options->file, "%5"PRIuMAX"%s", added + deleted,
1449                                 added + deleted ? " " : "");
1450                 show_graph(options->file, '+', add, add_c, reset);
1451                 show_graph(options->file, '-', del, del_c, reset);
1452                 fprintf(options->file, "\n");
1453         }
1454         for (i = count; i < data->nr; i++) {
1455                 uintmax_t added = data->files[i]->added;
1456                 uintmax_t deleted = data->files[i]->deleted;
1457                 if (!data->files[i]->is_renamed &&
1458                          (added + deleted == 0)) {
1459                         total_files--;
1460                         continue;
1461                 }
1462                 adds += added;
1463                 dels += deleted;
1464                 if (!extra_shown)
1465                         fprintf(options->file, "%s ...\n", line_prefix);
1466                 extra_shown = 1;
1467         }
1468         fprintf(options->file, "%s", line_prefix);
1469         fprintf(options->file,
1470                " %d files changed, %d insertions(+), %d deletions(-)\n",
1471                total_files, adds, dels);
1472 }
1473
1474 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1475 {
1476         int i, adds = 0, dels = 0, total_files = data->nr;
1477
1478         if (data->nr == 0)
1479                 return;
1480
1481         for (i = 0; i < data->nr; i++) {
1482                 if (!data->files[i]->is_binary &&
1483                     !data->files[i]->is_unmerged) {
1484                         int added = data->files[i]->added;
1485                         int deleted= data->files[i]->deleted;
1486                         if (!data->files[i]->is_renamed &&
1487                             (added + deleted == 0)) {
1488                                 total_files--;
1489                         } else {
1490                                 adds += added;
1491                                 dels += deleted;
1492                         }
1493                 }
1494         }
1495         if (options->output_prefix) {
1496                 struct strbuf *msg = NULL;
1497                 msg = options->output_prefix(options,
1498                                 options->output_prefix_data);
1499                 fprintf(options->file, "%s", msg->buf);
1500         }
1501         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
1502                total_files, adds, dels);
1503 }
1504
1505 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1506 {
1507         int i;
1508
1509         if (data->nr == 0)
1510                 return;
1511
1512         for (i = 0; i < data->nr; i++) {
1513                 struct diffstat_file *file = data->files[i];
1514
1515                 if (options->output_prefix) {
1516                         struct strbuf *msg = NULL;
1517                         msg = options->output_prefix(options,
1518                                         options->output_prefix_data);
1519                         fprintf(options->file, "%s", msg->buf);
1520                 }
1521
1522                 if (file->is_binary)
1523                         fprintf(options->file, "-\t-\t");
1524                 else
1525                         fprintf(options->file,
1526                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1527                                 file->added, file->deleted);
1528                 if (options->line_termination) {
1529                         fill_print_name(file);
1530                         if (!file->is_renamed)
1531                                 write_name_quoted(file->name, options->file,
1532                                                   options->line_termination);
1533                         else {
1534                                 fputs(file->print_name, options->file);
1535                                 putc(options->line_termination, options->file);
1536                         }
1537                 } else {
1538                         if (file->is_renamed) {
1539                                 putc('\0', options->file);
1540                                 write_name_quoted(file->from_name, options->file, '\0');
1541                         }
1542                         write_name_quoted(file->name, options->file, '\0');
1543                 }
1544         }
1545 }
1546
1547 struct dirstat_file {
1548         const char *name;
1549         unsigned long changed;
1550 };
1551
1552 struct dirstat_dir {
1553         struct dirstat_file *files;
1554         int alloc, nr, permille, cumulative;
1555 };
1556
1557 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1558                 unsigned long changed, const char *base, int baselen)
1559 {
1560         unsigned long this_dir = 0;
1561         unsigned int sources = 0;
1562         const char *line_prefix = "";
1563         struct strbuf *msg = NULL;
1564
1565         if (opt->output_prefix) {
1566                 msg = opt->output_prefix(opt, opt->output_prefix_data);
1567                 line_prefix = msg->buf;
1568         }
1569
1570         while (dir->nr) {
1571                 struct dirstat_file *f = dir->files;
1572                 int namelen = strlen(f->name);
1573                 unsigned long this;
1574                 char *slash;
1575
1576                 if (namelen < baselen)
1577                         break;
1578                 if (memcmp(f->name, base, baselen))
1579                         break;
1580                 slash = strchr(f->name + baselen, '/');
1581                 if (slash) {
1582                         int newbaselen = slash + 1 - f->name;
1583                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1584                         sources++;
1585                 } else {
1586                         this = f->changed;
1587                         dir->files++;
1588                         dir->nr--;
1589                         sources += 2;
1590                 }
1591                 this_dir += this;
1592         }
1593
1594         /*
1595          * We don't report dirstat's for
1596          *  - the top level
1597          *  - or cases where everything came from a single directory
1598          *    under this directory (sources == 1).
1599          */
1600         if (baselen && sources != 1) {
1601                 if (this_dir) {
1602                         int permille = this_dir * 1000 / changed;
1603                         if (permille >= dir->permille) {
1604                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1605                                         permille / 10, permille % 10, baselen, base);
1606                                 if (!dir->cumulative)
1607                                         return 0;
1608                         }
1609                 }
1610         }
1611         return this_dir;
1612 }
1613
1614 static int dirstat_compare(const void *_a, const void *_b)
1615 {
1616         const struct dirstat_file *a = _a;
1617         const struct dirstat_file *b = _b;
1618         return strcmp(a->name, b->name);
1619 }
1620
1621 static void show_dirstat(struct diff_options *options)
1622 {
1623         int i;
1624         unsigned long changed;
1625         struct dirstat_dir dir;
1626         struct diff_queue_struct *q = &diff_queued_diff;
1627
1628         dir.files = NULL;
1629         dir.alloc = 0;
1630         dir.nr = 0;
1631         dir.permille = options->dirstat_permille;
1632         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1633
1634         changed = 0;
1635         for (i = 0; i < q->nr; i++) {
1636                 struct diff_filepair *p = q->queue[i];
1637                 const char *name;
1638                 unsigned long copied, added, damage;
1639                 int content_changed;
1640
1641                 name = p->two->path ? p->two->path : p->one->path;
1642
1643                 if (p->one->sha1_valid && p->two->sha1_valid)
1644                         content_changed = hashcmp(p->one->sha1, p->two->sha1);
1645                 else
1646                         content_changed = 1;
1647
1648                 if (!content_changed) {
1649                         /*
1650                          * The SHA1 has not changed, so pre-/post-content is
1651                          * identical. We can therefore skip looking at the
1652                          * file contents altogether.
1653                          */
1654                         damage = 0;
1655                         goto found_damage;
1656                 }
1657
1658                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1659                         /*
1660                          * In --dirstat-by-file mode, we don't really need to
1661                          * look at the actual file contents at all.
1662                          * The fact that the SHA1 changed is enough for us to
1663                          * add this file to the list of results
1664                          * (with each file contributing equal damage).
1665                          */
1666                         damage = 1;
1667                         goto found_damage;
1668                 }
1669
1670                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1671                         diff_populate_filespec(p->one, 0);
1672                         diff_populate_filespec(p->two, 0);
1673                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1674                                                &copied, &added);
1675                         diff_free_filespec_data(p->one);
1676                         diff_free_filespec_data(p->two);
1677                 } else if (DIFF_FILE_VALID(p->one)) {
1678                         diff_populate_filespec(p->one, 1);
1679                         copied = added = 0;
1680                         diff_free_filespec_data(p->one);
1681                 } else if (DIFF_FILE_VALID(p->two)) {
1682                         diff_populate_filespec(p->two, 1);
1683                         copied = 0;
1684                         added = p->two->size;
1685                         diff_free_filespec_data(p->two);
1686                 } else
1687                         continue;
1688
1689                 /*
1690                  * Original minus copied is the removed material,
1691                  * added is the new material.  They are both damages
1692                  * made to the preimage.
1693                  * If the resulting damage is zero, we know that
1694                  * diffcore_count_changes() considers the two entries to
1695                  * be identical, but since content_changed is true, we
1696                  * know that there must have been _some_ kind of change,
1697                  * so we force all entries to have damage > 0.
1698                  */
1699                 damage = (p->one->size - copied) + added;
1700                 if (!damage)
1701                         damage = 1;
1702
1703 found_damage:
1704                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1705                 dir.files[dir.nr].name = name;
1706                 dir.files[dir.nr].changed = damage;
1707                 changed += damage;
1708                 dir.nr++;
1709         }
1710
1711         /* This can happen even with many files, if everything was renames */
1712         if (!changed)
1713                 return;
1714
1715         /* Show all directories with more than x% of the changes */
1716         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1717         gather_dirstat(options, &dir, changed, "", 0);
1718 }
1719
1720 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
1721 {
1722         int i;
1723         unsigned long changed;
1724         struct dirstat_dir dir;
1725
1726         if (data->nr == 0)
1727                 return;
1728
1729         dir.files = NULL;
1730         dir.alloc = 0;
1731         dir.nr = 0;
1732         dir.permille = options->dirstat_permille;
1733         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1734
1735         changed = 0;
1736         for (i = 0; i < data->nr; i++) {
1737                 struct diffstat_file *file = data->files[i];
1738                 unsigned long damage = file->added + file->deleted;
1739                 if (file->is_binary)
1740                         /*
1741                          * binary files counts bytes, not lines. Must find some
1742                          * way to normalize binary bytes vs. textual lines.
1743                          * The following heuristic assumes that there are 64
1744                          * bytes per "line".
1745                          * This is stupid and ugly, but very cheap...
1746                          */
1747                         damage = (damage + 63) / 64;
1748                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1749                 dir.files[dir.nr].name = file->name;
1750                 dir.files[dir.nr].changed = damage;
1751                 changed += damage;
1752                 dir.nr++;
1753         }
1754
1755         /* This can happen even with many files, if everything was renames */
1756         if (!changed)
1757                 return;
1758
1759         /* Show all directories with more than x% of the changes */
1760         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1761         gather_dirstat(options, &dir, changed, "", 0);
1762 }
1763
1764 static void free_diffstat_info(struct diffstat_t *diffstat)
1765 {
1766         int i;
1767         for (i = 0; i < diffstat->nr; i++) {
1768                 struct diffstat_file *f = diffstat->files[i];
1769                 if (f->name != f->print_name)
1770                         free(f->print_name);
1771                 free(f->name);
1772                 free(f->from_name);
1773                 free(f);
1774         }
1775         free(diffstat->files);
1776 }
1777
1778 struct checkdiff_t {
1779         const char *filename;
1780         int lineno;
1781         int conflict_marker_size;
1782         struct diff_options *o;
1783         unsigned ws_rule;
1784         unsigned status;
1785 };
1786
1787 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
1788 {
1789         char firstchar;
1790         int cnt;
1791
1792         if (len < marker_size + 1)
1793                 return 0;
1794         firstchar = line[0];
1795         switch (firstchar) {
1796         case '=': case '>': case '<': case '|':
1797                 break;
1798         default:
1799                 return 0;
1800         }
1801         for (cnt = 1; cnt < marker_size; cnt++)
1802                 if (line[cnt] != firstchar)
1803                         return 0;
1804         /* line[1] thru line[marker_size-1] are same as firstchar */
1805         if (len < marker_size + 1 || !isspace(line[marker_size]))
1806                 return 0;
1807         return 1;
1808 }
1809
1810 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1811 {
1812         struct checkdiff_t *data = priv;
1813         int marker_size = data->conflict_marker_size;
1814         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
1815         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
1816         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
1817         char *err;
1818         char *line_prefix = "";
1819         struct strbuf *msgbuf;
1820
1821         assert(data->o);
1822         if (data->o->output_prefix) {
1823                 msgbuf = data->o->output_prefix(data->o,
1824                         data->o->output_prefix_data);
1825                 line_prefix = msgbuf->buf;
1826         }
1827
1828         if (line[0] == '+') {
1829                 unsigned bad;
1830                 data->lineno++;
1831                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
1832                         data->status |= 1;
1833                         fprintf(data->o->file,
1834                                 "%s%s:%d: leftover conflict marker\n",
1835                                 line_prefix, data->filename, data->lineno);
1836                 }
1837                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1838                 if (!bad)
1839                         return;
1840                 data->status |= bad;
1841                 err = whitespace_error_string(bad);
1842                 fprintf(data->o->file, "%s%s:%d: %s.\n",
1843                         line_prefix, data->filename, data->lineno, err);
1844                 free(err);
1845                 emit_line(data->o, set, reset, line, 1);
1846                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1847                               data->o->file, set, reset, ws);
1848         } else if (line[0] == ' ') {
1849                 data->lineno++;
1850         } else if (line[0] == '@') {
1851                 char *plus = strchr(line, '+');
1852                 if (plus)
1853                         data->lineno = strtol(plus, NULL, 10) - 1;
1854                 else
1855                         die("invalid diff");
1856         }
1857 }
1858
1859 static unsigned char *deflate_it(char *data,
1860                                  unsigned long size,
1861                                  unsigned long *result_size)
1862 {
1863         int bound;
1864         unsigned char *deflated;
1865         git_zstream stream;
1866
1867         memset(&stream, 0, sizeof(stream));
1868         git_deflate_init(&stream, zlib_compression_level);
1869         bound = git_deflate_bound(&stream, size);
1870         deflated = xmalloc(bound);
1871         stream.next_out = deflated;
1872         stream.avail_out = bound;
1873
1874         stream.next_in = (unsigned char *)data;
1875         stream.avail_in = size;
1876         while (git_deflate(&stream, Z_FINISH) == Z_OK)
1877                 ; /* nothing */
1878         git_deflate_end(&stream);
1879         *result_size = stream.total_out;
1880         return deflated;
1881 }
1882
1883 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
1884 {
1885         void *cp;
1886         void *delta;
1887         void *deflated;
1888         void *data;
1889         unsigned long orig_size;
1890         unsigned long delta_size;
1891         unsigned long deflate_size;
1892         unsigned long data_size;
1893
1894         /* We could do deflated delta, or we could do just deflated two,
1895          * whichever is smaller.
1896          */
1897         delta = NULL;
1898         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1899         if (one->size && two->size) {
1900                 delta = diff_delta(one->ptr, one->size,
1901                                    two->ptr, two->size,
1902                                    &delta_size, deflate_size);
1903                 if (delta) {
1904                         void *to_free = delta;
1905                         orig_size = delta_size;
1906                         delta = deflate_it(delta, delta_size, &delta_size);
1907                         free(to_free);
1908                 }
1909         }
1910
1911         if (delta && delta_size < deflate_size) {
1912                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
1913                 free(deflated);
1914                 data = delta;
1915                 data_size = delta_size;
1916         }
1917         else {
1918                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
1919                 free(delta);
1920                 data = deflated;
1921                 data_size = deflate_size;
1922         }
1923
1924         /* emit data encoded in base85 */
1925         cp = data;
1926         while (data_size) {
1927                 int bytes = (52 < data_size) ? 52 : data_size;
1928                 char line[70];
1929                 data_size -= bytes;
1930                 if (bytes <= 26)
1931                         line[0] = bytes + 'A' - 1;
1932                 else
1933                         line[0] = bytes - 26 + 'a' - 1;
1934                 encode_85(line + 1, cp, bytes);
1935                 cp = (char *) cp + bytes;
1936                 fprintf(file, "%s", prefix);
1937                 fputs(line, file);
1938                 fputc('\n', file);
1939         }
1940         fprintf(file, "%s\n", prefix);
1941         free(data);
1942 }
1943
1944 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
1945 {
1946         fprintf(file, "%sGIT binary patch\n", prefix);
1947         emit_binary_diff_body(file, one, two, prefix);
1948         emit_binary_diff_body(file, two, one, prefix);
1949 }
1950
1951 static void diff_filespec_load_driver(struct diff_filespec *one)
1952 {
1953         /* Use already-loaded driver */
1954         if (one->driver)
1955                 return;
1956
1957         if (S_ISREG(one->mode))
1958                 one->driver = userdiff_find_by_path(one->path);
1959
1960         /* Fallback to default settings */
1961         if (!one->driver)
1962                 one->driver = userdiff_find_by_name("default");
1963 }
1964
1965 int diff_filespec_is_binary(struct diff_filespec *one)
1966 {
1967         if (one->is_binary == -1) {
1968                 diff_filespec_load_driver(one);
1969                 if (one->driver->binary != -1)
1970                         one->is_binary = one->driver->binary;
1971                 else {
1972                         if (!one->data && DIFF_FILE_VALID(one))
1973                                 diff_populate_filespec(one, 0);
1974                         if (one->data)
1975                                 one->is_binary = buffer_is_binary(one->data,
1976                                                 one->size);
1977                         if (one->is_binary == -1)
1978                                 one->is_binary = 0;
1979                 }
1980         }
1981         return one->is_binary;
1982 }
1983
1984 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1985 {
1986         diff_filespec_load_driver(one);
1987         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1988 }
1989
1990 static const char *userdiff_word_regex(struct diff_filespec *one)
1991 {
1992         diff_filespec_load_driver(one);
1993         return one->driver->word_regex;
1994 }
1995
1996 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1997 {
1998         if (!options->a_prefix)
1999                 options->a_prefix = a;
2000         if (!options->b_prefix)
2001                 options->b_prefix = b;
2002 }
2003
2004 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2005 {
2006         if (!DIFF_FILE_VALID(one))
2007                 return NULL;
2008
2009         diff_filespec_load_driver(one);
2010         return userdiff_get_textconv(one->driver);
2011 }
2012
2013 static void builtin_diff(const char *name_a,
2014                          const char *name_b,
2015                          struct diff_filespec *one,
2016                          struct diff_filespec *two,
2017                          const char *xfrm_msg,
2018                          int must_show_header,
2019                          struct diff_options *o,
2020                          int complete_rewrite)
2021 {
2022         mmfile_t mf1, mf2;
2023         const char *lbl[2];
2024         char *a_one, *b_two;
2025         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
2026         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2027         const char *a_prefix, *b_prefix;
2028         struct userdiff_driver *textconv_one = NULL;
2029         struct userdiff_driver *textconv_two = NULL;
2030         struct strbuf header = STRBUF_INIT;
2031         struct strbuf *msgbuf;
2032         char *line_prefix = "";
2033
2034         if (o->output_prefix) {
2035                 msgbuf = o->output_prefix(o, o->output_prefix_data);
2036                 line_prefix = msgbuf->buf;
2037         }
2038
2039         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2040                         (!one->mode || S_ISGITLINK(one->mode)) &&
2041                         (!two->mode || S_ISGITLINK(two->mode))) {
2042                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2043                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2044                 show_submodule_summary(o->file, one ? one->path : two->path,
2045                                 one->sha1, two->sha1, two->dirty_submodule,
2046                                 del, add, reset);
2047                 return;
2048         }
2049
2050         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2051                 textconv_one = get_textconv(one);
2052                 textconv_two = get_textconv(two);
2053         }
2054
2055         diff_set_mnemonic_prefix(o, "a/", "b/");
2056         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2057                 a_prefix = o->b_prefix;
2058                 b_prefix = o->a_prefix;
2059         } else {
2060                 a_prefix = o->a_prefix;
2061                 b_prefix = o->b_prefix;
2062         }
2063
2064         /* Never use a non-valid filename anywhere if at all possible */
2065         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2066         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2067
2068         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2069         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2070         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2071         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2072         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, set, a_one, b_two, reset);
2073         if (lbl[0][0] == '/') {
2074                 /* /dev/null */
2075                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, set, two->mode, reset);
2076                 if (xfrm_msg)
2077                         strbuf_addstr(&header, xfrm_msg);
2078                 must_show_header = 1;
2079         }
2080         else if (lbl[1][0] == '/') {
2081                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, set, one->mode, reset);
2082                 if (xfrm_msg)
2083                         strbuf_addstr(&header, xfrm_msg);
2084                 must_show_header = 1;
2085         }
2086         else {
2087                 if (one->mode != two->mode) {
2088                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, set, one->mode, reset);
2089                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, set, two->mode, reset);
2090                         must_show_header = 1;
2091                 }
2092                 if (xfrm_msg)
2093                         strbuf_addstr(&header, xfrm_msg);
2094
2095                 /*
2096                  * we do not run diff between different kind
2097                  * of objects.
2098                  */
2099                 if ((one->mode ^ two->mode) & S_IFMT)
2100                         goto free_ab_and_return;
2101                 if (complete_rewrite &&
2102                     (textconv_one || !diff_filespec_is_binary(one)) &&
2103                     (textconv_two || !diff_filespec_is_binary(two))) {
2104                         fprintf(o->file, "%s", header.buf);
2105                         strbuf_reset(&header);
2106                         emit_rewrite_diff(name_a, name_b, one, two,
2107                                                 textconv_one, textconv_two, o);
2108                         o->found_changes = 1;
2109                         goto free_ab_and_return;
2110                 }
2111         }
2112
2113         if (o->irreversible_delete && lbl[1][0] == '/') {
2114                 fprintf(o->file, "%s", header.buf);
2115                 strbuf_reset(&header);
2116                 goto free_ab_and_return;
2117         } else if (!DIFF_OPT_TST(o, TEXT) &&
2118             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2119               (!textconv_two && diff_filespec_is_binary(two)) )) {
2120                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2121                         die("unable to read files to diff");
2122                 /* Quite common confusing case */
2123                 if (mf1.size == mf2.size &&
2124                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2125                         if (must_show_header)
2126                                 fprintf(o->file, "%s", header.buf);
2127                         goto free_ab_and_return;
2128                 }
2129                 fprintf(o->file, "%s", header.buf);
2130                 strbuf_reset(&header);
2131                 if (DIFF_OPT_TST(o, BINARY))
2132                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2133                 else
2134                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2135                                 line_prefix, lbl[0], lbl[1]);
2136                 o->found_changes = 1;
2137         } else {
2138                 /* Crazy xdl interfaces.. */
2139                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2140                 xpparam_t xpp;
2141                 xdemitconf_t xecfg;
2142                 struct emit_callback ecbdata;
2143                 const struct userdiff_funcname *pe;
2144
2145                 if (!DIFF_XDL_TST(o, WHITESPACE_FLAGS) || must_show_header) {
2146                         fprintf(o->file, "%s", header.buf);
2147                         strbuf_reset(&header);
2148                 }
2149
2150                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2151                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2152
2153                 pe = diff_funcname_pattern(one);
2154                 if (!pe)
2155                         pe = diff_funcname_pattern(two);
2156
2157                 memset(&xpp, 0, sizeof(xpp));
2158                 memset(&xecfg, 0, sizeof(xecfg));
2159                 memset(&ecbdata, 0, sizeof(ecbdata));
2160                 ecbdata.label_path = lbl;
2161                 ecbdata.color_diff = want_color(o->use_color);
2162                 ecbdata.found_changesp = &o->found_changes;
2163                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
2164                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2165                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2166                 ecbdata.opt = o;
2167                 ecbdata.header = header.len ? &header : NULL;
2168                 xpp.flags = o->xdl_opts;
2169                 xecfg.ctxlen = o->context;
2170                 xecfg.interhunkctxlen = o->interhunkcontext;
2171                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2172                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2173                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2174                 if (pe)
2175                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2176                 if (!diffopts)
2177                         ;
2178                 else if (!prefixcmp(diffopts, "--unified="))
2179                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
2180                 else if (!prefixcmp(diffopts, "-u"))
2181                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
2182                 if (o->word_diff) {
2183                         int i;
2184
2185                         ecbdata.diff_words =
2186                                 xcalloc(1, sizeof(struct diff_words_data));
2187                         ecbdata.diff_words->type = o->word_diff;
2188                         ecbdata.diff_words->opt = o;
2189                         if (!o->word_regex)
2190                                 o->word_regex = userdiff_word_regex(one);
2191                         if (!o->word_regex)
2192                                 o->word_regex = userdiff_word_regex(two);
2193                         if (!o->word_regex)
2194                                 o->word_regex = diff_word_regex_cfg;
2195                         if (o->word_regex) {
2196                                 ecbdata.diff_words->word_regex = (regex_t *)
2197                                         xmalloc(sizeof(regex_t));
2198                                 if (regcomp(ecbdata.diff_words->word_regex,
2199                                                 o->word_regex,
2200                                                 REG_EXTENDED | REG_NEWLINE))
2201                                         die ("Invalid regular expression: %s",
2202                                                         o->word_regex);
2203                         }
2204                         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2205                                 if (o->word_diff == diff_words_styles[i].type) {
2206                                         ecbdata.diff_words->style =
2207                                                 &diff_words_styles[i];
2208                                         break;
2209                                 }
2210                         }
2211                         if (want_color(o->use_color)) {
2212                                 struct diff_words_style *st = ecbdata.diff_words->style;
2213                                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2214                                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2215                                 st->ctx.color = diff_get_color_opt(o, DIFF_PLAIN);
2216                         }
2217                 }
2218                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2219                               &xpp, &xecfg);
2220                 if (o->word_diff)
2221                         free_diff_words_data(&ecbdata);
2222                 if (textconv_one)
2223                         free(mf1.ptr);
2224                 if (textconv_two)
2225                         free(mf2.ptr);
2226                 xdiff_clear_find_func(&xecfg);
2227         }
2228
2229  free_ab_and_return:
2230         strbuf_release(&header);
2231         diff_free_filespec_data(one);
2232         diff_free_filespec_data(two);
2233         free(a_one);
2234         free(b_two);
2235         return;
2236 }
2237
2238 static void builtin_diffstat(const char *name_a, const char *name_b,
2239                              struct diff_filespec *one,
2240                              struct diff_filespec *two,
2241                              struct diffstat_t *diffstat,
2242                              struct diff_options *o,
2243                              int complete_rewrite)
2244 {
2245         mmfile_t mf1, mf2;
2246         struct diffstat_file *data;
2247
2248         data = diffstat_add(diffstat, name_a, name_b);
2249
2250         if (!one || !two) {
2251                 data->is_unmerged = 1;
2252                 return;
2253         }
2254
2255         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2256                 data->is_binary = 1;
2257                 data->added = diff_filespec_size(two);
2258                 data->deleted = diff_filespec_size(one);
2259         }
2260
2261         else if (complete_rewrite) {
2262                 diff_populate_filespec(one, 0);
2263                 diff_populate_filespec(two, 0);
2264                 data->deleted = count_lines(one->data, one->size);
2265                 data->added = count_lines(two->data, two->size);
2266         }
2267
2268         else {
2269                 /* Crazy xdl interfaces.. */
2270                 xpparam_t xpp;
2271                 xdemitconf_t xecfg;
2272
2273                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2274                         die("unable to read files to diff");
2275
2276                 memset(&xpp, 0, sizeof(xpp));
2277                 memset(&xecfg, 0, sizeof(xecfg));
2278                 xpp.flags = o->xdl_opts;
2279                 xecfg.ctxlen = o->context;
2280                 xecfg.interhunkctxlen = o->interhunkcontext;
2281                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2282                               &xpp, &xecfg);
2283         }
2284
2285         diff_free_filespec_data(one);
2286         diff_free_filespec_data(two);
2287 }
2288
2289 static void builtin_checkdiff(const char *name_a, const char *name_b,
2290                               const char *attr_path,
2291                               struct diff_filespec *one,
2292                               struct diff_filespec *two,
2293                               struct diff_options *o)
2294 {
2295         mmfile_t mf1, mf2;
2296         struct checkdiff_t data;
2297
2298         if (!two)
2299                 return;
2300
2301         memset(&data, 0, sizeof(data));
2302         data.filename = name_b ? name_b : name_a;
2303         data.lineno = 0;
2304         data.o = o;
2305         data.ws_rule = whitespace_rule(attr_path);
2306         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2307
2308         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2309                 die("unable to read files to diff");
2310
2311         /*
2312          * All the other codepaths check both sides, but not checking
2313          * the "old" side here is deliberate.  We are checking the newly
2314          * introduced changes, and as long as the "new" side is text, we
2315          * can and should check what it introduces.
2316          */
2317         if (diff_filespec_is_binary(two))
2318                 goto free_and_return;
2319         else {
2320                 /* Crazy xdl interfaces.. */
2321                 xpparam_t xpp;
2322                 xdemitconf_t xecfg;
2323
2324                 memset(&xpp, 0, sizeof(xpp));
2325                 memset(&xecfg, 0, sizeof(xecfg));
2326                 xecfg.ctxlen = 1; /* at least one context line */
2327                 xpp.flags = 0;
2328                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2329                               &xpp, &xecfg);
2330
2331                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2332                         struct emit_callback ecbdata;
2333                         int blank_at_eof;
2334
2335                         ecbdata.ws_rule = data.ws_rule;
2336                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2337                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2338
2339                         if (blank_at_eof) {
2340                                 static char *err;
2341                                 if (!err)
2342                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2343                                 fprintf(o->file, "%s:%d: %s.\n",
2344                                         data.filename, blank_at_eof, err);
2345                                 data.status = 1; /* report errors */
2346                         }
2347                 }
2348         }
2349  free_and_return:
2350         diff_free_filespec_data(one);
2351         diff_free_filespec_data(two);
2352         if (data.status)
2353                 DIFF_OPT_SET(o, CHECK_FAILED);
2354 }
2355
2356 struct diff_filespec *alloc_filespec(const char *path)
2357 {
2358         int namelen = strlen(path);
2359         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2360
2361         memset(spec, 0, sizeof(*spec));
2362         spec->path = (char *)(spec + 1);
2363         memcpy(spec->path, path, namelen+1);
2364         spec->count = 1;
2365         spec->is_binary = -1;
2366         return spec;
2367 }
2368
2369 void free_filespec(struct diff_filespec *spec)
2370 {
2371         if (!--spec->count) {
2372                 diff_free_filespec_data(spec);
2373                 free(spec);
2374         }
2375 }
2376
2377 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2378                    unsigned short mode)
2379 {
2380         if (mode) {
2381                 spec->mode = canon_mode(mode);
2382                 hashcpy(spec->sha1, sha1);
2383                 spec->sha1_valid = !is_null_sha1(sha1);
2384         }
2385 }
2386
2387 /*
2388  * Given a name and sha1 pair, if the index tells us the file in
2389  * the work tree has that object contents, return true, so that
2390  * prepare_temp_file() does not have to inflate and extract.
2391  */
2392 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2393 {
2394         struct cache_entry *ce;
2395         struct stat st;
2396         int pos, len;
2397
2398         /*
2399          * We do not read the cache ourselves here, because the
2400          * benchmark with my previous version that always reads cache
2401          * shows that it makes things worse for diff-tree comparing
2402          * two linux-2.6 kernel trees in an already checked out work
2403          * tree.  This is because most diff-tree comparisons deal with
2404          * only a small number of files, while reading the cache is
2405          * expensive for a large project, and its cost outweighs the
2406          * savings we get by not inflating the object to a temporary
2407          * file.  Practically, this code only helps when we are used
2408          * by diff-cache --cached, which does read the cache before
2409          * calling us.
2410          */
2411         if (!active_cache)
2412                 return 0;
2413
2414         /* We want to avoid the working directory if our caller
2415          * doesn't need the data in a normal file, this system
2416          * is rather slow with its stat/open/mmap/close syscalls,
2417          * and the object is contained in a pack file.  The pack
2418          * is probably already open and will be faster to obtain
2419          * the data through than the working directory.  Loose
2420          * objects however would tend to be slower as they need
2421          * to be individually opened and inflated.
2422          */
2423         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2424                 return 0;
2425
2426         len = strlen(name);
2427         pos = cache_name_pos(name, len);
2428         if (pos < 0)
2429                 return 0;
2430         ce = active_cache[pos];
2431
2432         /*
2433          * This is not the sha1 we are looking for, or
2434          * unreusable because it is not a regular file.
2435          */
2436         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2437                 return 0;
2438
2439         /*
2440          * If ce is marked as "assume unchanged", there is no
2441          * guarantee that work tree matches what we are looking for.
2442          */
2443         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2444                 return 0;
2445
2446         /*
2447          * If ce matches the file in the work tree, we can reuse it.
2448          */
2449         if (ce_uptodate(ce) ||
2450             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2451                 return 1;
2452
2453         return 0;
2454 }
2455
2456 static int populate_from_stdin(struct diff_filespec *s)
2457 {
2458         struct strbuf buf = STRBUF_INIT;
2459         size_t size = 0;
2460
2461         if (strbuf_read(&buf, 0, 0) < 0)
2462                 return error("error while reading from stdin %s",
2463                                      strerror(errno));
2464
2465         s->should_munmap = 0;
2466         s->data = strbuf_detach(&buf, &size);
2467         s->size = size;
2468         s->should_free = 1;
2469         return 0;
2470 }
2471
2472 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2473 {
2474         int len;
2475         char *data = xmalloc(100), *dirty = "";
2476
2477         /* Are we looking at the work tree? */
2478         if (s->dirty_submodule)
2479                 dirty = "-dirty";
2480
2481         len = snprintf(data, 100,
2482                        "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2483         s->data = data;
2484         s->size = len;
2485         s->should_free = 1;
2486         if (size_only) {
2487                 s->data = NULL;
2488                 free(data);
2489         }
2490         return 0;
2491 }
2492
2493 /*
2494  * While doing rename detection and pickaxe operation, we may need to
2495  * grab the data for the blob (or file) for our own in-core comparison.
2496  * diff_filespec has data and size fields for this purpose.
2497  */
2498 int diff_populate_filespec(struct diff_filespec *s, int size_only)
2499 {
2500         int err = 0;
2501         if (!DIFF_FILE_VALID(s))
2502                 die("internal error: asking to populate invalid file.");
2503         if (S_ISDIR(s->mode))
2504                 return -1;
2505
2506         if (s->data)
2507                 return 0;
2508
2509         if (size_only && 0 < s->size)
2510                 return 0;
2511
2512         if (S_ISGITLINK(s->mode))
2513                 return diff_populate_gitlink(s, size_only);
2514
2515         if (!s->sha1_valid ||
2516             reuse_worktree_file(s->path, s->sha1, 0)) {
2517                 struct strbuf buf = STRBUF_INIT;
2518                 struct stat st;
2519                 int fd;
2520
2521                 if (!strcmp(s->path, "-"))
2522                         return populate_from_stdin(s);
2523
2524                 if (lstat(s->path, &st) < 0) {
2525                         if (errno == ENOENT) {
2526                         err_empty:
2527                                 err = -1;
2528                         empty:
2529                                 s->data = (char *)"";
2530                                 s->size = 0;
2531                                 return err;
2532                         }
2533                 }
2534                 s->size = xsize_t(st.st_size);
2535                 if (!s->size)
2536                         goto empty;
2537                 if (S_ISLNK(st.st_mode)) {
2538                         struct strbuf sb = STRBUF_INIT;
2539
2540                         if (strbuf_readlink(&sb, s->path, s->size))
2541                                 goto err_empty;
2542                         s->size = sb.len;
2543                         s->data = strbuf_detach(&sb, NULL);
2544                         s->should_free = 1;
2545                         return 0;
2546                 }
2547                 if (size_only)
2548                         return 0;
2549                 fd = open(s->path, O_RDONLY);
2550                 if (fd < 0)
2551                         goto err_empty;
2552                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2553                 close(fd);
2554                 s->should_munmap = 1;
2555
2556                 /*
2557                  * Convert from working tree format to canonical git format
2558                  */
2559                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2560                         size_t size = 0;
2561                         munmap(s->data, s->size);
2562                         s->should_munmap = 0;
2563                         s->data = strbuf_detach(&buf, &size);
2564                         s->size = size;
2565                         s->should_free = 1;
2566                 }
2567         }
2568         else {
2569                 enum object_type type;
2570                 if (size_only) {
2571                         type = sha1_object_info(s->sha1, &s->size);
2572                         if (type < 0)
2573                                 die("unable to read %s", sha1_to_hex(s->sha1));
2574                 } else {
2575                         s->data = read_sha1_file(s->sha1, &type, &s->size);
2576                         if (!s->data)
2577                                 die("unable to read %s", sha1_to_hex(s->sha1));
2578                         s->should_free = 1;
2579                 }
2580         }
2581         return 0;
2582 }
2583
2584 void diff_free_filespec_blob(struct diff_filespec *s)
2585 {
2586         if (s->should_free)
2587                 free(s->data);
2588         else if (s->should_munmap)
2589                 munmap(s->data, s->size);
2590
2591         if (s->should_free || s->should_munmap) {
2592                 s->should_free = s->should_munmap = 0;
2593                 s->data = NULL;
2594         }
2595 }
2596
2597 void diff_free_filespec_data(struct diff_filespec *s)
2598 {
2599         diff_free_filespec_blob(s);
2600         free(s->cnt_data);
2601         s->cnt_data = NULL;
2602 }
2603
2604 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2605                            void *blob,
2606                            unsigned long size,
2607                            const unsigned char *sha1,
2608                            int mode)
2609 {
2610         int fd;
2611         struct strbuf buf = STRBUF_INIT;
2612         struct strbuf template = STRBUF_INIT;
2613         char *path_dup = xstrdup(path);
2614         const char *base = basename(path_dup);
2615
2616         /* Generate "XXXXXX_basename.ext" */
2617         strbuf_addstr(&template, "XXXXXX_");
2618         strbuf_addstr(&template, base);
2619
2620         fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2621                         strlen(base) + 1);
2622         if (fd < 0)
2623                 die_errno("unable to create temp-file");
2624         if (convert_to_working_tree(path,
2625                         (const char *)blob, (size_t)size, &buf)) {
2626                 blob = buf.buf;
2627                 size = buf.len;
2628         }
2629         if (write_in_full(fd, blob, size) != size)
2630                 die_errno("unable to write temp-file");
2631         close(fd);
2632         temp->name = temp->tmp_path;
2633         strcpy(temp->hex, sha1_to_hex(sha1));
2634         temp->hex[40] = 0;
2635         sprintf(temp->mode, "%06o", mode);
2636         strbuf_release(&buf);
2637         strbuf_release(&template);
2638         free(path_dup);
2639 }
2640
2641 static struct diff_tempfile *prepare_temp_file(const char *name,
2642                 struct diff_filespec *one)
2643 {
2644         struct diff_tempfile *temp = claim_diff_tempfile();
2645
2646         if (!DIFF_FILE_VALID(one)) {
2647         not_a_valid_file:
2648                 /* A '-' entry produces this for file-2, and
2649                  * a '+' entry produces this for file-1.
2650                  */
2651                 temp->name = "/dev/null";
2652                 strcpy(temp->hex, ".");
2653                 strcpy(temp->mode, ".");
2654                 return temp;
2655         }
2656
2657         if (!remove_tempfile_installed) {
2658                 atexit(remove_tempfile);
2659                 sigchain_push_common(remove_tempfile_on_signal);
2660                 remove_tempfile_installed = 1;
2661         }
2662
2663         if (!one->sha1_valid ||
2664             reuse_worktree_file(name, one->sha1, 1)) {
2665                 struct stat st;
2666                 if (lstat(name, &st) < 0) {
2667                         if (errno == ENOENT)
2668                                 goto not_a_valid_file;
2669                         die_errno("stat(%s)", name);
2670                 }
2671                 if (S_ISLNK(st.st_mode)) {
2672                         struct strbuf sb = STRBUF_INIT;
2673                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2674                                 die_errno("readlink(%s)", name);
2675                         prep_temp_blob(name, temp, sb.buf, sb.len,
2676                                        (one->sha1_valid ?
2677                                         one->sha1 : null_sha1),
2678                                        (one->sha1_valid ?
2679                                         one->mode : S_IFLNK));
2680                         strbuf_release(&sb);
2681                 }
2682                 else {
2683                         /* we can borrow from the file in the work tree */
2684                         temp->name = name;
2685                         if (!one->sha1_valid)
2686                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2687                         else
2688                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2689                         /* Even though we may sometimes borrow the
2690                          * contents from the work tree, we always want
2691                          * one->mode.  mode is trustworthy even when
2692                          * !(one->sha1_valid), as long as
2693                          * DIFF_FILE_VALID(one).
2694                          */
2695                         sprintf(temp->mode, "%06o", one->mode);
2696                 }
2697                 return temp;
2698         }
2699         else {
2700                 if (diff_populate_filespec(one, 0))
2701                         die("cannot read data blob for %s", one->path);
2702                 prep_temp_blob(name, temp, one->data, one->size,
2703                                one->sha1, one->mode);
2704         }
2705         return temp;
2706 }
2707
2708 /* An external diff command takes:
2709  *
2710  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2711  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2712  *
2713  */
2714 static void run_external_diff(const char *pgm,
2715                               const char *name,
2716                               const char *other,
2717                               struct diff_filespec *one,
2718                               struct diff_filespec *two,
2719                               const char *xfrm_msg,
2720                               int complete_rewrite)
2721 {
2722         const char *spawn_arg[10];
2723         int retval;
2724         const char **arg = &spawn_arg[0];
2725
2726         if (one && two) {
2727                 struct diff_tempfile *temp_one, *temp_two;
2728                 const char *othername = (other ? other : name);
2729                 temp_one = prepare_temp_file(name, one);
2730                 temp_two = prepare_temp_file(othername, two);
2731                 *arg++ = pgm;
2732                 *arg++ = name;
2733                 *arg++ = temp_one->name;
2734                 *arg++ = temp_one->hex;
2735                 *arg++ = temp_one->mode;
2736                 *arg++ = temp_two->name;
2737                 *arg++ = temp_two->hex;
2738                 *arg++ = temp_two->mode;
2739                 if (other) {
2740                         *arg++ = other;
2741                         *arg++ = xfrm_msg;
2742                 }
2743         } else {
2744                 *arg++ = pgm;
2745                 *arg++ = name;
2746         }
2747         *arg = NULL;
2748         fflush(NULL);
2749         retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2750         remove_tempfile();
2751         if (retval) {
2752                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2753                 exit(1);
2754         }
2755 }
2756
2757 static int similarity_index(struct diff_filepair *p)
2758 {
2759         return p->score * 100 / MAX_SCORE;
2760 }
2761
2762 static void fill_metainfo(struct strbuf *msg,
2763                           const char *name,
2764                           const char *other,
2765                           struct diff_filespec *one,
2766                           struct diff_filespec *two,
2767                           struct diff_options *o,
2768                           struct diff_filepair *p,
2769                           int *must_show_header,
2770                           int use_color)
2771 {
2772         const char *set = diff_get_color(use_color, DIFF_METAINFO);
2773         const char *reset = diff_get_color(use_color, DIFF_RESET);
2774         struct strbuf *msgbuf;
2775         char *line_prefix = "";
2776
2777         *must_show_header = 1;
2778         if (o->output_prefix) {
2779                 msgbuf = o->output_prefix(o, o->output_prefix_data);
2780                 line_prefix = msgbuf->buf;
2781         }
2782         strbuf_init(msg, PATH_MAX * 2 + 300);
2783         switch (p->status) {
2784         case DIFF_STATUS_COPIED:
2785                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2786                             line_prefix, set, similarity_index(p));
2787                 strbuf_addf(msg, "%s\n%s%scopy from ",
2788                             reset,  line_prefix, set);
2789                 quote_c_style(name, msg, NULL, 0);
2790                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
2791                 quote_c_style(other, msg, NULL, 0);
2792                 strbuf_addf(msg, "%s\n", reset);
2793                 break;
2794         case DIFF_STATUS_RENAMED:
2795                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2796                             line_prefix, set, similarity_index(p));
2797                 strbuf_addf(msg, "%s\n%s%srename from ",
2798                             reset, line_prefix, set);
2799                 quote_c_style(name, msg, NULL, 0);
2800                 strbuf_addf(msg, "%s\n%s%srename to ",
2801                             reset, line_prefix, set);
2802                 quote_c_style(other, msg, NULL, 0);
2803                 strbuf_addf(msg, "%s\n", reset);
2804                 break;
2805         case DIFF_STATUS_MODIFIED:
2806                 if (p->score) {
2807                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
2808                                     line_prefix,
2809                                     set, similarity_index(p), reset);
2810                         break;
2811                 }
2812                 /* fallthru */
2813         default:
2814                 *must_show_header = 0;
2815         }
2816         if (one && two && hashcmp(one->sha1, two->sha1)) {
2817                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2818
2819                 if (DIFF_OPT_TST(o, BINARY)) {
2820                         mmfile_t mf;
2821                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2822                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2823                                 abbrev = 40;
2824                 }
2825                 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
2826                             find_unique_abbrev(one->sha1, abbrev));
2827                 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
2828                 if (one->mode == two->mode)
2829                         strbuf_addf(msg, " %06o", one->mode);
2830                 strbuf_addf(msg, "%s\n", reset);
2831         }
2832 }
2833
2834 static void run_diff_cmd(const char *pgm,
2835                          const char *name,
2836                          const char *other,
2837                          const char *attr_path,
2838                          struct diff_filespec *one,
2839                          struct diff_filespec *two,
2840                          struct strbuf *msg,
2841                          struct diff_options *o,
2842                          struct diff_filepair *p)
2843 {
2844         const char *xfrm_msg = NULL;
2845         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2846         int must_show_header = 0;
2847
2848         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2849                 pgm = NULL;
2850         else {
2851                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2852                 if (drv && drv->external)
2853                         pgm = drv->external;
2854         }
2855
2856         if (msg) {
2857                 /*
2858                  * don't use colors when the header is intended for an
2859                  * external diff driver
2860                  */
2861                 fill_metainfo(msg, name, other, one, two, o, p,
2862                               &must_show_header,
2863                               want_color(o->use_color) && !pgm);
2864                 xfrm_msg = msg->len ? msg->buf : NULL;
2865         }
2866
2867         if (pgm) {
2868                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2869                                   complete_rewrite);
2870                 return;
2871         }
2872         if (one && two)
2873                 builtin_diff(name, other ? other : name,
2874                              one, two, xfrm_msg, must_show_header,
2875                              o, complete_rewrite);
2876         else
2877                 fprintf(o->file, "* Unmerged path %s\n", name);
2878 }
2879
2880 static void diff_fill_sha1_info(struct diff_filespec *one)
2881 {
2882         if (DIFF_FILE_VALID(one)) {
2883                 if (!one->sha1_valid) {
2884                         struct stat st;
2885                         if (!strcmp(one->path, "-")) {
2886                                 hashcpy(one->sha1, null_sha1);
2887                                 return;
2888                         }
2889                         if (lstat(one->path, &st) < 0)
2890                                 die_errno("stat '%s'", one->path);
2891                         if (index_path(one->sha1, one->path, &st, 0))
2892                                 die("cannot hash %s", one->path);
2893                 }
2894         }
2895         else
2896                 hashclr(one->sha1);
2897 }
2898
2899 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2900 {
2901         /* Strip the prefix but do not molest /dev/null and absolute paths */
2902         if (*namep && **namep != '/') {
2903                 *namep += prefix_length;
2904                 if (**namep == '/')
2905                         ++*namep;
2906         }
2907         if (*otherp && **otherp != '/') {
2908                 *otherp += prefix_length;
2909                 if (**otherp == '/')
2910                         ++*otherp;
2911         }
2912 }
2913
2914 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2915 {
2916         const char *pgm = external_diff();
2917         struct strbuf msg;
2918         struct diff_filespec *one = p->one;
2919         struct diff_filespec *two = p->two;
2920         const char *name;
2921         const char *other;
2922         const char *attr_path;
2923
2924         name  = p->one->path;
2925         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2926         attr_path = name;
2927         if (o->prefix_length)
2928                 strip_prefix(o->prefix_length, &name, &other);
2929
2930         if (DIFF_PAIR_UNMERGED(p)) {
2931                 run_diff_cmd(pgm, name, NULL, attr_path,
2932                              NULL, NULL, NULL, o, p);
2933                 return;
2934         }
2935
2936         diff_fill_sha1_info(one);
2937         diff_fill_sha1_info(two);
2938
2939         if (!pgm &&
2940             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2941             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2942                 /*
2943                  * a filepair that changes between file and symlink
2944                  * needs to be split into deletion and creation.
2945                  */
2946                 struct diff_filespec *null = alloc_filespec(two->path);
2947                 run_diff_cmd(NULL, name, other, attr_path,
2948                              one, null, &msg, o, p);
2949                 free(null);
2950                 strbuf_release(&msg);
2951
2952                 null = alloc_filespec(one->path);
2953                 run_diff_cmd(NULL, name, other, attr_path,
2954                              null, two, &msg, o, p);
2955                 free(null);
2956         }
2957         else
2958                 run_diff_cmd(pgm, name, other, attr_path,
2959                              one, two, &msg, o, p);
2960
2961         strbuf_release(&msg);
2962 }
2963
2964 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2965                          struct diffstat_t *diffstat)
2966 {
2967         const char *name;
2968         const char *other;
2969         int complete_rewrite = 0;
2970
2971         if (DIFF_PAIR_UNMERGED(p)) {
2972                 /* unmerged */
2973                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2974                 return;
2975         }
2976
2977         name = p->one->path;
2978         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2979
2980         if (o->prefix_length)
2981                 strip_prefix(o->prefix_length, &name, &other);
2982
2983         diff_fill_sha1_info(p->one);
2984         diff_fill_sha1_info(p->two);
2985
2986         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2987                 complete_rewrite = 1;
2988         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2989 }
2990
2991 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2992 {
2993         const char *name;
2994         const char *other;
2995         const char *attr_path;
2996
2997         if (DIFF_PAIR_UNMERGED(p)) {
2998                 /* unmerged */
2999                 return;
3000         }
3001
3002         name = p->one->path;
3003         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3004         attr_path = other ? other : name;
3005
3006         if (o->prefix_length)
3007                 strip_prefix(o->prefix_length, &name, &other);
3008
3009         diff_fill_sha1_info(p->one);
3010         diff_fill_sha1_info(p->two);
3011
3012         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3013 }
3014
3015 void diff_setup(struct diff_options *options)
3016 {
3017         memcpy(options, &default_diff_options, sizeof(*options));
3018
3019         options->file = stdout;
3020
3021         options->line_termination = '\n';
3022         options->break_opt = -1;
3023         options->rename_limit = -1;
3024         options->dirstat_permille = diff_dirstat_permille_default;
3025         options->context = 3;
3026
3027         options->change = diff_change;
3028         options->add_remove = diff_addremove;
3029         options->use_color = diff_use_color_default;
3030         options->detect_rename = diff_detect_rename_default;
3031
3032         if (diff_no_prefix) {
3033                 options->a_prefix = options->b_prefix = "";
3034         } else if (!diff_mnemonic_prefix) {
3035                 options->a_prefix = "a/";
3036                 options->b_prefix = "b/";
3037         }
3038 }
3039
3040 int diff_setup_done(struct diff_options *options)
3041 {
3042         int count = 0;
3043
3044         if (options->output_format & DIFF_FORMAT_NAME)
3045                 count++;
3046         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3047                 count++;
3048         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3049                 count++;
3050         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3051                 count++;
3052         if (count > 1)
3053                 die("--name-only, --name-status, --check and -s are mutually exclusive");
3054
3055         /*
3056          * Most of the time we can say "there are changes"
3057          * only by checking if there are changed paths, but
3058          * --ignore-whitespace* options force us to look
3059          * inside contents.
3060          */
3061
3062         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3063             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3064             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3065                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3066         else
3067                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3068
3069         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3070                 options->detect_rename = DIFF_DETECT_COPY;
3071
3072         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3073                 options->prefix = NULL;
3074         if (options->prefix)
3075                 options->prefix_length = strlen(options->prefix);
3076         else
3077                 options->prefix_length = 0;
3078
3079         if (options->output_format & (DIFF_FORMAT_NAME |
3080                                       DIFF_FORMAT_NAME_STATUS |
3081                                       DIFF_FORMAT_CHECKDIFF |
3082                                       DIFF_FORMAT_NO_OUTPUT))
3083                 options->output_format &= ~(DIFF_FORMAT_RAW |
3084                                             DIFF_FORMAT_NUMSTAT |
3085                                             DIFF_FORMAT_DIFFSTAT |
3086                                             DIFF_FORMAT_SHORTSTAT |
3087                                             DIFF_FORMAT_DIRSTAT |
3088                                             DIFF_FORMAT_SUMMARY |
3089                                             DIFF_FORMAT_PATCH);
3090
3091         /*
3092          * These cases always need recursive; we do not drop caller-supplied
3093          * recursive bits for other formats here.
3094          */
3095         if (options->output_format & (DIFF_FORMAT_PATCH |
3096                                       DIFF_FORMAT_NUMSTAT |
3097                                       DIFF_FORMAT_DIFFSTAT |
3098                                       DIFF_FORMAT_SHORTSTAT |
3099                                       DIFF_FORMAT_DIRSTAT |
3100                                       DIFF_FORMAT_SUMMARY |
3101                                       DIFF_FORMAT_CHECKDIFF))
3102                 DIFF_OPT_SET(options, RECURSIVE);
3103         /*
3104          * Also pickaxe would not work very well if you do not say recursive
3105          */
3106         if (options->pickaxe)
3107                 DIFF_OPT_SET(options, RECURSIVE);
3108         /*
3109          * When patches are generated, submodules diffed against the work tree
3110          * must be checked for dirtiness too so it can be shown in the output
3111          */
3112         if (options->output_format & DIFF_FORMAT_PATCH)
3113                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3114
3115         if (options->detect_rename && options->rename_limit < 0)
3116                 options->rename_limit = diff_rename_limit_default;
3117         if (options->setup & DIFF_SETUP_USE_CACHE) {
3118                 if (!active_cache)
3119                         /* read-cache does not die even when it fails
3120                          * so it is safe for us to do this here.  Also
3121                          * it does not smudge active_cache or active_nr
3122                          * when it fails, so we do not have to worry about
3123                          * cleaning it up ourselves either.
3124                          */
3125                         read_cache();
3126         }
3127         if (options->abbrev <= 0 || 40 < options->abbrev)
3128                 options->abbrev = 40; /* full */
3129
3130         /*
3131          * It does not make sense to show the first hit we happened
3132          * to have found.  It does not make sense not to return with
3133          * exit code in such a case either.
3134          */
3135         if (DIFF_OPT_TST(options, QUICK)) {
3136                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3137                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3138         }
3139
3140         return 0;
3141 }
3142
3143 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3144 {
3145         char c, *eq;
3146         int len;
3147
3148         if (*arg != '-')
3149                 return 0;
3150         c = *++arg;
3151         if (!c)
3152                 return 0;
3153         if (c == arg_short) {
3154                 c = *++arg;
3155                 if (!c)
3156                         return 1;
3157                 if (val && isdigit(c)) {
3158                         char *end;
3159                         int n = strtoul(arg, &end, 10);
3160                         if (*end)
3161                                 return 0;
3162                         *val = n;
3163                         return 1;
3164                 }
3165                 return 0;
3166         }
3167         if (c != '-')
3168                 return 0;
3169         arg++;
3170         eq = strchr(arg, '=');
3171         if (eq)
3172                 len = eq - arg;
3173         else
3174                 len = strlen(arg);
3175         if (!len || strncmp(arg, arg_long, len))
3176                 return 0;
3177         if (eq) {
3178                 int n;
3179                 char *end;
3180                 if (!isdigit(*++eq))
3181                         return 0;
3182                 n = strtoul(eq, &end, 10);
3183                 if (*end)
3184                         return 0;
3185                 *val = n;
3186         }
3187         return 1;
3188 }
3189
3190 static int diff_scoreopt_parse(const char *opt);
3191
3192 static inline int short_opt(char opt, const char **argv,
3193                             const char **optarg)
3194 {
3195         const char *arg = argv[0];
3196         if (arg[0] != '-' || arg[1] != opt)
3197                 return 0;
3198         if (arg[2] != '\0') {
3199                 *optarg = arg + 2;
3200                 return 1;
3201         }
3202         if (!argv[1])
3203                 die("Option '%c' requires a value", opt);
3204         *optarg = argv[1];
3205         return 2;
3206 }
3207
3208 int parse_long_opt(const char *opt, const char **argv,
3209                    const char **optarg)
3210 {
3211         const char *arg = argv[0];
3212         if (arg[0] != '-' || arg[1] != '-')
3213                 return 0;
3214         arg += strlen("--");
3215         if (prefixcmp(arg, opt))
3216                 return 0;
3217         arg += strlen(opt);
3218         if (*arg == '=') { /* sticked form: --option=value */
3219                 *optarg = arg + 1;
3220                 return 1;
3221         }
3222         if (*arg != '\0')
3223                 return 0;
3224         /* separate form: --option value */
3225         if (!argv[1])
3226                 die("Option '--%s' requires a value", opt);
3227         *optarg = argv[1];
3228         return 2;
3229 }
3230
3231 static int stat_opt(struct diff_options *options, const char **av)
3232 {
3233         const char *arg = av[0];
3234         char *end;
3235         int width = options->stat_width;
3236         int name_width = options->stat_name_width;
3237         int count = options->stat_count;
3238         int argcount = 1;
3239
3240         arg += strlen("--stat");
3241         end = (char *)arg;
3242
3243         switch (*arg) {
3244         case '-':
3245                 if (!prefixcmp(arg, "-width")) {
3246                         arg += strlen("-width");
3247                         if (*arg == '=')
3248                                 width = strtoul(arg + 1, &end, 10);
3249                         else if (!*arg && !av[1])
3250                                 die("Option '--stat-width' requires a value");
3251                         else if (!*arg) {
3252                                 width = strtoul(av[1], &end, 10);
3253                                 argcount = 2;
3254                         }
3255                 } else if (!prefixcmp(arg, "-name-width")) {
3256                         arg += strlen("-name-width");
3257                         if (*arg == '=')
3258                                 name_width = strtoul(arg + 1, &end, 10);
3259                         else if (!*arg && !av[1])
3260                                 die("Option '--stat-name-width' requires a value");
3261                         else if (!*arg) {
3262                                 name_width = strtoul(av[1], &end, 10);
3263                                 argcount = 2;
3264                         }
3265                 } else if (!prefixcmp(arg, "-count")) {
3266                         arg += strlen("-count");
3267                         if (*arg == '=')
3268                                 count = strtoul(arg + 1, &end, 10);
3269                         else if (!*arg && !av[1])
3270                                 die("Option '--stat-count' requires a value");
3271                         else if (!*arg) {
3272                                 count = strtoul(av[1], &end, 10);
3273                                 argcount = 2;
3274                         }
3275                 }
3276                 break;
3277         case '=':
3278                 width = strtoul(arg+1, &end, 10);
3279                 if (*end == ',')
3280                         name_width = strtoul(end+1, &end, 10);
3281                 if (*end == ',')
3282                         count = strtoul(end+1, &end, 10);
3283         }
3284
3285         /* Important! This checks all the error cases! */
3286         if (*end)
3287                 return 0;
3288         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3289         options->stat_name_width = name_width;
3290         options->stat_width = width;
3291         options->stat_count = count;
3292         return argcount;
3293 }
3294
3295 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3296 {
3297         struct strbuf errmsg = STRBUF_INIT;
3298         if (parse_dirstat_params(options, params, &errmsg))
3299                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3300                     errmsg.buf);
3301         strbuf_release(&errmsg);
3302         /*
3303          * The caller knows a dirstat-related option is given from the command
3304          * line; allow it to say "return this_function();"
3305          */
3306         options->output_format |= DIFF_FORMAT_DIRSTAT;
3307         return 1;
3308 }
3309
3310 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3311 {
3312         const char *arg = av[0];
3313         const char *optarg;
3314         int argcount;
3315
3316         /* Output format options */
3317         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch"))
3318                 options->output_format |= DIFF_FORMAT_PATCH;
3319         else if (opt_arg(arg, 'U', "unified", &options->context))
3320                 options->output_format |= DIFF_FORMAT_PATCH;
3321         else if (!strcmp(arg, "--raw"))
3322                 options->output_format |= DIFF_FORMAT_RAW;
3323         else if (!strcmp(arg, "--patch-with-raw"))
3324                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
3325         else if (!strcmp(arg, "--numstat"))
3326                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3327         else if (!strcmp(arg, "--shortstat"))
3328                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3329         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3330                 return parse_dirstat_opt(options, "");
3331         else if (!prefixcmp(arg, "-X"))
3332                 return parse_dirstat_opt(options, arg + 2);
3333         else if (!prefixcmp(arg, "--dirstat="))
3334                 return parse_dirstat_opt(options, arg + 10);
3335         else if (!strcmp(arg, "--cumulative"))
3336                 return parse_dirstat_opt(options, "cumulative");
3337         else if (!strcmp(arg, "--dirstat-by-file"))
3338                 return parse_dirstat_opt(options, "files");
3339         else if (!prefixcmp(arg, "--dirstat-by-file=")) {
3340                 parse_dirstat_opt(options, "files");
3341                 return parse_dirstat_opt(options, arg + 18);
3342         }
3343         else if (!strcmp(arg, "--check"))
3344                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3345         else if (!strcmp(arg, "--summary"))
3346                 options->output_format |= DIFF_FORMAT_SUMMARY;
3347         else if (!strcmp(arg, "--patch-with-stat"))
3348                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
3349         else if (!strcmp(arg, "--name-only"))
3350                 options->output_format |= DIFF_FORMAT_NAME;
3351         else if (!strcmp(arg, "--name-status"))
3352                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3353         else if (!strcmp(arg, "-s"))
3354                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3355         else if (!prefixcmp(arg, "--stat"))
3356                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3357                 return stat_opt(options, av);
3358
3359         /* renames options */
3360         else if (!prefixcmp(arg, "-B") || !prefixcmp(arg, "--break-rewrites=") ||
3361                  !strcmp(arg, "--break-rewrites")) {
3362                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3363                         return error("invalid argument to -B: %s", arg+2);
3364         }
3365         else if (!prefixcmp(arg, "-M") || !prefixcmp(arg, "--find-renames=") ||
3366                  !strcmp(arg, "--find-renames")) {
3367                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3368                         return error("invalid argument to -M: %s", arg+2);
3369                 options->detect_rename = DIFF_DETECT_RENAME;
3370         }
3371         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3372                 options->irreversible_delete = 1;
3373         }
3374         else if (!prefixcmp(arg, "-C") || !prefixcmp(arg, "--find-copies=") ||
3375                  !strcmp(arg, "--find-copies")) {
3376                 if (options->detect_rename == DIFF_DETECT_COPY)
3377                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3378                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3379                         return error("invalid argument to -C: %s", arg+2);
3380                 options->detect_rename = DIFF_DETECT_COPY;
3381         }
3382         else if (!strcmp(arg, "--no-renames"))
3383                 options->detect_rename = 0;
3384         else if (!strcmp(arg, "--relative"))
3385                 DIFF_OPT_SET(options, RELATIVE_NAME);
3386         else if (!prefixcmp(arg, "--relative=")) {
3387                 DIFF_OPT_SET(options, RELATIVE_NAME);
3388                 options->prefix = arg + 11;
3389         }
3390
3391         /* xdiff options */
3392         else if (!strcmp(arg, "--minimal"))
3393                 DIFF_XDL_SET(options, NEED_MINIMAL);
3394         else if (!strcmp(arg, "--no-minimal"))
3395                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3396         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3397                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3398         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3399                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3400         else if (!strcmp(arg, "--ignore-space-at-eol"))
3401                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3402         else if (!strcmp(arg, "--patience"))
3403                 DIFF_XDL_SET(options, PATIENCE_DIFF);
3404         else if (!strcmp(arg, "--histogram"))
3405                 DIFF_XDL_SET(options, HISTOGRAM_DIFF);
3406
3407         /* flags options */
3408         else if (!strcmp(arg, "--binary")) {
3409                 options->output_format |= DIFF_FORMAT_PATCH;
3410                 DIFF_OPT_SET(options, BINARY);
3411         }
3412         else if (!strcmp(arg, "--full-index"))
3413                 DIFF_OPT_SET(options, FULL_INDEX);
3414         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3415                 DIFF_OPT_SET(options, TEXT);
3416         else if (!strcmp(arg, "-R"))
3417                 DIFF_OPT_SET(options, REVERSE_DIFF);
3418         else if (!strcmp(arg, "--find-copies-harder"))
3419                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3420         else if (!strcmp(arg, "--follow"))
3421                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3422         else if (!strcmp(arg, "--color"))
3423                 options->use_color = 1;
3424         else if (!prefixcmp(arg, "--color=")) {
3425                 int value = git_config_colorbool(NULL, arg+8);
3426                 if (value < 0)
3427                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3428                 options->use_color = value;
3429         }
3430         else if (!strcmp(arg, "--no-color"))
3431                 options->use_color = 0;
3432         else if (!strcmp(arg, "--color-words")) {
3433                 options->use_color = 1;
3434                 options->word_diff = DIFF_WORDS_COLOR;
3435         }
3436         else if (!prefixcmp(arg, "--color-words=")) {
3437                 options->use_color = 1;
3438                 options->word_diff = DIFF_WORDS_COLOR;
3439                 options->word_regex = arg + 14;
3440         }
3441         else if (!strcmp(arg, "--word-diff")) {
3442                 if (options->word_diff == DIFF_WORDS_NONE)
3443                         options->word_diff = DIFF_WORDS_PLAIN;
3444         }
3445         else if (!prefixcmp(arg, "--word-diff=")) {
3446                 const char *type = arg + 12;
3447                 if (!strcmp(type, "plain"))
3448                         options->word_diff = DIFF_WORDS_PLAIN;
3449                 else if (!strcmp(type, "color")) {
3450                         options->use_color = 1;
3451                         options->word_diff = DIFF_WORDS_COLOR;
3452                 }
3453                 else if (!strcmp(type, "porcelain"))
3454                         options->word_diff = DIFF_WORDS_PORCELAIN;
3455                 else if (!strcmp(type, "none"))
3456                         options->word_diff = DIFF_WORDS_NONE;
3457                 else
3458                         die("bad --word-diff argument: %s", type);
3459         }
3460         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3461                 if (options->word_diff == DIFF_WORDS_NONE)
3462                         options->word_diff = DIFF_WORDS_PLAIN;
3463                 options->word_regex = optarg;
3464                 return argcount;
3465         }
3466         else if (!strcmp(arg, "--exit-code"))
3467                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3468         else if (!strcmp(arg, "--quiet"))
3469                 DIFF_OPT_SET(options, QUICK);
3470         else if (!strcmp(arg, "--ext-diff"))
3471                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3472         else if (!strcmp(arg, "--no-ext-diff"))
3473                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3474         else if (!strcmp(arg, "--textconv"))
3475                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3476         else if (!strcmp(arg, "--no-textconv"))
3477                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3478         else if (!strcmp(arg, "--ignore-submodules")) {
3479                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3480                 handle_ignore_submodules_arg(options, "all");
3481         } else if (!prefixcmp(arg, "--ignore-submodules=")) {
3482                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3483                 handle_ignore_submodules_arg(options, arg + 20);
3484         } else if (!strcmp(arg, "--submodule"))
3485                 DIFF_OPT_SET(options, SUBMODULE_LOG);
3486         else if (!prefixcmp(arg, "--submodule=")) {
3487                 if (!strcmp(arg + 12, "log"))
3488                         DIFF_OPT_SET(options, SUBMODULE_LOG);
3489         }
3490
3491         /* misc options */
3492         else if (!strcmp(arg, "-z"))
3493                 options->line_termination = 0;
3494         else if ((argcount = short_opt('l', av, &optarg))) {
3495                 options->rename_limit = strtoul(optarg, NULL, 10);
3496                 return argcount;
3497         }
3498         else if ((argcount = short_opt('S', av, &optarg))) {
3499                 options->pickaxe = optarg;
3500                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3501                 return argcount;
3502         } else if ((argcount = short_opt('G', av, &optarg))) {
3503                 options->pickaxe = optarg;
3504                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3505                 return argcount;
3506         }
3507         else if (!strcmp(arg, "--pickaxe-all"))
3508                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3509         else if (!strcmp(arg, "--pickaxe-regex"))
3510                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3511         else if ((argcount = short_opt('O', av, &optarg))) {
3512                 options->orderfile = optarg;
3513                 return argcount;
3514         }
3515         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3516                 options->filter = optarg;
3517                 return argcount;
3518         }
3519         else if (!strcmp(arg, "--abbrev"))
3520                 options->abbrev = DEFAULT_ABBREV;
3521         else if (!prefixcmp(arg, "--abbrev=")) {
3522                 options->abbrev = strtoul(arg + 9, NULL, 10);
3523                 if (options->abbrev < MINIMUM_ABBREV)
3524                         options->abbrev = MINIMUM_ABBREV;
3525                 else if (40 < options->abbrev)
3526                         options->abbrev = 40;
3527         }
3528         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3529                 options->a_prefix = optarg;
3530                 return argcount;
3531         }
3532         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3533                 options->b_prefix = optarg;
3534                 return argcount;
3535         }
3536         else if (!strcmp(arg, "--no-prefix"))
3537                 options->a_prefix = options->b_prefix = "";
3538         else if (opt_arg(arg, '\0', "inter-hunk-context",
3539                          &options->interhunkcontext))
3540                 ;
3541         else if (!strcmp(arg, "-W"))
3542                 DIFF_OPT_SET(options, FUNCCONTEXT);
3543         else if (!strcmp(arg, "--function-context"))
3544                 DIFF_OPT_SET(options, FUNCCONTEXT);
3545         else if (!strcmp(arg, "--no-function-context"))
3546                 DIFF_OPT_CLR(options, FUNCCONTEXT);
3547         else if ((argcount = parse_long_opt("output", av, &optarg))) {
3548                 options->file = fopen(optarg, "w");
3549                 if (!options->file)
3550                         die_errno("Could not open '%s'", optarg);
3551                 options->close_file = 1;
3552                 return argcount;
3553         } else
3554                 return 0;
3555         return 1;
3556 }
3557
3558 int parse_rename_score(const char **cp_p)
3559 {
3560         unsigned long num, scale;
3561         int ch, dot;
3562         const char *cp = *cp_p;
3563
3564         num = 0;
3565         scale = 1;
3566         dot = 0;
3567         for (;;) {
3568                 ch = *cp;
3569                 if ( !dot && ch == '.' ) {
3570                         scale = 1;
3571                         dot = 1;
3572                 } else if ( ch == '%' ) {
3573                         scale = dot ? scale*100 : 100;
3574                         cp++;   /* % is always at the end */
3575                         break;
3576                 } else if ( ch >= '0' && ch <= '9' ) {
3577                         if ( scale < 100000 ) {
3578                                 scale *= 10;
3579                                 num = (num*10) + (ch-'0');
3580                         }
3581                 } else {
3582                         break;
3583                 }
3584                 cp++;
3585         }
3586         *cp_p = cp;
3587
3588         /* user says num divided by scale and we say internally that
3589          * is MAX_SCORE * num / scale.
3590          */
3591         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
3592 }
3593
3594 static int diff_scoreopt_parse(const char *opt)
3595 {
3596         int opt1, opt2, cmd;
3597
3598         if (*opt++ != '-')
3599                 return -1;
3600         cmd = *opt++;
3601         if (cmd == '-') {
3602                 /* convert the long-form arguments into short-form versions */
3603                 if (!prefixcmp(opt, "break-rewrites")) {
3604                         opt += strlen("break-rewrites");
3605                         if (*opt == 0 || *opt++ == '=')
3606                                 cmd = 'B';
3607                 } else if (!prefixcmp(opt, "find-copies")) {
3608                         opt += strlen("find-copies");
3609                         if (*opt == 0 || *opt++ == '=')
3610                                 cmd = 'C';
3611                 } else if (!prefixcmp(opt, "find-renames")) {
3612                         opt += strlen("find-renames");
3613                         if (*opt == 0 || *opt++ == '=')
3614                                 cmd = 'M';
3615                 }
3616         }
3617         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
3618                 return -1; /* that is not a -M, -C nor -B option */
3619
3620         opt1 = parse_rename_score(&opt);
3621         if (cmd != 'B')
3622                 opt2 = 0;
3623         else {
3624                 if (*opt == 0)
3625                         opt2 = 0;
3626                 else if (*opt != '/')
3627                         return -1; /* we expect -B80/99 or -B80 */
3628                 else {
3629                         opt++;
3630                         opt2 = parse_rename_score(&opt);
3631                 }
3632         }
3633         if (*opt != 0)
3634                 return -1;
3635         return opt1 | (opt2 << 16);
3636 }
3637
3638 struct diff_queue_struct diff_queued_diff;
3639
3640 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
3641 {
3642         if (queue->alloc <= queue->nr) {
3643                 queue->alloc = alloc_nr(queue->alloc);
3644                 queue->queue = xrealloc(queue->queue,
3645                                         sizeof(dp) * queue->alloc);
3646         }
3647         queue->queue[queue->nr++] = dp;
3648 }
3649
3650 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
3651                                  struct diff_filespec *one,
3652                                  struct diff_filespec *two)
3653 {
3654         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3655         dp->one = one;
3656         dp->two = two;
3657         if (queue)
3658                 diff_q(queue, dp);
3659         return dp;
3660 }
3661
3662 void diff_free_filepair(struct diff_filepair *p)
3663 {
3664         free_filespec(p->one);
3665         free_filespec(p->two);
3666         free(p);
3667 }
3668
3669 /* This is different from find_unique_abbrev() in that
3670  * it stuffs the result with dots for alignment.
3671  */
3672 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3673 {
3674         int abblen;
3675         const char *abbrev;
3676         if (len == 40)
3677                 return sha1_to_hex(sha1);
3678
3679         abbrev = find_unique_abbrev(sha1, len);
3680         abblen = strlen(abbrev);
3681         if (abblen < 37) {
3682                 static char hex[41];
3683                 if (len < abblen && abblen <= len + 2)
3684                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3685                 else
3686                         sprintf(hex, "%s...", abbrev);
3687                 return hex;
3688         }
3689         return sha1_to_hex(sha1);
3690 }
3691
3692 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3693 {
3694         int line_termination = opt->line_termination;
3695         int inter_name_termination = line_termination ? '\t' : '\0';
3696         if (opt->output_prefix) {
3697                 struct strbuf *msg = NULL;
3698                 msg = opt->output_prefix(opt, opt->output_prefix_data);
3699                 fprintf(opt->file, "%s", msg->buf);
3700         }
3701
3702         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3703                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3704                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
3705                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3706         }
3707         if (p->score) {
3708                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3709                         inter_name_termination);
3710         } else {
3711                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
3712         }
3713
3714         if (p->status == DIFF_STATUS_COPIED ||
3715             p->status == DIFF_STATUS_RENAMED) {
3716                 const char *name_a, *name_b;
3717                 name_a = p->one->path;
3718                 name_b = p->two->path;
3719                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3720                 write_name_quoted(name_a, opt->file, inter_name_termination);
3721                 write_name_quoted(name_b, opt->file, line_termination);
3722         } else {
3723                 const char *name_a, *name_b;
3724                 name_a = p->one->mode ? p->one->path : p->two->path;
3725                 name_b = NULL;
3726                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3727                 write_name_quoted(name_a, opt->file, line_termination);
3728         }
3729 }
3730
3731 int diff_unmodified_pair(struct diff_filepair *p)
3732 {
3733         /* This function is written stricter than necessary to support
3734          * the currently implemented transformers, but the idea is to
3735          * let transformers to produce diff_filepairs any way they want,
3736          * and filter and clean them up here before producing the output.
3737          */
3738         struct diff_filespec *one = p->one, *two = p->two;
3739
3740         if (DIFF_PAIR_UNMERGED(p))
3741                 return 0; /* unmerged is interesting */
3742
3743         /* deletion, addition, mode or type change
3744          * and rename are all interesting.
3745          */
3746         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
3747             DIFF_PAIR_MODE_CHANGED(p) ||
3748             strcmp(one->path, two->path))
3749                 return 0;
3750
3751         /* both are valid and point at the same path.  that is, we are
3752          * dealing with a change.
3753          */
3754         if (one->sha1_valid && two->sha1_valid &&
3755             !hashcmp(one->sha1, two->sha1) &&
3756             !one->dirty_submodule && !two->dirty_submodule)
3757                 return 1; /* no change */
3758         if (!one->sha1_valid && !two->sha1_valid)
3759                 return 1; /* both look at the same file on the filesystem. */
3760         return 0;
3761 }
3762
3763 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
3764 {
3765         if (diff_unmodified_pair(p))
3766                 return;
3767
3768         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3769             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3770                 return; /* no tree diffs in patch format */
3771
3772         run_diff(p, o);
3773 }
3774
3775 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
3776                             struct diffstat_t *diffstat)
3777 {
3778         if (diff_unmodified_pair(p))
3779                 return;
3780
3781         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3782             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3783                 return; /* no useful stat for tree diffs */
3784
3785         run_diffstat(p, o, diffstat);
3786 }
3787
3788 static void diff_flush_checkdiff(struct diff_filepair *p,
3789                 struct diff_options *o)
3790 {
3791         if (diff_unmodified_pair(p))
3792                 return;
3793
3794         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3795             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3796                 return; /* nothing to check in tree diffs */
3797
3798         run_checkdiff(p, o);
3799 }
3800
3801 int diff_queue_is_empty(void)
3802 {
3803         struct diff_queue_struct *q = &diff_queued_diff;
3804         int i;
3805         for (i = 0; i < q->nr; i++)
3806                 if (!diff_unmodified_pair(q->queue[i]))
3807                         return 0;
3808         return 1;
3809 }
3810
3811 #if DIFF_DEBUG
3812 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
3813 {
3814         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
3815                 x, one ? one : "",
3816                 s->path,
3817                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
3818                 s->mode,
3819                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
3820         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
3821                 x, one ? one : "",
3822                 s->size, s->xfrm_flags);
3823 }
3824
3825 void diff_debug_filepair(const struct diff_filepair *p, int i)
3826 {
3827         diff_debug_filespec(p->one, i, "one");
3828         diff_debug_filespec(p->two, i, "two");
3829         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
3830                 p->score, p->status ? p->status : '?',
3831                 p->one->rename_used, p->broken_pair);
3832 }
3833
3834 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
3835 {
3836         int i;
3837         if (msg)
3838                 fprintf(stderr, "%s\n", msg);
3839         fprintf(stderr, "q->nr = %d\n", q->nr);
3840         for (i = 0; i < q->nr; i++) {
3841                 struct diff_filepair *p = q->queue[i];
3842                 diff_debug_filepair(p, i);
3843         }
3844 }
3845 #endif
3846
3847 static void diff_resolve_rename_copy(void)
3848 {
3849         int i;
3850         struct diff_filepair *p;
3851         struct diff_queue_struct *q = &diff_queued_diff;
3852
3853         diff_debug_queue("resolve-rename-copy", q);
3854
3855         for (i = 0; i < q->nr; i++) {
3856                 p = q->queue[i];
3857                 p->status = 0; /* undecided */
3858                 if (DIFF_PAIR_UNMERGED(p))
3859                         p->status = DIFF_STATUS_UNMERGED;
3860                 else if (!DIFF_FILE_VALID(p->one))
3861                         p->status = DIFF_STATUS_ADDED;
3862                 else if (!DIFF_FILE_VALID(p->two))
3863                         p->status = DIFF_STATUS_DELETED;
3864                 else if (DIFF_PAIR_TYPE_CHANGED(p))
3865                         p->status = DIFF_STATUS_TYPE_CHANGED;
3866
3867                 /* from this point on, we are dealing with a pair
3868                  * whose both sides are valid and of the same type, i.e.
3869                  * either in-place edit or rename/copy edit.
3870                  */
3871                 else if (DIFF_PAIR_RENAME(p)) {
3872                         /*
3873                          * A rename might have re-connected a broken
3874                          * pair up, causing the pathnames to be the
3875                          * same again. If so, that's not a rename at
3876                          * all, just a modification..
3877                          *
3878                          * Otherwise, see if this source was used for
3879                          * multiple renames, in which case we decrement
3880                          * the count, and call it a copy.
3881                          */
3882                         if (!strcmp(p->one->path, p->two->path))
3883                                 p->status = DIFF_STATUS_MODIFIED;
3884                         else if (--p->one->rename_used > 0)
3885                                 p->status = DIFF_STATUS_COPIED;
3886                         else
3887                                 p->status = DIFF_STATUS_RENAMED;
3888                 }
3889                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
3890                          p->one->mode != p->two->mode ||
3891                          p->one->dirty_submodule ||
3892                          p->two->dirty_submodule ||
3893                          is_null_sha1(p->one->sha1))
3894                         p->status = DIFF_STATUS_MODIFIED;
3895                 else {
3896                         /* This is a "no-change" entry and should not
3897                          * happen anymore, but prepare for broken callers.
3898                          */
3899                         error("feeding unmodified %s to diffcore",
3900                               p->one->path);
3901                         p->status = DIFF_STATUS_UNKNOWN;
3902                 }
3903         }
3904         diff_debug_queue("resolve-rename-copy done", q);
3905 }
3906
3907 static int check_pair_status(struct diff_filepair *p)
3908 {
3909         switch (p->status) {
3910         case DIFF_STATUS_UNKNOWN:
3911                 return 0;
3912         case 0:
3913                 die("internal error in diff-resolve-rename-copy");
3914         default:
3915                 return 1;
3916         }
3917 }
3918
3919 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3920 {
3921         int fmt = opt->output_format;
3922
3923         if (fmt & DIFF_FORMAT_CHECKDIFF)
3924                 diff_flush_checkdiff(p, opt);
3925         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3926                 diff_flush_raw(p, opt);
3927         else if (fmt & DIFF_FORMAT_NAME) {
3928                 const char *name_a, *name_b;
3929                 name_a = p->two->path;
3930                 name_b = NULL;
3931                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3932                 write_name_quoted(name_a, opt->file, opt->line_termination);
3933         }
3934 }
3935
3936 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3937 {
3938         if (fs->mode)
3939                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3940         else
3941                 fprintf(file, " %s ", newdelete);
3942         write_name_quoted(fs->path, file, '\n');
3943 }
3944
3945
3946 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
3947                 const char *line_prefix)
3948 {
3949         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3950                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
3951                         p->two->mode, show_name ? ' ' : '\n');
3952                 if (show_name) {
3953                         write_name_quoted(p->two->path, file, '\n');
3954                 }
3955         }
3956 }
3957
3958 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
3959                         const char *line_prefix)
3960 {
3961         char *names = pprint_rename(p->one->path, p->two->path);
3962
3963         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3964         free(names);
3965         show_mode_change(file, p, 0, line_prefix);
3966 }
3967
3968 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
3969 {
3970         FILE *file = opt->file;
3971         char *line_prefix = "";
3972
3973         if (opt->output_prefix) {
3974                 struct strbuf *buf = opt->output_prefix(opt, opt->output_prefix_data);
3975                 line_prefix = buf->buf;
3976         }
3977
3978         switch(p->status) {
3979         case DIFF_STATUS_DELETED:
3980                 fputs(line_prefix, file);
3981                 show_file_mode_name(file, "delete", p->one);
3982                 break;
3983         case DIFF_STATUS_ADDED:
3984                 fputs(line_prefix, file);
3985                 show_file_mode_name(file, "create", p->two);
3986                 break;
3987         case DIFF_STATUS_COPIED:
3988                 fputs(line_prefix, file);
3989                 show_rename_copy(file, "copy", p, line_prefix);
3990                 break;
3991         case DIFF_STATUS_RENAMED:
3992                 fputs(line_prefix, file);
3993                 show_rename_copy(file, "rename", p, line_prefix);
3994                 break;
3995         default:
3996                 if (p->score) {
3997                         fprintf(file, "%s rewrite ", line_prefix);
3998                         write_name_quoted(p->two->path, file, ' ');
3999                         fprintf(file, "(%d%%)\n", similarity_index(p));
4000                 }
4001                 show_mode_change(file, p, !p->score, line_prefix);
4002                 break;
4003         }
4004 }
4005
4006 struct patch_id_t {
4007         git_SHA_CTX *ctx;
4008         int patchlen;
4009 };
4010
4011 static int remove_space(char *line, int len)
4012 {
4013         int i;
4014         char *dst = line;
4015         unsigned char c;
4016
4017         for (i = 0; i < len; i++)
4018                 if (!isspace((c = line[i])))
4019                         *dst++ = c;
4020
4021         return dst - line;
4022 }
4023
4024 static void patch_id_consume(void *priv, char *line, unsigned long len)
4025 {
4026         struct patch_id_t *data = priv;
4027         int new_len;
4028
4029         /* Ignore line numbers when computing the SHA1 of the patch */
4030         if (!prefixcmp(line, "@@ -"))
4031                 return;
4032
4033         new_len = remove_space(line, len);
4034
4035         git_SHA1_Update(data->ctx, line, new_len);
4036         data->patchlen += new_len;
4037 }
4038
4039 /* returns 0 upon success, and writes result into sha1 */
4040 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
4041 {
4042         struct diff_queue_struct *q = &diff_queued_diff;
4043         int i;
4044         git_SHA_CTX ctx;
4045         struct patch_id_t data;
4046         char buffer[PATH_MAX * 4 + 20];
4047
4048         git_SHA1_Init(&ctx);
4049         memset(&data, 0, sizeof(struct patch_id_t));
4050         data.ctx = &ctx;
4051
4052         for (i = 0; i < q->nr; i++) {
4053                 xpparam_t xpp;
4054                 xdemitconf_t xecfg;
4055                 mmfile_t mf1, mf2;
4056                 struct diff_filepair *p = q->queue[i];
4057                 int len1, len2;
4058
4059                 memset(&xpp, 0, sizeof(xpp));
4060                 memset(&xecfg, 0, sizeof(xecfg));
4061                 if (p->status == 0)
4062                         return error("internal diff status error");
4063                 if (p->status == DIFF_STATUS_UNKNOWN)
4064                         continue;
4065                 if (diff_unmodified_pair(p))
4066                         continue;
4067                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4068                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4069                         continue;
4070                 if (DIFF_PAIR_UNMERGED(p))
4071                         continue;
4072
4073                 diff_fill_sha1_info(p->one);
4074                 diff_fill_sha1_info(p->two);
4075                 if (fill_mmfile(&mf1, p->one) < 0 ||
4076                                 fill_mmfile(&mf2, p->two) < 0)
4077                         return error("unable to read files to diff");
4078
4079                 len1 = remove_space(p->one->path, strlen(p->one->path));
4080                 len2 = remove_space(p->two->path, strlen(p->two->path));
4081                 if (p->one->mode == 0)
4082                         len1 = snprintf(buffer, sizeof(buffer),
4083                                         "diff--gita/%.*sb/%.*s"
4084                                         "newfilemode%06o"
4085                                         "---/dev/null"
4086                                         "+++b/%.*s",
4087                                         len1, p->one->path,
4088                                         len2, p->two->path,
4089                                         p->two->mode,
4090                                         len2, p->two->path);
4091                 else if (p->two->mode == 0)
4092                         len1 = snprintf(buffer, sizeof(buffer),
4093                                         "diff--gita/%.*sb/%.*s"
4094                                         "deletedfilemode%06o"
4095                                         "---a/%.*s"
4096                                         "+++/dev/null",
4097                                         len1, p->one->path,
4098                                         len2, p->two->path,
4099                                         p->one->mode,
4100                                         len1, p->one->path);
4101                 else
4102                         len1 = snprintf(buffer, sizeof(buffer),
4103                                         "diff--gita/%.*sb/%.*s"
4104                                         "---a/%.*s"
4105                                         "+++b/%.*s",
4106                                         len1, p->one->path,
4107                                         len2, p->two->path,
4108                                         len1, p->one->path,
4109                                         len2, p->two->path);
4110                 git_SHA1_Update(&ctx, buffer, len1);
4111
4112                 if (diff_filespec_is_binary(p->one) ||
4113                     diff_filespec_is_binary(p->two)) {
4114                         git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
4115                         git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
4116                         continue;
4117                 }
4118
4119                 xpp.flags = 0;
4120                 xecfg.ctxlen = 3;
4121                 xecfg.flags = 0;
4122                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4123                               &xpp, &xecfg);
4124         }
4125
4126         git_SHA1_Final(sha1, &ctx);
4127         return 0;
4128 }
4129
4130 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
4131 {
4132         struct diff_queue_struct *q = &diff_queued_diff;
4133         int i;
4134         int result = diff_get_patch_id(options, sha1);
4135
4136         for (i = 0; i < q->nr; i++)
4137                 diff_free_filepair(q->queue[i]);
4138
4139         free(q->queue);
4140         DIFF_QUEUE_CLEAR(q);
4141
4142         return result;
4143 }
4144
4145 static int is_summary_empty(const struct diff_queue_struct *q)
4146 {
4147         int i;
4148
4149         for (i = 0; i < q->nr; i++) {
4150                 const struct diff_filepair *p = q->queue[i];
4151
4152                 switch (p->status) {
4153                 case DIFF_STATUS_DELETED:
4154                 case DIFF_STATUS_ADDED:
4155                 case DIFF_STATUS_COPIED:
4156                 case DIFF_STATUS_RENAMED:
4157                         return 0;
4158                 default:
4159                         if (p->score)
4160                                 return 0;
4161                         if (p->one->mode && p->two->mode &&
4162                             p->one->mode != p->two->mode)
4163                                 return 0;
4164                         break;
4165                 }
4166         }
4167         return 1;
4168 }
4169
4170 static const char rename_limit_warning[] =
4171 "inexact rename detection was skipped due to too many files.";
4172
4173 static const char degrade_cc_to_c_warning[] =
4174 "only found copies from modified paths due to too many files.";
4175
4176 static const char rename_limit_advice[] =
4177 "you may want to set your %s variable to at least "
4178 "%d and retry the command.";
4179
4180 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4181 {
4182         if (degraded_cc)
4183                 warning(degrade_cc_to_c_warning);
4184         else if (needed)
4185                 warning(rename_limit_warning);
4186         else
4187                 return;
4188         if (0 < needed && needed < 32767)
4189                 warning(rename_limit_advice, varname, needed);
4190 }
4191
4192 void diff_flush(struct diff_options *options)
4193 {
4194         struct diff_queue_struct *q = &diff_queued_diff;
4195         int i, output_format = options->output_format;
4196         int separator = 0;
4197         int dirstat_by_line = 0;
4198
4199         /*
4200          * Order: raw, stat, summary, patch
4201          * or:    name/name-status/checkdiff (other bits clear)
4202          */
4203         if (!q->nr)
4204                 goto free_queue;
4205
4206         if (output_format & (DIFF_FORMAT_RAW |
4207                              DIFF_FORMAT_NAME |
4208                              DIFF_FORMAT_NAME_STATUS |
4209                              DIFF_FORMAT_CHECKDIFF)) {
4210                 for (i = 0; i < q->nr; i++) {
4211                         struct diff_filepair *p = q->queue[i];
4212                         if (check_pair_status(p))
4213                                 flush_one_pair(p, options);
4214                 }
4215                 separator++;
4216         }
4217
4218         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4219                 dirstat_by_line = 1;
4220
4221         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4222             dirstat_by_line) {
4223                 struct diffstat_t diffstat;
4224
4225                 memset(&diffstat, 0, sizeof(struct diffstat_t));
4226                 for (i = 0; i < q->nr; i++) {
4227                         struct diff_filepair *p = q->queue[i];
4228                         if (check_pair_status(p))
4229                                 diff_flush_stat(p, options, &diffstat);
4230                 }
4231                 if (output_format & DIFF_FORMAT_NUMSTAT)
4232                         show_numstat(&diffstat, options);
4233                 if (output_format & DIFF_FORMAT_DIFFSTAT)
4234                         show_stats(&diffstat, options);
4235                 if (output_format & DIFF_FORMAT_SHORTSTAT)
4236                         show_shortstats(&diffstat, options);
4237                 if (output_format & DIFF_FORMAT_DIRSTAT)
4238                         show_dirstat_by_line(&diffstat, options);
4239                 free_diffstat_info(&diffstat);
4240                 separator++;
4241         }
4242         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4243                 show_dirstat(options);
4244
4245         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4246                 for (i = 0; i < q->nr; i++) {
4247                         diff_summary(options, q->queue[i]);
4248                 }
4249                 separator++;
4250         }
4251
4252         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4253             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4254             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4255                 /*
4256                  * run diff_flush_patch for the exit status. setting
4257                  * options->file to /dev/null should be safe, becaue we
4258                  * aren't supposed to produce any output anyway.
4259                  */
4260                 if (options->close_file)
4261                         fclose(options->file);
4262                 options->file = fopen("/dev/null", "w");
4263                 if (!options->file)
4264                         die_errno("Could not open /dev/null");
4265                 options->close_file = 1;
4266                 for (i = 0; i < q->nr; i++) {
4267                         struct diff_filepair *p = q->queue[i];
4268                         if (check_pair_status(p))
4269                                 diff_flush_patch(p, options);
4270                         if (options->found_changes)
4271                                 break;
4272                 }
4273         }
4274
4275         if (output_format & DIFF_FORMAT_PATCH) {
4276                 if (separator) {
4277                         putc(options->line_termination, options->file);
4278                         if (options->stat_sep) {
4279                                 /* attach patch instead of inline */
4280                                 fputs(options->stat_sep, options->file);
4281                         }
4282                 }
4283
4284                 for (i = 0; i < q->nr; i++) {
4285                         struct diff_filepair *p = q->queue[i];
4286                         if (check_pair_status(p))
4287                                 diff_flush_patch(p, options);
4288                 }
4289         }
4290
4291         if (output_format & DIFF_FORMAT_CALLBACK)
4292                 options->format_callback(q, options, options->format_callback_data);
4293
4294         for (i = 0; i < q->nr; i++)
4295                 diff_free_filepair(q->queue[i]);
4296 free_queue:
4297         free(q->queue);
4298         DIFF_QUEUE_CLEAR(q);
4299         if (options->close_file)
4300                 fclose(options->file);
4301
4302         /*
4303          * Report the content-level differences with HAS_CHANGES;
4304          * diff_addremove/diff_change does not set the bit when
4305          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4306          */
4307         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4308                 if (options->found_changes)
4309                         DIFF_OPT_SET(options, HAS_CHANGES);
4310                 else
4311                         DIFF_OPT_CLR(options, HAS_CHANGES);
4312         }
4313 }
4314
4315 static void diffcore_apply_filter(const char *filter)
4316 {
4317         int i;
4318         struct diff_queue_struct *q = &diff_queued_diff;
4319         struct diff_queue_struct outq;
4320         DIFF_QUEUE_CLEAR(&outq);
4321
4322         if (!filter)
4323                 return;
4324
4325         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
4326                 int found;
4327                 for (i = found = 0; !found && i < q->nr; i++) {
4328                         struct diff_filepair *p = q->queue[i];
4329                         if (((p->status == DIFF_STATUS_MODIFIED) &&
4330                              ((p->score &&
4331                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
4332                               (!p->score &&
4333                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
4334                             ((p->status != DIFF_STATUS_MODIFIED) &&
4335                              strchr(filter, p->status)))
4336                                 found++;
4337                 }
4338                 if (found)
4339                         return;
4340
4341                 /* otherwise we will clear the whole queue
4342                  * by copying the empty outq at the end of this
4343                  * function, but first clear the current entries
4344                  * in the queue.
4345                  */
4346                 for (i = 0; i < q->nr; i++)
4347                         diff_free_filepair(q->queue[i]);
4348         }
4349         else {
4350                 /* Only the matching ones */
4351                 for (i = 0; i < q->nr; i++) {
4352                         struct diff_filepair *p = q->queue[i];
4353
4354                         if (((p->status == DIFF_STATUS_MODIFIED) &&
4355                              ((p->score &&
4356                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
4357                               (!p->score &&
4358                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
4359                             ((p->status != DIFF_STATUS_MODIFIED) &&
4360                              strchr(filter, p->status)))
4361                                 diff_q(&outq, p);
4362                         else
4363                                 diff_free_filepair(p);
4364                 }
4365         }
4366         free(q->queue);
4367         *q = outq;
4368 }
4369
4370 /* Check whether two filespecs with the same mode and size are identical */
4371 static int diff_filespec_is_identical(struct diff_filespec *one,
4372                                       struct diff_filespec *two)
4373 {
4374         if (S_ISGITLINK(one->mode))
4375                 return 0;
4376         if (diff_populate_filespec(one, 0))
4377                 return 0;
4378         if (diff_populate_filespec(two, 0))
4379                 return 0;
4380         return !memcmp(one->data, two->data, one->size);
4381 }
4382
4383 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4384 {
4385         int i;
4386         struct diff_queue_struct *q = &diff_queued_diff;
4387         struct diff_queue_struct outq;
4388         DIFF_QUEUE_CLEAR(&outq);
4389
4390         for (i = 0; i < q->nr; i++) {
4391                 struct diff_filepair *p = q->queue[i];
4392
4393                 /*
4394                  * 1. Entries that come from stat info dirtiness
4395                  *    always have both sides (iow, not create/delete),
4396                  *    one side of the object name is unknown, with
4397                  *    the same mode and size.  Keep the ones that
4398                  *    do not match these criteria.  They have real
4399                  *    differences.
4400                  *
4401                  * 2. At this point, the file is known to be modified,
4402                  *    with the same mode and size, and the object
4403                  *    name of one side is unknown.  Need to inspect
4404                  *    the identical contents.
4405                  */
4406                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
4407                     !DIFF_FILE_VALID(p->two) ||
4408                     (p->one->sha1_valid && p->two->sha1_valid) ||
4409                     (p->one->mode != p->two->mode) ||
4410                     diff_populate_filespec(p->one, 1) ||
4411                     diff_populate_filespec(p->two, 1) ||
4412                     (p->one->size != p->two->size) ||
4413                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4414                         diff_q(&outq, p);
4415                 else {
4416                         /*
4417                          * The caller can subtract 1 from skip_stat_unmatch
4418                          * to determine how many paths were dirty only
4419                          * due to stat info mismatch.
4420                          */
4421                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4422                                 diffopt->skip_stat_unmatch++;
4423                         diff_free_filepair(p);
4424                 }
4425         }
4426         free(q->queue);
4427         *q = outq;
4428 }
4429
4430 static int diffnamecmp(const void *a_, const void *b_)
4431 {
4432         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4433         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4434         const char *name_a, *name_b;
4435
4436         name_a = a->one ? a->one->path : a->two->path;
4437         name_b = b->one ? b->one->path : b->two->path;
4438         return strcmp(name_a, name_b);
4439 }
4440
4441 void diffcore_fix_diff_index(struct diff_options *options)
4442 {
4443         struct diff_queue_struct *q = &diff_queued_diff;
4444         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4445 }
4446
4447 void diffcore_std(struct diff_options *options)
4448 {
4449         if (options->skip_stat_unmatch)
4450                 diffcore_skip_stat_unmatch(options);
4451         if (!options->found_follow) {
4452                 /* See try_to_follow_renames() in tree-diff.c */
4453                 if (options->break_opt != -1)
4454                         diffcore_break(options->break_opt);
4455                 if (options->detect_rename)
4456                         diffcore_rename(options);
4457                 if (options->break_opt != -1)
4458                         diffcore_merge_broken();
4459         }
4460         if (options->pickaxe)
4461                 diffcore_pickaxe(options);
4462         if (options->orderfile)
4463                 diffcore_order(options->orderfile);
4464         if (!options->found_follow)
4465                 /* See try_to_follow_renames() in tree-diff.c */
4466                 diff_resolve_rename_copy();
4467         diffcore_apply_filter(options->filter);
4468
4469         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4470                 DIFF_OPT_SET(options, HAS_CHANGES);
4471         else
4472                 DIFF_OPT_CLR(options, HAS_CHANGES);
4473
4474         options->found_follow = 0;
4475 }
4476
4477 int diff_result_code(struct diff_options *opt, int status)
4478 {
4479         int result = 0;
4480
4481         diff_warn_rename_limit("diff.renamelimit",
4482                                opt->needed_rename_limit,
4483                                opt->degraded_cc_to_c);
4484         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4485             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4486                 return status;
4487         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4488             DIFF_OPT_TST(opt, HAS_CHANGES))
4489                 result |= 01;
4490         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4491             DIFF_OPT_TST(opt, CHECK_FAILED))
4492                 result |= 02;
4493         return result;
4494 }
4495
4496 int diff_can_quit_early(struct diff_options *opt)
4497 {
4498         return (DIFF_OPT_TST(opt, QUICK) &&
4499                 !opt->filter &&
4500                 DIFF_OPT_TST(opt, HAS_CHANGES));
4501 }
4502
4503 /*
4504  * Shall changes to this submodule be ignored?
4505  *
4506  * Submodule changes can be configured to be ignored separately for each path,
4507  * but that configuration can be overridden from the command line.
4508  */
4509 static int is_submodule_ignored(const char *path, struct diff_options *options)
4510 {
4511         int ignored = 0;
4512         unsigned orig_flags = options->flags;
4513         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4514                 set_diffopt_flags_from_submodule_config(options, path);
4515         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4516                 ignored = 1;
4517         options->flags = orig_flags;
4518         return ignored;
4519 }
4520
4521 void diff_addremove(struct diff_options *options,
4522                     int addremove, unsigned mode,
4523                     const unsigned char *sha1,
4524                     const char *concatpath, unsigned dirty_submodule)
4525 {
4526         struct diff_filespec *one, *two;
4527
4528         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4529                 return;
4530
4531         /* This may look odd, but it is a preparation for
4532          * feeding "there are unchanged files which should
4533          * not produce diffs, but when you are doing copy
4534          * detection you would need them, so here they are"
4535          * entries to the diff-core.  They will be prefixed
4536          * with something like '=' or '*' (I haven't decided
4537          * which but should not make any difference).
4538          * Feeding the same new and old to diff_change()
4539          * also has the same effect.
4540          * Before the final output happens, they are pruned after
4541          * merged into rename/copy pairs as appropriate.
4542          */
4543         if (DIFF_OPT_TST(options, REVERSE_DIFF))
4544                 addremove = (addremove == '+' ? '-' :
4545                              addremove == '-' ? '+' : addremove);
4546
4547         if (options->prefix &&
4548             strncmp(concatpath, options->prefix, options->prefix_length))
4549                 return;
4550
4551         one = alloc_filespec(concatpath);
4552         two = alloc_filespec(concatpath);
4553
4554         if (addremove != '+')
4555                 fill_filespec(one, sha1, mode);
4556         if (addremove != '-') {
4557                 fill_filespec(two, sha1, mode);
4558                 two->dirty_submodule = dirty_submodule;
4559         }
4560
4561         diff_queue(&diff_queued_diff, one, two);
4562         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4563                 DIFF_OPT_SET(options, HAS_CHANGES);
4564 }
4565
4566 void diff_change(struct diff_options *options,
4567                  unsigned old_mode, unsigned new_mode,
4568                  const unsigned char *old_sha1,
4569                  const unsigned char *new_sha1,
4570                  const char *concatpath,
4571                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4572 {
4573         struct diff_filespec *one, *two;
4574
4575         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4576             is_submodule_ignored(concatpath, options))
4577                 return;
4578
4579         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4580                 unsigned tmp;
4581                 const unsigned char *tmp_c;
4582                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4583                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4584                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
4585                         new_dirty_submodule = tmp;
4586         }
4587
4588         if (options->prefix &&
4589             strncmp(concatpath, options->prefix, options->prefix_length))
4590                 return;
4591
4592         one = alloc_filespec(concatpath);
4593         two = alloc_filespec(concatpath);
4594         fill_filespec(one, old_sha1, old_mode);
4595         fill_filespec(two, new_sha1, new_mode);
4596         one->dirty_submodule = old_dirty_submodule;
4597         two->dirty_submodule = new_dirty_submodule;
4598
4599         diff_queue(&diff_queued_diff, one, two);
4600         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4601                 DIFF_OPT_SET(options, HAS_CHANGES);
4602 }
4603
4604 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
4605 {
4606         struct diff_filepair *pair;
4607         struct diff_filespec *one, *two;
4608
4609         if (options->prefix &&
4610             strncmp(path, options->prefix, options->prefix_length))
4611                 return NULL;
4612
4613         one = alloc_filespec(path);
4614         two = alloc_filespec(path);
4615         pair = diff_queue(&diff_queued_diff, one, two);
4616         pair->is_unmerged = 1;
4617         return pair;
4618 }
4619
4620 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
4621                 size_t *outsize)
4622 {
4623         struct diff_tempfile *temp;
4624         const char *argv[3];
4625         const char **arg = argv;
4626         struct child_process child;
4627         struct strbuf buf = STRBUF_INIT;
4628         int err = 0;
4629
4630         temp = prepare_temp_file(spec->path, spec);
4631         *arg++ = pgm;
4632         *arg++ = temp->name;
4633         *arg = NULL;
4634
4635         memset(&child, 0, sizeof(child));
4636         child.use_shell = 1;
4637         child.argv = argv;
4638         child.out = -1;
4639         if (start_command(&child)) {
4640                 remove_tempfile();
4641                 return NULL;
4642         }
4643
4644         if (strbuf_read(&buf, child.out, 0) < 0)
4645                 err = error("error reading from textconv command '%s'", pgm);
4646         close(child.out);
4647
4648         if (finish_command(&child) || err) {
4649                 strbuf_release(&buf);
4650                 remove_tempfile();
4651                 return NULL;
4652         }
4653         remove_tempfile();
4654
4655         return strbuf_detach(&buf, outsize);
4656 }
4657
4658 size_t fill_textconv(struct userdiff_driver *driver,
4659                      struct diff_filespec *df,
4660                      char **outbuf)
4661 {
4662         size_t size;
4663
4664         if (!driver || !driver->textconv) {
4665                 if (!DIFF_FILE_VALID(df)) {
4666                         *outbuf = "";
4667                         return 0;
4668                 }
4669                 if (diff_populate_filespec(df, 0))
4670                         die("unable to read files to diff");
4671                 *outbuf = df->data;
4672                 return df->size;
4673         }
4674
4675         if (driver->textconv_cache && df->sha1_valid) {
4676                 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
4677                                           &size);
4678                 if (*outbuf)
4679                         return size;
4680         }
4681
4682         *outbuf = run_textconv(driver->textconv, df, &size);
4683         if (!*outbuf)
4684                 die("unable to read files to diff");
4685
4686         if (driver->textconv_cache && df->sha1_valid) {
4687                 /* ignore errors, as we might be in a readonly repository */
4688                 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
4689                                 size);
4690                 /*
4691                  * we could save up changes and flush them all at the end,
4692                  * but we would need an extra call after all diffing is done.
4693                  * Since generating a cache entry is the slow path anyway,
4694                  * this extra overhead probably isn't a big deal.
4695                  */
4696                 notes_cache_write(driver->textconv_cache);
4697         }
4698
4699         return size;
4700 }