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