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