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, count;
1447         int width, name_width, graph_width, number_width = 0, bin_width = 0;
1448         const char *reset, *add_c, *del_c;
1449         const char *line_prefix = "";
1450         int extra_shown = 0;
1451         struct strbuf *msg = NULL;
1452
1453         if (data->nr == 0)
1454                 return;
1455
1456         if (options->output_prefix) {
1457                 msg = options->output_prefix(options, options->output_prefix_data);
1458                 line_prefix = msg->buf;
1459         }
1460
1461         count = options->stat_count ? options->stat_count : data->nr;
1462
1463         reset = diff_get_color_opt(options, DIFF_RESET);
1464         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1465         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1466
1467         /*
1468          * Find the longest filename and max number of changes
1469          */
1470         for (i = 0; (i < count) && (i < data->nr); i++) {
1471                 struct diffstat_file *file = data->files[i];
1472                 uintmax_t change = file->added + file->deleted;
1473                 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_unmerged) {
1484                         /* "Unmerged" is 8 characters */
1485                         bin_width = bin_width < 8 ? 8 : bin_width;
1486                         continue;
1487                 }
1488                 if (file->is_binary) {
1489                         /* "Bin XXX -> YYY bytes" */
1490                         int w = 14 + decimal_width(file->added)
1491                                 + decimal_width(file->deleted);
1492                         bin_width = bin_width < w ? w : bin_width;
1493                         /* Display change counts aligned with "Bin" */
1494                         number_width = 3;
1495                         continue;
1496                 }
1497
1498                 if (max_change < change)
1499                         max_change = change;
1500         }
1501         count = i; /* min(count, data->nr) */
1502
1503         /*
1504          * We have width = stat_width or term_columns() columns total.
1505          * We want a maximum of min(max_len, stat_name_width) for the name part.
1506          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1507          * We also need 1 for " " and 4 + decimal_width(max_change)
1508          * for " | NNNN " and one the empty column at the end, altogether
1509          * 6 + decimal_width(max_change).
1510          *
1511          * If there's not enough space, we will use the smaller of
1512          * stat_name_width (if set) and 5/8*width for the filename,
1513          * and the rest for constant elements + graph part, but no more
1514          * than stat_graph_width for the graph part.
1515          * (5/8 gives 50 for filename and 30 for the constant parts + graph
1516          * for the standard terminal size).
1517          *
1518          * In other words: stat_width limits the maximum width, and
1519          * stat_name_width fixes the maximum width of the filename,
1520          * and is also used to divide available columns if there
1521          * aren't enough.
1522          *
1523          * Binary files are displayed with "Bin XXX -> YYY bytes"
1524          * instead of the change count and graph. This part is treated
1525          * similarly to the graph part, except that it is not
1526          * "scaled". If total width is too small to accomodate the
1527          * guaranteed minimum width of the filename part and the
1528          * separators and this message, this message will "overflow"
1529          * making the line longer than the maximum width.
1530          */
1531
1532         if (options->stat_width == -1)
1533                 width = term_columns() - options->output_prefix_length;
1534         else
1535                 width = options->stat_width ? options->stat_width : 80;
1536         number_width = decimal_width(max_change) > number_width ?
1537                 decimal_width(max_change) : number_width;
1538
1539         if (options->stat_graph_width == -1)
1540                 options->stat_graph_width = diff_stat_graph_width;
1541
1542         /*
1543          * Guarantee 3/8*16==6 for the graph part
1544          * and 5/8*16==10 for the filename part
1545          */
1546         if (width < 16 + 6 + number_width)
1547                 width = 16 + 6 + number_width;
1548
1549         /*
1550          * First assign sizes that are wanted, ignoring available width.
1551          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1552          * starting from "XXX" should fit in graph_width.
1553          */
1554         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1555         if (options->stat_graph_width &&
1556             options->stat_graph_width < graph_width)
1557                 graph_width = options->stat_graph_width;
1558
1559         name_width = (options->stat_name_width > 0 &&
1560                       options->stat_name_width < max_len) ?
1561                 options->stat_name_width : max_len;
1562
1563         /*
1564          * Adjust adjustable widths not to exceed maximum width
1565          */
1566         if (name_width + number_width + 6 + graph_width > width) {
1567                 if (graph_width > width * 3/8 - number_width - 6) {
1568                         graph_width = width * 3/8 - number_width - 6;
1569                         if (graph_width < 6)
1570                                 graph_width = 6;
1571                 }
1572
1573                 if (options->stat_graph_width &&
1574                     graph_width > options->stat_graph_width)
1575                         graph_width = options->stat_graph_width;
1576                 if (name_width > width - number_width - 6 - graph_width)
1577                         name_width = width - number_width - 6 - graph_width;
1578                 else
1579                         graph_width = width - number_width - 6 - name_width;
1580         }
1581
1582         /*
1583          * From here name_width is the width of the name area,
1584          * and graph_width is the width of the graph area.
1585          * max_change is used to scale graph properly.
1586          */
1587         for (i = 0; i < count; i++) {
1588                 const char *prefix = "";
1589                 char *name = data->files[i]->print_name;
1590                 uintmax_t added = data->files[i]->added;
1591                 uintmax_t deleted = data->files[i]->deleted;
1592                 int name_len;
1593
1594                 if (!data->files[i]->is_renamed &&
1595                          (added + deleted == 0)) {
1596                         total_files--;
1597                         continue;
1598                 }
1599                 /*
1600                  * "scale" the filename
1601                  */
1602                 len = name_width;
1603                 name_len = strlen(name);
1604                 if (name_width < name_len) {
1605                         char *slash;
1606                         prefix = "...";
1607                         len -= 3;
1608                         name += name_len - len;
1609                         slash = strchr(name, '/');
1610                         if (slash)
1611                                 name = slash;
1612                 }
1613
1614                 if (data->files[i]->is_binary) {
1615                         fprintf(options->file, "%s", line_prefix);
1616                         show_name(options->file, prefix, name, len);
1617                         fprintf(options->file, " %*s", number_width, "Bin");
1618                         if (!added && !deleted) {
1619                                 putc('\n', options->file);
1620                                 continue;
1621                         }
1622                         fprintf(options->file, " %s%"PRIuMAX"%s",
1623                                 del_c, deleted, reset);
1624                         fprintf(options->file, " -> ");
1625                         fprintf(options->file, "%s%"PRIuMAX"%s",
1626                                 add_c, added, reset);
1627                         fprintf(options->file, " bytes");
1628                         fprintf(options->file, "\n");
1629                         continue;
1630                 }
1631                 else if (data->files[i]->is_unmerged) {
1632                         fprintf(options->file, "%s", line_prefix);
1633                         show_name(options->file, prefix, name, len);
1634                         fprintf(options->file, " Unmerged\n");
1635                         continue;
1636                 }
1637
1638                 /*
1639                  * scale the add/delete
1640                  */
1641                 add = added;
1642                 del = deleted;
1643                 adds += add;
1644                 dels += del;
1645
1646                 if (graph_width <= max_change) {
1647                         int total = add + del;
1648
1649                         total = scale_linear(add + del, graph_width, max_change);
1650                         if (total < 2 && add && del)
1651                                 /* width >= 2 due to the sanity check */
1652                                 total = 2;
1653                         if (add < del) {
1654                                 add = scale_linear(add, graph_width, max_change);
1655                                 del = total - add;
1656                         } else {
1657                                 del = scale_linear(del, graph_width, max_change);
1658                                 add = total - del;
1659                         }
1660                 }
1661                 fprintf(options->file, "%s", line_prefix);
1662                 show_name(options->file, prefix, name, len);
1663                 fprintf(options->file, " %*"PRIuMAX"%s",
1664                         number_width, added + deleted,
1665                         added + deleted ? " " : "");
1666                 show_graph(options->file, '+', add, add_c, reset);
1667                 show_graph(options->file, '-', del, del_c, reset);
1668                 fprintf(options->file, "\n");
1669         }
1670         for (i = count; i < data->nr; i++) {
1671                 uintmax_t added = data->files[i]->added;
1672                 uintmax_t deleted = data->files[i]->deleted;
1673                 if (!data->files[i]->is_renamed &&
1674                          (added + deleted == 0)) {
1675                         total_files--;
1676                         continue;
1677                 }
1678                 adds += added;
1679                 dels += deleted;
1680                 if (!extra_shown)
1681                         fprintf(options->file, "%s ...\n", line_prefix);
1682                 extra_shown = 1;
1683         }
1684         fprintf(options->file, "%s", line_prefix);
1685         print_stat_summary(options->file, total_files, adds, dels);
1686 }
1687
1688 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1689 {
1690         int i, adds = 0, dels = 0, total_files = data->nr;
1691
1692         if (data->nr == 0)
1693                 return;
1694
1695         for (i = 0; i < data->nr; i++) {
1696                 int added = data->files[i]->added;
1697                 int deleted= data->files[i]->deleted;
1698
1699                 if (data->files[i]->is_unmerged)
1700                         continue;
1701                 if (!data->files[i]->is_renamed && (added + deleted == 0)) {
1702                         total_files--;
1703                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
1704                         adds += added;
1705                         dels += deleted;
1706                 }
1707         }
1708         if (options->output_prefix) {
1709                 struct strbuf *msg = NULL;
1710                 msg = options->output_prefix(options,
1711                                 options->output_prefix_data);
1712                 fprintf(options->file, "%s", msg->buf);
1713         }
1714         print_stat_summary(options->file, total_files, adds, dels);
1715 }
1716
1717 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1718 {
1719         int i;
1720
1721         if (data->nr == 0)
1722                 return;
1723
1724         for (i = 0; i < data->nr; i++) {
1725                 struct diffstat_file *file = data->files[i];
1726
1727                 if (options->output_prefix) {
1728                         struct strbuf *msg = NULL;
1729                         msg = options->output_prefix(options,
1730                                         options->output_prefix_data);
1731                         fprintf(options->file, "%s", msg->buf);
1732                 }
1733
1734                 if (file->is_binary)
1735                         fprintf(options->file, "-\t-\t");
1736                 else
1737                         fprintf(options->file,
1738                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1739                                 file->added, file->deleted);
1740                 if (options->line_termination) {
1741                         fill_print_name(file);
1742                         if (!file->is_renamed)
1743                                 write_name_quoted(file->name, options->file,
1744                                                   options->line_termination);
1745                         else {
1746                                 fputs(file->print_name, options->file);
1747                                 putc(options->line_termination, options->file);
1748                         }
1749                 } else {
1750                         if (file->is_renamed) {
1751                                 putc('\0', options->file);
1752                                 write_name_quoted(file->from_name, options->file, '\0');
1753                         }
1754                         write_name_quoted(file->name, options->file, '\0');
1755                 }
1756         }
1757 }
1758
1759 struct dirstat_file {
1760         const char *name;
1761         unsigned long changed;
1762 };
1763
1764 struct dirstat_dir {
1765         struct dirstat_file *files;
1766         int alloc, nr, permille, cumulative;
1767 };
1768
1769 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1770                 unsigned long changed, const char *base, int baselen)
1771 {
1772         unsigned long this_dir = 0;
1773         unsigned int sources = 0;
1774         const char *line_prefix = "";
1775         struct strbuf *msg = NULL;
1776
1777         if (opt->output_prefix) {
1778                 msg = opt->output_prefix(opt, opt->output_prefix_data);
1779                 line_prefix = msg->buf;
1780         }
1781
1782         while (dir->nr) {
1783                 struct dirstat_file *f = dir->files;
1784                 int namelen = strlen(f->name);
1785                 unsigned long this;
1786                 char *slash;
1787
1788                 if (namelen < baselen)
1789                         break;
1790                 if (memcmp(f->name, base, baselen))
1791                         break;
1792                 slash = strchr(f->name + baselen, '/');
1793                 if (slash) {
1794                         int newbaselen = slash + 1 - f->name;
1795                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1796                         sources++;
1797                 } else {
1798                         this = f->changed;
1799                         dir->files++;
1800                         dir->nr--;
1801                         sources += 2;
1802                 }
1803                 this_dir += this;
1804         }
1805
1806         /*
1807          * We don't report dirstat's for
1808          *  - the top level
1809          *  - or cases where everything came from a single directory
1810          *    under this directory (sources == 1).
1811          */
1812         if (baselen && sources != 1) {
1813                 if (this_dir) {
1814                         int permille = this_dir * 1000 / changed;
1815                         if (permille >= dir->permille) {
1816                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1817                                         permille / 10, permille % 10, baselen, base);
1818                                 if (!dir->cumulative)
1819                                         return 0;
1820                         }
1821                 }
1822         }
1823         return this_dir;
1824 }
1825
1826 static int dirstat_compare(const void *_a, const void *_b)
1827 {
1828         const struct dirstat_file *a = _a;
1829         const struct dirstat_file *b = _b;
1830         return strcmp(a->name, b->name);
1831 }
1832
1833 static void show_dirstat(struct diff_options *options)
1834 {
1835         int i;
1836         unsigned long changed;
1837         struct dirstat_dir dir;
1838         struct diff_queue_struct *q = &diff_queued_diff;
1839
1840         dir.files = NULL;
1841         dir.alloc = 0;
1842         dir.nr = 0;
1843         dir.permille = options->dirstat_permille;
1844         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1845
1846         changed = 0;
1847         for (i = 0; i < q->nr; i++) {
1848                 struct diff_filepair *p = q->queue[i];
1849                 const char *name;
1850                 unsigned long copied, added, damage;
1851                 int content_changed;
1852
1853                 name = p->two->path ? p->two->path : p->one->path;
1854
1855                 if (p->one->sha1_valid && p->two->sha1_valid)
1856                         content_changed = hashcmp(p->one->sha1, p->two->sha1);
1857                 else
1858                         content_changed = 1;
1859
1860                 if (!content_changed) {
1861                         /*
1862                          * The SHA1 has not changed, so pre-/post-content is
1863                          * identical. We can therefore skip looking at the
1864                          * file contents altogether.
1865                          */
1866                         damage = 0;
1867                         goto found_damage;
1868                 }
1869
1870                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1871                         /*
1872                          * In --dirstat-by-file mode, we don't really need to
1873                          * look at the actual file contents at all.
1874                          * The fact that the SHA1 changed is enough for us to
1875                          * add this file to the list of results
1876                          * (with each file contributing equal damage).
1877                          */
1878                         damage = 1;
1879                         goto found_damage;
1880                 }
1881
1882                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1883                         diff_populate_filespec(p->one, 0);
1884                         diff_populate_filespec(p->two, 0);
1885                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1886                                                &copied, &added);
1887                         diff_free_filespec_data(p->one);
1888                         diff_free_filespec_data(p->two);
1889                 } else if (DIFF_FILE_VALID(p->one)) {
1890                         diff_populate_filespec(p->one, 1);
1891                         copied = added = 0;
1892                         diff_free_filespec_data(p->one);
1893                 } else if (DIFF_FILE_VALID(p->two)) {
1894                         diff_populate_filespec(p->two, 1);
1895                         copied = 0;
1896                         added = p->two->size;
1897                         diff_free_filespec_data(p->two);
1898                 } else
1899                         continue;
1900
1901                 /*
1902                  * Original minus copied is the removed material,
1903                  * added is the new material.  They are both damages
1904                  * made to the preimage.
1905                  * If the resulting damage is zero, we know that
1906                  * diffcore_count_changes() considers the two entries to
1907                  * be identical, but since content_changed is true, we
1908                  * know that there must have been _some_ kind of change,
1909                  * so we force all entries to have damage > 0.
1910                  */
1911                 damage = (p->one->size - copied) + added;
1912                 if (!damage)
1913                         damage = 1;
1914
1915 found_damage:
1916                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1917                 dir.files[dir.nr].name = name;
1918                 dir.files[dir.nr].changed = damage;
1919                 changed += damage;
1920                 dir.nr++;
1921         }
1922
1923         /* This can happen even with many files, if everything was renames */
1924         if (!changed)
1925                 return;
1926
1927         /* Show all directories with more than x% of the changes */
1928         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1929         gather_dirstat(options, &dir, changed, "", 0);
1930 }
1931
1932 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
1933 {
1934         int i;
1935         unsigned long changed;
1936         struct dirstat_dir dir;
1937
1938         if (data->nr == 0)
1939                 return;
1940
1941         dir.files = NULL;
1942         dir.alloc = 0;
1943         dir.nr = 0;
1944         dir.permille = options->dirstat_permille;
1945         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1946
1947         changed = 0;
1948         for (i = 0; i < data->nr; i++) {
1949                 struct diffstat_file *file = data->files[i];
1950                 unsigned long damage = file->added + file->deleted;
1951                 if (file->is_binary)
1952                         /*
1953                          * binary files counts bytes, not lines. Must find some
1954                          * way to normalize binary bytes vs. textual lines.
1955                          * The following heuristic assumes that there are 64
1956                          * bytes per "line".
1957                          * This is stupid and ugly, but very cheap...
1958                          */
1959                         damage = (damage + 63) / 64;
1960                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1961                 dir.files[dir.nr].name = file->name;
1962                 dir.files[dir.nr].changed = damage;
1963                 changed += damage;
1964                 dir.nr++;
1965         }
1966
1967         /* This can happen even with many files, if everything was renames */
1968         if (!changed)
1969                 return;
1970
1971         /* Show all directories with more than x% of the changes */
1972         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1973         gather_dirstat(options, &dir, changed, "", 0);
1974 }
1975
1976 static void free_diffstat_info(struct diffstat_t *diffstat)
1977 {
1978         int i;
1979         for (i = 0; i < diffstat->nr; i++) {
1980                 struct diffstat_file *f = diffstat->files[i];
1981                 if (f->name != f->print_name)
1982                         free(f->print_name);
1983                 free(f->name);
1984                 free(f->from_name);
1985                 free(f);
1986         }
1987         free(diffstat->files);
1988 }
1989
1990 struct checkdiff_t {
1991         const char *filename;
1992         int lineno;
1993         int conflict_marker_size;
1994         struct diff_options *o;
1995         unsigned ws_rule;
1996         unsigned status;
1997 };
1998
1999 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2000 {
2001         char firstchar;
2002         int cnt;
2003
2004         if (len < marker_size + 1)
2005                 return 0;
2006         firstchar = line[0];
2007         switch (firstchar) {
2008         case '=': case '>': case '<': case '|':
2009                 break;
2010         default:
2011                 return 0;
2012         }
2013         for (cnt = 1; cnt < marker_size; cnt++)
2014                 if (line[cnt] != firstchar)
2015                         return 0;
2016         /* line[1] thru line[marker_size-1] are same as firstchar */
2017         if (len < marker_size + 1 || !isspace(line[marker_size]))
2018                 return 0;
2019         return 1;
2020 }
2021
2022 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2023 {
2024         struct checkdiff_t *data = priv;
2025         int marker_size = data->conflict_marker_size;
2026         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2027         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2028         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2029         char *err;
2030         char *line_prefix = "";
2031         struct strbuf *msgbuf;
2032
2033         assert(data->o);
2034         if (data->o->output_prefix) {
2035                 msgbuf = data->o->output_prefix(data->o,
2036                         data->o->output_prefix_data);
2037                 line_prefix = msgbuf->buf;
2038         }
2039
2040         if (line[0] == '+') {
2041                 unsigned bad;
2042                 data->lineno++;
2043                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2044                         data->status |= 1;
2045                         fprintf(data->o->file,
2046                                 "%s%s:%d: leftover conflict marker\n",
2047                                 line_prefix, data->filename, data->lineno);
2048                 }
2049                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2050                 if (!bad)
2051                         return;
2052                 data->status |= bad;
2053                 err = whitespace_error_string(bad);
2054                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2055                         line_prefix, data->filename, data->lineno, err);
2056                 free(err);
2057                 emit_line(data->o, set, reset, line, 1);
2058                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2059                               data->o->file, set, reset, ws);
2060         } else if (line[0] == ' ') {
2061                 data->lineno++;
2062         } else if (line[0] == '@') {
2063                 char *plus = strchr(line, '+');
2064                 if (plus)
2065                         data->lineno = strtol(plus, NULL, 10) - 1;
2066                 else
2067                         die("invalid diff");
2068         }
2069 }
2070
2071 static unsigned char *deflate_it(char *data,
2072                                  unsigned long size,
2073                                  unsigned long *result_size)
2074 {
2075         int bound;
2076         unsigned char *deflated;
2077         git_zstream stream;
2078
2079         memset(&stream, 0, sizeof(stream));
2080         git_deflate_init(&stream, zlib_compression_level);
2081         bound = git_deflate_bound(&stream, size);
2082         deflated = xmalloc(bound);
2083         stream.next_out = deflated;
2084         stream.avail_out = bound;
2085
2086         stream.next_in = (unsigned char *)data;
2087         stream.avail_in = size;
2088         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2089                 ; /* nothing */
2090         git_deflate_end(&stream);
2091         *result_size = stream.total_out;
2092         return deflated;
2093 }
2094
2095 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
2096 {
2097         void *cp;
2098         void *delta;
2099         void *deflated;
2100         void *data;
2101         unsigned long orig_size;
2102         unsigned long delta_size;
2103         unsigned long deflate_size;
2104         unsigned long data_size;
2105
2106         /* We could do deflated delta, or we could do just deflated two,
2107          * whichever is smaller.
2108          */
2109         delta = NULL;
2110         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2111         if (one->size && two->size) {
2112                 delta = diff_delta(one->ptr, one->size,
2113                                    two->ptr, two->size,
2114                                    &delta_size, deflate_size);
2115                 if (delta) {
2116                         void *to_free = delta;
2117                         orig_size = delta_size;
2118                         delta = deflate_it(delta, delta_size, &delta_size);
2119                         free(to_free);
2120                 }
2121         }
2122
2123         if (delta && delta_size < deflate_size) {
2124                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2125                 free(deflated);
2126                 data = delta;
2127                 data_size = delta_size;
2128         }
2129         else {
2130                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
2131                 free(delta);
2132                 data = deflated;
2133                 data_size = deflate_size;
2134         }
2135
2136         /* emit data encoded in base85 */
2137         cp = data;
2138         while (data_size) {
2139                 int bytes = (52 < data_size) ? 52 : data_size;
2140                 char line[70];
2141                 data_size -= bytes;
2142                 if (bytes <= 26)
2143                         line[0] = bytes + 'A' - 1;
2144                 else
2145                         line[0] = bytes - 26 + 'a' - 1;
2146                 encode_85(line + 1, cp, bytes);
2147                 cp = (char *) cp + bytes;
2148                 fprintf(file, "%s", prefix);
2149                 fputs(line, file);
2150                 fputc('\n', file);
2151         }
2152         fprintf(file, "%s\n", prefix);
2153         free(data);
2154 }
2155
2156 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
2157 {
2158         fprintf(file, "%sGIT binary patch\n", prefix);
2159         emit_binary_diff_body(file, one, two, prefix);
2160         emit_binary_diff_body(file, two, one, prefix);
2161 }
2162
2163 int diff_filespec_is_binary(struct diff_filespec *one)
2164 {
2165         if (one->is_binary == -1) {
2166                 diff_filespec_load_driver(one);
2167                 if (one->driver->binary != -1)
2168                         one->is_binary = one->driver->binary;
2169                 else {
2170                         if (!one->data && DIFF_FILE_VALID(one))
2171                                 diff_populate_filespec(one, 0);
2172                         if (one->data)
2173                                 one->is_binary = buffer_is_binary(one->data,
2174                                                 one->size);
2175                         if (one->is_binary == -1)
2176                                 one->is_binary = 0;
2177                 }
2178         }
2179         return one->is_binary;
2180 }
2181
2182 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2183 {
2184         diff_filespec_load_driver(one);
2185         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2186 }
2187
2188 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2189 {
2190         if (!options->a_prefix)
2191                 options->a_prefix = a;
2192         if (!options->b_prefix)
2193                 options->b_prefix = b;
2194 }
2195
2196 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2197 {
2198         if (!DIFF_FILE_VALID(one))
2199                 return NULL;
2200
2201         diff_filespec_load_driver(one);
2202         return userdiff_get_textconv(one->driver);
2203 }
2204
2205 static void builtin_diff(const char *name_a,
2206                          const char *name_b,
2207                          struct diff_filespec *one,
2208                          struct diff_filespec *two,
2209                          const char *xfrm_msg,
2210                          int must_show_header,
2211                          struct diff_options *o,
2212                          int complete_rewrite)
2213 {
2214         mmfile_t mf1, mf2;
2215         const char *lbl[2];
2216         char *a_one, *b_two;
2217         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
2218         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2219         const char *a_prefix, *b_prefix;
2220         struct userdiff_driver *textconv_one = NULL;
2221         struct userdiff_driver *textconv_two = NULL;
2222         struct strbuf header = STRBUF_INIT;
2223         struct strbuf *msgbuf;
2224         char *line_prefix = "";
2225
2226         if (o->output_prefix) {
2227                 msgbuf = o->output_prefix(o, o->output_prefix_data);
2228                 line_prefix = msgbuf->buf;
2229         }
2230
2231         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2232                         (!one->mode || S_ISGITLINK(one->mode)) &&
2233                         (!two->mode || S_ISGITLINK(two->mode))) {
2234                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2235                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2236                 show_submodule_summary(o->file, one ? one->path : two->path,
2237                                 one->sha1, two->sha1, two->dirty_submodule,
2238                                 del, add, reset);
2239                 return;
2240         }
2241
2242         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2243                 textconv_one = get_textconv(one);
2244                 textconv_two = get_textconv(two);
2245         }
2246
2247         diff_set_mnemonic_prefix(o, "a/", "b/");
2248         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2249                 a_prefix = o->b_prefix;
2250                 b_prefix = o->a_prefix;
2251         } else {
2252                 a_prefix = o->a_prefix;
2253                 b_prefix = o->b_prefix;
2254         }
2255
2256         /* Never use a non-valid filename anywhere if at all possible */
2257         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2258         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2259
2260         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2261         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2262         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2263         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2264         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, set, a_one, b_two, reset);
2265         if (lbl[0][0] == '/') {
2266                 /* /dev/null */
2267                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, set, two->mode, reset);
2268                 if (xfrm_msg)
2269                         strbuf_addstr(&header, xfrm_msg);
2270                 must_show_header = 1;
2271         }
2272         else if (lbl[1][0] == '/') {
2273                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, set, one->mode, reset);
2274                 if (xfrm_msg)
2275                         strbuf_addstr(&header, xfrm_msg);
2276                 must_show_header = 1;
2277         }
2278         else {
2279                 if (one->mode != two->mode) {
2280                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, set, one->mode, reset);
2281                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, set, two->mode, reset);
2282                         must_show_header = 1;
2283                 }
2284                 if (xfrm_msg)
2285                         strbuf_addstr(&header, xfrm_msg);
2286
2287                 /*
2288                  * we do not run diff between different kind
2289                  * of objects.
2290                  */
2291                 if ((one->mode ^ two->mode) & S_IFMT)
2292                         goto free_ab_and_return;
2293                 if (complete_rewrite &&
2294                     (textconv_one || !diff_filespec_is_binary(one)) &&
2295                     (textconv_two || !diff_filespec_is_binary(two))) {
2296                         fprintf(o->file, "%s", header.buf);
2297                         strbuf_reset(&header);
2298                         emit_rewrite_diff(name_a, name_b, one, two,
2299                                                 textconv_one, textconv_two, o);
2300                         o->found_changes = 1;
2301                         goto free_ab_and_return;
2302                 }
2303         }
2304
2305         if (o->irreversible_delete && lbl[1][0] == '/') {
2306                 fprintf(o->file, "%s", header.buf);
2307                 strbuf_reset(&header);
2308                 goto free_ab_and_return;
2309         } else if (!DIFF_OPT_TST(o, TEXT) &&
2310             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2311               (!textconv_two && diff_filespec_is_binary(two)) )) {
2312                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2313                         die("unable to read files to diff");
2314                 /* Quite common confusing case */
2315                 if (mf1.size == mf2.size &&
2316                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2317                         if (must_show_header)
2318                                 fprintf(o->file, "%s", header.buf);
2319                         goto free_ab_and_return;
2320                 }
2321                 fprintf(o->file, "%s", header.buf);
2322                 strbuf_reset(&header);
2323                 if (DIFF_OPT_TST(o, BINARY))
2324                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2325                 else
2326                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2327                                 line_prefix, lbl[0], lbl[1]);
2328                 o->found_changes = 1;
2329         } else {
2330                 /* Crazy xdl interfaces.. */
2331                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2332                 xpparam_t xpp;
2333                 xdemitconf_t xecfg;
2334                 struct emit_callback ecbdata;
2335                 const struct userdiff_funcname *pe;
2336
2337                 if (must_show_header) {
2338                         fprintf(o->file, "%s", header.buf);
2339                         strbuf_reset(&header);
2340                 }
2341
2342                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2343                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2344
2345                 pe = diff_funcname_pattern(one);
2346                 if (!pe)
2347                         pe = diff_funcname_pattern(two);
2348
2349                 memset(&xpp, 0, sizeof(xpp));
2350                 memset(&xecfg, 0, sizeof(xecfg));
2351                 memset(&ecbdata, 0, sizeof(ecbdata));
2352                 ecbdata.label_path = lbl;
2353                 ecbdata.color_diff = want_color(o->use_color);
2354                 ecbdata.found_changesp = &o->found_changes;
2355                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
2356                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2357                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2358                 ecbdata.opt = o;
2359                 ecbdata.header = header.len ? &header : NULL;
2360                 xpp.flags = o->xdl_opts;
2361                 xecfg.ctxlen = o->context;
2362                 xecfg.interhunkctxlen = o->interhunkcontext;
2363                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2364                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2365                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2366                 if (pe)
2367                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2368                 if (!diffopts)
2369                         ;
2370                 else if (!prefixcmp(diffopts, "--unified="))
2371                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
2372                 else if (!prefixcmp(diffopts, "-u"))
2373                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
2374                 if (o->word_diff)
2375                         init_diff_words_data(&ecbdata, o, one, two);
2376                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2377                               &xpp, &xecfg);
2378                 if (o->word_diff)
2379                         free_diff_words_data(&ecbdata);
2380                 if (textconv_one)
2381                         free(mf1.ptr);
2382                 if (textconv_two)
2383                         free(mf2.ptr);
2384                 xdiff_clear_find_func(&xecfg);
2385         }
2386
2387  free_ab_and_return:
2388         strbuf_release(&header);
2389         diff_free_filespec_data(one);
2390         diff_free_filespec_data(two);
2391         free(a_one);
2392         free(b_two);
2393         return;
2394 }
2395
2396 static void builtin_diffstat(const char *name_a, const char *name_b,
2397                              struct diff_filespec *one,
2398                              struct diff_filespec *two,
2399                              struct diffstat_t *diffstat,
2400                              struct diff_options *o,
2401                              int complete_rewrite)
2402 {
2403         mmfile_t mf1, mf2;
2404         struct diffstat_file *data;
2405         int same_contents;
2406
2407         data = diffstat_add(diffstat, name_a, name_b);
2408
2409         if (!one || !two) {
2410                 data->is_unmerged = 1;
2411                 return;
2412         }
2413
2414         same_contents = !hashcmp(one->sha1, two->sha1);
2415
2416         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2417                 data->is_binary = 1;
2418                 if (same_contents) {
2419                         data->added = 0;
2420                         data->deleted = 0;
2421                 } else {
2422                         data->added = diff_filespec_size(two);
2423                         data->deleted = diff_filespec_size(one);
2424                 }
2425         }
2426
2427         else if (complete_rewrite) {
2428                 diff_populate_filespec(one, 0);
2429                 diff_populate_filespec(two, 0);
2430                 data->deleted = count_lines(one->data, one->size);
2431                 data->added = count_lines(two->data, two->size);
2432         }
2433
2434         else if (!same_contents) {
2435                 /* Crazy xdl interfaces.. */
2436                 xpparam_t xpp;
2437                 xdemitconf_t xecfg;
2438
2439                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2440                         die("unable to read files to diff");
2441
2442                 memset(&xpp, 0, sizeof(xpp));
2443                 memset(&xecfg, 0, sizeof(xecfg));
2444                 xpp.flags = o->xdl_opts;
2445                 xecfg.ctxlen = o->context;
2446                 xecfg.interhunkctxlen = o->interhunkcontext;
2447                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2448                               &xpp, &xecfg);
2449         }
2450
2451         diff_free_filespec_data(one);
2452         diff_free_filespec_data(two);
2453 }
2454
2455 static void builtin_checkdiff(const char *name_a, const char *name_b,
2456                               const char *attr_path,
2457                               struct diff_filespec *one,
2458                               struct diff_filespec *two,
2459                               struct diff_options *o)
2460 {
2461         mmfile_t mf1, mf2;
2462         struct checkdiff_t data;
2463
2464         if (!two)
2465                 return;
2466
2467         memset(&data, 0, sizeof(data));
2468         data.filename = name_b ? name_b : name_a;
2469         data.lineno = 0;
2470         data.o = o;
2471         data.ws_rule = whitespace_rule(attr_path);
2472         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2473
2474         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2475                 die("unable to read files to diff");
2476
2477         /*
2478          * All the other codepaths check both sides, but not checking
2479          * the "old" side here is deliberate.  We are checking the newly
2480          * introduced changes, and as long as the "new" side is text, we
2481          * can and should check what it introduces.
2482          */
2483         if (diff_filespec_is_binary(two))
2484                 goto free_and_return;
2485         else {
2486                 /* Crazy xdl interfaces.. */
2487                 xpparam_t xpp;
2488                 xdemitconf_t xecfg;
2489
2490                 memset(&xpp, 0, sizeof(xpp));
2491                 memset(&xecfg, 0, sizeof(xecfg));
2492                 xecfg.ctxlen = 1; /* at least one context line */
2493                 xpp.flags = 0;
2494                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2495                               &xpp, &xecfg);
2496
2497                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2498                         struct emit_callback ecbdata;
2499                         int blank_at_eof;
2500
2501                         ecbdata.ws_rule = data.ws_rule;
2502                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2503                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2504
2505                         if (blank_at_eof) {
2506                                 static char *err;
2507                                 if (!err)
2508                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2509                                 fprintf(o->file, "%s:%d: %s.\n",
2510                                         data.filename, blank_at_eof, err);
2511                                 data.status = 1; /* report errors */
2512                         }
2513                 }
2514         }
2515  free_and_return:
2516         diff_free_filespec_data(one);
2517         diff_free_filespec_data(two);
2518         if (data.status)
2519                 DIFF_OPT_SET(o, CHECK_FAILED);
2520 }
2521
2522 struct diff_filespec *alloc_filespec(const char *path)
2523 {
2524         int namelen = strlen(path);
2525         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2526
2527         memset(spec, 0, sizeof(*spec));
2528         spec->path = (char *)(spec + 1);
2529         memcpy(spec->path, path, namelen+1);
2530         spec->count = 1;
2531         spec->is_binary = -1;
2532         return spec;
2533 }
2534
2535 void free_filespec(struct diff_filespec *spec)
2536 {
2537         if (!--spec->count) {
2538                 diff_free_filespec_data(spec);
2539                 free(spec);
2540         }
2541 }
2542
2543 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2544                    unsigned short mode)
2545 {
2546         if (mode) {
2547                 spec->mode = canon_mode(mode);
2548                 hashcpy(spec->sha1, sha1);
2549                 spec->sha1_valid = !is_null_sha1(sha1);
2550         }
2551 }
2552
2553 /*
2554  * Given a name and sha1 pair, if the index tells us the file in
2555  * the work tree has that object contents, return true, so that
2556  * prepare_temp_file() does not have to inflate and extract.
2557  */
2558 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2559 {
2560         struct cache_entry *ce;
2561         struct stat st;
2562         int pos, len;
2563
2564         /*
2565          * We do not read the cache ourselves here, because the
2566          * benchmark with my previous version that always reads cache
2567          * shows that it makes things worse for diff-tree comparing
2568          * two linux-2.6 kernel trees in an already checked out work
2569          * tree.  This is because most diff-tree comparisons deal with
2570          * only a small number of files, while reading the cache is
2571          * expensive for a large project, and its cost outweighs the
2572          * savings we get by not inflating the object to a temporary
2573          * file.  Practically, this code only helps when we are used
2574          * by diff-cache --cached, which does read the cache before
2575          * calling us.
2576          */
2577         if (!active_cache)
2578                 return 0;
2579
2580         /* We want to avoid the working directory if our caller
2581          * doesn't need the data in a normal file, this system
2582          * is rather slow with its stat/open/mmap/close syscalls,
2583          * and the object is contained in a pack file.  The pack
2584          * is probably already open and will be faster to obtain
2585          * the data through than the working directory.  Loose
2586          * objects however would tend to be slower as they need
2587          * to be individually opened and inflated.
2588          */
2589         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2590                 return 0;
2591
2592         len = strlen(name);
2593         pos = cache_name_pos(name, len);
2594         if (pos < 0)
2595                 return 0;
2596         ce = active_cache[pos];
2597
2598         /*
2599          * This is not the sha1 we are looking for, or
2600          * unreusable because it is not a regular file.
2601          */
2602         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2603                 return 0;
2604
2605         /*
2606          * If ce is marked as "assume unchanged", there is no
2607          * guarantee that work tree matches what we are looking for.
2608          */
2609         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2610                 return 0;
2611
2612         /*
2613          * If ce matches the file in the work tree, we can reuse it.
2614          */
2615         if (ce_uptodate(ce) ||
2616             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2617                 return 1;
2618
2619         return 0;
2620 }
2621
2622 static int populate_from_stdin(struct diff_filespec *s)
2623 {
2624         struct strbuf buf = STRBUF_INIT;
2625         size_t size = 0;
2626
2627         if (strbuf_read(&buf, 0, 0) < 0)
2628                 return error("error while reading from stdin %s",
2629                                      strerror(errno));
2630
2631         s->should_munmap = 0;
2632         s->data = strbuf_detach(&buf, &size);
2633         s->size = size;
2634         s->should_free = 1;
2635         return 0;
2636 }
2637
2638 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2639 {
2640         int len;
2641         char *data = xmalloc(100), *dirty = "";
2642
2643         /* Are we looking at the work tree? */
2644         if (s->dirty_submodule)
2645                 dirty = "-dirty";
2646
2647         len = snprintf(data, 100,
2648                        "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2649         s->data = data;
2650         s->size = len;
2651         s->should_free = 1;
2652         if (size_only) {
2653                 s->data = NULL;
2654                 free(data);
2655         }
2656         return 0;
2657 }
2658
2659 /*
2660  * While doing rename detection and pickaxe operation, we may need to
2661  * grab the data for the blob (or file) for our own in-core comparison.
2662  * diff_filespec has data and size fields for this purpose.
2663  */
2664 int diff_populate_filespec(struct diff_filespec *s, int size_only)
2665 {
2666         int err = 0;
2667         if (!DIFF_FILE_VALID(s))
2668                 die("internal error: asking to populate invalid file.");
2669         if (S_ISDIR(s->mode))
2670                 return -1;
2671
2672         if (s->data)
2673                 return 0;
2674
2675         if (size_only && 0 < s->size)
2676                 return 0;
2677
2678         if (S_ISGITLINK(s->mode))
2679                 return diff_populate_gitlink(s, size_only);
2680
2681         if (!s->sha1_valid ||
2682             reuse_worktree_file(s->path, s->sha1, 0)) {
2683                 struct strbuf buf = STRBUF_INIT;
2684                 struct stat st;
2685                 int fd;
2686
2687                 if (!strcmp(s->path, "-"))
2688                         return populate_from_stdin(s);
2689
2690                 if (lstat(s->path, &st) < 0) {
2691                         if (errno == ENOENT) {
2692                         err_empty:
2693                                 err = -1;
2694                         empty:
2695                                 s->data = (char *)"";
2696                                 s->size = 0;
2697                                 return err;
2698                         }
2699                 }
2700                 s->size = xsize_t(st.st_size);
2701                 if (!s->size)
2702                         goto empty;
2703                 if (S_ISLNK(st.st_mode)) {
2704                         struct strbuf sb = STRBUF_INIT;
2705
2706                         if (strbuf_readlink(&sb, s->path, s->size))
2707                                 goto err_empty;
2708                         s->size = sb.len;
2709                         s->data = strbuf_detach(&sb, NULL);
2710                         s->should_free = 1;
2711                         return 0;
2712                 }
2713                 if (size_only)
2714                         return 0;
2715                 fd = open(s->path, O_RDONLY);
2716                 if (fd < 0)
2717                         goto err_empty;
2718                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2719                 close(fd);
2720                 s->should_munmap = 1;
2721
2722                 /*
2723                  * Convert from working tree format to canonical git format
2724                  */
2725                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2726                         size_t size = 0;
2727                         munmap(s->data, s->size);
2728                         s->should_munmap = 0;
2729                         s->data = strbuf_detach(&buf, &size);
2730                         s->size = size;
2731                         s->should_free = 1;
2732                 }
2733         }
2734         else {
2735                 enum object_type type;
2736                 if (size_only) {
2737                         type = sha1_object_info(s->sha1, &s->size);
2738                         if (type < 0)
2739                                 die("unable to read %s", sha1_to_hex(s->sha1));
2740                 } else {
2741                         s->data = read_sha1_file(s->sha1, &type, &s->size);
2742                         if (!s->data)
2743                                 die("unable to read %s", sha1_to_hex(s->sha1));
2744                         s->should_free = 1;
2745                 }
2746         }
2747         return 0;
2748 }
2749
2750 void diff_free_filespec_blob(struct diff_filespec *s)
2751 {
2752         if (s->should_free)
2753                 free(s->data);
2754         else if (s->should_munmap)
2755                 munmap(s->data, s->size);
2756
2757         if (s->should_free || s->should_munmap) {
2758                 s->should_free = s->should_munmap = 0;
2759                 s->data = NULL;
2760         }
2761 }
2762
2763 void diff_free_filespec_data(struct diff_filespec *s)
2764 {
2765         diff_free_filespec_blob(s);
2766         free(s->cnt_data);
2767         s->cnt_data = NULL;
2768 }
2769
2770 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2771                            void *blob,
2772                            unsigned long size,
2773                            const unsigned char *sha1,
2774                            int mode)
2775 {
2776         int fd;
2777         struct strbuf buf = STRBUF_INIT;
2778         struct strbuf template = STRBUF_INIT;
2779         char *path_dup = xstrdup(path);
2780         const char *base = basename(path_dup);
2781
2782         /* Generate "XXXXXX_basename.ext" */
2783         strbuf_addstr(&template, "XXXXXX_");
2784         strbuf_addstr(&template, base);
2785
2786         fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2787                         strlen(base) + 1);
2788         if (fd < 0)
2789                 die_errno("unable to create temp-file");
2790         if (convert_to_working_tree(path,
2791                         (const char *)blob, (size_t)size, &buf)) {
2792                 blob = buf.buf;
2793                 size = buf.len;
2794         }
2795         if (write_in_full(fd, blob, size) != size)
2796                 die_errno("unable to write temp-file");
2797         close(fd);
2798         temp->name = temp->tmp_path;
2799         strcpy(temp->hex, sha1_to_hex(sha1));
2800         temp->hex[40] = 0;
2801         sprintf(temp->mode, "%06o", mode);
2802         strbuf_release(&buf);
2803         strbuf_release(&template);
2804         free(path_dup);
2805 }
2806
2807 static struct diff_tempfile *prepare_temp_file(const char *name,
2808                 struct diff_filespec *one)
2809 {
2810         struct diff_tempfile *temp = claim_diff_tempfile();
2811
2812         if (!DIFF_FILE_VALID(one)) {
2813         not_a_valid_file:
2814                 /* A '-' entry produces this for file-2, and
2815                  * a '+' entry produces this for file-1.
2816                  */
2817                 temp->name = "/dev/null";
2818                 strcpy(temp->hex, ".");
2819                 strcpy(temp->mode, ".");
2820                 return temp;
2821         }
2822
2823         if (!remove_tempfile_installed) {
2824                 atexit(remove_tempfile);
2825                 sigchain_push_common(remove_tempfile_on_signal);
2826                 remove_tempfile_installed = 1;
2827         }
2828
2829         if (!one->sha1_valid ||
2830             reuse_worktree_file(name, one->sha1, 1)) {
2831                 struct stat st;
2832                 if (lstat(name, &st) < 0) {
2833                         if (errno == ENOENT)
2834                                 goto not_a_valid_file;
2835                         die_errno("stat(%s)", name);
2836                 }
2837                 if (S_ISLNK(st.st_mode)) {
2838                         struct strbuf sb = STRBUF_INIT;
2839                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2840                                 die_errno("readlink(%s)", name);
2841                         prep_temp_blob(name, temp, sb.buf, sb.len,
2842                                        (one->sha1_valid ?
2843                                         one->sha1 : null_sha1),
2844                                        (one->sha1_valid ?
2845                                         one->mode : S_IFLNK));
2846                         strbuf_release(&sb);
2847                 }
2848                 else {
2849                         /* we can borrow from the file in the work tree */
2850                         temp->name = name;
2851                         if (!one->sha1_valid)
2852                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2853                         else
2854                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2855                         /* Even though we may sometimes borrow the
2856                          * contents from the work tree, we always want
2857                          * one->mode.  mode is trustworthy even when
2858                          * !(one->sha1_valid), as long as
2859                          * DIFF_FILE_VALID(one).
2860                          */
2861                         sprintf(temp->mode, "%06o", one->mode);
2862                 }
2863                 return temp;
2864         }
2865         else {
2866                 if (diff_populate_filespec(one, 0))
2867                         die("cannot read data blob for %s", one->path);
2868                 prep_temp_blob(name, temp, one->data, one->size,
2869                                one->sha1, one->mode);
2870         }
2871         return temp;
2872 }
2873
2874 /* An external diff command takes:
2875  *
2876  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2877  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2878  *
2879  */
2880 static void run_external_diff(const char *pgm,
2881                               const char *name,
2882                               const char *other,
2883                               struct diff_filespec *one,
2884                               struct diff_filespec *two,
2885                               const char *xfrm_msg,
2886                               int complete_rewrite)
2887 {
2888         const char *spawn_arg[10];
2889         int retval;
2890         const char **arg = &spawn_arg[0];
2891
2892         if (one && two) {
2893                 struct diff_tempfile *temp_one, *temp_two;
2894                 const char *othername = (other ? other : name);
2895                 temp_one = prepare_temp_file(name, one);
2896                 temp_two = prepare_temp_file(othername, two);
2897                 *arg++ = pgm;
2898                 *arg++ = name;
2899                 *arg++ = temp_one->name;
2900                 *arg++ = temp_one->hex;
2901                 *arg++ = temp_one->mode;
2902                 *arg++ = temp_two->name;
2903                 *arg++ = temp_two->hex;
2904                 *arg++ = temp_two->mode;
2905                 if (other) {
2906                         *arg++ = other;
2907                         *arg++ = xfrm_msg;
2908                 }
2909         } else {
2910                 *arg++ = pgm;
2911                 *arg++ = name;
2912         }
2913         *arg = NULL;
2914         fflush(NULL);
2915         retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2916         remove_tempfile();
2917         if (retval) {
2918                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2919                 exit(1);
2920         }
2921 }
2922
2923 static int similarity_index(struct diff_filepair *p)
2924 {
2925         return p->score * 100 / MAX_SCORE;
2926 }
2927
2928 static void fill_metainfo(struct strbuf *msg,
2929                           const char *name,
2930                           const char *other,
2931                           struct diff_filespec *one,
2932                           struct diff_filespec *two,
2933                           struct diff_options *o,
2934                           struct diff_filepair *p,
2935                           int *must_show_header,
2936                           int use_color)
2937 {
2938         const char *set = diff_get_color(use_color, DIFF_METAINFO);
2939         const char *reset = diff_get_color(use_color, DIFF_RESET);
2940         struct strbuf *msgbuf;
2941         char *line_prefix = "";
2942
2943         *must_show_header = 1;
2944         if (o->output_prefix) {
2945                 msgbuf = o->output_prefix(o, o->output_prefix_data);
2946                 line_prefix = msgbuf->buf;
2947         }
2948         strbuf_init(msg, PATH_MAX * 2 + 300);
2949         switch (p->status) {
2950         case DIFF_STATUS_COPIED:
2951                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2952                             line_prefix, set, similarity_index(p));
2953                 strbuf_addf(msg, "%s\n%s%scopy from ",
2954                             reset,  line_prefix, set);
2955                 quote_c_style(name, msg, NULL, 0);
2956                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
2957                 quote_c_style(other, msg, NULL, 0);
2958                 strbuf_addf(msg, "%s\n", reset);
2959                 break;
2960         case DIFF_STATUS_RENAMED:
2961                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2962                             line_prefix, set, similarity_index(p));
2963                 strbuf_addf(msg, "%s\n%s%srename from ",
2964                             reset, line_prefix, set);
2965                 quote_c_style(name, msg, NULL, 0);
2966                 strbuf_addf(msg, "%s\n%s%srename to ",
2967                             reset, line_prefix, set);
2968                 quote_c_style(other, msg, NULL, 0);
2969                 strbuf_addf(msg, "%s\n", reset);
2970                 break;
2971         case DIFF_STATUS_MODIFIED:
2972                 if (p->score) {
2973                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
2974                                     line_prefix,
2975                                     set, similarity_index(p), reset);
2976                         break;
2977                 }
2978                 /* fallthru */
2979         default:
2980                 *must_show_header = 0;
2981         }
2982         if (one && two && hashcmp(one->sha1, two->sha1)) {
2983                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2984
2985                 if (DIFF_OPT_TST(o, BINARY)) {
2986                         mmfile_t mf;
2987                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2988                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2989                                 abbrev = 40;
2990                 }
2991                 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
2992                             find_unique_abbrev(one->sha1, abbrev));
2993                 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
2994                 if (one->mode == two->mode)
2995                         strbuf_addf(msg, " %06o", one->mode);
2996                 strbuf_addf(msg, "%s\n", reset);
2997         }
2998 }
2999
3000 static void run_diff_cmd(const char *pgm,
3001                          const char *name,
3002                          const char *other,
3003                          const char *attr_path,
3004                          struct diff_filespec *one,
3005                          struct diff_filespec *two,
3006                          struct strbuf *msg,
3007                          struct diff_options *o,
3008                          struct diff_filepair *p)
3009 {
3010         const char *xfrm_msg = NULL;
3011         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3012         int must_show_header = 0;
3013
3014         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3015                 pgm = NULL;
3016         else {
3017                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3018                 if (drv && drv->external)
3019                         pgm = drv->external;
3020         }
3021
3022         if (msg) {
3023                 /*
3024                  * don't use colors when the header is intended for an
3025                  * external diff driver
3026                  */
3027                 fill_metainfo(msg, name, other, one, two, o, p,
3028                               &must_show_header,
3029                               want_color(o->use_color) && !pgm);
3030                 xfrm_msg = msg->len ? msg->buf : NULL;
3031         }
3032
3033         if (pgm) {
3034                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3035                                   complete_rewrite);
3036                 return;
3037         }
3038         if (one && two)
3039                 builtin_diff(name, other ? other : name,
3040                              one, two, xfrm_msg, must_show_header,
3041                              o, complete_rewrite);
3042         else
3043                 fprintf(o->file, "* Unmerged path %s\n", name);
3044 }
3045
3046 static void diff_fill_sha1_info(struct diff_filespec *one)
3047 {
3048         if (DIFF_FILE_VALID(one)) {
3049                 if (!one->sha1_valid) {
3050                         struct stat st;
3051                         if (!strcmp(one->path, "-")) {
3052                                 hashcpy(one->sha1, null_sha1);
3053                                 return;
3054                         }
3055                         if (lstat(one->path, &st) < 0)
3056                                 die_errno("stat '%s'", one->path);
3057                         if (index_path(one->sha1, one->path, &st, 0))
3058                                 die("cannot hash %s", one->path);
3059                 }
3060         }
3061         else
3062                 hashclr(one->sha1);
3063 }
3064
3065 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3066 {
3067         /* Strip the prefix but do not molest /dev/null and absolute paths */
3068         if (*namep && **namep != '/') {
3069                 *namep += prefix_length;
3070                 if (**namep == '/')
3071                         ++*namep;
3072         }
3073         if (*otherp && **otherp != '/') {
3074                 *otherp += prefix_length;
3075                 if (**otherp == '/')
3076                         ++*otherp;
3077         }
3078 }
3079
3080 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3081 {
3082         const char *pgm = external_diff();
3083         struct strbuf msg;
3084         struct diff_filespec *one = p->one;
3085         struct diff_filespec *two = p->two;
3086         const char *name;
3087         const char *other;
3088         const char *attr_path;
3089
3090         name  = p->one->path;
3091         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3092         attr_path = name;
3093         if (o->prefix_length)
3094                 strip_prefix(o->prefix_length, &name, &other);
3095
3096         if (DIFF_PAIR_UNMERGED(p)) {
3097                 run_diff_cmd(pgm, name, NULL, attr_path,
3098                              NULL, NULL, NULL, o, p);
3099                 return;
3100         }
3101
3102         diff_fill_sha1_info(one);
3103         diff_fill_sha1_info(two);
3104
3105         if (!pgm &&
3106             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3107             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3108                 /*
3109                  * a filepair that changes between file and symlink
3110                  * needs to be split into deletion and creation.
3111                  */
3112                 struct diff_filespec *null = alloc_filespec(two->path);
3113                 run_diff_cmd(NULL, name, other, attr_path,
3114                              one, null, &msg, o, p);
3115                 free(null);
3116                 strbuf_release(&msg);
3117
3118                 null = alloc_filespec(one->path);
3119                 run_diff_cmd(NULL, name, other, attr_path,
3120                              null, two, &msg, o, p);
3121                 free(null);
3122         }
3123         else
3124                 run_diff_cmd(pgm, name, other, attr_path,
3125                              one, two, &msg, o, p);
3126
3127         strbuf_release(&msg);
3128 }
3129
3130 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3131                          struct diffstat_t *diffstat)
3132 {
3133         const char *name;
3134         const char *other;
3135         int complete_rewrite = 0;
3136
3137         if (DIFF_PAIR_UNMERGED(p)) {
3138                 /* unmerged */
3139                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
3140                 return;
3141         }
3142
3143         name = p->one->path;
3144         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3145
3146         if (o->prefix_length)
3147                 strip_prefix(o->prefix_length, &name, &other);
3148
3149         diff_fill_sha1_info(p->one);
3150         diff_fill_sha1_info(p->two);
3151
3152         if (p->status == DIFF_STATUS_MODIFIED && p->score)
3153                 complete_rewrite = 1;
3154         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
3155 }
3156
3157 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3158 {
3159         const char *name;
3160         const char *other;
3161         const char *attr_path;
3162
3163         if (DIFF_PAIR_UNMERGED(p)) {
3164                 /* unmerged */
3165                 return;
3166         }
3167
3168         name = p->one->path;
3169         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3170         attr_path = other ? other : name;
3171
3172         if (o->prefix_length)
3173                 strip_prefix(o->prefix_length, &name, &other);
3174
3175         diff_fill_sha1_info(p->one);
3176         diff_fill_sha1_info(p->two);
3177
3178         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3179 }
3180
3181 void diff_setup(struct diff_options *options)
3182 {
3183         memcpy(options, &default_diff_options, sizeof(*options));
3184
3185         options->file = stdout;
3186
3187         options->line_termination = '\n';
3188         options->break_opt = -1;
3189         options->rename_limit = -1;
3190         options->dirstat_permille = diff_dirstat_permille_default;
3191         options->context = 3;
3192         DIFF_OPT_SET(options, RENAME_EMPTY);
3193
3194         options->change = diff_change;
3195         options->add_remove = diff_addremove;
3196         options->use_color = diff_use_color_default;
3197         options->detect_rename = diff_detect_rename_default;
3198
3199         if (diff_no_prefix) {
3200                 options->a_prefix = options->b_prefix = "";
3201         } else if (!diff_mnemonic_prefix) {
3202                 options->a_prefix = "a/";
3203                 options->b_prefix = "b/";
3204         }
3205 }
3206
3207 int diff_setup_done(struct diff_options *options)
3208 {
3209         int count = 0;
3210
3211         if (options->output_format & DIFF_FORMAT_NAME)
3212                 count++;
3213         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3214                 count++;
3215         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3216                 count++;
3217         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3218                 count++;
3219         if (count > 1)
3220                 die("--name-only, --name-status, --check and -s are mutually exclusive");
3221
3222         /*
3223          * Most of the time we can say "there are changes"
3224          * only by checking if there are changed paths, but
3225          * --ignore-whitespace* options force us to look
3226          * inside contents.
3227          */
3228
3229         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3230             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3231             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3232                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3233         else
3234                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3235
3236         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3237                 options->detect_rename = DIFF_DETECT_COPY;
3238
3239         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3240                 options->prefix = NULL;
3241         if (options->prefix)
3242                 options->prefix_length = strlen(options->prefix);
3243         else
3244                 options->prefix_length = 0;
3245
3246         if (options->output_format & (DIFF_FORMAT_NAME |
3247                                       DIFF_FORMAT_NAME_STATUS |
3248                                       DIFF_FORMAT_CHECKDIFF |
3249                                       DIFF_FORMAT_NO_OUTPUT))
3250                 options->output_format &= ~(DIFF_FORMAT_RAW |
3251                                             DIFF_FORMAT_NUMSTAT |
3252                                             DIFF_FORMAT_DIFFSTAT |
3253                                             DIFF_FORMAT_SHORTSTAT |
3254                                             DIFF_FORMAT_DIRSTAT |
3255                                             DIFF_FORMAT_SUMMARY |
3256                                             DIFF_FORMAT_PATCH);
3257
3258         /*
3259          * These cases always need recursive; we do not drop caller-supplied
3260          * recursive bits for other formats here.
3261          */
3262         if (options->output_format & (DIFF_FORMAT_PATCH |
3263                                       DIFF_FORMAT_NUMSTAT |
3264                                       DIFF_FORMAT_DIFFSTAT |
3265                                       DIFF_FORMAT_SHORTSTAT |
3266                                       DIFF_FORMAT_DIRSTAT |
3267                                       DIFF_FORMAT_SUMMARY |
3268                                       DIFF_FORMAT_CHECKDIFF))
3269                 DIFF_OPT_SET(options, RECURSIVE);
3270         /*
3271          * Also pickaxe would not work very well if you do not say recursive
3272          */
3273         if (options->pickaxe)
3274                 DIFF_OPT_SET(options, RECURSIVE);
3275         /*
3276          * When patches are generated, submodules diffed against the work tree
3277          * must be checked for dirtiness too so it can be shown in the output
3278          */
3279         if (options->output_format & DIFF_FORMAT_PATCH)
3280                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3281
3282         if (options->detect_rename && options->rename_limit < 0)
3283                 options->rename_limit = diff_rename_limit_default;
3284         if (options->setup & DIFF_SETUP_USE_CACHE) {
3285                 if (!active_cache)
3286                         /* read-cache does not die even when it fails
3287                          * so it is safe for us to do this here.  Also
3288                          * it does not smudge active_cache or active_nr
3289                          * when it fails, so we do not have to worry about
3290                          * cleaning it up ourselves either.
3291                          */
3292                         read_cache();
3293         }
3294         if (options->abbrev <= 0 || 40 < options->abbrev)
3295                 options->abbrev = 40; /* full */
3296
3297         /*
3298          * It does not make sense to show the first hit we happened
3299          * to have found.  It does not make sense not to return with
3300          * exit code in such a case either.
3301          */
3302         if (DIFF_OPT_TST(options, QUICK)) {
3303                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3304                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3305         }
3306
3307         return 0;
3308 }
3309
3310 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3311 {
3312         char c, *eq;
3313         int len;
3314
3315         if (*arg != '-')
3316                 return 0;
3317         c = *++arg;
3318         if (!c)
3319                 return 0;
3320         if (c == arg_short) {
3321                 c = *++arg;
3322                 if (!c)
3323                         return 1;
3324                 if (val && isdigit(c)) {
3325                         char *end;
3326                         int n = strtoul(arg, &end, 10);
3327                         if (*end)
3328                                 return 0;
3329                         *val = n;
3330                         return 1;
3331                 }
3332                 return 0;
3333         }
3334         if (c != '-')
3335                 return 0;
3336         arg++;
3337         eq = strchr(arg, '=');
3338         if (eq)
3339                 len = eq - arg;
3340         else
3341                 len = strlen(arg);
3342         if (!len || strncmp(arg, arg_long, len))
3343                 return 0;
3344         if (eq) {
3345                 int n;
3346                 char *end;
3347                 if (!isdigit(*++eq))
3348                         return 0;
3349                 n = strtoul(eq, &end, 10);
3350                 if (*end)
3351                         return 0;
3352                 *val = n;
3353         }
3354         return 1;
3355 }
3356
3357 static int diff_scoreopt_parse(const char *opt);
3358
3359 static inline int short_opt(char opt, const char **argv,
3360                             const char **optarg)
3361 {
3362         const char *arg = argv[0];
3363         if (arg[0] != '-' || arg[1] != opt)
3364                 return 0;
3365         if (arg[2] != '\0') {
3366                 *optarg = arg + 2;
3367                 return 1;
3368         }
3369         if (!argv[1])
3370                 die("Option '%c' requires a value", opt);
3371         *optarg = argv[1];
3372         return 2;
3373 }
3374
3375 int parse_long_opt(const char *opt, const char **argv,
3376                    const char **optarg)
3377 {
3378         const char *arg = argv[0];
3379         if (arg[0] != '-' || arg[1] != '-')
3380                 return 0;
3381         arg += strlen("--");
3382         if (prefixcmp(arg, opt))
3383                 return 0;
3384         arg += strlen(opt);
3385         if (*arg == '=') { /* sticked form: --option=value */
3386                 *optarg = arg + 1;
3387                 return 1;
3388         }
3389         if (*arg != '\0')
3390                 return 0;
3391         /* separate form: --option value */
3392         if (!argv[1])
3393                 die("Option '--%s' requires a value", opt);
3394         *optarg = argv[1];
3395         return 2;
3396 }
3397
3398 static int stat_opt(struct diff_options *options, const char **av)
3399 {
3400         const char *arg = av[0];
3401         char *end;
3402         int width = options->stat_width;
3403         int name_width = options->stat_name_width;
3404         int graph_width = options->stat_graph_width;
3405         int count = options->stat_count;
3406         int argcount = 1;
3407
3408         arg += strlen("--stat");
3409         end = (char *)arg;
3410
3411         switch (*arg) {
3412         case '-':
3413                 if (!prefixcmp(arg, "-width")) {
3414                         arg += strlen("-width");
3415                         if (*arg == '=')
3416                                 width = strtoul(arg + 1, &end, 10);
3417                         else if (!*arg && !av[1])
3418                                 die("Option '--stat-width' requires a value");
3419                         else if (!*arg) {
3420                                 width = strtoul(av[1], &end, 10);
3421                                 argcount = 2;
3422                         }
3423                 } else if (!prefixcmp(arg, "-name-width")) {
3424                         arg += strlen("-name-width");
3425                         if (*arg == '=')
3426                                 name_width = strtoul(arg + 1, &end, 10);
3427                         else if (!*arg && !av[1])
3428                                 die("Option '--stat-name-width' requires a value");
3429                         else if (!*arg) {
3430                                 name_width = strtoul(av[1], &end, 10);
3431                                 argcount = 2;
3432                         }
3433                 } else if (!prefixcmp(arg, "-graph-width")) {
3434                         arg += strlen("-graph-width");
3435                         if (*arg == '=')
3436                                 graph_width = strtoul(arg + 1, &end, 10);
3437                         else if (!*arg && !av[1])
3438                                 die("Option '--stat-graph-width' requires a value");
3439                         else if (!*arg) {
3440                                 graph_width = strtoul(av[1], &end, 10);
3441                                 argcount = 2;
3442                         }
3443                 } else if (!prefixcmp(arg, "-count")) {
3444                         arg += strlen("-count");
3445                         if (*arg == '=')
3446                                 count = strtoul(arg + 1, &end, 10);
3447                         else if (!*arg && !av[1])
3448                                 die("Option '--stat-count' requires a value");
3449                         else if (!*arg) {
3450                                 count = strtoul(av[1], &end, 10);
3451                                 argcount = 2;
3452                         }
3453                 }
3454                 break;
3455         case '=':
3456                 width = strtoul(arg+1, &end, 10);
3457                 if (*end == ',')
3458                         name_width = strtoul(end+1, &end, 10);
3459                 if (*end == ',')
3460                         count = strtoul(end+1, &end, 10);
3461         }
3462
3463         /* Important! This checks all the error cases! */
3464         if (*end)
3465                 return 0;
3466         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3467         options->stat_name_width = name_width;
3468         options->stat_graph_width = graph_width;
3469         options->stat_width = width;
3470         options->stat_count = count;
3471         return argcount;
3472 }
3473
3474 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3475 {
3476         struct strbuf errmsg = STRBUF_INIT;
3477         if (parse_dirstat_params(options, params, &errmsg))
3478                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3479                     errmsg.buf);
3480         strbuf_release(&errmsg);
3481         /*
3482          * The caller knows a dirstat-related option is given from the command
3483          * line; allow it to say "return this_function();"
3484          */
3485         options->output_format |= DIFF_FORMAT_DIRSTAT;
3486         return 1;
3487 }
3488
3489 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3490 {
3491         const char *arg = av[0];
3492         const char *optarg;
3493         int argcount;
3494
3495         /* Output format options */
3496         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch"))
3497                 options->output_format |= DIFF_FORMAT_PATCH;
3498         else if (opt_arg(arg, 'U', "unified", &options->context))
3499                 options->output_format |= DIFF_FORMAT_PATCH;
3500         else if (!strcmp(arg, "--raw"))
3501                 options->output_format |= DIFF_FORMAT_RAW;
3502         else if (!strcmp(arg, "--patch-with-raw"))
3503                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
3504         else if (!strcmp(arg, "--numstat"))
3505                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3506         else if (!strcmp(arg, "--shortstat"))
3507                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3508         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3509                 return parse_dirstat_opt(options, "");
3510         else if (!prefixcmp(arg, "-X"))
3511                 return parse_dirstat_opt(options, arg + 2);
3512         else if (!prefixcmp(arg, "--dirstat="))
3513                 return parse_dirstat_opt(options, arg + 10);
3514         else if (!strcmp(arg, "--cumulative"))
3515                 return parse_dirstat_opt(options, "cumulative");
3516         else if (!strcmp(arg, "--dirstat-by-file"))
3517                 return parse_dirstat_opt(options, "files");
3518         else if (!prefixcmp(arg, "--dirstat-by-file=")) {
3519                 parse_dirstat_opt(options, "files");
3520                 return parse_dirstat_opt(options, arg + 18);
3521         }
3522         else if (!strcmp(arg, "--check"))
3523                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3524         else if (!strcmp(arg, "--summary"))
3525                 options->output_format |= DIFF_FORMAT_SUMMARY;
3526         else if (!strcmp(arg, "--patch-with-stat"))
3527                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
3528         else if (!strcmp(arg, "--name-only"))
3529                 options->output_format |= DIFF_FORMAT_NAME;
3530         else if (!strcmp(arg, "--name-status"))
3531                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3532         else if (!strcmp(arg, "-s"))
3533                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3534         else if (!prefixcmp(arg, "--stat"))
3535                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3536                 return stat_opt(options, av);
3537
3538         /* renames options */
3539         else if (!prefixcmp(arg, "-B") || !prefixcmp(arg, "--break-rewrites=") ||
3540                  !strcmp(arg, "--break-rewrites")) {
3541                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3542                         return error("invalid argument to -B: %s", arg+2);
3543         }
3544         else if (!prefixcmp(arg, "-M") || !prefixcmp(arg, "--find-renames=") ||
3545                  !strcmp(arg, "--find-renames")) {
3546                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3547                         return error("invalid argument to -M: %s", arg+2);
3548                 options->detect_rename = DIFF_DETECT_RENAME;
3549         }
3550         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3551                 options->irreversible_delete = 1;
3552         }
3553         else if (!prefixcmp(arg, "-C") || !prefixcmp(arg, "--find-copies=") ||
3554                  !strcmp(arg, "--find-copies")) {
3555                 if (options->detect_rename == DIFF_DETECT_COPY)
3556                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3557                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3558                         return error("invalid argument to -C: %s", arg+2);
3559                 options->detect_rename = DIFF_DETECT_COPY;
3560         }
3561         else if (!strcmp(arg, "--no-renames"))
3562                 options->detect_rename = 0;
3563         else if (!strcmp(arg, "--rename-empty"))
3564                 DIFF_OPT_SET(options, RENAME_EMPTY);
3565         else if (!strcmp(arg, "--no-rename-empty"))
3566                 DIFF_OPT_CLR(options, RENAME_EMPTY);
3567         else if (!strcmp(arg, "--relative"))
3568                 DIFF_OPT_SET(options, RELATIVE_NAME);
3569         else if (!prefixcmp(arg, "--relative=")) {
3570                 DIFF_OPT_SET(options, RELATIVE_NAME);
3571                 options->prefix = arg + 11;
3572         }
3573
3574         /* xdiff options */
3575         else if (!strcmp(arg, "--minimal"))
3576                 DIFF_XDL_SET(options, NEED_MINIMAL);
3577         else if (!strcmp(arg, "--no-minimal"))
3578                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3579         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3580                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3581         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3582                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3583         else if (!strcmp(arg, "--ignore-space-at-eol"))
3584                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3585         else if (!strcmp(arg, "--patience"))
3586                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3587         else if (!strcmp(arg, "--histogram"))
3588                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3589
3590         /* flags options */
3591         else if (!strcmp(arg, "--binary")) {
3592                 options->output_format |= DIFF_FORMAT_PATCH;
3593                 DIFF_OPT_SET(options, BINARY);
3594         }
3595         else if (!strcmp(arg, "--full-index"))
3596                 DIFF_OPT_SET(options, FULL_INDEX);
3597         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3598                 DIFF_OPT_SET(options, TEXT);
3599         else if (!strcmp(arg, "-R"))
3600                 DIFF_OPT_SET(options, REVERSE_DIFF);
3601         else if (!strcmp(arg, "--find-copies-harder"))
3602                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3603         else if (!strcmp(arg, "--follow"))
3604                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3605         else if (!strcmp(arg, "--color"))
3606                 options->use_color = 1;
3607         else if (!prefixcmp(arg, "--color=")) {
3608                 int value = git_config_colorbool(NULL, arg+8);
3609                 if (value < 0)
3610                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3611                 options->use_color = value;
3612         }
3613         else if (!strcmp(arg, "--no-color"))
3614                 options->use_color = 0;
3615         else if (!strcmp(arg, "--color-words")) {
3616                 options->use_color = 1;
3617                 options->word_diff = DIFF_WORDS_COLOR;
3618         }
3619         else if (!prefixcmp(arg, "--color-words=")) {
3620                 options->use_color = 1;
3621                 options->word_diff = DIFF_WORDS_COLOR;
3622                 options->word_regex = arg + 14;
3623         }
3624         else if (!strcmp(arg, "--word-diff")) {
3625                 if (options->word_diff == DIFF_WORDS_NONE)
3626                         options->word_diff = DIFF_WORDS_PLAIN;
3627         }
3628         else if (!prefixcmp(arg, "--word-diff=")) {
3629                 const char *type = arg + 12;
3630                 if (!strcmp(type, "plain"))
3631                         options->word_diff = DIFF_WORDS_PLAIN;
3632                 else if (!strcmp(type, "color")) {
3633                         options->use_color = 1;
3634                         options->word_diff = DIFF_WORDS_COLOR;
3635                 }
3636                 else if (!strcmp(type, "porcelain"))
3637                         options->word_diff = DIFF_WORDS_PORCELAIN;
3638                 else if (!strcmp(type, "none"))
3639                         options->word_diff = DIFF_WORDS_NONE;
3640                 else
3641                         die("bad --word-diff argument: %s", type);
3642         }
3643         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3644                 if (options->word_diff == DIFF_WORDS_NONE)
3645                         options->word_diff = DIFF_WORDS_PLAIN;
3646                 options->word_regex = optarg;
3647                 return argcount;
3648         }
3649         else if (!strcmp(arg, "--exit-code"))
3650                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3651         else if (!strcmp(arg, "--quiet"))
3652                 DIFF_OPT_SET(options, QUICK);
3653         else if (!strcmp(arg, "--ext-diff"))
3654                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3655         else if (!strcmp(arg, "--no-ext-diff"))
3656                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3657         else if (!strcmp(arg, "--textconv"))
3658                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3659         else if (!strcmp(arg, "--no-textconv"))
3660                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3661         else if (!strcmp(arg, "--ignore-submodules")) {
3662                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3663                 handle_ignore_submodules_arg(options, "all");
3664         } else if (!prefixcmp(arg, "--ignore-submodules=")) {
3665                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3666                 handle_ignore_submodules_arg(options, arg + 20);
3667         } else if (!strcmp(arg, "--submodule"))
3668                 DIFF_OPT_SET(options, SUBMODULE_LOG);
3669         else if (!prefixcmp(arg, "--submodule=")) {
3670                 if (!strcmp(arg + 12, "log"))
3671                         DIFF_OPT_SET(options, SUBMODULE_LOG);
3672         }
3673
3674         /* misc options */
3675         else if (!strcmp(arg, "-z"))
3676                 options->line_termination = 0;
3677         else if ((argcount = short_opt('l', av, &optarg))) {
3678                 options->rename_limit = strtoul(optarg, NULL, 10);
3679                 return argcount;
3680         }
3681         else if ((argcount = short_opt('S', av, &optarg))) {
3682                 options->pickaxe = optarg;
3683                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3684                 return argcount;
3685         } else if ((argcount = short_opt('G', av, &optarg))) {
3686                 options->pickaxe = optarg;
3687                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3688                 return argcount;
3689         }
3690         else if (!strcmp(arg, "--pickaxe-all"))
3691                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3692         else if (!strcmp(arg, "--pickaxe-regex"))
3693                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3694         else if ((argcount = short_opt('O', av, &optarg))) {
3695                 options->orderfile = optarg;
3696                 return argcount;
3697         }
3698         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3699                 options->filter = optarg;
3700                 return argcount;
3701         }
3702         else if (!strcmp(arg, "--abbrev"))
3703                 options->abbrev = DEFAULT_ABBREV;
3704         else if (!prefixcmp(arg, "--abbrev=")) {
3705                 options->abbrev = strtoul(arg + 9, NULL, 10);
3706                 if (options->abbrev < MINIMUM_ABBREV)
3707                         options->abbrev = MINIMUM_ABBREV;
3708                 else if (40 < options->abbrev)
3709                         options->abbrev = 40;
3710         }
3711         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3712                 options->a_prefix = optarg;
3713                 return argcount;
3714         }
3715         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3716                 options->b_prefix = optarg;
3717                 return argcount;
3718         }
3719         else if (!strcmp(arg, "--no-prefix"))
3720                 options->a_prefix = options->b_prefix = "";
3721         else if (opt_arg(arg, '\0', "inter-hunk-context",
3722                          &options->interhunkcontext))
3723                 ;
3724         else if (!strcmp(arg, "-W"))
3725                 DIFF_OPT_SET(options, FUNCCONTEXT);
3726         else if (!strcmp(arg, "--function-context"))
3727                 DIFF_OPT_SET(options, FUNCCONTEXT);
3728         else if (!strcmp(arg, "--no-function-context"))
3729                 DIFF_OPT_CLR(options, FUNCCONTEXT);
3730         else if ((argcount = parse_long_opt("output", av, &optarg))) {
3731                 options->file = fopen(optarg, "w");
3732                 if (!options->file)
3733                         die_errno("Could not open '%s'", optarg);
3734                 options->close_file = 1;
3735                 return argcount;
3736         } else
3737                 return 0;
3738         return 1;
3739 }
3740
3741 int parse_rename_score(const char **cp_p)
3742 {
3743         unsigned long num, scale;
3744         int ch, dot;
3745         const char *cp = *cp_p;
3746
3747         num = 0;
3748         scale = 1;
3749         dot = 0;
3750         for (;;) {
3751                 ch = *cp;
3752                 if ( !dot && ch == '.' ) {
3753                         scale = 1;
3754                         dot = 1;
3755                 } else if ( ch == '%' ) {
3756                         scale = dot ? scale*100 : 100;
3757                         cp++;   /* % is always at the end */
3758                         break;
3759                 } else if ( ch >= '0' && ch <= '9' ) {
3760                         if ( scale < 100000 ) {
3761                                 scale *= 10;
3762                                 num = (num*10) + (ch-'0');
3763                         }
3764                 } else {
3765                         break;
3766                 }
3767                 cp++;
3768         }
3769         *cp_p = cp;
3770
3771         /* user says num divided by scale and we say internally that
3772          * is MAX_SCORE * num / scale.
3773          */
3774         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
3775 }
3776
3777 static int diff_scoreopt_parse(const char *opt)
3778 {
3779         int opt1, opt2, cmd;
3780
3781         if (*opt++ != '-')
3782                 return -1;
3783         cmd = *opt++;
3784         if (cmd == '-') {
3785                 /* convert the long-form arguments into short-form versions */
3786                 if (!prefixcmp(opt, "break-rewrites")) {
3787                         opt += strlen("break-rewrites");
3788                         if (*opt == 0 || *opt++ == '=')
3789                                 cmd = 'B';
3790                 } else if (!prefixcmp(opt, "find-copies")) {
3791                         opt += strlen("find-copies");
3792                         if (*opt == 0 || *opt++ == '=')
3793                                 cmd = 'C';
3794                 } else if (!prefixcmp(opt, "find-renames")) {
3795                         opt += strlen("find-renames");
3796                         if (*opt == 0 || *opt++ == '=')
3797                                 cmd = 'M';
3798                 }
3799         }
3800         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
3801                 return -1; /* that is not a -M, -C nor -B option */
3802
3803         opt1 = parse_rename_score(&opt);
3804         if (cmd != 'B')
3805                 opt2 = 0;
3806         else {
3807                 if (*opt == 0)
3808                         opt2 = 0;
3809                 else if (*opt != '/')
3810                         return -1; /* we expect -B80/99 or -B80 */
3811                 else {
3812                         opt++;
3813                         opt2 = parse_rename_score(&opt);
3814                 }
3815         }
3816         if (*opt != 0)
3817                 return -1;
3818         return opt1 | (opt2 << 16);
3819 }
3820
3821 struct diff_queue_struct diff_queued_diff;
3822
3823 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
3824 {
3825         if (queue->alloc <= queue->nr) {
3826                 queue->alloc = alloc_nr(queue->alloc);
3827                 queue->queue = xrealloc(queue->queue,
3828                                         sizeof(dp) * queue->alloc);
3829         }
3830         queue->queue[queue->nr++] = dp;
3831 }
3832
3833 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
3834                                  struct diff_filespec *one,
3835                                  struct diff_filespec *two)
3836 {
3837         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3838         dp->one = one;
3839         dp->two = two;
3840         if (queue)
3841                 diff_q(queue, dp);
3842         return dp;
3843 }
3844
3845 void diff_free_filepair(struct diff_filepair *p)
3846 {
3847         free_filespec(p->one);
3848         free_filespec(p->two);
3849         free(p);
3850 }
3851
3852 /* This is different from find_unique_abbrev() in that
3853  * it stuffs the result with dots for alignment.
3854  */
3855 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3856 {
3857         int abblen;
3858         const char *abbrev;
3859         if (len == 40)
3860                 return sha1_to_hex(sha1);
3861
3862         abbrev = find_unique_abbrev(sha1, len);
3863         abblen = strlen(abbrev);
3864         if (abblen < 37) {
3865                 static char hex[41];
3866                 if (len < abblen && abblen <= len + 2)
3867                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3868                 else
3869                         sprintf(hex, "%s...", abbrev);
3870                 return hex;
3871         }
3872         return sha1_to_hex(sha1);
3873 }
3874
3875 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3876 {
3877         int line_termination = opt->line_termination;
3878         int inter_name_termination = line_termination ? '\t' : '\0';
3879         if (opt->output_prefix) {
3880                 struct strbuf *msg = NULL;
3881                 msg = opt->output_prefix(opt, opt->output_prefix_data);
3882                 fprintf(opt->file, "%s", msg->buf);
3883         }
3884
3885         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3886                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3887                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
3888                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3889         }
3890         if (p->score) {
3891                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3892                         inter_name_termination);
3893         } else {
3894                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
3895         }
3896
3897         if (p->status == DIFF_STATUS_COPIED ||
3898             p->status == DIFF_STATUS_RENAMED) {
3899                 const char *name_a, *name_b;
3900                 name_a = p->one->path;
3901                 name_b = p->two->path;
3902                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3903                 write_name_quoted(name_a, opt->file, inter_name_termination);
3904                 write_name_quoted(name_b, opt->file, line_termination);
3905         } else {
3906                 const char *name_a, *name_b;
3907                 name_a = p->one->mode ? p->one->path : p->two->path;
3908                 name_b = NULL;
3909                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3910                 write_name_quoted(name_a, opt->file, line_termination);
3911         }
3912 }
3913
3914 int diff_unmodified_pair(struct diff_filepair *p)
3915 {
3916         /* This function is written stricter than necessary to support
3917          * the currently implemented transformers, but the idea is to
3918          * let transformers to produce diff_filepairs any way they want,
3919          * and filter and clean them up here before producing the output.
3920          */
3921         struct diff_filespec *one = p->one, *two = p->two;
3922
3923         if (DIFF_PAIR_UNMERGED(p))
3924                 return 0; /* unmerged is interesting */
3925
3926         /* deletion, addition, mode or type change
3927          * and rename are all interesting.
3928          */
3929         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
3930             DIFF_PAIR_MODE_CHANGED(p) ||
3931             strcmp(one->path, two->path))
3932                 return 0;
3933
3934         /* both are valid and point at the same path.  that is, we are
3935          * dealing with a change.
3936          */
3937         if (one->sha1_valid && two->sha1_valid &&
3938             !hashcmp(one->sha1, two->sha1) &&
3939             !one->dirty_submodule && !two->dirty_submodule)
3940                 return 1; /* no change */
3941         if (!one->sha1_valid && !two->sha1_valid)
3942                 return 1; /* both look at the same file on the filesystem. */
3943         return 0;
3944 }
3945
3946 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
3947 {
3948         if (diff_unmodified_pair(p))
3949                 return;
3950
3951         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3952             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3953                 return; /* no tree diffs in patch format */
3954
3955         run_diff(p, o);
3956 }
3957
3958 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
3959                             struct diffstat_t *diffstat)
3960 {
3961         if (diff_unmodified_pair(p))
3962                 return;
3963
3964         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3965             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3966                 return; /* no useful stat for tree diffs */
3967
3968         run_diffstat(p, o, diffstat);
3969 }
3970
3971 static void diff_flush_checkdiff(struct diff_filepair *p,
3972                 struct diff_options *o)
3973 {
3974         if (diff_unmodified_pair(p))
3975                 return;
3976
3977         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3978             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3979                 return; /* nothing to check in tree diffs */
3980
3981         run_checkdiff(p, o);
3982 }
3983
3984 int diff_queue_is_empty(void)
3985 {
3986         struct diff_queue_struct *q = &diff_queued_diff;
3987         int i;
3988         for (i = 0; i < q->nr; i++)
3989                 if (!diff_unmodified_pair(q->queue[i]))
3990                         return 0;
3991         return 1;
3992 }
3993
3994 #if DIFF_DEBUG
3995 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
3996 {
3997         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
3998                 x, one ? one : "",
3999                 s->path,
4000                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
4001                 s->mode,
4002                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
4003         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
4004                 x, one ? one : "",
4005                 s->size, s->xfrm_flags);
4006 }
4007
4008 void diff_debug_filepair(const struct diff_filepair *p, int i)
4009 {
4010         diff_debug_filespec(p->one, i, "one");
4011         diff_debug_filespec(p->two, i, "two");
4012         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4013                 p->score, p->status ? p->status : '?',
4014                 p->one->rename_used, p->broken_pair);
4015 }
4016
4017 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4018 {
4019         int i;
4020         if (msg)
4021                 fprintf(stderr, "%s\n", msg);
4022         fprintf(stderr, "q->nr = %d\n", q->nr);
4023         for (i = 0; i < q->nr; i++) {
4024                 struct diff_filepair *p = q->queue[i];
4025                 diff_debug_filepair(p, i);
4026         }
4027 }
4028 #endif
4029
4030 static void diff_resolve_rename_copy(void)
4031 {
4032         int i;
4033         struct diff_filepair *p;
4034         struct diff_queue_struct *q = &diff_queued_diff;
4035
4036         diff_debug_queue("resolve-rename-copy", q);
4037
4038         for (i = 0; i < q->nr; i++) {
4039                 p = q->queue[i];
4040                 p->status = 0; /* undecided */
4041                 if (DIFF_PAIR_UNMERGED(p))
4042                         p->status = DIFF_STATUS_UNMERGED;
4043                 else if (!DIFF_FILE_VALID(p->one))
4044                         p->status = DIFF_STATUS_ADDED;
4045                 else if (!DIFF_FILE_VALID(p->two))
4046                         p->status = DIFF_STATUS_DELETED;
4047                 else if (DIFF_PAIR_TYPE_CHANGED(p))
4048                         p->status = DIFF_STATUS_TYPE_CHANGED;
4049
4050                 /* from this point on, we are dealing with a pair
4051                  * whose both sides are valid and of the same type, i.e.
4052                  * either in-place edit or rename/copy edit.
4053                  */
4054                 else if (DIFF_PAIR_RENAME(p)) {
4055                         /*
4056                          * A rename might have re-connected a broken
4057                          * pair up, causing the pathnames to be the
4058                          * same again. If so, that's not a rename at
4059                          * all, just a modification..
4060                          *
4061                          * Otherwise, see if this source was used for
4062                          * multiple renames, in which case we decrement
4063                          * the count, and call it a copy.
4064                          */
4065                         if (!strcmp(p->one->path, p->two->path))
4066                                 p->status = DIFF_STATUS_MODIFIED;
4067                         else if (--p->one->rename_used > 0)
4068                                 p->status = DIFF_STATUS_COPIED;
4069                         else
4070                                 p->status = DIFF_STATUS_RENAMED;
4071                 }
4072                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
4073                          p->one->mode != p->two->mode ||
4074                          p->one->dirty_submodule ||
4075                          p->two->dirty_submodule ||
4076                          is_null_sha1(p->one->sha1))
4077                         p->status = DIFF_STATUS_MODIFIED;
4078                 else {
4079                         /* This is a "no-change" entry and should not
4080                          * happen anymore, but prepare for broken callers.
4081                          */
4082                         error("feeding unmodified %s to diffcore",
4083                               p->one->path);
4084                         p->status = DIFF_STATUS_UNKNOWN;
4085                 }
4086         }
4087         diff_debug_queue("resolve-rename-copy done", q);
4088 }
4089
4090 static int check_pair_status(struct diff_filepair *p)
4091 {
4092         switch (p->status) {
4093         case DIFF_STATUS_UNKNOWN:
4094                 return 0;
4095         case 0:
4096                 die("internal error in diff-resolve-rename-copy");
4097         default:
4098                 return 1;
4099         }
4100 }
4101
4102 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4103 {
4104         int fmt = opt->output_format;
4105
4106         if (fmt & DIFF_FORMAT_CHECKDIFF)
4107                 diff_flush_checkdiff(p, opt);
4108         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4109                 diff_flush_raw(p, opt);
4110         else if (fmt & DIFF_FORMAT_NAME) {
4111                 const char *name_a, *name_b;
4112                 name_a = p->two->path;
4113                 name_b = NULL;
4114                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4115                 write_name_quoted(name_a, opt->file, opt->line_termination);
4116         }
4117 }
4118
4119 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4120 {
4121         if (fs->mode)
4122                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4123         else
4124                 fprintf(file, " %s ", newdelete);
4125         write_name_quoted(fs->path, file, '\n');
4126 }
4127
4128
4129 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4130                 const char *line_prefix)
4131 {
4132         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4133                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4134                         p->two->mode, show_name ? ' ' : '\n');
4135                 if (show_name) {
4136                         write_name_quoted(p->two->path, file, '\n');
4137                 }
4138         }
4139 }
4140
4141 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4142                         const char *line_prefix)
4143 {
4144         char *names = pprint_rename(p->one->path, p->two->path);
4145
4146         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4147         free(names);
4148         show_mode_change(file, p, 0, line_prefix);
4149 }
4150
4151 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4152 {
4153         FILE *file = opt->file;
4154         char *line_prefix = "";
4155
4156         if (opt->output_prefix) {
4157                 struct strbuf *buf = opt->output_prefix(opt, opt->output_prefix_data);
4158                 line_prefix = buf->buf;
4159         }
4160
4161         switch(p->status) {
4162         case DIFF_STATUS_DELETED:
4163                 fputs(line_prefix, file);
4164                 show_file_mode_name(file, "delete", p->one);
4165                 break;
4166         case DIFF_STATUS_ADDED:
4167                 fputs(line_prefix, file);
4168                 show_file_mode_name(file, "create", p->two);
4169                 break;
4170         case DIFF_STATUS_COPIED:
4171                 fputs(line_prefix, file);
4172                 show_rename_copy(file, "copy", p, line_prefix);
4173                 break;
4174         case DIFF_STATUS_RENAMED:
4175                 fputs(line_prefix, file);
4176                 show_rename_copy(file, "rename", p, line_prefix);
4177                 break;
4178         default:
4179                 if (p->score) {
4180                         fprintf(file, "%s rewrite ", line_prefix);
4181                         write_name_quoted(p->two->path, file, ' ');
4182                         fprintf(file, "(%d%%)\n", similarity_index(p));
4183                 }
4184                 show_mode_change(file, p, !p->score, line_prefix);
4185                 break;
4186         }
4187 }
4188
4189 struct patch_id_t {
4190         git_SHA_CTX *ctx;
4191         int patchlen;
4192 };
4193
4194 static int remove_space(char *line, int len)
4195 {
4196         int i;
4197         char *dst = line;
4198         unsigned char c;
4199
4200         for (i = 0; i < len; i++)
4201                 if (!isspace((c = line[i])))
4202                         *dst++ = c;
4203
4204         return dst - line;
4205 }
4206
4207 static void patch_id_consume(void *priv, char *line, unsigned long len)
4208 {
4209         struct patch_id_t *data = priv;
4210         int new_len;
4211
4212         /* Ignore line numbers when computing the SHA1 of the patch */
4213         if (!prefixcmp(line, "@@ -"))
4214                 return;
4215
4216         new_len = remove_space(line, len);
4217
4218         git_SHA1_Update(data->ctx, line, new_len);
4219         data->patchlen += new_len;
4220 }
4221
4222 /* returns 0 upon success, and writes result into sha1 */
4223 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
4224 {
4225         struct diff_queue_struct *q = &diff_queued_diff;
4226         int i;
4227         git_SHA_CTX ctx;
4228         struct patch_id_t data;
4229         char buffer[PATH_MAX * 4 + 20];
4230
4231         git_SHA1_Init(&ctx);
4232         memset(&data, 0, sizeof(struct patch_id_t));
4233         data.ctx = &ctx;
4234
4235         for (i = 0; i < q->nr; i++) {
4236                 xpparam_t xpp;
4237                 xdemitconf_t xecfg;
4238                 mmfile_t mf1, mf2;
4239                 struct diff_filepair *p = q->queue[i];
4240                 int len1, len2;
4241
4242                 memset(&xpp, 0, sizeof(xpp));
4243                 memset(&xecfg, 0, sizeof(xecfg));
4244                 if (p->status == 0)
4245                         return error("internal diff status error");
4246                 if (p->status == DIFF_STATUS_UNKNOWN)
4247                         continue;
4248                 if (diff_unmodified_pair(p))
4249                         continue;
4250                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4251                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4252                         continue;
4253                 if (DIFF_PAIR_UNMERGED(p))
4254                         continue;
4255
4256                 diff_fill_sha1_info(p->one);
4257                 diff_fill_sha1_info(p->two);
4258                 if (fill_mmfile(&mf1, p->one) < 0 ||
4259                                 fill_mmfile(&mf2, p->two) < 0)
4260                         return error("unable to read files to diff");
4261
4262                 len1 = remove_space(p->one->path, strlen(p->one->path));
4263                 len2 = remove_space(p->two->path, strlen(p->two->path));
4264                 if (p->one->mode == 0)
4265                         len1 = snprintf(buffer, sizeof(buffer),
4266                                         "diff--gita/%.*sb/%.*s"
4267                                         "newfilemode%06o"
4268                                         "---/dev/null"
4269                                         "+++b/%.*s",
4270                                         len1, p->one->path,
4271                                         len2, p->two->path,
4272                                         p->two->mode,
4273                                         len2, p->two->path);
4274                 else if (p->two->mode == 0)
4275                         len1 = snprintf(buffer, sizeof(buffer),
4276                                         "diff--gita/%.*sb/%.*s"
4277                                         "deletedfilemode%06o"
4278                                         "---a/%.*s"
4279                                         "+++/dev/null",
4280                                         len1, p->one->path,
4281                                         len2, p->two->path,
4282                                         p->one->mode,
4283                                         len1, p->one->path);
4284                 else
4285                         len1 = snprintf(buffer, sizeof(buffer),
4286                                         "diff--gita/%.*sb/%.*s"
4287                                         "---a/%.*s"
4288                                         "+++b/%.*s",
4289                                         len1, p->one->path,
4290                                         len2, p->two->path,
4291                                         len1, p->one->path,
4292                                         len2, p->two->path);
4293                 git_SHA1_Update(&ctx, buffer, len1);
4294
4295                 if (diff_filespec_is_binary(p->one) ||
4296                     diff_filespec_is_binary(p->two)) {
4297                         git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
4298                         git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
4299                         continue;
4300                 }
4301
4302                 xpp.flags = 0;
4303                 xecfg.ctxlen = 3;
4304                 xecfg.flags = 0;
4305                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4306                               &xpp, &xecfg);
4307         }
4308
4309         git_SHA1_Final(sha1, &ctx);
4310         return 0;
4311 }
4312
4313 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
4314 {
4315         struct diff_queue_struct *q = &diff_queued_diff;
4316         int i;
4317         int result = diff_get_patch_id(options, sha1);
4318
4319         for (i = 0; i < q->nr; i++)
4320                 diff_free_filepair(q->queue[i]);
4321
4322         free(q->queue);
4323         DIFF_QUEUE_CLEAR(q);
4324
4325         return result;
4326 }
4327
4328 static int is_summary_empty(const struct diff_queue_struct *q)
4329 {
4330         int i;
4331
4332         for (i = 0; i < q->nr; i++) {
4333                 const struct diff_filepair *p = q->queue[i];
4334
4335                 switch (p->status) {
4336                 case DIFF_STATUS_DELETED:
4337                 case DIFF_STATUS_ADDED:
4338                 case DIFF_STATUS_COPIED:
4339                 case DIFF_STATUS_RENAMED:
4340                         return 0;
4341                 default:
4342                         if (p->score)
4343                                 return 0;
4344                         if (p->one->mode && p->two->mode &&
4345                             p->one->mode != p->two->mode)
4346                                 return 0;
4347                         break;
4348                 }
4349         }
4350         return 1;
4351 }
4352
4353 static const char rename_limit_warning[] =
4354 "inexact rename detection was skipped due to too many files.";
4355
4356 static const char degrade_cc_to_c_warning[] =
4357 "only found copies from modified paths due to too many files.";
4358
4359 static const char rename_limit_advice[] =
4360 "you may want to set your %s variable to at least "
4361 "%d and retry the command.";
4362
4363 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4364 {
4365         if (degraded_cc)
4366                 warning(degrade_cc_to_c_warning);
4367         else if (needed)
4368                 warning(rename_limit_warning);
4369         else
4370                 return;
4371         if (0 < needed && needed < 32767)
4372                 warning(rename_limit_advice, varname, needed);
4373 }
4374
4375 void diff_flush(struct diff_options *options)
4376 {
4377         struct diff_queue_struct *q = &diff_queued_diff;
4378         int i, output_format = options->output_format;
4379         int separator = 0;
4380         int dirstat_by_line = 0;
4381
4382         /*
4383          * Order: raw, stat, summary, patch
4384          * or:    name/name-status/checkdiff (other bits clear)
4385          */
4386         if (!q->nr)
4387                 goto free_queue;
4388
4389         if (output_format & (DIFF_FORMAT_RAW |
4390                              DIFF_FORMAT_NAME |
4391                              DIFF_FORMAT_NAME_STATUS |
4392                              DIFF_FORMAT_CHECKDIFF)) {
4393                 for (i = 0; i < q->nr; i++) {
4394                         struct diff_filepair *p = q->queue[i];
4395                         if (check_pair_status(p))
4396                                 flush_one_pair(p, options);
4397                 }
4398                 separator++;
4399         }
4400
4401         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4402                 dirstat_by_line = 1;
4403
4404         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4405             dirstat_by_line) {
4406                 struct diffstat_t diffstat;
4407
4408                 memset(&diffstat, 0, sizeof(struct diffstat_t));
4409                 for (i = 0; i < q->nr; i++) {
4410                         struct diff_filepair *p = q->queue[i];
4411                         if (check_pair_status(p))
4412                                 diff_flush_stat(p, options, &diffstat);
4413                 }
4414                 if (output_format & DIFF_FORMAT_NUMSTAT)
4415                         show_numstat(&diffstat, options);
4416                 if (output_format & DIFF_FORMAT_DIFFSTAT)
4417                         show_stats(&diffstat, options);
4418                 if (output_format & DIFF_FORMAT_SHORTSTAT)
4419                         show_shortstats(&diffstat, options);
4420                 if (output_format & DIFF_FORMAT_DIRSTAT)
4421                         show_dirstat_by_line(&diffstat, options);
4422                 free_diffstat_info(&diffstat);
4423                 separator++;
4424         }
4425         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4426                 show_dirstat(options);
4427
4428         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4429                 for (i = 0; i < q->nr; i++) {
4430                         diff_summary(options, q->queue[i]);
4431                 }
4432                 separator++;
4433         }
4434
4435         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4436             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4437             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4438                 /*
4439                  * run diff_flush_patch for the exit status. setting
4440                  * options->file to /dev/null should be safe, becaue we
4441                  * aren't supposed to produce any output anyway.
4442                  */
4443                 if (options->close_file)
4444                         fclose(options->file);
4445                 options->file = fopen("/dev/null", "w");
4446                 if (!options->file)
4447                         die_errno("Could not open /dev/null");
4448                 options->close_file = 1;
4449                 for (i = 0; i < q->nr; i++) {
4450                         struct diff_filepair *p = q->queue[i];
4451                         if (check_pair_status(p))
4452                                 diff_flush_patch(p, options);
4453                         if (options->found_changes)
4454                                 break;
4455                 }
4456         }
4457
4458         if (output_format & DIFF_FORMAT_PATCH) {
4459                 if (separator) {
4460                         if (options->output_prefix) {
4461                                 struct strbuf *msg = NULL;
4462                                 msg = options->output_prefix(options,
4463                                         options->output_prefix_data);
4464                                 fwrite(msg->buf, msg->len, 1, stdout);
4465                         }
4466                         putc(options->line_termination, options->file);
4467                         if (options->stat_sep) {
4468                                 /* attach patch instead of inline */
4469                                 fputs(options->stat_sep, options->file);
4470                         }
4471                 }
4472
4473                 for (i = 0; i < q->nr; i++) {
4474                         struct diff_filepair *p = q->queue[i];
4475                         if (check_pair_status(p))
4476                                 diff_flush_patch(p, options);
4477                 }
4478         }
4479
4480         if (output_format & DIFF_FORMAT_CALLBACK)
4481                 options->format_callback(q, options, options->format_callback_data);
4482
4483         for (i = 0; i < q->nr; i++)
4484                 diff_free_filepair(q->queue[i]);
4485 free_queue:
4486         free(q->queue);
4487         DIFF_QUEUE_CLEAR(q);
4488         if (options->close_file)
4489                 fclose(options->file);
4490
4491         /*
4492          * Report the content-level differences with HAS_CHANGES;
4493          * diff_addremove/diff_change does not set the bit when
4494          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4495          */
4496         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4497                 if (options->found_changes)
4498                         DIFF_OPT_SET(options, HAS_CHANGES);
4499                 else
4500                         DIFF_OPT_CLR(options, HAS_CHANGES);
4501         }
4502 }
4503
4504 static void diffcore_apply_filter(const char *filter)
4505 {
4506         int i;
4507         struct diff_queue_struct *q = &diff_queued_diff;
4508         struct diff_queue_struct outq;
4509         DIFF_QUEUE_CLEAR(&outq);
4510
4511         if (!filter)
4512                 return;
4513
4514         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
4515                 int found;
4516                 for (i = found = 0; !found && i < q->nr; i++) {
4517                         struct diff_filepair *p = q->queue[i];
4518                         if (((p->status == DIFF_STATUS_MODIFIED) &&
4519                              ((p->score &&
4520                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
4521                               (!p->score &&
4522                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
4523                             ((p->status != DIFF_STATUS_MODIFIED) &&
4524                              strchr(filter, p->status)))
4525                                 found++;
4526                 }
4527                 if (found)
4528                         return;
4529
4530                 /* otherwise we will clear the whole queue
4531                  * by copying the empty outq at the end of this
4532                  * function, but first clear the current entries
4533                  * in the queue.
4534                  */
4535                 for (i = 0; i < q->nr; i++)
4536                         diff_free_filepair(q->queue[i]);
4537         }
4538         else {
4539                 /* Only the matching ones */
4540                 for (i = 0; i < q->nr; i++) {
4541                         struct diff_filepair *p = q->queue[i];
4542
4543                         if (((p->status == DIFF_STATUS_MODIFIED) &&
4544                              ((p->score &&
4545                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
4546                               (!p->score &&
4547                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
4548                             ((p->status != DIFF_STATUS_MODIFIED) &&
4549                              strchr(filter, p->status)))
4550                                 diff_q(&outq, p);
4551                         else
4552                                 diff_free_filepair(p);
4553                 }
4554         }
4555         free(q->queue);
4556         *q = outq;
4557 }
4558
4559 /* Check whether two filespecs with the same mode and size are identical */
4560 static int diff_filespec_is_identical(struct diff_filespec *one,
4561                                       struct diff_filespec *two)
4562 {
4563         if (S_ISGITLINK(one->mode))
4564                 return 0;
4565         if (diff_populate_filespec(one, 0))
4566                 return 0;
4567         if (diff_populate_filespec(two, 0))
4568                 return 0;
4569         return !memcmp(one->data, two->data, one->size);
4570 }
4571
4572 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4573 {
4574         int i;
4575         struct diff_queue_struct *q = &diff_queued_diff;
4576         struct diff_queue_struct outq;
4577         DIFF_QUEUE_CLEAR(&outq);
4578
4579         for (i = 0; i < q->nr; i++) {
4580                 struct diff_filepair *p = q->queue[i];
4581
4582                 /*
4583                  * 1. Entries that come from stat info dirtiness
4584                  *    always have both sides (iow, not create/delete),
4585                  *    one side of the object name is unknown, with
4586                  *    the same mode and size.  Keep the ones that
4587                  *    do not match these criteria.  They have real
4588                  *    differences.
4589                  *
4590                  * 2. At this point, the file is known to be modified,
4591                  *    with the same mode and size, and the object
4592                  *    name of one side is unknown.  Need to inspect
4593                  *    the identical contents.
4594                  */
4595                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
4596                     !DIFF_FILE_VALID(p->two) ||
4597                     (p->one->sha1_valid && p->two->sha1_valid) ||
4598                     (p->one->mode != p->two->mode) ||
4599                     diff_populate_filespec(p->one, 1) ||
4600                     diff_populate_filespec(p->two, 1) ||
4601                     (p->one->size != p->two->size) ||
4602                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4603                         diff_q(&outq, p);
4604                 else {
4605                         /*
4606                          * The caller can subtract 1 from skip_stat_unmatch
4607                          * to determine how many paths were dirty only
4608                          * due to stat info mismatch.
4609                          */
4610                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4611                                 diffopt->skip_stat_unmatch++;
4612                         diff_free_filepair(p);
4613                 }
4614         }
4615         free(q->queue);
4616         *q = outq;
4617 }
4618
4619 static int diffnamecmp(const void *a_, const void *b_)
4620 {
4621         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4622         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4623         const char *name_a, *name_b;
4624
4625         name_a = a->one ? a->one->path : a->two->path;
4626         name_b = b->one ? b->one->path : b->two->path;
4627         return strcmp(name_a, name_b);
4628 }
4629
4630 void diffcore_fix_diff_index(struct diff_options *options)
4631 {
4632         struct diff_queue_struct *q = &diff_queued_diff;
4633         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4634 }
4635
4636 void diffcore_std(struct diff_options *options)
4637 {
4638         if (options->skip_stat_unmatch)
4639                 diffcore_skip_stat_unmatch(options);
4640         if (!options->found_follow) {
4641                 /* See try_to_follow_renames() in tree-diff.c */
4642                 if (options->break_opt != -1)
4643                         diffcore_break(options->break_opt);
4644                 if (options->detect_rename)
4645                         diffcore_rename(options);
4646                 if (options->break_opt != -1)
4647                         diffcore_merge_broken();
4648         }
4649         if (options->pickaxe)
4650                 diffcore_pickaxe(options);
4651         if (options->orderfile)
4652                 diffcore_order(options->orderfile);
4653         if (!options->found_follow)
4654                 /* See try_to_follow_renames() in tree-diff.c */
4655                 diff_resolve_rename_copy();
4656         diffcore_apply_filter(options->filter);
4657
4658         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4659                 DIFF_OPT_SET(options, HAS_CHANGES);
4660         else
4661                 DIFF_OPT_CLR(options, HAS_CHANGES);
4662
4663         options->found_follow = 0;
4664 }
4665
4666 int diff_result_code(struct diff_options *opt, int status)
4667 {
4668         int result = 0;
4669
4670         diff_warn_rename_limit("diff.renamelimit",
4671                                opt->needed_rename_limit,
4672                                opt->degraded_cc_to_c);
4673         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4674             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4675                 return status;
4676         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4677             DIFF_OPT_TST(opt, HAS_CHANGES))
4678                 result |= 01;
4679         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4680             DIFF_OPT_TST(opt, CHECK_FAILED))
4681                 result |= 02;
4682         return result;
4683 }
4684
4685 int diff_can_quit_early(struct diff_options *opt)
4686 {
4687         return (DIFF_OPT_TST(opt, QUICK) &&
4688                 !opt->filter &&
4689                 DIFF_OPT_TST(opt, HAS_CHANGES));
4690 }
4691
4692 /*
4693  * Shall changes to this submodule be ignored?
4694  *
4695  * Submodule changes can be configured to be ignored separately for each path,
4696  * but that configuration can be overridden from the command line.
4697  */
4698 static int is_submodule_ignored(const char *path, struct diff_options *options)
4699 {
4700         int ignored = 0;
4701         unsigned orig_flags = options->flags;
4702         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4703                 set_diffopt_flags_from_submodule_config(options, path);
4704         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4705                 ignored = 1;
4706         options->flags = orig_flags;
4707         return ignored;
4708 }
4709
4710 void diff_addremove(struct diff_options *options,
4711                     int addremove, unsigned mode,
4712                     const unsigned char *sha1,
4713                     const char *concatpath, unsigned dirty_submodule)
4714 {
4715         struct diff_filespec *one, *two;
4716
4717         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4718                 return;
4719
4720         /* This may look odd, but it is a preparation for
4721          * feeding "there are unchanged files which should
4722          * not produce diffs, but when you are doing copy
4723          * detection you would need them, so here they are"
4724          * entries to the diff-core.  They will be prefixed
4725          * with something like '=' or '*' (I haven't decided
4726          * which but should not make any difference).
4727          * Feeding the same new and old to diff_change()
4728          * also has the same effect.
4729          * Before the final output happens, they are pruned after
4730          * merged into rename/copy pairs as appropriate.
4731          */
4732         if (DIFF_OPT_TST(options, REVERSE_DIFF))
4733                 addremove = (addremove == '+' ? '-' :
4734                              addremove == '-' ? '+' : addremove);
4735
4736         if (options->prefix &&
4737             strncmp(concatpath, options->prefix, options->prefix_length))
4738                 return;
4739
4740         one = alloc_filespec(concatpath);
4741         two = alloc_filespec(concatpath);
4742
4743         if (addremove != '+')
4744                 fill_filespec(one, sha1, mode);
4745         if (addremove != '-') {
4746                 fill_filespec(two, sha1, mode);
4747                 two->dirty_submodule = dirty_submodule;
4748         }
4749
4750         diff_queue(&diff_queued_diff, one, two);
4751         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4752                 DIFF_OPT_SET(options, HAS_CHANGES);
4753 }
4754
4755 void diff_change(struct diff_options *options,
4756                  unsigned old_mode, unsigned new_mode,
4757                  const unsigned char *old_sha1,
4758                  const unsigned char *new_sha1,
4759                  const char *concatpath,
4760                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4761 {
4762         struct diff_filespec *one, *two;
4763
4764         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4765             is_submodule_ignored(concatpath, options))
4766                 return;
4767
4768         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4769                 unsigned tmp;
4770                 const unsigned char *tmp_c;
4771                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4772                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4773                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
4774                         new_dirty_submodule = tmp;
4775         }
4776
4777         if (options->prefix &&
4778             strncmp(concatpath, options->prefix, options->prefix_length))
4779                 return;
4780
4781         one = alloc_filespec(concatpath);
4782         two = alloc_filespec(concatpath);
4783         fill_filespec(one, old_sha1, old_mode);
4784         fill_filespec(two, new_sha1, new_mode);
4785         one->dirty_submodule = old_dirty_submodule;
4786         two->dirty_submodule = new_dirty_submodule;
4787
4788         diff_queue(&diff_queued_diff, one, two);
4789         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4790                 DIFF_OPT_SET(options, HAS_CHANGES);
4791 }
4792
4793 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
4794 {
4795         struct diff_filepair *pair;
4796         struct diff_filespec *one, *two;
4797
4798         if (options->prefix &&
4799             strncmp(path, options->prefix, options->prefix_length))
4800                 return NULL;
4801
4802         one = alloc_filespec(path);
4803         two = alloc_filespec(path);
4804         pair = diff_queue(&diff_queued_diff, one, two);
4805         pair->is_unmerged = 1;
4806         return pair;
4807 }
4808
4809 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
4810                 size_t *outsize)
4811 {
4812         struct diff_tempfile *temp;
4813         const char *argv[3];
4814         const char **arg = argv;
4815         struct child_process child;
4816         struct strbuf buf = STRBUF_INIT;
4817         int err = 0;
4818
4819         temp = prepare_temp_file(spec->path, spec);
4820         *arg++ = pgm;
4821         *arg++ = temp->name;
4822         *arg = NULL;
4823
4824         memset(&child, 0, sizeof(child));
4825         child.use_shell = 1;
4826         child.argv = argv;
4827         child.out = -1;
4828         if (start_command(&child)) {
4829                 remove_tempfile();
4830                 return NULL;
4831         }
4832
4833         if (strbuf_read(&buf, child.out, 0) < 0)
4834                 err = error("error reading from textconv command '%s'", pgm);
4835         close(child.out);
4836
4837         if (finish_command(&child) || err) {
4838                 strbuf_release(&buf);
4839                 remove_tempfile();
4840                 return NULL;
4841         }
4842         remove_tempfile();
4843
4844         return strbuf_detach(&buf, outsize);
4845 }
4846
4847 size_t fill_textconv(struct userdiff_driver *driver,
4848                      struct diff_filespec *df,
4849                      char **outbuf)
4850 {
4851         size_t size;
4852
4853         if (!driver || !driver->textconv) {
4854                 if (!DIFF_FILE_VALID(df)) {
4855                         *outbuf = "";
4856                         return 0;
4857                 }
4858                 if (diff_populate_filespec(df, 0))
4859                         die("unable to read files to diff");
4860                 *outbuf = df->data;
4861                 return df->size;
4862         }
4863
4864         if (driver->textconv_cache && df->sha1_valid) {
4865                 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
4866                                           &size);
4867                 if (*outbuf)
4868                         return size;
4869         }
4870
4871         *outbuf = run_textconv(driver->textconv, df, &size);
4872         if (!*outbuf)
4873                 die("unable to read files to diff");
4874
4875         if (driver->textconv_cache && df->sha1_valid) {
4876                 /* ignore errors, as we might be in a readonly repository */
4877                 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
4878                                 size);
4879                 /*
4880                  * we could save up changes and flush them all at the end,
4881                  * but we would need an extra call after all diffing is done.
4882                  * Since generating a cache entry is the slow path anyway,
4883                  * this extra overhead probably isn't a big deal.
4884                  */
4885                 notes_cache_write(driver->textconv_cache);
4886         }
4887
4888         return size;
4889 }