diff.c

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "quote.h"
   6 #include "diff.h"
   7 #include "diffcore.h"
   8 #include "delta.h"
   9 #include "xdiff-interface.h"
  10 #include "color.h"
  11 #include "attr.h"
  12 #include "run-command.h"
  13 #include "utf8.h"
  14 #include "userdiff.h"
  15 #include "sigchain.h"
  16 #include "submodule.h"
  17 #include "ll-merge.h"
  18
  19 #ifdef NO_FAST_WORKING_DIRECTORY
  20 #define FAST_WORKING_DIRECTORY 0
  21 #else
  22 #define FAST_WORKING_DIRECTORY 1
  23 #endif
  24
  25 static int diff_detect_rename_default;
  26 static int diff_rename_limit_default = 400;
  27 static int diff_suppress_blank_empty;
  28 int diff_use_color_default = -1;
  29 static const char *diff_word_regex_cfg;
  30 static const char *external_diff_cmd_cfg;
  31 int diff_auto_refresh_index = 1;
  32 static int diff_mnemonic_prefix;
  33 static int diff_no_prefix;
  34 static int diff_stat_graph_width;
  35 static int diff_dirstat_permille_default = 30;
  36 static struct diff_options default_diff_options;
  37
  38 static char diff_colors[][COLOR_MAXLEN] = {
  39         GIT_COLOR_RESET,
  40         GIT_COLOR_NORMAL,       /* PLAIN */
  41         GIT_COLOR_BOLD,         /* METAINFO */
  42         GIT_COLOR_CYAN,         /* FRAGINFO */
  43         GIT_COLOR_RED,          /* OLD */
  44         GIT_COLOR_GREEN,        /* NEW */
  45         GIT_COLOR_YELLOW,       /* COMMIT */
  46         GIT_COLOR_BG_RED,       /* WHITESPACE */
  47         GIT_COLOR_NORMAL,       /* FUNCINFO */
  48 };
  49
  50 static int parse_diff_color_slot(const char *var, int ofs)
  51 {
  52         if (!strcasecmp(var+ofs, "plain"))
  53                 return DIFF_PLAIN;
  54         if (!strcasecmp(var+ofs, "meta"))
  55                 return DIFF_METAINFO;
  56         if (!strcasecmp(var+ofs, "frag"))
  57                 return DIFF_FRAGINFO;
  58         if (!strcasecmp(var+ofs, "old"))
  59                 return DIFF_FILE_OLD;
  60         if (!strcasecmp(var+ofs, "new"))
  61                 return DIFF_FILE_NEW;
  62         if (!strcasecmp(var+ofs, "commit"))
  63                 return DIFF_COMMIT;
  64         if (!strcasecmp(var+ofs, "whitespace"))
  65                 return DIFF_WHITESPACE;
  66         if (!strcasecmp(var+ofs, "func"))
  67                 return DIFF_FUNCINFO;
  68         return -1;
  69 }
  70
  71 static int parse_dirstat_params(struct diff_options *options, const char *params,
  72                                 struct strbuf *errmsg)
  73 {
  74         const char *p = params;
  75         int p_len, ret = 0;
  76
  77         while (*p) {
  78                 p_len = strchrnul(p, ',') - p;
  79                 if (!memcmp(p, "changes", p_len)) {
  80                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  81                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
  82                 } else if (!memcmp(p, "lines", p_len)) {
  83                         DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
  84                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
  85                 } else if (!memcmp(p, "files", p_len)) {
  86                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  87                         DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
  88                 } else if (!memcmp(p, "noncumulative", p_len)) {
  89                         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
  90                 } else if (!memcmp(p, "cumulative", p_len)) {
  91                         DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
  92                 } else if (isdigit(*p)) {
  93                         char *end;
  94                         int permille = strtoul(p, &end, 10) * 10;
  95                         if (*end == '.' && isdigit(*++end)) {
  96                                 /* only use first digit */
  97                                 permille += *end - '0';
  98                                 /* .. and ignore any further digits */
  99                                 while (isdigit(*++end))
 100                                         ; /* nothing */
 101                         }
 102                         if (end - p == p_len)
 103                                 options->dirstat_permille = permille;
 104                         else {
 105                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%.*s'\n"),
 106                                             p_len, p);
 107                                 ret++;
 108                         }
 109                 } else {
 110                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%.*s'\n"),
 111                                     p_len, p);
 112                         ret++;
 113                 }
 114
 115                 p += p_len;
 116
 117                 if (*p)
 118                         p++; /* more parameters, swallow separator */
 119         }
 120         return ret;
 121 }
 122
 123 static int git_config_rename(const char *var, const char *value)
 124 {
 125         if (!value)
 126                 return DIFF_DETECT_RENAME;
 127         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 128                 return  DIFF_DETECT_COPY;
 129         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 130 }
 131
 132 /*
 133  * These are to give UI layer defaults.
 134  * The core-level commands such as git-diff-files should
 135  * never be affected by the setting of diff.renames
 136  * the user happens to have in the configuration file.
 137  */
 138 int git_diff_ui_config(const char *var, const char *value, void *cb)
 139 {
 140         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 141                 diff_use_color_default = git_config_colorbool(var, value);
 142                 return 0;
 143         }
 144         if (!strcmp(var, "diff.renames")) {
 145                 diff_detect_rename_default = git_config_rename(var, value);
 146                 return 0;
 147         }
 148         if (!strcmp(var, "diff.autorefreshindex")) {
 149                 diff_auto_refresh_index = git_config_bool(var, value);
 150                 return 0;
 151         }
 152         if (!strcmp(var, "diff.mnemonicprefix")) {
 153                 diff_mnemonic_prefix = git_config_bool(var, value);
 154                 return 0;
 155         }
 156         if (!strcmp(var, "diff.noprefix")) {
 157                 diff_no_prefix = git_config_bool(var, value);
 158                 return 0;
 159         }
 160         if (!strcmp(var, "diff.statgraphwidth")) {
 161                 diff_stat_graph_width = git_config_int(var, value);
 162                 return 0;
 163         }
 164         if (!strcmp(var, "diff.external"))
 165                 return git_config_string(&external_diff_cmd_cfg, var, value);
 166         if (!strcmp(var, "diff.wordregex"))
 167                 return git_config_string(&diff_word_regex_cfg, var, value);
 168
 169         if (!strcmp(var, "diff.ignoresubmodules"))
 170                 handle_ignore_submodules_arg(&default_diff_options, value);
 171
 172         if (git_color_config(var, value, cb) < 0)
 173                 return -1;
 174
 175         return git_diff_basic_config(var, value, cb);
 176 }
 177
 178 int git_diff_basic_config(const char *var, const char *value, void *cb)
 179 {
 180         if (!strcmp(var, "diff.renamelimit")) {
 181                 diff_rename_limit_default = git_config_int(var, value);
 182                 return 0;
 183         }
 184
 185         if (userdiff_config(var, value) < 0)
 186                 return -1;
 187
 188         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 189                 int slot = parse_diff_color_slot(var, 11);
 190                 if (slot < 0)
 191                         return 0;
 192                 if (!value)
 193                         return config_error_nonbool(var);
 194                 color_parse(value, var, diff_colors[slot]);
 195                 return 0;
 196         }
 197
 198         /* like GNU diff's --suppress-blank-empty option  */
 199         if (!strcmp(var, "diff.suppressblankempty") ||
 200                         /* for backwards compatibility */
 201                         !strcmp(var, "diff.suppress-blank-empty")) {
 202                 diff_suppress_blank_empty = git_config_bool(var, value);
 203                 return 0;
 204         }
 205
 206         if (!strcmp(var, "diff.dirstat")) {
 207                 struct strbuf errmsg = STRBUF_INIT;
 208                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 209                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 210                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 211                                 errmsg.buf);
 212                 strbuf_release(&errmsg);
 213                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 214                 return 0;
 215         }
 216
 217         if (!prefixcmp(var, "submodule."))
 218                 return parse_submodule_config_option(var, value);
 219
 220         return git_default_config(var, value, cb);
 221 }
 222
 223 static char *quote_two(const char *one, const char *two)
 224 {
 225         int need_one = quote_c_style(one, NULL, NULL, 1);
 226         int need_two = quote_c_style(two, NULL, NULL, 1);
 227         struct strbuf res = STRBUF_INIT;
 228
 229         if (need_one + need_two) {
 230                 strbuf_addch(&res, '"');
 231                 quote_c_style(one, &res, NULL, 1);
 232                 quote_c_style(two, &res, NULL, 1);
 233                 strbuf_addch(&res, '"');
 234         } else {
 235                 strbuf_addstr(&res, one);
 236                 strbuf_addstr(&res, two);
 237         }
 238         return strbuf_detach(&res, NULL);
 239 }
 240
 241 static const char *external_diff(void)
 242 {
 243         static const char *external_diff_cmd = NULL;
 244         static int done_preparing = 0;
 245
 246         if (done_preparing)
 247                 return external_diff_cmd;
 248         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 249         if (!external_diff_cmd)
 250                 external_diff_cmd = external_diff_cmd_cfg;
 251         done_preparing = 1;
 252         return external_diff_cmd;
 253 }
 254
 255 static struct diff_tempfile {
 256         const char *name; /* filename external diff should read from */
 257         char hex[41];
 258         char mode[10];
 259         char tmp_path[PATH_MAX];
 260 } diff_temp[2];
 261
 262 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 263
 264 struct emit_callback {
 265         int color_diff;
 266         unsigned ws_rule;
 267         int blank_at_eof_in_preimage;
 268         int blank_at_eof_in_postimage;
 269         int lno_in_preimage;
 270         int lno_in_postimage;
 271         sane_truncate_fn truncate;
 272         const char **label_path;
 273         struct diff_words_data *diff_words;
 274         struct diff_options *opt;
 275         int *found_changesp;
 276         struct strbuf *header;
 277 };
 278
 279 static int count_lines(const char *data, int size)
 280 {
 281         int count, ch, completely_empty = 1, nl_just_seen = 0;
 282         count = 0;
 283         while (0 < size--) {
 284                 ch = *data++;
 285                 if (ch == '\n') {
 286                         count++;
 287                         nl_just_seen = 1;
 288                         completely_empty = 0;
 289                 }
 290                 else {
 291                         nl_just_seen = 0;
 292                         completely_empty = 0;
 293                 }
 294         }
 295         if (completely_empty)
 296                 return 0;
 297         if (!nl_just_seen)
 298                 count++; /* no trailing newline */
 299         return count;
 300 }
 301
 302 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 303 {
 304         if (!DIFF_FILE_VALID(one)) {
 305                 mf->ptr = (char *)""; /* does not matter */
 306                 mf->size = 0;
 307                 return 0;
 308         }
 309         else if (diff_populate_filespec(one, 0))
 310                 return -1;
 311
 312         mf->ptr = one->data;
 313         mf->size = one->size;
 314         return 0;
 315 }
 316
 317 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 318 static unsigned long diff_filespec_size(struct diff_filespec *one)
 319 {
 320         if (!DIFF_FILE_VALID(one))
 321                 return 0;
 322         diff_populate_filespec(one, 1);
 323         return one->size;
 324 }
 325
 326 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 327 {
 328         char *ptr = mf->ptr;
 329         long size = mf->size;
 330         int cnt = 0;
 331
 332         if (!size)
 333                 return cnt;
 334         ptr += size - 1; /* pointing at the very end */
 335         if (*ptr != '\n')
 336                 ; /* incomplete line */
 337         else
 338                 ptr--; /* skip the last LF */
 339         while (mf->ptr < ptr) {
 340                 char *prev_eol;
 341                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 342                         if (*prev_eol == '\n')
 343                                 break;
 344                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 345                         break;
 346                 cnt++;
 347                 ptr = prev_eol - 1;
 348         }
 349         return cnt;
 350 }
 351
 352 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 353                                struct emit_callback *ecbdata)
 354 {
 355         int l1, l2, at;
 356         unsigned ws_rule = ecbdata->ws_rule;
 357         l1 = count_trailing_blank(mf1, ws_rule);
 358         l2 = count_trailing_blank(mf2, ws_rule);
 359         if (l2 <= l1) {
 360                 ecbdata->blank_at_eof_in_preimage = 0;
 361                 ecbdata->blank_at_eof_in_postimage = 0;
 362                 return;
 363         }
 364         at = count_lines(mf1->ptr, mf1->size);
 365         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 366
 367         at = count_lines(mf2->ptr, mf2->size);
 368         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 369 }
 370
 371 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 372                         int first, const char *line, int len)
 373 {
 374         int has_trailing_newline, has_trailing_carriage_return;
 375         int nofirst;
 376         FILE *file = o->file;
 377
 378         if (o->output_prefix) {
 379                 struct strbuf *msg = NULL;
 380                 msg = o->output_prefix(o, o->output_prefix_data);
 381                 assert(msg);
 382                 fwrite(msg->buf, msg->len, 1, file);
 383         }
 384
 385         if (len == 0) {
 386                 has_trailing_newline = (first == '\n');
 387                 has_trailing_carriage_return = (!has_trailing_newline &&
 388                                                 (first == '\r'));
 389                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 390         } else {
 391                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 392                 if (has_trailing_newline)
 393                         len--;
 394                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 395                 if (has_trailing_carriage_return)
 396                         len--;
 397                 nofirst = 0;
 398         }
 399
 400         if (len || !nofirst) {
 401                 fputs(set, file);
 402                 if (!nofirst)
 403                         fputc(first, file);
 404                 fwrite(line, len, 1, file);
 405                 fputs(reset, file);
 406         }
 407         if (has_trailing_carriage_return)
 408                 fputc('\r', file);
 409         if (has_trailing_newline)
 410                 fputc('\n', file);
 411 }
 412
 413 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 414                       const char *line, int len)
 415 {
 416         emit_line_0(o, set, reset, line[0], line+1, len-1);
 417 }
 418
 419 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
 420 {
 421         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
 422               ecbdata->blank_at_eof_in_preimage &&
 423               ecbdata->blank_at_eof_in_postimage &&
 424               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
 425               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
 426                 return 0;
 427         return ws_blank_line(line, len, ecbdata->ws_rule);
 428 }
 429
 430 static void emit_add_line(const char *reset,
 431                           struct emit_callback *ecbdata,
 432                           const char *line, int len)
 433 {
 434         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 435         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 436
 437         if (!*ws)
 438                 emit_line_0(ecbdata->opt, set, reset, '+', line, len);
 439         else if (new_blank_line_at_eof(ecbdata, line, len))
 440                 /* Blank line at EOF - paint '+' as well */
 441                 emit_line_0(ecbdata->opt, ws, reset, '+', line, len);
 442         else {
 443                 /* Emit just the prefix, then the rest. */
 444                 emit_line_0(ecbdata->opt, set, reset, '+', "", 0);
 445                 ws_check_emit(line, len, ecbdata->ws_rule,
 446                               ecbdata->opt->file, set, reset, ws);
 447         }
 448 }
 449
 450 static void emit_hunk_header(struct emit_callback *ecbdata,
 451                              const char *line, int len)
 452 {
 453         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 454         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
 455         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
 456         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 457         static const char atat[2] = { '@', '@' };
 458         const char *cp, *ep;
 459         struct strbuf msgbuf = STRBUF_INIT;
 460         int org_len = len;
 461         int i = 1;
 462
 463         /*
 464          * As a hunk header must begin with "@@ -<old>, +<new> @@",
 465          * it always is at least 10 bytes long.
 466          */
 467         if (len < 10 ||
 468             memcmp(line, atat, 2) ||
 469             !(ep = memmem(line + 2, len - 2, atat, 2))) {
 470                 emit_line(ecbdata->opt, plain, reset, line, len);
 471                 return;
 472         }
 473         ep += 2; /* skip over @@ */
 474
 475         /* The hunk header in fraginfo color */
 476         strbuf_add(&msgbuf, frag, strlen(frag));
 477         strbuf_add(&msgbuf, line, ep - line);
 478         strbuf_add(&msgbuf, reset, strlen(reset));
 479
 480         /*
 481          * trailing "\r\n"
 482          */
 483         for ( ; i < 3; i++)
 484                 if (line[len - i] == '\r' || line[len - i] == '\n')
 485                         len--;
 486
 487         /* blank before the func header */
 488         for (cp = ep; ep - line < len; ep++)
 489                 if (*ep != ' ' && *ep != '\t')
 490                         break;
 491         if (ep != cp) {
 492                 strbuf_add(&msgbuf, plain, strlen(plain));
 493                 strbuf_add(&msgbuf, cp, ep - cp);
 494                 strbuf_add(&msgbuf, reset, strlen(reset));
 495         }
 496
 497         if (ep < line + len) {
 498                 strbuf_add(&msgbuf, func, strlen(func));
 499                 strbuf_add(&msgbuf, ep, line + len - ep);
 500                 strbuf_add(&msgbuf, reset, strlen(reset));
 501         }
 502
 503         strbuf_add(&msgbuf, line + len, org_len - len);
 504         emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
 505         strbuf_release(&msgbuf);
 506 }
 507
 508 static struct diff_tempfile *claim_diff_tempfile(void) {
 509         int i;
 510         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
 511                 if (!diff_temp[i].name)
 512                         return diff_temp + i;
 513         die("BUG: diff is failing to clean up its tempfiles");
 514 }
 515
 516 static int remove_tempfile_installed;
 517
 518 static void remove_tempfile(void)
 519 {
 520         int i;
 521         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
 522                 if (diff_temp[i].name == diff_temp[i].tmp_path)
 523                         unlink_or_warn(diff_temp[i].name);
 524                 diff_temp[i].name = NULL;
 525         }
 526 }
 527
 528 static void remove_tempfile_on_signal(int signo)
 529 {
 530         remove_tempfile();
 531         sigchain_pop(signo);
 532         raise(signo);
 533 }
 534
 535 static void print_line_count(FILE *file, int count)
 536 {
 537         switch (count) {
 538         case 0:
 539                 fprintf(file, "0,0");
 540                 break;
 541         case 1:
 542                 fprintf(file, "1");
 543                 break;
 544         default:
 545                 fprintf(file, "1,%d", count);
 546                 break;
 547         }
 548 }
 549
 550 static void emit_rewrite_lines(struct emit_callback *ecb,
 551                                int prefix, const char *data, int size)
 552 {
 553         const char *endp = NULL;
 554         static const char *nneof = " No newline at end of file\n";
 555         const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
 556         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
 557
 558         while (0 < size) {
 559                 int len;
 560
 561                 endp = memchr(data, '\n', size);
 562                 len = endp ? (endp - data + 1) : size;
 563                 if (prefix != '+') {
 564                         ecb->lno_in_preimage++;
 565                         emit_line_0(ecb->opt, old, reset, '-',
 566                                     data, len);
 567                 } else {
 568                         ecb->lno_in_postimage++;
 569                         emit_add_line(reset, ecb, data, len);
 570                 }
 571                 size -= len;
 572                 data += len;
 573         }
 574         if (!endp) {
 575                 const char *plain = diff_get_color(ecb->color_diff,
 576                                                    DIFF_PLAIN);
 577                 emit_line_0(ecb->opt, plain, reset, '\\',
 578                             nneof, strlen(nneof));
 579         }
 580 }
 581
 582 static void emit_rewrite_diff(const char *name_a,
 583                               const char *name_b,
 584                               struct diff_filespec *one,
 585                               struct diff_filespec *two,
 586                               struct userdiff_driver *textconv_one,
 587                               struct userdiff_driver *textconv_two,
 588                               struct diff_options *o)
 589 {
 590         int lc_a, lc_b;
 591         const char *name_a_tab, *name_b_tab;
 592         const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
 593         const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
 594         const char *reset = diff_get_color(o->use_color, DIFF_RESET);
 595         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 596         const char *a_prefix, *b_prefix;
 597         char *data_one, *data_two;
 598         size_t size_one, size_two;
 599         struct emit_callback ecbdata;
 600         char *line_prefix = "";
 601         struct strbuf *msgbuf;
 602
 603         if (o && o->output_prefix) {
 604                 msgbuf = o->output_prefix(o, o->output_prefix_data);
 605                 line_prefix = msgbuf->buf;
 606         }
 607
 608         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 609                 a_prefix = o->b_prefix;
 610                 b_prefix = o->a_prefix;
 611         } else {
 612                 a_prefix = o->a_prefix;
 613                 b_prefix = o->b_prefix;
 614         }
 615
 616         name_a += (*name_a == '/');
 617         name_b += (*name_b == '/');
 618         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 619         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 620
 621         strbuf_reset(&a_name);
 622         strbuf_reset(&b_name);
 623         quote_two_c_style(&a_name, a_prefix, name_a, 0);
 624         quote_two_c_style(&b_name, b_prefix, name_b, 0);
 625
 626         size_one = fill_textconv(textconv_one, one, &data_one);
 627         size_two = fill_textconv(textconv_two, two, &data_two);
 628
 629         memset(&ecbdata, 0, sizeof(ecbdata));
 630         ecbdata.color_diff = want_color(o->use_color);
 631         ecbdata.found_changesp = &o->found_changes;
 632         ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
 633         ecbdata.opt = o;
 634         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
 635                 mmfile_t mf1, mf2;
 636                 mf1.ptr = (char *)data_one;
 637                 mf2.ptr = (char *)data_two;
 638                 mf1.size = size_one;
 639                 mf2.size = size_two;
 640                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
 641         }
 642         ecbdata.lno_in_preimage = 1;
 643         ecbdata.lno_in_postimage = 1;
 644
 645         lc_a = count_lines(data_one, size_one);
 646         lc_b = count_lines(data_two, size_two);
 647         fprintf(o->file,
 648                 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
 649                 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
 650                 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
 651                 line_prefix, fraginfo);
 652         if (!o->irreversible_delete)
 653                 print_line_count(o->file, lc_a);
 654         else
 655                 fprintf(o->file, "?,?");
 656         fprintf(o->file, " +");
 657         print_line_count(o->file, lc_b);
 658         fprintf(o->file, " @@%s\n", reset);
 659         if (lc_a && !o->irreversible_delete)
 660                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
 661         if (lc_b)
 662                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
 663         if (textconv_one)
 664                 free((char *)data_one);
 665         if (textconv_two)
 666                 free((char *)data_two);
 667 }
 668
 669 struct diff_words_buffer {
 670         mmfile_t text;
 671         long alloc;
 672         struct diff_words_orig {
 673                 const char *begin, *end;
 674         } *orig;
 675         int orig_nr, orig_alloc;
 676 };
 677
 678 static void diff_words_append(char *line, unsigned long len,
 679                 struct diff_words_buffer *buffer)
 680 {
 681         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
 682         line++;
 683         len--;
 684         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 685         buffer->text.size += len;
 686         buffer->text.ptr[buffer->text.size] = '\0';
 687 }
 688
 689 struct diff_words_style_elem {
 690         const char *prefix;
 691         const char *suffix;
 692         const char *color; /* NULL; filled in by the setup code if
 693                             * color is enabled */
 694 };
 695
 696 struct diff_words_style {
 697         enum diff_words_type type;
 698         struct diff_words_style_elem new, old, ctx;
 699         const char *newline;
 700 };
 701
 702 static struct diff_words_style diff_words_styles[] = {
 703         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
 704         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
 705         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
 706 };
 707
 708 struct diff_words_data {
 709         struct diff_words_buffer minus, plus;
 710         const char *current_plus;
 711         int last_minus;
 712         struct diff_options *opt;
 713         regex_t *word_regex;
 714         enum diff_words_type type;
 715         struct diff_words_style *style;
 716 };
 717
 718 static int fn_out_diff_words_write_helper(FILE *fp,
 719                                           struct diff_words_style_elem *st_el,
 720                                           const char *newline,
 721                                           size_t count, const char *buf,
 722                                           const char *line_prefix)
 723 {
 724         int print = 0;
 725
 726         while (count) {
 727                 char *p = memchr(buf, '\n', count);
 728                 if (print)
 729                         fputs(line_prefix, fp);
 730                 if (p != buf) {
 731                         if (st_el->color && fputs(st_el->color, fp) < 0)
 732                                 return -1;
 733                         if (fputs(st_el->prefix, fp) < 0 ||
 734                             fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
 735                             fputs(st_el->suffix, fp) < 0)
 736                                 return -1;
 737                         if (st_el->color && *st_el->color
 738                             && fputs(GIT_COLOR_RESET, fp) < 0)
 739                                 return -1;
 740                 }
 741                 if (!p)
 742                         return 0;
 743                 if (fputs(newline, fp) < 0)
 744                         return -1;
 745                 count -= p + 1 - buf;
 746                 buf = p + 1;
 747                 print = 1;
 748         }
 749         return 0;
 750 }
 751
 752 /*
 753  * '--color-words' algorithm can be described as:
 754  *
 755  *   1. collect a the minus/plus lines of a diff hunk, divided into
 756  *      minus-lines and plus-lines;
 757  *
 758  *   2. break both minus-lines and plus-lines into words and
 759  *      place them into two mmfile_t with one word for each line;
 760  *
 761  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
 762  *
 763  * And for the common parts of the both file, we output the plus side text.
 764  * diff_words->current_plus is used to trace the current position of the plus file
 765  * which printed. diff_words->last_minus is used to trace the last minus word
 766  * printed.
 767  *
 768  * For '--graph' to work with '--color-words', we need to output the graph prefix
 769  * on each line of color words output. Generally, there are two conditions on
 770  * which we should output the prefix.
 771  *
 772  *   1. diff_words->last_minus == 0 &&
 773  *      diff_words->current_plus == diff_words->plus.text.ptr
 774  *
 775  *      that is: the plus text must start as a new line, and if there is no minus
 776  *      word printed, a graph prefix must be printed.
 777  *
 778  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
 779  *      *(diff_words->current_plus - 1) == '\n'
 780  *
 781  *      that is: a graph prefix must be printed following a '\n'
 782  */
 783 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
 784 {
 785         if ((diff_words->last_minus == 0 &&
 786                 diff_words->current_plus == diff_words->plus.text.ptr) ||
 787                 (diff_words->current_plus > diff_words->plus.text.ptr &&
 788                 *(diff_words->current_plus - 1) == '\n')) {
 789                 return 1;
 790         } else {
 791                 return 0;
 792         }
 793 }
 794
 795 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 796 {
 797         struct diff_words_data *diff_words = priv;
 798         struct diff_words_style *style = diff_words->style;
 799         int minus_first, minus_len, plus_first, plus_len;
 800         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 801         struct diff_options *opt = diff_words->opt;
 802         struct strbuf *msgbuf;
 803         char *line_prefix = "";
 804
 805         if (line[0] != '@' || parse_hunk_header(line, len,
 806                         &minus_first, &minus_len, &plus_first, &plus_len))
 807                 return;
 808
 809         assert(opt);
 810         if (opt->output_prefix) {
 811                 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
 812                 line_prefix = msgbuf->buf;
 813         }
 814
 815         /* POSIX requires that first be decremented by one if len == 0... */
 816         if (minus_len) {
 817                 minus_begin = diff_words->minus.orig[minus_first].begin;
 818                 minus_end =
 819                         diff_words->minus.orig[minus_first + minus_len - 1].end;
 820         } else
 821                 minus_begin = minus_end =
 822                         diff_words->minus.orig[minus_first].end;
 823
 824         if (plus_len) {
 825                 plus_begin = diff_words->plus.orig[plus_first].begin;
 826                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
 827         } else
 828                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
 829
 830         if (color_words_output_graph_prefix(diff_words)) {
 831                 fputs(line_prefix, diff_words->opt->file);
 832         }
 833         if (diff_words->current_plus != plus_begin) {
 834                 fn_out_diff_words_write_helper(diff_words->opt->file,
 835                                 &style->ctx, style->newline,
 836                                 plus_begin - diff_words->current_plus,
 837                                 diff_words->current_plus, line_prefix);
 838                 if (*(plus_begin - 1) == '\n')
 839                         fputs(line_prefix, diff_words->opt->file);
 840         }
 841         if (minus_begin != minus_end) {
 842                 fn_out_diff_words_write_helper(diff_words->opt->file,
 843                                 &style->old, style->newline,
 844                                 minus_end - minus_begin, minus_begin,
 845                                 line_prefix);
 846         }
 847         if (plus_begin != plus_end) {
 848                 fn_out_diff_words_write_helper(diff_words->opt->file,
 849                                 &style->new, style->newline,
 850                                 plus_end - plus_begin, plus_begin,
 851                                 line_prefix);
 852         }
 853
 854         diff_words->current_plus = plus_end;
 855         diff_words->last_minus = minus_first;
 856 }
 857
 858 /* This function starts looking at *begin, and returns 0 iff a word was found. */
 859 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
 860                 int *begin, int *end)
 861 {
 862         if (word_regex && *begin < buffer->size) {
 863                 regmatch_t match[1];
 864                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
 865                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
 866                                         '\n', match[0].rm_eo - match[0].rm_so);
 867                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
 868                         *begin += match[0].rm_so;
 869                         return *begin >= *end;
 870                 }
 871                 return -1;
 872         }
 873
 874         /* find the next word */
 875         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
 876                 (*begin)++;
 877         if (*begin >= buffer->size)
 878                 return -1;
 879
 880         /* find the end of the word */
 881         *end = *begin + 1;
 882         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
 883                 (*end)++;
 884
 885         return 0;
 886 }
 887
 888 /*
 889  * This function splits the words in buffer->text, stores the list with
 890  * newline separator into out, and saves the offsets of the original words
 891  * in buffer->orig.
 892  */
 893 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
 894                 regex_t *word_regex)
 895 {
 896         int i, j;
 897         long alloc = 0;
 898
 899         out->size = 0;
 900         out->ptr = NULL;
 901
 902         /* fake an empty "0th" word */
 903         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
 904         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
 905         buffer->orig_nr = 1;
 906
 907         for (i = 0; i < buffer->text.size; i++) {
 908                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
 909                         return;
 910
 911                 /* store original boundaries */
 912                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
 913                                 buffer->orig_alloc);
 914                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
 915                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
 916                 buffer->orig_nr++;
 917
 918                 /* store one word */
 919                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
 920                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
 921                 out->ptr[out->size + j - i] = '\n';
 922                 out->size += j - i + 1;
 923
 924                 i = j - 1;
 925         }
 926 }
 927
 928 /* this executes the word diff on the accumulated buffers */
 929 static void diff_words_show(struct diff_words_data *diff_words)
 930 {
 931         xpparam_t xpp;
 932         xdemitconf_t xecfg;
 933         mmfile_t minus, plus;
 934         struct diff_words_style *style = diff_words->style;
 935
 936         struct diff_options *opt = diff_words->opt;
 937         struct strbuf *msgbuf;
 938         char *line_prefix = "";
 939
 940         assert(opt);
 941         if (opt->output_prefix) {
 942                 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
 943                 line_prefix = msgbuf->buf;
 944         }
 945
 946         /* special case: only removal */
 947         if (!diff_words->plus.text.size) {
 948                 fputs(line_prefix, diff_words->opt->file);
 949                 fn_out_diff_words_write_helper(diff_words->opt->file,
 950                         &style->old, style->newline,
 951                         diff_words->minus.text.size,
 952                         diff_words->minus.text.ptr, line_prefix);
 953                 diff_words->minus.text.size = 0;
 954                 return;
 955         }
 956
 957         diff_words->current_plus = diff_words->plus.text.ptr;
 958         diff_words->last_minus = 0;
 959
 960         memset(&xpp, 0, sizeof(xpp));
 961         memset(&xecfg, 0, sizeof(xecfg));
 962         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
 963         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
 964         xpp.flags = 0;
 965         /* as only the hunk header will be parsed, we need a 0-context */
 966         xecfg.ctxlen = 0;
 967         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
 968                       &xpp, &xecfg);
 969         free(minus.ptr);
 970         free(plus.ptr);
 971         if (diff_words->current_plus != diff_words->plus.text.ptr +
 972                         diff_words->plus.text.size) {
 973                 if (color_words_output_graph_prefix(diff_words))
 974                         fputs(line_prefix, diff_words->opt->file);
 975                 fn_out_diff_words_write_helper(diff_words->opt->file,
 976                         &style->ctx, style->newline,
 977                         diff_words->plus.text.ptr + diff_words->plus.text.size
 978                         - diff_words->current_plus, diff_words->current_plus,
 979                         line_prefix);
 980         }
 981         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 982 }
 983
 984 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
 985 static void diff_words_flush(struct emit_callback *ecbdata)
 986 {
 987         if (ecbdata->diff_words->minus.text.size ||
 988             ecbdata->diff_words->plus.text.size)
 989                 diff_words_show(ecbdata->diff_words);
 990 }
 991
 992 static void diff_filespec_load_driver(struct diff_filespec *one)
 993 {
 994         /* Use already-loaded driver */
 995         if (one->driver)
 996                 return;
 997
 998         if (S_ISREG(one->mode))
 999                 one->driver = userdiff_find_by_path(one->path);
1000
1001         /* Fallback to default settings */
1002         if (!one->driver)
1003                 one->driver = userdiff_find_by_name("default");
1004 }
1005
1006 static const char *userdiff_word_regex(struct diff_filespec *one)
1007 {
1008         diff_filespec_load_driver(one);
1009         return one->driver->word_regex;
1010 }
1011
1012 static void init_diff_words_data(struct emit_callback *ecbdata,
1013                                  struct diff_options *orig_opts,
1014                                  struct diff_filespec *one,
1015                                  struct diff_filespec *two)
1016 {
1017         int i;
1018         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1019         memcpy(o, orig_opts, sizeof(struct diff_options));
1020
1021         ecbdata->diff_words =
1022                 xcalloc(1, sizeof(struct diff_words_data));
1023         ecbdata->diff_words->type = o->word_diff;
1024         ecbdata->diff_words->opt = o;
1025         if (!o->word_regex)
1026                 o->word_regex = userdiff_word_regex(one);
1027         if (!o->word_regex)
1028                 o->word_regex = userdiff_word_regex(two);
1029         if (!o->word_regex)
1030                 o->word_regex = diff_word_regex_cfg;
1031         if (o->word_regex) {
1032                 ecbdata->diff_words->word_regex = (regex_t *)
1033                         xmalloc(sizeof(regex_t));
1034                 if (regcomp(ecbdata->diff_words->word_regex,
1035                             o->word_regex,
1036                             REG_EXTENDED | REG_NEWLINE))
1037                         die ("Invalid regular expression: %s",
1038                              o->word_regex);
1039         }
1040         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1041                 if (o->word_diff == diff_words_styles[i].type) {
1042                         ecbdata->diff_words->style =
1043                                 &diff_words_styles[i];
1044                         break;
1045                 }
1046         }
1047         if (want_color(o->use_color)) {
1048                 struct diff_words_style *st = ecbdata->diff_words->style;
1049                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1050                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1051                 st->ctx.color = diff_get_color_opt(o, DIFF_PLAIN);
1052         }
1053 }
1054
1055 static void free_diff_words_data(struct emit_callback *ecbdata)
1056 {
1057         if (ecbdata->diff_words) {
1058                 diff_words_flush(ecbdata);
1059                 free (ecbdata->diff_words->opt);
1060                 free (ecbdata->diff_words->minus.text.ptr);
1061                 free (ecbdata->diff_words->minus.orig);
1062                 free (ecbdata->diff_words->plus.text.ptr);
1063                 free (ecbdata->diff_words->plus.orig);
1064                 if (ecbdata->diff_words->word_regex) {
1065                         regfree(ecbdata->diff_words->word_regex);
1066                         free(ecbdata->diff_words->word_regex);
1067                 }
1068                 free(ecbdata->diff_words);
1069                 ecbdata->diff_words = NULL;
1070         }
1071 }
1072
1073 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1074 {
1075         if (want_color(diff_use_color))
1076                 return diff_colors[ix];
1077         return "";
1078 }
1079
1080 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1081 {
1082         const char *cp;
1083         unsigned long allot;
1084         size_t l = len;
1085
1086         if (ecb->truncate)
1087                 return ecb->truncate(line, len);
1088         cp = line;
1089         allot = l;
1090         while (0 < l) {
1091                 (void) utf8_width(&cp, &l);
1092                 if (!cp)
1093                         break; /* truncated in the middle? */
1094         }
1095         return allot - l;
1096 }
1097
1098 static void find_lno(const char *line, struct emit_callback *ecbdata)
1099 {
1100         const char *p;
1101         ecbdata->lno_in_preimage = 0;
1102         ecbdata->lno_in_postimage = 0;
1103         p = strchr(line, '-');
1104         if (!p)
1105                 return; /* cannot happen */
1106         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1107         p = strchr(p, '+');
1108         if (!p)
1109                 return; /* cannot happen */
1110         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1111 }
1112
1113 static void fn_out_consume(void *priv, char *line, unsigned long len)
1114 {
1115         struct emit_callback *ecbdata = priv;
1116         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1117         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
1118         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1119         struct diff_options *o = ecbdata->opt;
1120         char *line_prefix = "";
1121         struct strbuf *msgbuf;
1122
1123         if (o && o->output_prefix) {
1124                 msgbuf = o->output_prefix(o, o->output_prefix_data);
1125                 line_prefix = msgbuf->buf;
1126         }
1127
1128         if (ecbdata->header) {
1129                 fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
1130                 strbuf_reset(ecbdata->header);
1131                 ecbdata->header = NULL;
1132         }
1133         *(ecbdata->found_changesp) = 1;
1134
1135         if (ecbdata->label_path[0]) {
1136                 const char *name_a_tab, *name_b_tab;
1137
1138                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1139                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1140
1141                 fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1142                         line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1143                 fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1144                         line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1145                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1146         }
1147
1148         if (diff_suppress_blank_empty
1149             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1150                 line[0] = '\n';
1151                 len = 1;
1152         }
1153
1154         if (line[0] == '@') {
1155                 if (ecbdata->diff_words)
1156                         diff_words_flush(ecbdata);
1157                 len = sane_truncate_line(ecbdata, line, len);
1158                 find_lno(line, ecbdata);
1159                 emit_hunk_header(ecbdata, line, len);
1160                 if (line[len-1] != '\n')
1161                         putc('\n', ecbdata->opt->file);
1162                 return;
1163         }
1164
1165         if (len < 1) {
1166                 emit_line(ecbdata->opt, reset, reset, line, len);
1167                 if (ecbdata->diff_words
1168                     && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1169                         fputs("~\n", ecbdata->opt->file);
1170                 return;
1171         }
1172
1173         if (ecbdata->diff_words) {
1174                 if (line[0] == '-') {
1175                         diff_words_append(line, len,
1176                                           &ecbdata->diff_words->minus);
1177                         return;
1178                 } else if (line[0] == '+') {
1179                         diff_words_append(line, len,
1180                                           &ecbdata->diff_words->plus);
1181                         return;
1182                 } else if (!prefixcmp(line, "\\ ")) {
1183                         /*
1184                          * Eat the "no newline at eof" marker as if we
1185                          * saw a "+" or "-" line with nothing on it,
1186                          * and return without diff_words_flush() to
1187                          * defer processing. If this is the end of
1188                          * preimage, more "+" lines may come after it.
1189                          */
1190                         return;
1191                 }
1192                 diff_words_flush(ecbdata);
1193                 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1194                         emit_line(ecbdata->opt, plain, reset, line, len);
1195                         fputs("~\n", ecbdata->opt->file);
1196                 } else {
1197                         /*
1198                          * Skip the prefix character, if any.  With
1199                          * diff_suppress_blank_empty, there may be
1200                          * none.
1201                          */
1202                         if (line[0] != '\n') {
1203                               line++;
1204                               len--;
1205                         }
1206                         emit_line(ecbdata->opt, plain, reset, line, len);
1207                 }
1208                 return;
1209         }
1210
1211         if (line[0] != '+') {
1212                 const char *color =
1213                         diff_get_color(ecbdata->color_diff,
1214                                        line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
1215                 ecbdata->lno_in_preimage++;
1216                 if (line[0] == ' ')
1217                         ecbdata->lno_in_postimage++;
1218                 emit_line(ecbdata->opt, color, reset, line, len);
1219         } else {
1220                 ecbdata->lno_in_postimage++;
1221                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1222         }
1223 }
1224
1225 static char *pprint_rename(const char *a, const char *b)
1226 {
1227         const char *old = a;
1228         const char *new = b;
1229         struct strbuf name = STRBUF_INIT;
1230         int pfx_length, sfx_length;
1231         int len_a = strlen(a);
1232         int len_b = strlen(b);
1233         int a_midlen, b_midlen;
1234         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1235         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1236
1237         if (qlen_a || qlen_b) {
1238                 quote_c_style(a, &name, NULL, 0);
1239                 strbuf_addstr(&name, " => ");
1240                 quote_c_style(b, &name, NULL, 0);
1241                 return strbuf_detach(&name, NULL);
1242         }
1243
1244         /* Find common prefix */
1245         pfx_length = 0;
1246         while (*old && *new && *old == *new) {
1247                 if (*old == '/')
1248                         pfx_length = old - a + 1;
1249                 old++;
1250                 new++;
1251         }
1252
1253         /* Find common suffix */
1254         old = a + len_a;
1255         new = b + len_b;
1256         sfx_length = 0;
1257         while (a <= old && b <= new && *old == *new) {
1258                 if (*old == '/')
1259                         sfx_length = len_a - (old - a);
1260                 old--;
1261                 new--;
1262         }
1263
1264         /*
1265          * pfx{mid-a => mid-b}sfx
1266          * {pfx-a => pfx-b}sfx
1267          * pfx{sfx-a => sfx-b}
1268          * name-a => name-b
1269          */
1270         a_midlen = len_a - pfx_length - sfx_length;
1271         b_midlen = len_b - pfx_length - sfx_length;
1272         if (a_midlen < 0)
1273                 a_midlen = 0;
1274         if (b_midlen < 0)
1275                 b_midlen = 0;
1276
1277         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1278         if (pfx_length + sfx_length) {
1279                 strbuf_add(&name, a, pfx_length);
1280                 strbuf_addch(&name, '{');
1281         }
1282         strbuf_add(&name, a + pfx_length, a_midlen);
1283         strbuf_addstr(&name, " => ");
1284         strbuf_add(&name, b + pfx_length, b_midlen);
1285         if (pfx_length + sfx_length) {
1286                 strbuf_addch(&name, '}');
1287                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1288         }
1289         return strbuf_detach(&name, NULL);
1290 }
1291
1292 struct diffstat_t {
1293         int nr;
1294         int alloc;
1295         struct diffstat_file {
1296                 char *from_name;
1297                 char *name;
1298                 char *print_name;
1299                 unsigned is_unmerged:1;
1300                 unsigned is_binary:1;
1301                 unsigned is_renamed:1;
1302                 uintmax_t added, deleted;
1303         } **files;
1304 };
1305
1306 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1307                                           const char *name_a,
1308                                           const char *name_b)
1309 {
1310         struct diffstat_file *x;
1311         x = xcalloc(sizeof (*x), 1);
1312         if (diffstat->nr == diffstat->alloc) {
1313                 diffstat->alloc = alloc_nr(diffstat->alloc);
1314                 diffstat->files = xrealloc(diffstat->files,
1315                                 diffstat->alloc * sizeof(x));
1316         }
1317         diffstat->files[diffstat->nr++] = x;
1318         if (name_b) {
1319                 x->from_name = xstrdup(name_a);
1320                 x->name = xstrdup(name_b);
1321                 x->is_renamed = 1;
1322         }
1323         else {
1324                 x->from_name = NULL;
1325                 x->name = xstrdup(name_a);
1326         }
1327         return x;
1328 }
1329
1330 static void diffstat_consume(void *priv, char *line, unsigned long len)
1331 {
1332         struct diffstat_t *diffstat = priv;
1333         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1334
1335         if (line[0] == '+')
1336                 x->added++;
1337         else if (line[0] == '-')
1338                 x->deleted++;
1339 }
1340
1341 const char mime_boundary_leader[] = "------------";
1342
1343 static int scale_linear(int it, int width, int max_change)
1344 {
1345         if (!it)
1346                 return 0;
1347         /*
1348          * make sure that at least one '-' or '+' is printed if
1349          * there is any change to this path. The easiest way is to
1350          * scale linearly as if the alloted width is one column shorter
1351          * than it is, and then add 1 to the result.
1352          */
1353         return 1 + (it * (width - 1) / max_change);
1354 }
1355
1356 static void show_name(FILE *file,
1357                       const char *prefix, const char *name, int len)
1358 {
1359         fprintf(file, " %s%-*s |", prefix, len, name);
1360 }
1361
1362 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1363 {
1364         if (cnt <= 0)
1365                 return;
1366         fprintf(file, "%s", set);
1367         while (cnt--)
1368                 putc(ch, file);
1369         fprintf(file, "%s", reset);
1370 }
1371
1372 static void fill_print_name(struct diffstat_file *file)
1373 {
1374         char *pname;
1375
1376         if (file->print_name)
1377                 return;
1378
1379         if (!file->is_renamed) {
1380                 struct strbuf buf = STRBUF_INIT;
1381                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1382                         pname = strbuf_detach(&buf, NULL);
1383                 } else {
1384                         pname = file->name;
1385                         strbuf_release(&buf);
1386                 }
1387         } else {
1388                 pname = pprint_rename(file->from_name, file->name);
1389         }
1390         file->print_name = pname;
1391 }
1392
1393 int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1394 {
1395         struct strbuf sb = STRBUF_INIT;
1396         int ret;
1397
1398         if (!files) {
1399                 assert(insertions == 0 && deletions == 0);
1400                 return fputs(_(" 0 files changed\n"), fp);
1401         }
1402
1403         strbuf_addf(&sb,
1404                     Q_(" %d file changed", " %d files changed", files),
1405                     files);
1406
1407         /*
1408          * For binary diff, the caller may want to print "x files
1409          * changed" with insertions == 0 && deletions == 0.
1410          *
1411          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1412          * is probably less confusing (i.e skip over "2 files changed
1413          * but nothing about added/removed lines? Is this a bug in Git?").
1414          */
1415         if (insertions || deletions == 0) {
1416                 /*
1417                  * TRANSLATORS: "+" in (+) is a line addition marker;
1418                  * do not translate it.
1419                  */
1420                 strbuf_addf(&sb,
1421                             Q_(", %d insertion(+)", ", %d insertions(+)",
1422                                insertions),
1423                             insertions);
1424         }
1425
1426         if (deletions || insertions == 0) {
1427                 /*
1428                  * TRANSLATORS: "-" in (-) is a line removal marker;
1429                  * do not translate it.
1430                  */
1431                 strbuf_addf(&sb,
1432                             Q_(", %d deletion(-)", ", %d deletions(-)",
1433                                deletions),
1434                             deletions);
1435         }
1436         strbuf_addch(&sb, '\n');
1437         ret = fputs(sb.buf, fp);
1438         strbuf_release(&sb);
1439         return ret;
1440 }
1441
1442 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1443 {
1444         int i, len, add, del, adds = 0, dels = 0;
1445         uintmax_t max_change = 0, max_len = 0;
1446         int total_files = data->nr;
1447         int width, name_width, graph_width, number_width = 4, count;
1448         const char *reset, *add_c, *del_c;
1449         const char *line_prefix = "";
1450         int extra_shown = 0;
1451         struct strbuf *msg = NULL;
1452
1453         if (data->nr == 0)
1454                 return;
1455
1456         if (options->output_prefix) {
1457                 msg = options->output_prefix(options, options->output_prefix_data);
1458                 line_prefix = msg->buf;
1459         }
1460
1461         count = options->stat_count ? options->stat_count : data->nr;
1462
1463         reset = diff_get_color_opt(options, DIFF_RESET);
1464         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1465         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1466
1467         /*
1468          * Find the longest filename and max number of changes
1469          */
1470         for (i = 0; (i < count) && (i < data->nr); i++) {
1471                 struct diffstat_file *file = data->files[i];
1472                 uintmax_t change = file->added + file->deleted;
1473                 if (!data->files[i]->is_renamed &&
1474                          (change == 0)) {
1475                         count++; /* not shown == room for one more */
1476                         continue;
1477                 }
1478                 fill_print_name(file);
1479                 len = strlen(file->print_name);
1480                 if (max_len < len)
1481                         max_len = len;
1482
1483                 if (file->is_binary || file->is_unmerged)
1484                         continue;
1485                 if (max_change < change)
1486                         max_change = change;
1487         }
1488         count = i; /* min(count, data->nr) */
1489
1490         /*
1491          * We have width = stat_width or term_columns() columns total.
1492          * We want a maximum of min(max_len, stat_name_width) for the name part.
1493          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1494          * We also need 1 for " " and 4 + decimal_width(max_change)
1495          * for " | NNNN " and one the empty column at the end, altogether
1496          * 6 + decimal_width(max_change).
1497          *
1498          * If there's not enough space, we will use the smaller of
1499          * stat_name_width (if set) and 5/8*width for the filename,
1500          * and the rest for constant elements + graph part, but no more
1501          * than stat_graph_width for the graph part.
1502          * (5/8 gives 50 for filename and 30 for the constant parts + graph
1503          * for the standard terminal size).
1504          *
1505          * In other words: stat_width limits the maximum width, and
1506          * stat_name_width fixes the maximum width of the filename,
1507          * and is also used to divide available columns if there
1508          * aren't enough.
1509          */
1510
1511         if (options->stat_width == -1)
1512                 width = term_columns();
1513         else
1514                 width = options->stat_width ? options->stat_width : 80;
1515
1516         if (options->stat_graph_width == -1)
1517                 options->stat_graph_width = diff_stat_graph_width;
1518
1519         /*
1520          * Guarantee 3/8*16==6 for the graph part
1521          * and 5/8*16==10 for the filename part
1522          */
1523         if (width < 16 + 6 + number_width)
1524                 width = 16 + 6 + number_width;
1525
1526         /*
1527          * First assign sizes that are wanted, ignoring available width.
1528          */
1529         graph_width = (options->stat_graph_width &&
1530                        options->stat_graph_width < max_change) ?
1531                 options->stat_graph_width : max_change;
1532         name_width = (options->stat_name_width > 0 &&
1533                       options->stat_name_width < max_len) ?
1534                 options->stat_name_width : max_len;
1535
1536         /*
1537          * Adjust adjustable widths not to exceed maximum width
1538          */
1539         if (name_width + number_width + 6 + graph_width > width) {
1540                 if (graph_width > width * 3/8 - number_width - 6)
1541                         graph_width = width * 3/8 - number_width - 6;
1542                 if (options->stat_graph_width &&
1543                     graph_width > options->stat_graph_width)
1544                         graph_width = options->stat_graph_width;
1545                 if (name_width > width - number_width - 6 - graph_width)
1546                         name_width = width - number_width - 6 - graph_width;
1547                 else
1548                         graph_width = width - number_width - 6 - name_width;
1549         }
1550
1551         /*
1552          * From here name_width is the width of the name area,
1553          * and graph_width is the width of the graph area.
1554          * max_change is used to scale graph properly.
1555          */
1556         for (i = 0; i < count; i++) {
1557                 const char *prefix = "";
1558                 char *name = data->files[i]->print_name;
1559                 uintmax_t added = data->files[i]->added;
1560                 uintmax_t deleted = data->files[i]->deleted;
1561                 int name_len;
1562
1563                 if (!data->files[i]->is_renamed &&
1564                          (added + deleted == 0)) {
1565                         total_files--;
1566                         continue;
1567                 }
1568                 /*
1569                  * "scale" the filename
1570                  */
1571                 len = name_width;
1572                 name_len = strlen(name);
1573                 if (name_width < name_len) {
1574                         char *slash;
1575                         prefix = "...";
1576                         len -= 3;
1577                         name += name_len - len;
1578                         slash = strchr(name, '/');
1579                         if (slash)
1580                                 name = slash;
1581                 }
1582
1583                 if (data->files[i]->is_binary) {
1584                         fprintf(options->file, "%s", line_prefix);
1585                         show_name(options->file, prefix, name, len);
1586                         fprintf(options->file, "  Bin ");
1587                         fprintf(options->file, "%s%"PRIuMAX"%s",
1588                                 del_c, deleted, reset);
1589                         fprintf(options->file, " -> ");
1590                         fprintf(options->file, "%s%"PRIuMAX"%s",
1591                                 add_c, added, reset);
1592                         fprintf(options->file, " bytes");
1593                         fprintf(options->file, "\n");
1594                         continue;
1595                 }
1596                 else if (data->files[i]->is_unmerged) {
1597                         fprintf(options->file, "%s", line_prefix);
1598                         show_name(options->file, prefix, name, len);
1599                         fprintf(options->file, "  Unmerged\n");
1600                         continue;
1601                 }
1602
1603                 /*
1604                  * scale the add/delete
1605                  */
1606                 add = added;
1607                 del = deleted;
1608                 adds += add;
1609                 dels += del;
1610
1611                 if (graph_width <= max_change) {
1612                         int total = add + del;
1613
1614                         total = scale_linear(add + del, graph_width, max_change);
1615                         if (total < 2 && add && del)
1616                                 /* width >= 2 due to the sanity check */
1617                                 total = 2;
1618                         if (add < del) {
1619                                 add = scale_linear(add, graph_width, max_change);
1620                                 del = total - add;
1621                         } else {
1622                                 del = scale_linear(del, graph_width, max_change);
1623                                 add = total - del;
1624                         }
1625                 }
1626                 fprintf(options->file, "%s", line_prefix);
1627                 show_name(options->file, prefix, name, len);
1628                 fprintf(options->file, "%5"PRIuMAX"%s", added + deleted,
1629                                 added + deleted ? " " : "");
1630                 show_graph(options->file, '+', add, add_c, reset);
1631                 show_graph(options->file, '-', del, del_c, reset);
1632                 fprintf(options->file, "\n");
1633         }
1634         for (i = count; i < data->nr; i++) {
1635                 uintmax_t added = data->files[i]->added;
1636                 uintmax_t deleted = data->files[i]->deleted;
1637                 if (!data->files[i]->is_renamed &&
1638                          (added + deleted == 0)) {
1639                         total_files--;
1640                         continue;
1641                 }
1642                 adds += added;
1643                 dels += deleted;
1644                 if (!extra_shown)
1645                         fprintf(options->file, "%s ...\n", line_prefix);
1646                 extra_shown = 1;
1647         }
1648         fprintf(options->file, "%s", line_prefix);
1649         print_stat_summary(options->file, total_files, adds, dels);
1650 }
1651
1652 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1653 {
1654         int i, adds = 0, dels = 0, total_files = data->nr;
1655
1656         if (data->nr == 0)
1657                 return;
1658
1659         for (i = 0; i < data->nr; i++) {
1660                 if (!data->files[i]->is_binary &&
1661                     !data->files[i]->is_unmerged) {
1662                         int added = data->files[i]->added;
1663                         int deleted= data->files[i]->deleted;
1664                         if (!data->files[i]->is_renamed &&
1665                             (added + deleted == 0)) {
1666                                 total_files--;
1667                         } else {
1668                                 adds += added;
1669                                 dels += deleted;
1670                         }
1671                 }
1672         }
1673         if (options->output_prefix) {
1674                 struct strbuf *msg = NULL;
1675                 msg = options->output_prefix(options,
1676                                 options->output_prefix_data);
1677                 fprintf(options->file, "%s", msg->buf);
1678         }
1679         print_stat_summary(options->file, total_files, adds, dels);
1680 }
1681
1682 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1683 {
1684         int i;
1685
1686         if (data->nr == 0)
1687                 return;
1688
1689         for (i = 0; i < data->nr; i++) {
1690                 struct diffstat_file *file = data->files[i];
1691
1692                 if (options->output_prefix) {
1693                         struct strbuf *msg = NULL;
1694                         msg = options->output_prefix(options,
1695                                         options->output_prefix_data);
1696                         fprintf(options->file, "%s", msg->buf);
1697                 }
1698
1699                 if (file->is_binary)
1700                         fprintf(options->file, "-\t-\t");
1701                 else
1702                         fprintf(options->file,
1703                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1704                                 file->added, file->deleted);
1705                 if (options->line_termination) {
1706                         fill_print_name(file);
1707                         if (!file->is_renamed)
1708                                 write_name_quoted(file->name, options->file,
1709                                                   options->line_termination);
1710                         else {
1711                                 fputs(file->print_name, options->file);
1712                                 putc(options->line_termination, options->file);
1713                         }
1714                 } else {
1715                         if (file->is_renamed) {
1716                                 putc('\0', options->file);
1717                                 write_name_quoted(file->from_name, options->file, '\0');
1718                         }
1719                         write_name_quoted(file->name, options->file, '\0');
1720                 }
1721         }
1722 }
1723
1724 struct dirstat_file {
1725         const char *name;
1726         unsigned long changed;
1727 };
1728
1729 struct dirstat_dir {
1730         struct dirstat_file *files;
1731         int alloc, nr, permille, cumulative;
1732 };
1733
1734 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1735                 unsigned long changed, const char *base, int baselen)
1736 {
1737         unsigned long this_dir = 0;
1738         unsigned int sources = 0;
1739         const char *line_prefix = "";
1740         struct strbuf *msg = NULL;
1741
1742         if (opt->output_prefix) {
1743                 msg = opt->output_prefix(opt, opt->output_prefix_data);
1744                 line_prefix = msg->buf;
1745         }
1746
1747         while (dir->nr) {
1748                 struct dirstat_file *f = dir->files;
1749                 int namelen = strlen(f->name);
1750                 unsigned long this;
1751                 char *slash;
1752
1753                 if (namelen < baselen)
1754                         break;
1755                 if (memcmp(f->name, base, baselen))
1756                         break;
1757                 slash = strchr(f->name + baselen, '/');
1758                 if (slash) {
1759                         int newbaselen = slash + 1 - f->name;
1760                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1761                         sources++;
1762                 } else {
1763                         this = f->changed;
1764                         dir->files++;
1765                         dir->nr--;
1766                         sources += 2;
1767                 }
1768                 this_dir += this;
1769         }
1770
1771         /*
1772          * We don't report dirstat's for
1773          *  - the top level
1774          *  - or cases where everything came from a single directory
1775          *    under this directory (sources == 1).
1776          */
1777         if (baselen && sources != 1) {
1778                 if (this_dir) {
1779                         int permille = this_dir * 1000 / changed;
1780                         if (permille >= dir->permille) {
1781                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1782                                         permille / 10, permille % 10, baselen, base);
1783                                 if (!dir->cumulative)
1784                                         return 0;
1785                         }
1786                 }
1787         }
1788         return this_dir;
1789 }
1790
1791 static int dirstat_compare(const void *_a, const void *_b)
1792 {
1793         const struct dirstat_file *a = _a;
1794         const struct dirstat_file *b = _b;
1795         return strcmp(a->name, b->name);
1796 }
1797
1798 static void show_dirstat(struct diff_options *options)
1799 {
1800         int i;
1801         unsigned long changed;
1802         struct dirstat_dir dir;
1803         struct diff_queue_struct *q = &diff_queued_diff;
1804
1805         dir.files = NULL;
1806         dir.alloc = 0;
1807         dir.nr = 0;
1808         dir.permille = options->dirstat_permille;
1809         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1810
1811         changed = 0;
1812         for (i = 0; i < q->nr; i++) {
1813                 struct diff_filepair *p = q->queue[i];
1814                 const char *name;
1815                 unsigned long copied, added, damage;
1816                 int content_changed;
1817
1818                 name = p->two->path ? p->two->path : p->one->path;
1819
1820                 if (p->one->sha1_valid && p->two->sha1_valid)
1821                         content_changed = hashcmp(p->one->sha1, p->two->sha1);
1822                 else
1823                         content_changed = 1;
1824
1825                 if (!content_changed) {
1826                         /*
1827                          * The SHA1 has not changed, so pre-/post-content is
1828                          * identical. We can therefore skip looking at the
1829                          * file contents altogether.
1830                          */
1831                         damage = 0;
1832                         goto found_damage;
1833                 }
1834
1835                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1836                         /*
1837                          * In --dirstat-by-file mode, we don't really need to
1838                          * look at the actual file contents at all.
1839                          * The fact that the SHA1 changed is enough for us to
1840                          * add this file to the list of results
1841                          * (with each file contributing equal damage).
1842                          */
1843                         damage = 1;
1844                         goto found_damage;
1845                 }
1846
1847                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1848                         diff_populate_filespec(p->one, 0);
1849                         diff_populate_filespec(p->two, 0);
1850                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1851                                                &copied, &added);
1852                         diff_free_filespec_data(p->one);
1853                         diff_free_filespec_data(p->two);
1854                 } else if (DIFF_FILE_VALID(p->one)) {
1855                         diff_populate_filespec(p->one, 1);
1856                         copied = added = 0;
1857                         diff_free_filespec_data(p->one);
1858                 } else if (DIFF_FILE_VALID(p->two)) {
1859                         diff_populate_filespec(p->two, 1);
1860                         copied = 0;
1861                         added = p->two->size;
1862                         diff_free_filespec_data(p->two);
1863                 } else
1864                         continue;
1865
1866                 /*
1867                  * Original minus copied is the removed material,
1868                  * added is the new material.  They are both damages
1869                  * made to the preimage.
1870                  * If the resulting damage is zero, we know that
1871                  * diffcore_count_changes() considers the two entries to
1872                  * be identical, but since content_changed is true, we
1873                  * know that there must have been _some_ kind of change,
1874                  * so we force all entries to have damage > 0.
1875                  */
1876                 damage = (p->one->size - copied) + added;
1877                 if (!damage)
1878                         damage = 1;
1879
1880 found_damage:
1881                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1882                 dir.files[dir.nr].name = name;
1883                 dir.files[dir.nr].changed = damage;
1884                 changed += damage;
1885                 dir.nr++;
1886         }
1887
1888         /* This can happen even with many files, if everything was renames */
1889         if (!changed)
1890                 return;
1891
1892         /* Show all directories with more than x% of the changes */
1893         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1894         gather_dirstat(options, &dir, changed, "", 0);
1895 }
1896
1897 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
1898 {
1899         int i;
1900         unsigned long changed;
1901         struct dirstat_dir dir;
1902
1903         if (data->nr == 0)
1904                 return;
1905
1906         dir.files = NULL;
1907         dir.alloc = 0;
1908         dir.nr = 0;
1909         dir.permille = options->dirstat_permille;
1910         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1911
1912         changed = 0;
1913         for (i = 0; i < data->nr; i++) {
1914                 struct diffstat_file *file = data->files[i];
1915                 unsigned long damage = file->added + file->deleted;
1916                 if (file->is_binary)
1917                         /*
1918                          * binary files counts bytes, not lines. Must find some
1919                          * way to normalize binary bytes vs. textual lines.
1920                          * The following heuristic assumes that there are 64
1921                          * bytes per "line".
1922                          * This is stupid and ugly, but very cheap...
1923                          */
1924                         damage = (damage + 63) / 64;
1925                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1926                 dir.files[dir.nr].name = file->name;
1927                 dir.files[dir.nr].changed = damage;
1928                 changed += damage;
1929                 dir.nr++;
1930         }
1931
1932         /* This can happen even with many files, if everything was renames */
1933         if (!changed)
1934                 return;
1935
1936         /* Show all directories with more than x% of the changes */
1937         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1938         gather_dirstat(options, &dir, changed, "", 0);
1939 }
1940
1941 static void free_diffstat_info(struct diffstat_t *diffstat)
1942 {
1943         int i;
1944         for (i = 0; i < diffstat->nr; i++) {
1945                 struct diffstat_file *f = diffstat->files[i];
1946                 if (f->name != f->print_name)
1947                         free(f->print_name);
1948                 free(f->name);
1949                 free(f->from_name);
1950                 free(f);
1951         }
1952         free(diffstat->files);
1953 }
1954
1955 struct checkdiff_t {
1956         const char *filename;
1957         int lineno;
1958         int conflict_marker_size;
1959         struct diff_options *o;
1960         unsigned ws_rule;
1961         unsigned status;
1962 };
1963
1964 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
1965 {
1966         char firstchar;
1967         int cnt;
1968
1969         if (len < marker_size + 1)
1970                 return 0;
1971         firstchar = line[0];
1972         switch (firstchar) {
1973         case '=': case '>': case '<': case '|':
1974                 break;
1975         default:
1976                 return 0;
1977         }
1978         for (cnt = 1; cnt < marker_size; cnt++)
1979                 if (line[cnt] != firstchar)
1980                         return 0;
1981         /* line[1] thru line[marker_size-1] are same as firstchar */
1982         if (len < marker_size + 1 || !isspace(line[marker_size]))
1983                 return 0;
1984         return 1;
1985 }
1986
1987 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1988 {
1989         struct checkdiff_t *data = priv;
1990         int marker_size = data->conflict_marker_size;
1991         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
1992         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
1993         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
1994         char *err;
1995         char *line_prefix = "";
1996         struct strbuf *msgbuf;
1997
1998         assert(data->o);
1999         if (data->o->output_prefix) {
2000                 msgbuf = data->o->output_prefix(data->o,
2001                         data->o->output_prefix_data);
2002                 line_prefix = msgbuf->buf;
2003         }
2004
2005         if (line[0] == '+') {
2006                 unsigned bad;
2007                 data->lineno++;
2008                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2009                         data->status |= 1;
2010                         fprintf(data->o->file,
2011                                 "%s%s:%d: leftover conflict marker\n",
2012                                 line_prefix, data->filename, data->lineno);
2013                 }
2014                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2015                 if (!bad)
2016                         return;
2017                 data->status |= bad;
2018                 err = whitespace_error_string(bad);
2019                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2020                         line_prefix, data->filename, data->lineno, err);
2021                 free(err);
2022                 emit_line(data->o, set, reset, line, 1);
2023                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2024                               data->o->file, set, reset, ws);
2025         } else if (line[0] == ' ') {
2026                 data->lineno++;
2027         } else if (line[0] == '@') {
2028                 char *plus = strchr(line, '+');
2029                 if (plus)
2030                         data->lineno = strtol(plus, NULL, 10) - 1;
2031                 else
2032                         die("invalid diff");
2033         }
2034 }
2035
2036 static unsigned char *deflate_it(char *data,
2037                                  unsigned long size,
2038                                  unsigned long *result_size)
2039 {
2040         int bound;
2041         unsigned char *deflated;
2042         git_zstream stream;
2043
2044         memset(&stream, 0, sizeof(stream));
2045         git_deflate_init(&stream, zlib_compression_level);
2046         bound = git_deflate_bound(&stream, size);
2047         deflated = xmalloc(bound);
2048         stream.next_out = deflated;
2049         stream.avail_out = bound;
2050
2051         stream.next_in = (unsigned char *)data;
2052         stream.avail_in = size;
2053         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2054                 ; /* nothing */
2055         git_deflate_end(&stream);
2056         *result_size = stream.total_out;
2057         return deflated;
2058 }
2059
2060 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
2061 {
2062         void *cp;
2063         void *delta;
2064         void *deflated;
2065         void *data;
2066         unsigned long orig_size;
2067         unsigned long delta_size;
2068         unsigned long deflate_size;
2069         unsigned long data_size;
2070
2071         /* We could do deflated delta, or we could do just deflated two,
2072          * whichever is smaller.
2073          */
2074         delta = NULL;
2075         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2076         if (one->size && two->size) {
2077                 delta = diff_delta(one->ptr, one->size,
2078                                    two->ptr, two->size,
2079                                    &delta_size, deflate_size);
2080                 if (delta) {
2081                         void *to_free = delta;
2082                         orig_size = delta_size;
2083                         delta = deflate_it(delta, delta_size, &delta_size);
2084                         free(to_free);
2085                 }
2086         }
2087
2088         if (delta && delta_size < deflate_size) {
2089                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2090                 free(deflated);
2091                 data = delta;
2092                 data_size = delta_size;
2093         }
2094         else {
2095                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
2096                 free(delta);
2097                 data = deflated;
2098                 data_size = deflate_size;
2099         }
2100
2101         /* emit data encoded in base85 */
2102         cp = data;
2103         while (data_size) {
2104                 int bytes = (52 < data_size) ? 52 : data_size;
2105                 char line[70];
2106                 data_size -= bytes;
2107                 if (bytes <= 26)
2108                         line[0] = bytes + 'A' - 1;
2109                 else
2110                         line[0] = bytes - 26 + 'a' - 1;
2111                 encode_85(line + 1, cp, bytes);
2112                 cp = (char *) cp + bytes;
2113                 fprintf(file, "%s", prefix);
2114                 fputs(line, file);
2115                 fputc('\n', file);
2116         }
2117         fprintf(file, "%s\n", prefix);
2118         free(data);
2119 }
2120
2121 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
2122 {
2123         fprintf(file, "%sGIT binary patch\n", prefix);
2124         emit_binary_diff_body(file, one, two, prefix);
2125         emit_binary_diff_body(file, two, one, prefix);
2126 }
2127
2128 int diff_filespec_is_binary(struct diff_filespec *one)
2129 {
2130         if (one->is_binary == -1) {
2131                 diff_filespec_load_driver(one);
2132                 if (one->driver->binary != -1)
2133                         one->is_binary = one->driver->binary;
2134                 else {
2135                         if (!one->data && DIFF_FILE_VALID(one))
2136                                 diff_populate_filespec(one, 0);
2137                         if (one->data)
2138                                 one->is_binary = buffer_is_binary(one->data,
2139                                                 one->size);
2140                         if (one->is_binary == -1)
2141                                 one->is_binary = 0;
2142                 }
2143         }
2144         return one->is_binary;
2145 }
2146
2147 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2148 {
2149         diff_filespec_load_driver(one);
2150         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2151 }
2152
2153 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2154 {
2155         if (!options->a_prefix)
2156                 options->a_prefix = a;
2157         if (!options->b_prefix)
2158                 options->b_prefix = b;
2159 }
2160
2161 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2162 {
2163         if (!DIFF_FILE_VALID(one))
2164                 return NULL;
2165
2166         diff_filespec_load_driver(one);
2167         return userdiff_get_textconv(one->driver);
2168 }
2169
2170 static void builtin_diff(const char *name_a,
2171                          const char *name_b,
2172                          struct diff_filespec *one,
2173                          struct diff_filespec *two,
2174                          const char *xfrm_msg,
2175                          int must_show_header,
2176                          struct diff_options *o,
2177                          int complete_rewrite)
2178 {
2179         mmfile_t mf1, mf2;
2180         const char *lbl[2];
2181         char *a_one, *b_two;
2182         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
2183         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2184         const char *a_prefix, *b_prefix;
2185         struct userdiff_driver *textconv_one = NULL;
2186         struct userdiff_driver *textconv_two = NULL;
2187         struct strbuf header = STRBUF_INIT;
2188         struct strbuf *msgbuf;
2189         char *line_prefix = "";
2190
2191         if (o->output_prefix) {
2192                 msgbuf = o->output_prefix(o, o->output_prefix_data);
2193                 line_prefix = msgbuf->buf;
2194         }
2195
2196         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2197                         (!one->mode || S_ISGITLINK(one->mode)) &&
2198                         (!two->mode || S_ISGITLINK(two->mode))) {
2199                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2200                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2201                 show_submodule_summary(o->file, one ? one->path : two->path,
2202                                 one->sha1, two->sha1, two->dirty_submodule,
2203                                 del, add, reset);
2204                 return;
2205         }
2206
2207         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2208                 textconv_one = get_textconv(one);
2209                 textconv_two = get_textconv(two);
2210         }
2211
2212         diff_set_mnemonic_prefix(o, "a/", "b/");
2213         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2214                 a_prefix = o->b_prefix;
2215                 b_prefix = o->a_prefix;
2216         } else {
2217                 a_prefix = o->a_prefix;
2218                 b_prefix = o->b_prefix;
2219         }
2220
2221         /* Never use a non-valid filename anywhere if at all possible */
2222         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2223         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2224
2225         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2226         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2227         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2228         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2229         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, set, a_one, b_two, reset);
2230         if (lbl[0][0] == '/') {
2231                 /* /dev/null */
2232                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, set, two->mode, reset);
2233                 if (xfrm_msg)
2234                         strbuf_addstr(&header, xfrm_msg);
2235                 must_show_header = 1;
2236         }
2237         else if (lbl[1][0] == '/') {
2238                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, set, one->mode, reset);
2239                 if (xfrm_msg)
2240                         strbuf_addstr(&header, xfrm_msg);
2241                 must_show_header = 1;
2242         }
2243         else {
2244                 if (one->mode != two->mode) {
2245                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, set, one->mode, reset);
2246                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, set, two->mode, reset);
2247                         must_show_header = 1;
2248                 }
2249                 if (xfrm_msg)
2250                         strbuf_addstr(&header, xfrm_msg);
2251
2252                 /*
2253                  * we do not run diff between different kind
2254                  * of objects.
2255                  */
2256                 if ((one->mode ^ two->mode) & S_IFMT)
2257                         goto free_ab_and_return;
2258                 if (complete_rewrite &&
2259                     (textconv_one || !diff_filespec_is_binary(one)) &&
2260                     (textconv_two || !diff_filespec_is_binary(two))) {
2261                         fprintf(o->file, "%s", header.buf);
2262                         strbuf_reset(&header);
2263                         emit_rewrite_diff(name_a, name_b, one, two,
2264                                                 textconv_one, textconv_two, o);
2265                         o->found_changes = 1;
2266                         goto free_ab_and_return;
2267                 }
2268         }
2269
2270         if (o->irreversible_delete && lbl[1][0] == '/') {
2271                 fprintf(o->file, "%s", header.buf);
2272                 strbuf_reset(&header);
2273                 goto free_ab_and_return;
2274         } else if (!DIFF_OPT_TST(o, TEXT) &&
2275             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2276               (!textconv_two && diff_filespec_is_binary(two)) )) {
2277                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2278                         die("unable to read files to diff");
2279                 /* Quite common confusing case */
2280                 if (mf1.size == mf2.size &&
2281                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2282                         if (must_show_header)
2283                                 fprintf(o->file, "%s", header.buf);
2284                         goto free_ab_and_return;
2285                 }
2286                 fprintf(o->file, "%s", header.buf);
2287                 strbuf_reset(&header);
2288                 if (DIFF_OPT_TST(o, BINARY))
2289                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2290                 else
2291                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2292                                 line_prefix, lbl[0], lbl[1]);
2293                 o->found_changes = 1;
2294         } else {
2295                 /* Crazy xdl interfaces.. */
2296                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2297                 xpparam_t xpp;
2298                 xdemitconf_t xecfg;
2299                 struct emit_callback ecbdata;
2300                 const struct userdiff_funcname *pe;
2301
2302                 if (must_show_header) {
2303                         fprintf(o->file, "%s", header.buf);
2304                         strbuf_reset(&header);
2305                 }
2306
2307                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2308                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2309
2310                 pe = diff_funcname_pattern(one);
2311                 if (!pe)
2312                         pe = diff_funcname_pattern(two);
2313
2314                 memset(&xpp, 0, sizeof(xpp));
2315                 memset(&xecfg, 0, sizeof(xecfg));
2316                 memset(&ecbdata, 0, sizeof(ecbdata));
2317                 ecbdata.label_path = lbl;
2318                 ecbdata.color_diff = want_color(o->use_color);
2319                 ecbdata.found_changesp = &o->found_changes;
2320                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
2321                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2322                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2323                 ecbdata.opt = o;
2324                 ecbdata.header = header.len ? &header : NULL;
2325                 xpp.flags = o->xdl_opts;
2326                 xecfg.ctxlen = o->context;
2327                 xecfg.interhunkctxlen = o->interhunkcontext;
2328                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2329                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2330                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2331                 if (pe)
2332                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2333                 if (!diffopts)
2334                         ;
2335                 else if (!prefixcmp(diffopts, "--unified="))
2336                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
2337                 else if (!prefixcmp(diffopts, "-u"))
2338                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
2339                 if (o->word_diff)
2340                         init_diff_words_data(&ecbdata, o, one, two);
2341                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2342                               &xpp, &xecfg);
2343                 if (o->word_diff)
2344                         free_diff_words_data(&ecbdata);
2345                 if (textconv_one)
2346                         free(mf1.ptr);
2347                 if (textconv_two)
2348                         free(mf2.ptr);
2349                 xdiff_clear_find_func(&xecfg);
2350         }
2351
2352  free_ab_and_return:
2353         strbuf_release(&header);
2354         diff_free_filespec_data(one);
2355         diff_free_filespec_data(two);
2356         free(a_one);
2357         free(b_two);
2358         return;
2359 }
2360
2361 static void builtin_diffstat(const char *name_a, const char *name_b,
2362                              struct diff_filespec *one,
2363                              struct diff_filespec *two,
2364                              struct diffstat_t *diffstat,
2365                              struct diff_options *o,
2366                              int complete_rewrite)
2367 {
2368         mmfile_t mf1, mf2;
2369         struct diffstat_file *data;
2370
2371         data = diffstat_add(diffstat, name_a, name_b);
2372
2373         if (!one || !two) {
2374                 data->is_unmerged = 1;
2375                 return;
2376         }
2377
2378         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2379                 data->is_binary = 1;
2380                 data->added = diff_filespec_size(two);
2381                 data->deleted = diff_filespec_size(one);
2382         }
2383
2384         else if (complete_rewrite) {
2385                 diff_populate_filespec(one, 0);
2386                 diff_populate_filespec(two, 0);
2387                 data->deleted = count_lines(one->data, one->size);
2388                 data->added = count_lines(two->data, two->size);
2389         }
2390
2391         else {
2392                 /* Crazy xdl interfaces.. */
2393                 xpparam_t xpp;
2394                 xdemitconf_t xecfg;
2395
2396                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2397                         die("unable to read files to diff");
2398
2399                 memset(&xpp, 0, sizeof(xpp));
2400                 memset(&xecfg, 0, sizeof(xecfg));
2401                 xpp.flags = o->xdl_opts;
2402                 xecfg.ctxlen = o->context;
2403                 xecfg.interhunkctxlen = o->interhunkcontext;
2404                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2405                               &xpp, &xecfg);
2406         }
2407
2408         diff_free_filespec_data(one);
2409         diff_free_filespec_data(two);
2410 }
2411
2412 static void builtin_checkdiff(const char *name_a, const char *name_b,
2413                               const char *attr_path,
2414                               struct diff_filespec *one,
2415                               struct diff_filespec *two,
2416                               struct diff_options *o)
2417 {
2418         mmfile_t mf1, mf2;
2419         struct checkdiff_t data;
2420
2421         if (!two)
2422                 return;
2423
2424         memset(&data, 0, sizeof(data));
2425         data.filename = name_b ? name_b : name_a;
2426         data.lineno = 0;
2427         data.o = o;
2428         data.ws_rule = whitespace_rule(attr_path);
2429         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2430
2431         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2432                 die("unable to read files to diff");
2433
2434         /*
2435          * All the other codepaths check both sides, but not checking
2436          * the "old" side here is deliberate.  We are checking the newly
2437          * introduced changes, and as long as the "new" side is text, we
2438          * can and should check what it introduces.
2439          */
2440         if (diff_filespec_is_binary(two))
2441                 goto free_and_return;
2442         else {
2443                 /* Crazy xdl interfaces.. */
2444                 xpparam_t xpp;
2445                 xdemitconf_t xecfg;
2446
2447                 memset(&xpp, 0, sizeof(xpp));
2448                 memset(&xecfg, 0, sizeof(xecfg));
2449                 xecfg.ctxlen = 1; /* at least one context line */
2450                 xpp.flags = 0;
2451                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2452                               &xpp, &xecfg);
2453
2454                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2455                         struct emit_callback ecbdata;
2456                         int blank_at_eof;
2457
2458                         ecbdata.ws_rule = data.ws_rule;
2459                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2460                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2461
2462                         if (blank_at_eof) {
2463                                 static char *err;
2464                                 if (!err)
2465                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2466                                 fprintf(o->file, "%s:%d: %s.\n",
2467                                         data.filename, blank_at_eof, err);
2468                                 data.status = 1; /* report errors */
2469                         }
2470                 }
2471         }
2472  free_and_return:
2473         diff_free_filespec_data(one);
2474         diff_free_filespec_data(two);
2475         if (data.status)
2476                 DIFF_OPT_SET(o, CHECK_FAILED);
2477 }
2478
2479 struct diff_filespec *alloc_filespec(const char *path)
2480 {
2481         int namelen = strlen(path);
2482         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2483
2484         memset(spec, 0, sizeof(*spec));
2485         spec->path = (char *)(spec + 1);
2486         memcpy(spec->path, path, namelen+1);
2487         spec->count = 1;
2488         spec->is_binary = -1;
2489         return spec;
2490 }
2491
2492 void free_filespec(struct diff_filespec *spec)
2493 {
2494         if (!--spec->count) {
2495                 diff_free_filespec_data(spec);
2496                 free(spec);
2497         }
2498 }
2499
2500 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2501                    unsigned short mode)
2502 {
2503         if (mode) {
2504                 spec->mode = canon_mode(mode);
2505                 hashcpy(spec->sha1, sha1);
2506                 spec->sha1_valid = !is_null_sha1(sha1);
2507         }
2508 }
2509
2510 /*
2511  * Given a name and sha1 pair, if the index tells us the file in
2512  * the work tree has that object contents, return true, so that
2513  * prepare_temp_file() does not have to inflate and extract.
2514  */
2515 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2516 {
2517         struct cache_entry *ce;
2518         struct stat st;
2519         int pos, len;
2520
2521         /*
2522          * We do not read the cache ourselves here, because the
2523          * benchmark with my previous version that always reads cache
2524          * shows that it makes things worse for diff-tree comparing
2525          * two linux-2.6 kernel trees in an already checked out work
2526          * tree.  This is because most diff-tree comparisons deal with
2527          * only a small number of files, while reading the cache is
2528          * expensive for a large project, and its cost outweighs the
2529          * savings we get by not inflating the object to a temporary
2530          * file.  Practically, this code only helps when we are used
2531          * by diff-cache --cached, which does read the cache before
2532          * calling us.
2533          */
2534         if (!active_cache)
2535                 return 0;
2536
2537         /* We want to avoid the working directory if our caller
2538          * doesn't need the data in a normal file, this system
2539          * is rather slow with its stat/open/mmap/close syscalls,
2540          * and the object is contained in a pack file.  The pack
2541          * is probably already open and will be faster to obtain
2542          * the data through than the working directory.  Loose
2543          * objects however would tend to be slower as they need
2544          * to be individually opened and inflated.
2545          */
2546         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2547                 return 0;
2548
2549         len = strlen(name);
2550         pos = cache_name_pos(name, len);
2551         if (pos < 0)
2552                 return 0;
2553         ce = active_cache[pos];
2554
2555         /*
2556          * This is not the sha1 we are looking for, or
2557          * unreusable because it is not a regular file.
2558          */
2559         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2560                 return 0;
2561
2562         /*
2563          * If ce is marked as "assume unchanged", there is no
2564          * guarantee that work tree matches what we are looking for.
2565          */
2566         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2567                 return 0;
2568
2569         /*
2570          * If ce matches the file in the work tree, we can reuse it.
2571          */
2572         if (ce_uptodate(ce) ||
2573             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2574                 return 1;
2575
2576         return 0;
2577 }
2578
2579 static int populate_from_stdin(struct diff_filespec *s)
2580 {
2581         struct strbuf buf = STRBUF_INIT;
2582         size_t size = 0;
2583
2584         if (strbuf_read(&buf, 0, 0) < 0)
2585                 return error("error while reading from stdin %s",
2586                                      strerror(errno));
2587
2588         s->should_munmap = 0;
2589         s->data = strbuf_detach(&buf, &size);
2590         s->size = size;
2591         s->should_free = 1;
2592         return 0;
2593 }
2594
2595 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2596 {
2597         int len;
2598         char *data = xmalloc(100), *dirty = "";
2599
2600         /* Are we looking at the work tree? */
2601         if (s->dirty_submodule)
2602                 dirty = "-dirty";
2603
2604         len = snprintf(data, 100,
2605                        "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2606         s->data = data;
2607         s->size = len;
2608         s->should_free = 1;
2609         if (size_only) {
2610                 s->data = NULL;
2611                 free(data);
2612         }
2613         return 0;
2614 }
2615
2616 /*
2617  * While doing rename detection and pickaxe operation, we may need to
2618  * grab the data for the blob (or file) for our own in-core comparison.
2619  * diff_filespec has data and size fields for this purpose.
2620  */
2621 int diff_populate_filespec(struct diff_filespec *s, int size_only)
2622 {
2623         int err = 0;
2624         if (!DIFF_FILE_VALID(s))
2625                 die("internal error: asking to populate invalid file.");
2626         if (S_ISDIR(s->mode))
2627                 return -1;
2628
2629         if (s->data)
2630                 return 0;
2631
2632         if (size_only && 0 < s->size)
2633                 return 0;
2634
2635         if (S_ISGITLINK(s->mode))
2636                 return diff_populate_gitlink(s, size_only);
2637
2638         if (!s->sha1_valid ||
2639             reuse_worktree_file(s->path, s->sha1, 0)) {
2640                 struct strbuf buf = STRBUF_INIT;
2641                 struct stat st;
2642                 int fd;
2643
2644                 if (!strcmp(s->path, "-"))
2645                         return populate_from_stdin(s);
2646
2647                 if (lstat(s->path, &st) < 0) {
2648                         if (errno == ENOENT) {
2649                         err_empty:
2650                                 err = -1;
2651                         empty:
2652                                 s->data = (char *)"";
2653                                 s->size = 0;
2654                                 return err;
2655                         }
2656                 }
2657                 s->size = xsize_t(st.st_size);
2658                 if (!s->size)
2659                         goto empty;
2660                 if (S_ISLNK(st.st_mode)) {
2661                         struct strbuf sb = STRBUF_INIT;
2662
2663                         if (strbuf_readlink(&sb, s->path, s->size))
2664                                 goto err_empty;
2665                         s->size = sb.len;
2666                         s->data = strbuf_detach(&sb, NULL);
2667                         s->should_free = 1;
2668                         return 0;
2669                 }
2670                 if (size_only)
2671                         return 0;
2672                 fd = open(s->path, O_RDONLY);
2673                 if (fd < 0)
2674                         goto err_empty;
2675                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2676                 close(fd);
2677                 s->should_munmap = 1;
2678
2679                 /*
2680                  * Convert from working tree format to canonical git format
2681                  */
2682                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2683                         size_t size = 0;
2684                         munmap(s->data, s->size);
2685                         s->should_munmap = 0;
2686                         s->data = strbuf_detach(&buf, &size);
2687                         s->size = size;
2688                         s->should_free = 1;
2689                 }
2690         }
2691         else {
2692                 enum object_type type;
2693                 if (size_only) {
2694                         type = sha1_object_info(s->sha1, &s->size);
2695                         if (type < 0)
2696                                 die("unable to read %s", sha1_to_hex(s->sha1));
2697                 } else {
2698                         s->data = read_sha1_file(s->sha1, &type, &s->size);
2699                         if (!s->data)
2700                                 die("unable to read %s", sha1_to_hex(s->sha1));
2701                         s->should_free = 1;
2702                 }
2703         }
2704         return 0;
2705 }
2706
2707 void diff_free_filespec_blob(struct diff_filespec *s)
2708 {
2709         if (s->should_free)
2710                 free(s->data);
2711         else if (s->should_munmap)
2712                 munmap(s->data, s->size);
2713
2714         if (s->should_free || s->should_munmap) {
2715                 s->should_free = s->should_munmap = 0;
2716                 s->data = NULL;
2717         }
2718 }
2719
2720 void diff_free_filespec_data(struct diff_filespec *s)
2721 {
2722         diff_free_filespec_blob(s);
2723         free(s->cnt_data);
2724         s->cnt_data = NULL;
2725 }
2726
2727 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2728                            void *blob,
2729                            unsigned long size,
2730                            const unsigned char *sha1,
2731                            int mode)
2732 {
2733         int fd;
2734         struct strbuf buf = STRBUF_INIT;
2735         struct strbuf template = STRBUF_INIT;
2736         char *path_dup = xstrdup(path);
2737         const char *base = basename(path_dup);
2738
2739         /* Generate "XXXXXX_basename.ext" */
2740         strbuf_addstr(&template, "XXXXXX_");
2741         strbuf_addstr(&template, base);
2742
2743         fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2744                         strlen(base) + 1);
2745         if (fd < 0)
2746                 die_errno("unable to create temp-file");
2747         if (convert_to_working_tree(path,
2748                         (const char *)blob, (size_t)size, &buf)) {
2749                 blob = buf.buf;
2750                 size = buf.len;
2751         }
2752         if (write_in_full(fd, blob, size) != size)
2753                 die_errno("unable to write temp-file");
2754         close(fd);
2755         temp->name = temp->tmp_path;
2756         strcpy(temp->hex, sha1_to_hex(sha1));
2757         temp->hex[40] = 0;
2758         sprintf(temp->mode, "%06o", mode);
2759         strbuf_release(&buf);
2760         strbuf_release(&template);
2761         free(path_dup);
2762 }
2763
2764 static struct diff_tempfile *prepare_temp_file(const char *name,
2765                 struct diff_filespec *one)
2766 {
2767         struct diff_tempfile *temp = claim_diff_tempfile();
2768
2769         if (!DIFF_FILE_VALID(one)) {
2770         not_a_valid_file:
2771                 /* A '-' entry produces this for file-2, and
2772                  * a '+' entry produces this for file-1.
2773                  */
2774                 temp->name = "/dev/null";
2775                 strcpy(temp->hex, ".");
2776                 strcpy(temp->mode, ".");
2777                 return temp;
2778         }
2779
2780         if (!remove_tempfile_installed) {
2781                 atexit(remove_tempfile);
2782                 sigchain_push_common(remove_tempfile_on_signal);
2783                 remove_tempfile_installed = 1;
2784         }
2785
2786         if (!one->sha1_valid ||
2787             reuse_worktree_file(name, one->sha1, 1)) {
2788                 struct stat st;
2789                 if (lstat(name, &st) < 0) {
2790                         if (errno == ENOENT)
2791                                 goto not_a_valid_file;
2792                         die_errno("stat(%s)", name);
2793                 }
2794                 if (S_ISLNK(st.st_mode)) {
2795                         struct strbuf sb = STRBUF_INIT;
2796                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2797                                 die_errno("readlink(%s)", name);
2798                         prep_temp_blob(name, temp, sb.buf, sb.len,
2799                                        (one->sha1_valid ?
2800                                         one->sha1 : null_sha1),
2801                                        (one->sha1_valid ?
2802                                         one->mode : S_IFLNK));
2803                         strbuf_release(&sb);
2804                 }
2805                 else {
2806                         /* we can borrow from the file in the work tree */
2807                         temp->name = name;
2808                         if (!one->sha1_valid)
2809                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2810                         else
2811                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2812                         /* Even though we may sometimes borrow the
2813                          * contents from the work tree, we always want
2814                          * one->mode.  mode is trustworthy even when
2815                          * !(one->sha1_valid), as long as
2816                          * DIFF_FILE_VALID(one).
2817                          */
2818                         sprintf(temp->mode, "%06o", one->mode);
2819                 }
2820                 return temp;
2821         }
2822         else {
2823                 if (diff_populate_filespec(one, 0))
2824                         die("cannot read data blob for %s", one->path);
2825                 prep_temp_blob(name, temp, one->data, one->size,
2826                                one->sha1, one->mode);
2827         }
2828         return temp;
2829 }
2830
2831 /* An external diff command takes:
2832  *
2833  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2834  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2835  *
2836  */
2837 static void run_external_diff(const char *pgm,
2838                               const char *name,
2839                               const char *other,
2840                               struct diff_filespec *one,
2841                               struct diff_filespec *two,
2842                               const char *xfrm_msg,
2843                               int complete_rewrite)
2844 {
2845         const char *spawn_arg[10];
2846         int retval;
2847         const char **arg = &spawn_arg[0];
2848
2849         if (one && two) {
2850                 struct diff_tempfile *temp_one, *temp_two;
2851                 const char *othername = (other ? other : name);
2852                 temp_one = prepare_temp_file(name, one);
2853                 temp_two = prepare_temp_file(othername, two);
2854                 *arg++ = pgm;
2855                 *arg++ = name;
2856                 *arg++ = temp_one->name;
2857                 *arg++ = temp_one->hex;
2858                 *arg++ = temp_one->mode;
2859                 *arg++ = temp_two->name;
2860                 *arg++ = temp_two->hex;
2861                 *arg++ = temp_two->mode;
2862                 if (other) {
2863                         *arg++ = other;
2864                         *arg++ = xfrm_msg;
2865                 }
2866         } else {
2867                 *arg++ = pgm;
2868                 *arg++ = name;
2869         }
2870         *arg = NULL;
2871         fflush(NULL);
2872         retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2873         remove_tempfile();
2874         if (retval) {
2875                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2876                 exit(1);
2877         }
2878 }
2879
2880 static int similarity_index(struct diff_filepair *p)
2881 {
2882         return p->score * 100 / MAX_SCORE;
2883 }
2884
2885 static void fill_metainfo(struct strbuf *msg,
2886                           const char *name,
2887                           const char *other,
2888                           struct diff_filespec *one,
2889                           struct diff_filespec *two,
2890                           struct diff_options *o,
2891                           struct diff_filepair *p,
2892                           int *must_show_header,
2893                           int use_color)
2894 {
2895         const char *set = diff_get_color(use_color, DIFF_METAINFO);
2896         const char *reset = diff_get_color(use_color, DIFF_RESET);
2897         struct strbuf *msgbuf;
2898         char *line_prefix = "";
2899
2900         *must_show_header = 1;
2901         if (o->output_prefix) {
2902                 msgbuf = o->output_prefix(o, o->output_prefix_data);
2903                 line_prefix = msgbuf->buf;
2904         }
2905         strbuf_init(msg, PATH_MAX * 2 + 300);
2906         switch (p->status) {
2907         case DIFF_STATUS_COPIED:
2908                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2909                             line_prefix, set, similarity_index(p));
2910                 strbuf_addf(msg, "%s\n%s%scopy from ",
2911                             reset,  line_prefix, set);
2912                 quote_c_style(name, msg, NULL, 0);
2913                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
2914                 quote_c_style(other, msg, NULL, 0);
2915                 strbuf_addf(msg, "%s\n", reset);
2916                 break;
2917         case DIFF_STATUS_RENAMED:
2918                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2919                             line_prefix, set, similarity_index(p));
2920                 strbuf_addf(msg, "%s\n%s%srename from ",
2921                             reset, line_prefix, set);
2922                 quote_c_style(name, msg, NULL, 0);
2923                 strbuf_addf(msg, "%s\n%s%srename to ",
2924                             reset, line_prefix, set);
2925                 quote_c_style(other, msg, NULL, 0);
2926                 strbuf_addf(msg, "%s\n", reset);
2927                 break;
2928         case DIFF_STATUS_MODIFIED:
2929                 if (p->score) {
2930                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
2931                                     line_prefix,
2932                                     set, similarity_index(p), reset);
2933                         break;
2934                 }
2935                 /* fallthru */
2936         default:
2937                 *must_show_header = 0;
2938         }
2939         if (one && two && hashcmp(one->sha1, two->sha1)) {
2940                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2941
2942                 if (DIFF_OPT_TST(o, BINARY)) {
2943                         mmfile_t mf;
2944                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2945                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2946                                 abbrev = 40;
2947                 }
2948                 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
2949                             find_unique_abbrev(one->sha1, abbrev));
2950                 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
2951                 if (one->mode == two->mode)
2952                         strbuf_addf(msg, " %06o", one->mode);
2953                 strbuf_addf(msg, "%s\n", reset);
2954         }
2955 }
2956
2957 static void run_diff_cmd(const char *pgm,
2958                          const char *name,
2959                          const char *other,
2960                          const char *attr_path,
2961                          struct diff_filespec *one,
2962                          struct diff_filespec *two,
2963                          struct strbuf *msg,
2964                          struct diff_options *o,
2965                          struct diff_filepair *p)
2966 {
2967         const char *xfrm_msg = NULL;
2968         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2969         int must_show_header = 0;
2970
2971         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2972                 pgm = NULL;
2973         else {
2974                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2975                 if (drv && drv->external)
2976                         pgm = drv->external;
2977         }
2978
2979         if (msg) {
2980                 /*
2981                  * don't use colors when the header is intended for an
2982                  * external diff driver
2983                  */
2984                 fill_metainfo(msg, name, other, one, two, o, p,
2985                               &must_show_header,
2986                               want_color(o->use_color) && !pgm);
2987                 xfrm_msg = msg->len ? msg->buf : NULL;
2988         }
2989
2990         if (pgm) {
2991                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2992                                   complete_rewrite);
2993                 return;
2994         }
2995         if (one && two)
2996                 builtin_diff(name, other ? other : name,
2997                              one, two, xfrm_msg, must_show_header,
2998                              o, complete_rewrite);
2999         else
3000                 fprintf(o->file, "* Unmerged path %s\n", name);
3001 }
3002
3003 static void diff_fill_sha1_info(struct diff_filespec *one)
3004 {
3005         if (DIFF_FILE_VALID(one)) {
3006                 if (!one->sha1_valid) {
3007                         struct stat st;
3008                         if (!strcmp(one->path, "-")) {
3009                                 hashcpy(one->sha1, null_sha1);
3010                                 return;
3011                         }
3012                         if (lstat(one->path, &st) < 0)
3013                                 die_errno("stat '%s'", one->path);
3014                         if (index_path(one->sha1, one->path, &st, 0))
3015                                 die("cannot hash %s", one->path);
3016                 }
3017         }
3018         else
3019                 hashclr(one->sha1);
3020 }
3021
3022 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3023 {
3024         /* Strip the prefix but do not molest /dev/null and absolute paths */
3025         if (*namep && **namep != '/') {
3026                 *namep += prefix_length;
3027                 if (**namep == '/')
3028                         ++*namep;
3029         }
3030         if (*otherp && **otherp != '/') {
3031                 *otherp += prefix_length;
3032                 if (**otherp == '/')
3033                         ++*otherp;
3034         }
3035 }
3036
3037 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3038 {
3039         const char *pgm = external_diff();
3040         struct strbuf msg;
3041         struct diff_filespec *one = p->one;
3042         struct diff_filespec *two = p->two;
3043         const char *name;
3044         const char *other;
3045         const char *attr_path;
3046
3047         name  = p->one->path;
3048         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3049         attr_path = name;
3050         if (o->prefix_length)
3051                 strip_prefix(o->prefix_length, &name, &other);
3052
3053         if (DIFF_PAIR_UNMERGED(p)) {
3054                 run_diff_cmd(pgm, name, NULL, attr_path,
3055                              NULL, NULL, NULL, o, p);
3056                 return;
3057         }
3058
3059         diff_fill_sha1_info(one);
3060         diff_fill_sha1_info(two);
3061
3062         if (!pgm &&
3063             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3064             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3065                 /*
3066                  * a filepair that changes between file and symlink
3067                  * needs to be split into deletion and creation.
3068                  */
3069                 struct diff_filespec *null = alloc_filespec(two->path);
3070                 run_diff_cmd(NULL, name, other, attr_path,
3071                              one, null, &msg, o, p);
3072                 free(null);
3073                 strbuf_release(&msg);
3074
3075                 null = alloc_filespec(one->path);
3076                 run_diff_cmd(NULL, name, other, attr_path,
3077                              null, two, &msg, o, p);
3078                 free(null);
3079         }
3080         else
3081                 run_diff_cmd(pgm, name, other, attr_path,
3082                              one, two, &msg, o, p);
3083
3084         strbuf_release(&msg);
3085 }
3086
3087 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3088                          struct diffstat_t *diffstat)
3089 {
3090         const char *name;
3091         const char *other;
3092         int complete_rewrite = 0;
3093
3094         if (DIFF_PAIR_UNMERGED(p)) {
3095                 /* unmerged */
3096                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
3097                 return;
3098         }
3099
3100         name = p->one->path;
3101         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3102
3103         if (o->prefix_length)
3104                 strip_prefix(o->prefix_length, &name, &other);
3105
3106         diff_fill_sha1_info(p->one);
3107         diff_fill_sha1_info(p->two);
3108
3109         if (p->status == DIFF_STATUS_MODIFIED && p->score)
3110                 complete_rewrite = 1;
3111         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
3112 }
3113
3114 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3115 {
3116         const char *name;
3117         const char *other;
3118         const char *attr_path;
3119
3120         if (DIFF_PAIR_UNMERGED(p)) {
3121                 /* unmerged */
3122                 return;
3123         }
3124
3125         name = p->one->path;
3126         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3127         attr_path = other ? other : name;
3128
3129         if (o->prefix_length)
3130                 strip_prefix(o->prefix_length, &name, &other);
3131
3132         diff_fill_sha1_info(p->one);
3133         diff_fill_sha1_info(p->two);
3134
3135         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3136 }
3137
3138 void diff_setup(struct diff_options *options)
3139 {
3140         memcpy(options, &default_diff_options, sizeof(*options));
3141
3142         options->file = stdout;
3143
3144         options->line_termination = '\n';
3145         options->break_opt = -1;
3146         options->rename_limit = -1;
3147         options->dirstat_permille = diff_dirstat_permille_default;
3148         options->context = 3;
3149         DIFF_OPT_SET(options, RENAME_EMPTY);
3150
3151         options->change = diff_change;
3152         options->add_remove = diff_addremove;
3153         options->use_color = diff_use_color_default;
3154         options->detect_rename = diff_detect_rename_default;
3155
3156         if (diff_no_prefix) {
3157                 options->a_prefix = options->b_prefix = "";
3158         } else if (!diff_mnemonic_prefix) {
3159                 options->a_prefix = "a/";
3160                 options->b_prefix = "b/";
3161         }
3162 }
3163
3164 int diff_setup_done(struct diff_options *options)
3165 {
3166         int count = 0;
3167
3168         if (options->output_format & DIFF_FORMAT_NAME)
3169                 count++;
3170         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3171                 count++;
3172         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3173                 count++;
3174         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3175                 count++;
3176         if (count > 1)
3177                 die("--name-only, --name-status, --check and -s are mutually exclusive");
3178
3179         /*
3180          * Most of the time we can say "there are changes"
3181          * only by checking if there are changed paths, but
3182          * --ignore-whitespace* options force us to look
3183          * inside contents.
3184          */
3185
3186         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3187             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3188             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3189                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3190         else
3191                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3192
3193         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3194                 options->detect_rename = DIFF_DETECT_COPY;
3195
3196         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3197                 options->prefix = NULL;
3198         if (options->prefix)
3199                 options->prefix_length = strlen(options->prefix);
3200         else
3201                 options->prefix_length = 0;
3202
3203         if (options->output_format & (DIFF_FORMAT_NAME |
3204                                       DIFF_FORMAT_NAME_STATUS |
3205                                       DIFF_FORMAT_CHECKDIFF |
3206                                       DIFF_FORMAT_NO_OUTPUT))
3207                 options->output_format &= ~(DIFF_FORMAT_RAW |
3208                                             DIFF_FORMAT_NUMSTAT |
3209                                             DIFF_FORMAT_DIFFSTAT |
3210                                             DIFF_FORMAT_SHORTSTAT |
3211                                             DIFF_FORMAT_DIRSTAT |
3212                                             DIFF_FORMAT_SUMMARY |
3213                                             DIFF_FORMAT_PATCH);
3214
3215         /*
3216          * These cases always need recursive; we do not drop caller-supplied
3217          * recursive bits for other formats here.
3218          */
3219         if (options->output_format & (DIFF_FORMAT_PATCH |
3220                                       DIFF_FORMAT_NUMSTAT |
3221                                       DIFF_FORMAT_DIFFSTAT |
3222                                       DIFF_FORMAT_SHORTSTAT |
3223                                       DIFF_FORMAT_DIRSTAT |
3224                                       DIFF_FORMAT_SUMMARY |
3225                                       DIFF_FORMAT_CHECKDIFF))
3226                 DIFF_OPT_SET(options, RECURSIVE);
3227         /*
3228          * Also pickaxe would not work very well if you do not say recursive
3229          */
3230         if (options->pickaxe)
3231                 DIFF_OPT_SET(options, RECURSIVE);
3232         /*
3233          * When patches are generated, submodules diffed against the work tree
3234          * must be checked for dirtiness too so it can be shown in the output
3235          */
3236         if (options->output_format & DIFF_FORMAT_PATCH)
3237                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3238
3239         if (options->detect_rename && options->rename_limit < 0)
3240                 options->rename_limit = diff_rename_limit_default;
3241         if (options->setup & DIFF_SETUP_USE_CACHE) {
3242                 if (!active_cache)
3243                         /* read-cache does not die even when it fails
3244                          * so it is safe for us to do this here.  Also
3245                          * it does not smudge active_cache or active_nr
3246                          * when it fails, so we do not have to worry about
3247                          * cleaning it up ourselves either.
3248                          */
3249                         read_cache();
3250         }
3251         if (options->abbrev <= 0 || 40 < options->abbrev)
3252                 options->abbrev = 40; /* full */
3253
3254         /*
3255          * It does not make sense to show the first hit we happened
3256          * to have found.  It does not make sense not to return with
3257          * exit code in such a case either.
3258          */
3259         if (DIFF_OPT_TST(options, QUICK)) {
3260                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3261                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3262         }
3263
3264         return 0;
3265 }
3266
3267 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3268 {
3269         char c, *eq;
3270         int len;
3271
3272         if (*arg != '-')
3273                 return 0;
3274         c = *++arg;
3275         if (!c)
3276                 return 0;
3277         if (c == arg_short) {
3278                 c = *++arg;
3279                 if (!c)
3280                         return 1;
3281                 if (val && isdigit(c)) {
3282                         char *end;
3283                         int n = strtoul(arg, &end, 10);
3284                         if (*end)
3285                                 return 0;
3286                         *val = n;
3287                         return 1;
3288                 }
3289                 return 0;
3290         }
3291         if (c != '-')
3292                 return 0;
3293         arg++;
3294         eq = strchr(arg, '=');
3295         if (eq)
3296                 len = eq - arg;
3297         else
3298                 len = strlen(arg);
3299         if (!len || strncmp(arg, arg_long, len))
3300                 return 0;
3301         if (eq) {
3302                 int n;
3303                 char *end;
3304                 if (!isdigit(*++eq))
3305                         return 0;
3306                 n = strtoul(eq, &end, 10);
3307                 if (*end)
3308                         return 0;
3309                 *val = n;
3310         }
3311         return 1;
3312 }
3313
3314 static int diff_scoreopt_parse(const char *opt);
3315
3316 static inline int short_opt(char opt, const char **argv,
3317                             const char **optarg)
3318 {
3319         const char *arg = argv[0];
3320         if (arg[0] != '-' || arg[1] != opt)
3321                 return 0;
3322         if (arg[2] != '\0') {
3323                 *optarg = arg + 2;
3324                 return 1;
3325         }
3326         if (!argv[1])
3327                 die("Option '%c' requires a value", opt);
3328         *optarg = argv[1];
3329         return 2;
3330 }
3331
3332 int parse_long_opt(const char *opt, const char **argv,
3333                    const char **optarg)
3334 {
3335         const char *arg = argv[0];
3336         if (arg[0] != '-' || arg[1] != '-')
3337                 return 0;
3338         arg += strlen("--");
3339         if (prefixcmp(arg, opt))
3340                 return 0;
3341         arg += strlen(opt);
3342         if (*arg == '=') { /* sticked form: --option=value */
3343                 *optarg = arg + 1;
3344                 return 1;
3345         }
3346         if (*arg != '\0')
3347                 return 0;
3348         /* separate form: --option value */
3349         if (!argv[1])
3350                 die("Option '--%s' requires a value", opt);
3351         *optarg = argv[1];
3352         return 2;
3353 }
3354
3355 static int stat_opt(struct diff_options *options, const char **av)
3356 {
3357         const char *arg = av[0];
3358         char *end;
3359         int width = options->stat_width;
3360         int name_width = options->stat_name_width;
3361         int graph_width = options->stat_graph_width;
3362         int count = options->stat_count;
3363         int argcount = 1;
3364
3365         arg += strlen("--stat");
3366         end = (char *)arg;
3367
3368         switch (*arg) {
3369         case '-':
3370                 if (!prefixcmp(arg, "-width")) {
3371                         arg += strlen("-width");
3372                         if (*arg == '=')
3373                                 width = strtoul(arg + 1, &end, 10);
3374                         else if (!*arg && !av[1])
3375                                 die("Option '--stat-width' requires a value");
3376                         else if (!*arg) {
3377                                 width = strtoul(av[1], &end, 10);
3378                                 argcount = 2;
3379                         }
3380                 } else if (!prefixcmp(arg, "-name-width")) {
3381                         arg += strlen("-name-width");
3382                         if (*arg == '=')
3383                                 name_width = strtoul(arg + 1, &end, 10);
3384                         else if (!*arg && !av[1])
3385                                 die("Option '--stat-name-width' requires a value");
3386                         else if (!*arg) {
3387                                 name_width = strtoul(av[1], &end, 10);
3388                                 argcount = 2;
3389                         }
3390                 } else if (!prefixcmp(arg, "-graph-width")) {
3391                         arg += strlen("-graph-width");
3392                         if (*arg == '=')
3393                                 graph_width = strtoul(arg + 1, &end, 10);
3394                         else if (!*arg && !av[1])
3395                                 die("Option '--stat-graph-width' requires a value");
3396                         else if (!*arg) {
3397                                 graph_width = strtoul(av[1], &end, 10);
3398                                 argcount = 2;
3399                         }
3400                 } else if (!prefixcmp(arg, "-count")) {
3401                         arg += strlen("-count");
3402                         if (*arg == '=')
3403                                 count = strtoul(arg + 1, &end, 10);
3404                         else if (!*arg && !av[1])
3405                                 die("Option '--stat-count' requires a value");
3406                         else if (!*arg) {
3407                                 count = strtoul(av[1], &end, 10);
3408                                 argcount = 2;
3409                         }
3410                 }
3411                 break;
3412         case '=':
3413                 width = strtoul(arg+1, &end, 10);
3414                 if (*end == ',')
3415                         name_width = strtoul(end+1, &end, 10);
3416                 if (*end == ',')
3417                         count = strtoul(end+1, &end, 10);
3418         }
3419
3420         /* Important! This checks all the error cases! */
3421         if (*end)
3422                 return 0;
3423         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3424         options->stat_name_width = name_width;
3425         options->stat_graph_width = graph_width;
3426         options->stat_width = width;
3427         options->stat_count = count;
3428         return argcount;
3429 }
3430
3431 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3432 {
3433         struct strbuf errmsg = STRBUF_INIT;
3434         if (parse_dirstat_params(options, params, &errmsg))
3435                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3436                     errmsg.buf);
3437         strbuf_release(&errmsg);
3438         /*
3439          * The caller knows a dirstat-related option is given from the command
3440          * line; allow it to say "return this_function();"
3441          */
3442         options->output_format |= DIFF_FORMAT_DIRSTAT;
3443         return 1;
3444 }
3445
3446 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3447 {
3448         const char *arg = av[0];
3449         const char *optarg;
3450         int argcount;
3451
3452         /* Output format options */
3453         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch"))
3454                 options->output_format |= DIFF_FORMAT_PATCH;
3455         else if (opt_arg(arg, 'U', "unified", &options->context))
3456                 options->output_format |= DIFF_FORMAT_PATCH;
3457         else if (!strcmp(arg, "--raw"))
3458                 options->output_format |= DIFF_FORMAT_RAW;
3459         else if (!strcmp(arg, "--patch-with-raw"))
3460                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
3461         else if (!strcmp(arg, "--numstat"))
3462                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3463         else if (!strcmp(arg, "--shortstat"))
3464                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3465         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3466                 return parse_dirstat_opt(options, "");
3467         else if (!prefixcmp(arg, "-X"))
3468                 return parse_dirstat_opt(options, arg + 2);
3469         else if (!prefixcmp(arg, "--dirstat="))
3470                 return parse_dirstat_opt(options, arg + 10);
3471         else if (!strcmp(arg, "--cumulative"))
3472                 return parse_dirstat_opt(options, "cumulative");
3473         else if (!strcmp(arg, "--dirstat-by-file"))
3474                 return parse_dirstat_opt(options, "files");
3475         else if (!prefixcmp(arg, "--dirstat-by-file=")) {
3476                 parse_dirstat_opt(options, "files");
3477                 return parse_dirstat_opt(options, arg + 18);
3478         }
3479         else if (!strcmp(arg, "--check"))
3480                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3481         else if (!strcmp(arg, "--summary"))
3482                 options->output_format |= DIFF_FORMAT_SUMMARY;
3483         else if (!strcmp(arg, "--patch-with-stat"))
3484                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
3485         else if (!strcmp(arg, "--name-only"))
3486                 options->output_format |= DIFF_FORMAT_NAME;
3487         else if (!strcmp(arg, "--name-status"))
3488                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3489         else if (!strcmp(arg, "-s"))
3490                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3491         else if (!prefixcmp(arg, "--stat"))
3492                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3493                 return stat_opt(options, av);
3494
3495         /* renames options */
3496         else if (!prefixcmp(arg, "-B") || !prefixcmp(arg, "--break-rewrites=") ||
3497                  !strcmp(arg, "--break-rewrites")) {
3498                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3499                         return error("invalid argument to -B: %s", arg+2);
3500         }
3501         else if (!prefixcmp(arg, "-M") || !prefixcmp(arg, "--find-renames=") ||
3502                  !strcmp(arg, "--find-renames")) {
3503                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3504                         return error("invalid argument to -M: %s", arg+2);
3505                 options->detect_rename = DIFF_DETECT_RENAME;
3506         }
3507         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3508                 options->irreversible_delete = 1;
3509         }
3510         else if (!prefixcmp(arg, "-C") || !prefixcmp(arg, "--find-copies=") ||
3511                  !strcmp(arg, "--find-copies")) {
3512                 if (options->detect_rename == DIFF_DETECT_COPY)
3513                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3514                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3515                         return error("invalid argument to -C: %s", arg+2);
3516                 options->detect_rename = DIFF_DETECT_COPY;
3517         }
3518         else if (!strcmp(arg, "--no-renames"))
3519                 options->detect_rename = 0;
3520         else if (!strcmp(arg, "--rename-empty"))
3521                 DIFF_OPT_SET(options, RENAME_EMPTY);
3522         else if (!strcmp(arg, "--no-rename-empty"))
3523                 DIFF_OPT_CLR(options, RENAME_EMPTY);
3524         else if (!strcmp(arg, "--relative"))
3525                 DIFF_OPT_SET(options, RELATIVE_NAME);
3526         else if (!prefixcmp(arg, "--relative=")) {
3527                 DIFF_OPT_SET(options, RELATIVE_NAME);
3528                 options->prefix = arg + 11;
3529         }
3530
3531         /* xdiff options */
3532         else if (!strcmp(arg, "--minimal"))
3533                 DIFF_XDL_SET(options, NEED_MINIMAL);
3534         else if (!strcmp(arg, "--no-minimal"))
3535                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3536         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3537                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3538         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3539                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3540         else if (!strcmp(arg, "--ignore-space-at-eol"))
3541                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3542         else if (!strcmp(arg, "--patience"))
3543                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3544         else if (!strcmp(arg, "--histogram"))
3545                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3546
3547         /* flags options */
3548         else if (!strcmp(arg, "--binary")) {
3549                 options->output_format |= DIFF_FORMAT_PATCH;
3550                 DIFF_OPT_SET(options, BINARY);
3551         }
3552         else if (!strcmp(arg, "--full-index"))
3553                 DIFF_OPT_SET(options, FULL_INDEX);
3554         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3555                 DIFF_OPT_SET(options, TEXT);
3556         else if (!strcmp(arg, "-R"))
3557                 DIFF_OPT_SET(options, REVERSE_DIFF);
3558         else if (!strcmp(arg, "--find-copies-harder"))
3559                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3560         else if (!strcmp(arg, "--follow"))
3561                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3562         else if (!strcmp(arg, "--color"))
3563                 options->use_color = 1;
3564         else if (!prefixcmp(arg, "--color=")) {
3565                 int value = git_config_colorbool(NULL, arg+8);
3566                 if (value < 0)
3567                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3568                 options->use_color = value;
3569         }
3570         else if (!strcmp(arg, "--no-color"))
3571                 options->use_color = 0;
3572         else if (!strcmp(arg, "--color-words")) {
3573                 options->use_color = 1;
3574                 options->word_diff = DIFF_WORDS_COLOR;
3575         }
3576         else if (!prefixcmp(arg, "--color-words=")) {
3577                 options->use_color = 1;
3578                 options->word_diff = DIFF_WORDS_COLOR;
3579                 options->word_regex = arg + 14;
3580         }
3581         else if (!strcmp(arg, "--word-diff")) {
3582                 if (options->word_diff == DIFF_WORDS_NONE)
3583                         options->word_diff = DIFF_WORDS_PLAIN;
3584         }
3585         else if (!prefixcmp(arg, "--word-diff=")) {
3586                 const char *type = arg + 12;
3587                 if (!strcmp(type, "plain"))
3588                         options->word_diff = DIFF_WORDS_PLAIN;
3589                 else if (!strcmp(type, "color")) {
3590                         options->use_color = 1;
3591                         options->word_diff = DIFF_WORDS_COLOR;
3592                 }
3593                 else if (!strcmp(type, "porcelain"))
3594                         options->word_diff = DIFF_WORDS_PORCELAIN;
3595                 else if (!strcmp(type, "none"))
3596                         options->word_diff = DIFF_WORDS_NONE;
3597                 else
3598                         die("bad --word-diff argument: %s", type);
3599         }
3600         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3601                 if (options->word_diff == DIFF_WORDS_NONE)
3602                         options->word_diff = DIFF_WORDS_PLAIN;
3603                 options->word_regex = optarg;
3604                 return argcount;
3605         }
3606         else if (!strcmp(arg, "--exit-code"))
3607                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3608         else if (!strcmp(arg, "--quiet"))
3609                 DIFF_OPT_SET(options, QUICK);
3610         else if (!strcmp(arg, "--ext-diff"))
3611                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3612         else if (!strcmp(arg, "--no-ext-diff"))
3613                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3614         else if (!strcmp(arg, "--textconv"))
3615                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3616         else if (!strcmp(arg, "--no-textconv"))
3617                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3618         else if (!strcmp(arg, "--ignore-submodules")) {
3619                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3620                 handle_ignore_submodules_arg(options, "all");
3621         } else if (!prefixcmp(arg, "--ignore-submodules=")) {
3622                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3623                 handle_ignore_submodules_arg(options, arg + 20);
3624         } else if (!strcmp(arg, "--submodule"))
3625                 DIFF_OPT_SET(options, SUBMODULE_LOG);
3626         else if (!prefixcmp(arg, "--submodule=")) {
3627                 if (!strcmp(arg + 12, "log"))
3628                         DIFF_OPT_SET(options, SUBMODULE_LOG);
3629         }
3630
3631         /* misc options */
3632         else if (!strcmp(arg, "-z"))
3633                 options->line_termination = 0;
3634         else if ((argcount = short_opt('l', av, &optarg))) {
3635                 options->rename_limit = strtoul(optarg, NULL, 10);
3636                 return argcount;
3637         }
3638         else if ((argcount = short_opt('S', av, &optarg))) {
3639                 options->pickaxe = optarg;
3640                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3641                 return argcount;
3642         } else if ((argcount = short_opt('G', av, &optarg))) {
3643                 options->pickaxe = optarg;
3644                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3645                 return argcount;
3646         }
3647         else if (!strcmp(arg, "--pickaxe-all"))
3648                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3649         else if (!strcmp(arg, "--pickaxe-regex"))
3650                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3651         else if ((argcount = short_opt('O', av, &optarg))) {
3652                 options->orderfile = optarg;
3653                 return argcount;
3654         }
3655         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3656                 options->filter = optarg;
3657                 return argcount;
3658         }
3659         else if (!strcmp(arg, "--abbrev"))
3660                 options->abbrev = DEFAULT_ABBREV;
3661         else if (!prefixcmp(arg, "--abbrev=")) {
3662                 options->abbrev = strtoul(arg + 9, NULL, 10);
3663                 if (options->abbrev < MINIMUM_ABBREV)
3664                         options->abbrev = MINIMUM_ABBREV;
3665                 else if (40 < options->abbrev)
3666                         options->abbrev = 40;
3667         }
3668         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3669                 options->a_prefix = optarg;
3670                 return argcount;
3671         }
3672         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3673                 options->b_prefix = optarg;
3674                 return argcount;
3675         }
3676         else if (!strcmp(arg, "--no-prefix"))
3677                 options->a_prefix = options->b_prefix = "";
3678         else if (opt_arg(arg, '\0', "inter-hunk-context",
3679                          &options->interhunkcontext))
3680                 ;
3681         else if (!strcmp(arg, "-W"))
3682                 DIFF_OPT_SET(options, FUNCCONTEXT);
3683         else if (!strcmp(arg, "--function-context"))
3684                 DIFF_OPT_SET(options, FUNCCONTEXT);
3685         else if (!strcmp(arg, "--no-function-context"))
3686                 DIFF_OPT_CLR(options, FUNCCONTEXT);
3687         else if ((argcount = parse_long_opt("output", av, &optarg))) {
3688                 options->file = fopen(optarg, "w");
3689                 if (!options->file)
3690                         die_errno("Could not open '%s'", optarg);
3691                 options->close_file = 1;
3692                 return argcount;
3693         } else
3694                 return 0;
3695         return 1;
3696 }
3697
3698 int parse_rename_score(const char **cp_p)
3699 {
3700         unsigned long num, scale;
3701         int ch, dot;
3702         const char *cp = *cp_p;
3703
3704         num = 0;
3705         scale = 1;
3706         dot = 0;
3707         for (;;) {
3708                 ch = *cp;
3709                 if ( !dot && ch == '.' ) {
3710                         scale = 1;
3711                         dot = 1;
3712                 } else if ( ch == '%' ) {
3713                         scale = dot ? scale*100 : 100;
3714                         cp++;   /* % is always at the end */
3715                         break;
3716                 } else if ( ch >= '0' && ch <= '9' ) {
3717                         if ( scale < 100000 ) {
3718                                 scale *= 10;
3719                                 num = (num*10) + (ch-'0');
3720                         }
3721                 } else {
3722                         break;
3723                 }
3724                 cp++;
3725         }
3726         *cp_p = cp;
3727
3728         /* user says num divided by scale and we say internally that
3729          * is MAX_SCORE * num / scale.
3730          */
3731         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
3732 }
3733
3734 static int diff_scoreopt_parse(const char *opt)
3735 {
3736         int opt1, opt2, cmd;
3737
3738         if (*opt++ != '-')
3739                 return -1;
3740         cmd = *opt++;
3741         if (cmd == '-') {
3742                 /* convert the long-form arguments into short-form versions */
3743                 if (!prefixcmp(opt, "break-rewrites")) {
3744                         opt += strlen("break-rewrites");
3745                         if (*opt == 0 || *opt++ == '=')
3746                                 cmd = 'B';
3747                 } else if (!prefixcmp(opt, "find-copies")) {
3748                         opt += strlen("find-copies");
3749                         if (*opt == 0 || *opt++ == '=')
3750                                 cmd = 'C';
3751                 } else if (!prefixcmp(opt, "find-renames")) {
3752                         opt += strlen("find-renames");
3753                         if (*opt == 0 || *opt++ == '=')
3754                                 cmd = 'M';
3755                 }
3756         }
3757         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
3758                 return -1; /* that is not a -M, -C nor -B option */
3759
3760         opt1 = parse_rename_score(&opt);
3761         if (cmd != 'B')
3762                 opt2 = 0;
3763         else {
3764                 if (*opt == 0)
3765                         opt2 = 0;
3766                 else if (*opt != '/')
3767                         return -1; /* we expect -B80/99 or -B80 */
3768                 else {
3769                         opt++;
3770                         opt2 = parse_rename_score(&opt);
3771                 }
3772         }
3773         if (*opt != 0)
3774                 return -1;
3775         return opt1 | (opt2 << 16);
3776 }
3777
3778 struct diff_queue_struct diff_queued_diff;
3779
3780 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
3781 {
3782         if (queue->alloc <= queue->nr) {
3783                 queue->alloc = alloc_nr(queue->alloc);
3784                 queue->queue = xrealloc(queue->queue,
3785                                         sizeof(dp) * queue->alloc);
3786         }
3787         queue->queue[queue->nr++] = dp;
3788 }
3789
3790 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
3791                                  struct diff_filespec *one,
3792                                  struct diff_filespec *two)
3793 {
3794         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3795         dp->one = one;
3796         dp->two = two;
3797         if (queue)
3798                 diff_q(queue, dp);
3799         return dp;
3800 }
3801
3802 void diff_free_filepair(struct diff_filepair *p)
3803 {
3804         free_filespec(p->one);
3805         free_filespec(p->two);
3806         free(p);
3807 }
3808
3809 /* This is different from find_unique_abbrev() in that
3810  * it stuffs the result with dots for alignment.
3811  */
3812 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3813 {
3814         int abblen;
3815         const char *abbrev;
3816         if (len == 40)
3817                 return sha1_to_hex(sha1);
3818
3819         abbrev = find_unique_abbrev(sha1, len);
3820         abblen = strlen(abbrev);
3821         if (abblen < 37) {
3822                 static char hex[41];
3823                 if (len < abblen && abblen <= len + 2)
3824                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3825                 else
3826                         sprintf(hex, "%s...", abbrev);
3827                 return hex;
3828         }
3829         return sha1_to_hex(sha1);
3830 }
3831
3832 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3833 {
3834         int line_termination = opt->line_termination;
3835         int inter_name_termination = line_termination ? '\t' : '\0';
3836         if (opt->output_prefix) {
3837                 struct strbuf *msg = NULL;
3838                 msg = opt->output_prefix(opt, opt->output_prefix_data);
3839                 fprintf(opt->file, "%s", msg->buf);
3840         }
3841
3842         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3843                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3844                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
3845                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3846         }
3847         if (p->score) {
3848                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3849                         inter_name_termination);
3850         } else {
3851                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
3852         }
3853
3854         if (p->status == DIFF_STATUS_COPIED ||
3855             p->status == DIFF_STATUS_RENAMED) {
3856                 const char *name_a, *name_b;
3857                 name_a = p->one->path;
3858                 name_b = p->two->path;
3859                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3860                 write_name_quoted(name_a, opt->file, inter_name_termination);
3861                 write_name_quoted(name_b, opt->file, line_termination);
3862         } else {
3863                 const char *name_a, *name_b;
3864                 name_a = p->one->mode ? p->one->path : p->two->path;
3865                 name_b = NULL;
3866                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3867                 write_name_quoted(name_a, opt->file, line_termination);
3868         }
3869 }
3870
3871 int diff_unmodified_pair(struct diff_filepair *p)
3872 {
3873         /* This function is written stricter than necessary to support
3874          * the currently implemented transformers, but the idea is to
3875          * let transformers to produce diff_filepairs any way they want,
3876          * and filter and clean them up here before producing the output.
3877          */
3878         struct diff_filespec *one = p->one, *two = p->two;
3879
3880         if (DIFF_PAIR_UNMERGED(p))
3881                 return 0; /* unmerged is interesting */
3882
3883         /* deletion, addition, mode or type change
3884          * and rename are all interesting.
3885          */
3886         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
3887             DIFF_PAIR_MODE_CHANGED(p) ||
3888             strcmp(one->path, two->path))
3889                 return 0;
3890
3891         /* both are valid and point at the same path.  that is, we are
3892          * dealing with a change.
3893          */
3894         if (one->sha1_valid && two->sha1_valid &&
3895             !hashcmp(one->sha1, two->sha1) &&
3896             !one->dirty_submodule && !two->dirty_submodule)
3897                 return 1; /* no change */
3898         if (!one->sha1_valid && !two->sha1_valid)
3899                 return 1; /* both look at the same file on the filesystem. */
3900         return 0;
3901 }
3902
3903 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
3904 {
3905         if (diff_unmodified_pair(p))
3906                 return;
3907
3908         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3909             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3910                 return; /* no tree diffs in patch format */
3911
3912         run_diff(p, o);
3913 }
3914
3915 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
3916                             struct diffstat_t *diffstat)
3917 {
3918         if (diff_unmodified_pair(p))
3919                 return;
3920
3921         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3922             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3923                 return; /* no useful stat for tree diffs */
3924
3925         run_diffstat(p, o, diffstat);
3926 }
3927
3928 static void diff_flush_checkdiff(struct diff_filepair *p,
3929                 struct diff_options *o)
3930 {
3931         if (diff_unmodified_pair(p))
3932                 return;
3933
3934         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3935             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3936                 return; /* nothing to check in tree diffs */
3937
3938         run_checkdiff(p, o);
3939 }
3940
3941 int diff_queue_is_empty(void)
3942 {
3943         struct diff_queue_struct *q = &diff_queued_diff;
3944         int i;
3945         for (i = 0; i < q->nr; i++)
3946                 if (!diff_unmodified_pair(q->queue[i]))
3947                         return 0;
3948         return 1;
3949 }
3950
3951 #if DIFF_DEBUG
3952 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
3953 {
3954         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
3955                 x, one ? one : "",
3956                 s->path,
3957                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
3958                 s->mode,
3959                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
3960         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
3961                 x, one ? one : "",
3962                 s->size, s->xfrm_flags);
3963 }
3964
3965 void diff_debug_filepair(const struct diff_filepair *p, int i)
3966 {
3967         diff_debug_filespec(p->one, i, "one");
3968         diff_debug_filespec(p->two, i, "two");
3969         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
3970                 p->score, p->status ? p->status : '?',
3971                 p->one->rename_used, p->broken_pair);
3972 }
3973
3974 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
3975 {
3976         int i;
3977         if (msg)
3978                 fprintf(stderr, "%s\n", msg);
3979         fprintf(stderr, "q->nr = %d\n", q->nr);
3980         for (i = 0; i < q->nr; i++) {
3981                 struct diff_filepair *p = q->queue[i];
3982                 diff_debug_filepair(p, i);
3983         }
3984 }
3985 #endif
3986
3987 static void diff_resolve_rename_copy(void)
3988 {
3989         int i;
3990         struct diff_filepair *p;
3991         struct diff_queue_struct *q = &diff_queued_diff;
3992
3993         diff_debug_queue("resolve-rename-copy", q);
3994
3995         for (i = 0; i < q->nr; i++) {
3996                 p = q->queue[i];
3997                 p->status = 0; /* undecided */
3998                 if (DIFF_PAIR_UNMERGED(p))
3999                         p->status = DIFF_STATUS_UNMERGED;
4000                 else if (!DIFF_FILE_VALID(p->one))
4001                         p->status = DIFF_STATUS_ADDED;
4002                 else if (!DIFF_FILE_VALID(p->two))
4003                         p->status = DIFF_STATUS_DELETED;
4004                 else if (DIFF_PAIR_TYPE_CHANGED(p))
4005                         p->status = DIFF_STATUS_TYPE_CHANGED;
4006
4007                 /* from this point on, we are dealing with a pair
4008                  * whose both sides are valid and of the same type, i.e.
4009                  * either in-place edit or rename/copy edit.
4010                  */
4011                 else if (DIFF_PAIR_RENAME(p)) {
4012                         /*
4013                          * A rename might have re-connected a broken
4014                          * pair up, causing the pathnames to be the
4015                          * same again. If so, that's not a rename at
4016                          * all, just a modification..
4017                          *
4018                          * Otherwise, see if this source was used for
4019                          * multiple renames, in which case we decrement
4020                          * the count, and call it a copy.
4021                          */
4022                         if (!strcmp(p->one->path, p->two->path))
4023                                 p->status = DIFF_STATUS_MODIFIED;
4024                         else if (--p->one->rename_used > 0)
4025                                 p->status = DIFF_STATUS_COPIED;
4026                         else
4027                                 p->status = DIFF_STATUS_RENAMED;
4028                 }
4029                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
4030                          p->one->mode != p->two->mode ||
4031                          p->one->dirty_submodule ||
4032                          p->two->dirty_submodule ||
4033                          is_null_sha1(p->one->sha1))
4034                         p->status = DIFF_STATUS_MODIFIED;
4035                 else {
4036                         /* This is a "no-change" entry and should not
4037                          * happen anymore, but prepare for broken callers.
4038                          */
4039                         error("feeding unmodified %s to diffcore",
4040                               p->one->path);
4041                         p->status = DIFF_STATUS_UNKNOWN;
4042                 }
4043         }
4044         diff_debug_queue("resolve-rename-copy done", q);
4045 }
4046
4047 static int check_pair_status(struct diff_filepair *p)
4048 {
4049         switch (p->status) {
4050         case DIFF_STATUS_UNKNOWN:
4051                 return 0;
4052         case 0:
4053                 die("internal error in diff-resolve-rename-copy");
4054         default:
4055                 return 1;
4056         }
4057 }
4058
4059 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4060 {
4061         int fmt = opt->output_format;
4062
4063         if (fmt & DIFF_FORMAT_CHECKDIFF)
4064                 diff_flush_checkdiff(p, opt);
4065         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4066                 diff_flush_raw(p, opt);
4067         else if (fmt & DIFF_FORMAT_NAME) {
4068                 const char *name_a, *name_b;
4069                 name_a = p->two->path;
4070                 name_b = NULL;
4071                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4072                 write_name_quoted(name_a, opt->file, opt->line_termination);
4073         }
4074 }
4075
4076 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4077 {
4078         if (fs->mode)
4079                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4080         else
4081                 fprintf(file, " %s ", newdelete);
4082         write_name_quoted(fs->path, file, '\n');
4083 }
4084
4085
4086 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4087                 const char *line_prefix)
4088 {
4089         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4090                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4091                         p->two->mode, show_name ? ' ' : '\n');
4092                 if (show_name) {
4093                         write_name_quoted(p->two->path, file, '\n');
4094                 }
4095         }
4096 }
4097
4098 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4099                         const char *line_prefix)
4100 {
4101         char *names = pprint_rename(p->one->path, p->two->path);
4102
4103         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4104         free(names);
4105         show_mode_change(file, p, 0, line_prefix);
4106 }
4107
4108 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4109 {
4110         FILE *file = opt->file;
4111         char *line_prefix = "";
4112
4113         if (opt->output_prefix) {
4114                 struct strbuf *buf = opt->output_prefix(opt, opt->output_prefix_data);
4115                 line_prefix = buf->buf;
4116         }
4117
4118         switch(p->status) {
4119         case DIFF_STATUS_DELETED:
4120                 fputs(line_prefix, file);
4121                 show_file_mode_name(file, "delete", p->one);
4122                 break;
4123         case DIFF_STATUS_ADDED:
4124                 fputs(line_prefix, file);
4125                 show_file_mode_name(file, "create", p->two);
4126                 break;
4127         case DIFF_STATUS_COPIED:
4128                 fputs(line_prefix, file);
4129                 show_rename_copy(file, "copy", p, line_prefix);
4130                 break;
4131         case DIFF_STATUS_RENAMED:
4132                 fputs(line_prefix, file);
4133                 show_rename_copy(file, "rename", p, line_prefix);
4134                 break;
4135         default:
4136                 if (p->score) {
4137                         fprintf(file, "%s rewrite ", line_prefix);
4138                         write_name_quoted(p->two->path, file, ' ');
4139                         fprintf(file, "(%d%%)\n", similarity_index(p));
4140                 }
4141                 show_mode_change(file, p, !p->score, line_prefix);
4142                 break;
4143         }
4144 }
4145
4146 struct patch_id_t {
4147         git_SHA_CTX *ctx;
4148         int patchlen;
4149 };
4150
4151 static int remove_space(char *line, int len)
4152 {
4153         int i;
4154         char *dst = line;
4155         unsigned char c;
4156
4157         for (i = 0; i < len; i++)
4158                 if (!isspace((c = line[i])))
4159                         *dst++ = c;
4160
4161         return dst - line;
4162 }
4163
4164 static void patch_id_consume(void *priv, char *line, unsigned long len)
4165 {
4166         struct patch_id_t *data = priv;
4167         int new_len;
4168
4169         /* Ignore line numbers when computing the SHA1 of the patch */
4170         if (!prefixcmp(line, "@@ -"))
4171                 return;
4172
4173         new_len = remove_space(line, len);
4174
4175         git_SHA1_Update(data->ctx, line, new_len);
4176         data->patchlen += new_len;
4177 }
4178
4179 /* returns 0 upon success, and writes result into sha1 */
4180 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
4181 {
4182         struct diff_queue_struct *q = &diff_queued_diff;
4183         int i;
4184         git_SHA_CTX ctx;
4185         struct patch_id_t data;
4186         char buffer[PATH_MAX * 4 + 20];
4187
4188         git_SHA1_Init(&ctx);
4189         memset(&data, 0, sizeof(struct patch_id_t));
4190         data.ctx = &ctx;
4191
4192         for (i = 0; i < q->nr; i++) {
4193                 xpparam_t xpp;
4194                 xdemitconf_t xecfg;
4195                 mmfile_t mf1, mf2;
4196                 struct diff_filepair *p = q->queue[i];
4197                 int len1, len2;
4198
4199                 memset(&xpp, 0, sizeof(xpp));
4200                 memset(&xecfg, 0, sizeof(xecfg));
4201                 if (p->status == 0)
4202                         return error("internal diff status error");
4203                 if (p->status == DIFF_STATUS_UNKNOWN)
4204                         continue;
4205                 if (diff_unmodified_pair(p))
4206                         continue;
4207                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4208                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4209                         continue;
4210                 if (DIFF_PAIR_UNMERGED(p))
4211                         continue;
4212
4213                 diff_fill_sha1_info(p->one);
4214                 diff_fill_sha1_info(p->two);
4215                 if (fill_mmfile(&mf1, p->one) < 0 ||
4216                                 fill_mmfile(&mf2, p->two) < 0)
4217                         return error("unable to read files to diff");
4218
4219                 len1 = remove_space(p->one->path, strlen(p->one->path));
4220                 len2 = remove_space(p->two->path, strlen(p->two->path));
4221                 if (p->one->mode == 0)
4222                         len1 = snprintf(buffer, sizeof(buffer),
4223                                         "diff--gita/%.*sb/%.*s"
4224                                         "newfilemode%06o"
4225                                         "---/dev/null"
4226                                         "+++b/%.*s",
4227                                         len1, p->one->path,
4228                                         len2, p->two->path,
4229                                         p->two->mode,
4230                                         len2, p->two->path);
4231                 else if (p->two->mode == 0)
4232                         len1 = snprintf(buffer, sizeof(buffer),
4233                                         "diff--gita/%.*sb/%.*s"
4234                                         "deletedfilemode%06o"
4235                                         "---a/%.*s"
4236                                         "+++/dev/null",
4237                                         len1, p->one->path,
4238                                         len2, p->two->path,
4239                                         p->one->mode,
4240                                         len1, p->one->path);
4241                 else
4242                         len1 = snprintf(buffer, sizeof(buffer),
4243                                         "diff--gita/%.*sb/%.*s"
4244                                         "---a/%.*s"
4245                                         "+++b/%.*s",
4246                                         len1, p->one->path,
4247                                         len2, p->two->path,
4248                                         len1, p->one->path,
4249                                         len2, p->two->path);
4250                 git_SHA1_Update(&ctx, buffer, len1);
4251
4252                 if (diff_filespec_is_binary(p->one) ||
4253                     diff_filespec_is_binary(p->two)) {
4254                         git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
4255                         git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
4256                         continue;
4257                 }
4258
4259                 xpp.flags = 0;
4260                 xecfg.ctxlen = 3;
4261                 xecfg.flags = 0;
4262                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4263                               &xpp, &xecfg);
4264         }
4265
4266         git_SHA1_Final(sha1, &ctx);
4267         return 0;
4268 }
4269
4270 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
4271 {
4272         struct diff_queue_struct *q = &diff_queued_diff;
4273         int i;
4274         int result = diff_get_patch_id(options, sha1);
4275
4276         for (i = 0; i < q->nr; i++)
4277                 diff_free_filepair(q->queue[i]);
4278
4279         free(q->queue);
4280         DIFF_QUEUE_CLEAR(q);
4281
4282         return result;
4283 }
4284
4285 static int is_summary_empty(const struct diff_queue_struct *q)
4286 {
4287         int i;
4288
4289         for (i = 0; i < q->nr; i++) {
4290                 const struct diff_filepair *p = q->queue[i];
4291
4292                 switch (p->status) {
4293                 case DIFF_STATUS_DELETED:
4294                 case DIFF_STATUS_ADDED:
4295                 case DIFF_STATUS_COPIED:
4296                 case DIFF_STATUS_RENAMED:
4297                         return 0;
4298                 default:
4299                         if (p->score)
4300                                 return 0;
4301                         if (p->one->mode && p->two->mode &&
4302                             p->one->mode != p->two->mode)
4303                                 return 0;
4304                         break;
4305                 }
4306         }
4307         return 1;
4308 }
4309
4310 static const char rename_limit_warning[] =
4311 "inexact rename detection was skipped due to too many files.";
4312
4313 static const char degrade_cc_to_c_warning[] =
4314 "only found copies from modified paths due to too many files.";
4315
4316 static const char rename_limit_advice[] =
4317 "you may want to set your %s variable to at least "
4318 "%d and retry the command.";
4319
4320 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4321 {
4322         if (degraded_cc)
4323                 warning(degrade_cc_to_c_warning);
4324         else if (needed)
4325                 warning(rename_limit_warning);
4326         else
4327                 return;
4328         if (0 < needed && needed < 32767)
4329                 warning(rename_limit_advice, varname, needed);
4330 }
4331
4332 void diff_flush(struct diff_options *options)
4333 {
4334         struct diff_queue_struct *q = &diff_queued_diff;
4335         int i, output_format = options->output_format;
4336         int separator = 0;
4337         int dirstat_by_line = 0;
4338
4339         /*
4340          * Order: raw, stat, summary, patch
4341          * or:    name/name-status/checkdiff (other bits clear)
4342          */
4343         if (!q->nr)
4344                 goto free_queue;
4345
4346         if (output_format & (DIFF_FORMAT_RAW |
4347                              DIFF_FORMAT_NAME |
4348                              DIFF_FORMAT_NAME_STATUS |
4349                              DIFF_FORMAT_CHECKDIFF)) {
4350                 for (i = 0; i < q->nr; i++) {
4351                         struct diff_filepair *p = q->queue[i];
4352                         if (check_pair_status(p))
4353                                 flush_one_pair(p, options);
4354                 }
4355                 separator++;
4356         }
4357
4358         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4359                 dirstat_by_line = 1;
4360
4361         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4362             dirstat_by_line) {
4363                 struct diffstat_t diffstat;
4364
4365                 memset(&diffstat, 0, sizeof(struct diffstat_t));
4366                 for (i = 0; i < q->nr; i++) {
4367                         struct diff_filepair *p = q->queue[i];
4368                         if (check_pair_status(p))
4369                                 diff_flush_stat(p, options, &diffstat);
4370                 }
4371                 if (output_format & DIFF_FORMAT_NUMSTAT)
4372                         show_numstat(&diffstat, options);
4373                 if (output_format & DIFF_FORMAT_DIFFSTAT)
4374                         show_stats(&diffstat, options);
4375                 if (output_format & DIFF_FORMAT_SHORTSTAT)
4376                         show_shortstats(&diffstat, options);
4377                 if (output_format & DIFF_FORMAT_DIRSTAT)
4378                         show_dirstat_by_line(&diffstat, options);
4379                 free_diffstat_info(&diffstat);
4380                 separator++;
4381         }
4382         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4383                 show_dirstat(options);
4384
4385         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4386                 for (i = 0; i < q->nr; i++) {
4387                         diff_summary(options, q->queue[i]);
4388                 }
4389                 separator++;
4390         }
4391
4392         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4393             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4394             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4395                 /*
4396                  * run diff_flush_patch for the exit status. setting
4397                  * options->file to /dev/null should be safe, becaue we
4398                  * aren't supposed to produce any output anyway.
4399                  */
4400                 if (options->close_file)
4401                         fclose(options->file);
4402                 options->file = fopen("/dev/null", "w");
4403                 if (!options->file)
4404                         die_errno("Could not open /dev/null");
4405                 options->close_file = 1;
4406                 for (i = 0; i < q->nr; i++) {
4407                         struct diff_filepair *p = q->queue[i];
4408                         if (check_pair_status(p))
4409                                 diff_flush_patch(p, options);
4410                         if (options->found_changes)
4411                                 break;
4412                 }
4413         }
4414
4415         if (output_format & DIFF_FORMAT_PATCH) {
4416                 if (separator) {
4417                         putc(options->line_termination, options->file);
4418                         if (options->stat_sep) {
4419                                 /* attach patch instead of inline */
4420                                 fputs(options->stat_sep, options->file);
4421                         }
4422                 }
4423
4424                 for (i = 0; i < q->nr; i++) {
4425                         struct diff_filepair *p = q->queue[i];
4426                         if (check_pair_status(p))
4427                                 diff_flush_patch(p, options);
4428                 }
4429         }
4430
4431         if (output_format & DIFF_FORMAT_CALLBACK)
4432                 options->format_callback(q, options, options->format_callback_data);
4433
4434         for (i = 0; i < q->nr; i++)
4435                 diff_free_filepair(q->queue[i]);
4436 free_queue:
4437         free(q->queue);
4438         DIFF_QUEUE_CLEAR(q);
4439         if (options->close_file)
4440                 fclose(options->file);
4441
4442         /*
4443          * Report the content-level differences with HAS_CHANGES;
4444          * diff_addremove/diff_change does not set the bit when
4445          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4446          */
4447         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4448                 if (options->found_changes)
4449                         DIFF_OPT_SET(options, HAS_CHANGES);
4450                 else
4451                         DIFF_OPT_CLR(options, HAS_CHANGES);
4452         }
4453 }
4454
4455 static void diffcore_apply_filter(const char *filter)
4456 {
4457         int i;
4458         struct diff_queue_struct *q = &diff_queued_diff;
4459         struct diff_queue_struct outq;
4460         DIFF_QUEUE_CLEAR(&outq);
4461
4462         if (!filter)
4463                 return;
4464
4465         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
4466                 int found;
4467                 for (i = found = 0; !found && i < q->nr; i++) {
4468                         struct diff_filepair *p = q->queue[i];
4469                         if (((p->status == DIFF_STATUS_MODIFIED) &&
4470                              ((p->score &&
4471                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
4472                               (!p->score &&
4473                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
4474                             ((p->status != DIFF_STATUS_MODIFIED) &&
4475                              strchr(filter, p->status)))
4476                                 found++;
4477                 }
4478                 if (found)
4479                         return;
4480
4481                 /* otherwise we will clear the whole queue
4482                  * by copying the empty outq at the end of this
4483                  * function, but first clear the current entries
4484                  * in the queue.
4485                  */
4486                 for (i = 0; i < q->nr; i++)
4487                         diff_free_filepair(q->queue[i]);
4488         }
4489         else {
4490                 /* Only the matching ones */
4491                 for (i = 0; i < q->nr; i++) {
4492                         struct diff_filepair *p = q->queue[i];
4493
4494                         if (((p->status == DIFF_STATUS_MODIFIED) &&
4495                              ((p->score &&
4496                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
4497                               (!p->score &&
4498                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
4499                             ((p->status != DIFF_STATUS_MODIFIED) &&
4500                              strchr(filter, p->status)))
4501                                 diff_q(&outq, p);
4502                         else
4503                                 diff_free_filepair(p);
4504                 }
4505         }
4506         free(q->queue);
4507         *q = outq;
4508 }
4509
4510 /* Check whether two filespecs with the same mode and size are identical */
4511 static int diff_filespec_is_identical(struct diff_filespec *one,
4512                                       struct diff_filespec *two)
4513 {
4514         if (S_ISGITLINK(one->mode))
4515                 return 0;
4516         if (diff_populate_filespec(one, 0))
4517                 return 0;
4518         if (diff_populate_filespec(two, 0))
4519                 return 0;
4520         return !memcmp(one->data, two->data, one->size);
4521 }
4522
4523 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4524 {
4525         int i;
4526         struct diff_queue_struct *q = &diff_queued_diff;
4527         struct diff_queue_struct outq;
4528         DIFF_QUEUE_CLEAR(&outq);
4529
4530         for (i = 0; i < q->nr; i++) {
4531                 struct diff_filepair *p = q->queue[i];
4532
4533                 /*
4534                  * 1. Entries that come from stat info dirtiness
4535                  *    always have both sides (iow, not create/delete),
4536                  *    one side of the object name is unknown, with
4537                  *    the same mode and size.  Keep the ones that
4538                  *    do not match these criteria.  They have real
4539                  *    differences.
4540                  *
4541                  * 2. At this point, the file is known to be modified,
4542                  *    with the same mode and size, and the object
4543                  *    name of one side is unknown.  Need to inspect
4544                  *    the identical contents.
4545                  */
4546                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
4547                     !DIFF_FILE_VALID(p->two) ||
4548                     (p->one->sha1_valid && p->two->sha1_valid) ||
4549                     (p->one->mode != p->two->mode) ||
4550                     diff_populate_filespec(p->one, 1) ||
4551                     diff_populate_filespec(p->two, 1) ||
4552                     (p->one->size != p->two->size) ||
4553                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4554                         diff_q(&outq, p);
4555                 else {
4556                         /*
4557                          * The caller can subtract 1 from skip_stat_unmatch
4558                          * to determine how many paths were dirty only
4559                          * due to stat info mismatch.
4560                          */
4561                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4562                                 diffopt->skip_stat_unmatch++;
4563                         diff_free_filepair(p);
4564                 }
4565         }
4566         free(q->queue);
4567         *q = outq;
4568 }
4569
4570 static int diffnamecmp(const void *a_, const void *b_)
4571 {
4572         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4573         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4574         const char *name_a, *name_b;
4575
4576         name_a = a->one ? a->one->path : a->two->path;
4577         name_b = b->one ? b->one->path : b->two->path;
4578         return strcmp(name_a, name_b);
4579 }
4580
4581 void diffcore_fix_diff_index(struct diff_options *options)
4582 {
4583         struct diff_queue_struct *q = &diff_queued_diff;
4584         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4585 }
4586
4587 void diffcore_std(struct diff_options *options)
4588 {
4589         if (options->skip_stat_unmatch)
4590                 diffcore_skip_stat_unmatch(options);
4591         if (!options->found_follow) {
4592                 /* See try_to_follow_renames() in tree-diff.c */
4593                 if (options->break_opt != -1)
4594                         diffcore_break(options->break_opt);
4595                 if (options->detect_rename)
4596                         diffcore_rename(options);
4597                 if (options->break_opt != -1)
4598                         diffcore_merge_broken();
4599         }
4600         if (options->pickaxe)
4601                 diffcore_pickaxe(options);
4602         if (options->orderfile)
4603                 diffcore_order(options->orderfile);
4604         if (!options->found_follow)
4605                 /* See try_to_follow_renames() in tree-diff.c */
4606                 diff_resolve_rename_copy();
4607         diffcore_apply_filter(options->filter);
4608
4609         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4610                 DIFF_OPT_SET(options, HAS_CHANGES);
4611         else
4612                 DIFF_OPT_CLR(options, HAS_CHANGES);
4613
4614         options->found_follow = 0;
4615 }
4616
4617 int diff_result_code(struct diff_options *opt, int status)
4618 {
4619         int result = 0;
4620
4621         diff_warn_rename_limit("diff.renamelimit",
4622                                opt->needed_rename_limit,
4623                                opt->degraded_cc_to_c);
4624         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4625             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4626                 return status;
4627         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4628             DIFF_OPT_TST(opt, HAS_CHANGES))
4629                 result |= 01;
4630         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4631             DIFF_OPT_TST(opt, CHECK_FAILED))
4632                 result |= 02;
4633         return result;
4634 }
4635
4636 int diff_can_quit_early(struct diff_options *opt)
4637 {
4638         return (DIFF_OPT_TST(opt, QUICK) &&
4639                 !opt->filter &&
4640                 DIFF_OPT_TST(opt, HAS_CHANGES));
4641 }
4642
4643 /*
4644  * Shall changes to this submodule be ignored?
4645  *
4646  * Submodule changes can be configured to be ignored separately for each path,
4647  * but that configuration can be overridden from the command line.
4648  */
4649 static int is_submodule_ignored(const char *path, struct diff_options *options)
4650 {
4651         int ignored = 0;
4652         unsigned orig_flags = options->flags;
4653         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4654                 set_diffopt_flags_from_submodule_config(options, path);
4655         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4656                 ignored = 1;
4657         options->flags = orig_flags;
4658         return ignored;
4659 }
4660
4661 void diff_addremove(struct diff_options *options,
4662                     int addremove, unsigned mode,
4663                     const unsigned char *sha1,
4664                     const char *concatpath, unsigned dirty_submodule)
4665 {
4666         struct diff_filespec *one, *two;
4667
4668         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4669                 return;
4670
4671         /* This may look odd, but it is a preparation for
4672          * feeding "there are unchanged files which should
4673          * not produce diffs, but when you are doing copy
4674          * detection you would need them, so here they are"
4675          * entries to the diff-core.  They will be prefixed
4676          * with something like '=' or '*' (I haven't decided
4677          * which but should not make any difference).
4678          * Feeding the same new and old to diff_change()
4679          * also has the same effect.
4680          * Before the final output happens, they are pruned after
4681          * merged into rename/copy pairs as appropriate.
4682          */
4683         if (DIFF_OPT_TST(options, REVERSE_DIFF))
4684                 addremove = (addremove == '+' ? '-' :
4685                              addremove == '-' ? '+' : addremove);
4686
4687         if (options->prefix &&
4688             strncmp(concatpath, options->prefix, options->prefix_length))
4689                 return;
4690
4691         one = alloc_filespec(concatpath);
4692         two = alloc_filespec(concatpath);
4693
4694         if (addremove != '+')
4695                 fill_filespec(one, sha1, mode);
4696         if (addremove != '-') {
4697                 fill_filespec(two, sha1, mode);
4698                 two->dirty_submodule = dirty_submodule;
4699         }
4700
4701         diff_queue(&diff_queued_diff, one, two);
4702         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4703                 DIFF_OPT_SET(options, HAS_CHANGES);
4704 }
4705
4706 void diff_change(struct diff_options *options,
4707                  unsigned old_mode, unsigned new_mode,
4708                  const unsigned char *old_sha1,
4709                  const unsigned char *new_sha1,
4710                  const char *concatpath,
4711                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4712 {
4713         struct diff_filespec *one, *two;
4714
4715         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4716             is_submodule_ignored(concatpath, options))
4717                 return;
4718
4719         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4720                 unsigned tmp;
4721                 const unsigned char *tmp_c;
4722                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4723                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4724                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
4725                         new_dirty_submodule = tmp;
4726         }
4727
4728         if (options->prefix &&
4729             strncmp(concatpath, options->prefix, options->prefix_length))
4730                 return;
4731
4732         one = alloc_filespec(concatpath);
4733         two = alloc_filespec(concatpath);
4734         fill_filespec(one, old_sha1, old_mode);
4735         fill_filespec(two, new_sha1, new_mode);
4736         one->dirty_submodule = old_dirty_submodule;
4737         two->dirty_submodule = new_dirty_submodule;
4738
4739         diff_queue(&diff_queued_diff, one, two);
4740         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4741                 DIFF_OPT_SET(options, HAS_CHANGES);
4742 }
4743
4744 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
4745 {
4746         struct diff_filepair *pair;
4747         struct diff_filespec *one, *two;
4748
4749         if (options->prefix &&
4750             strncmp(path, options->prefix, options->prefix_length))
4751                 return NULL;
4752
4753         one = alloc_filespec(path);
4754         two = alloc_filespec(path);
4755         pair = diff_queue(&diff_queued_diff, one, two);
4756         pair->is_unmerged = 1;
4757         return pair;
4758 }
4759
4760 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
4761                 size_t *outsize)
4762 {
4763         struct diff_tempfile *temp;
4764         const char *argv[3];
4765         const char **arg = argv;
4766         struct child_process child;
4767         struct strbuf buf = STRBUF_INIT;
4768         int err = 0;
4769
4770         temp = prepare_temp_file(spec->path, spec);
4771         *arg++ = pgm;
4772         *arg++ = temp->name;
4773         *arg = NULL;
4774
4775         memset(&child, 0, sizeof(child));
4776         child.use_shell = 1;
4777         child.argv = argv;
4778         child.out = -1;
4779         if (start_command(&child)) {
4780                 remove_tempfile();
4781                 return NULL;
4782         }
4783
4784         if (strbuf_read(&buf, child.out, 0) < 0)
4785                 err = error("error reading from textconv command '%s'", pgm);
4786         close(child.out);
4787
4788         if (finish_command(&child) || err) {
4789                 strbuf_release(&buf);
4790                 remove_tempfile();
4791                 return NULL;
4792         }
4793         remove_tempfile();
4794
4795         return strbuf_detach(&buf, outsize);
4796 }
4797
4798 size_t fill_textconv(struct userdiff_driver *driver,
4799                      struct diff_filespec *df,
4800                      char **outbuf)
4801 {
4802         size_t size;
4803
4804         if (!driver || !driver->textconv) {
4805                 if (!DIFF_FILE_VALID(df)) {
4806                         *outbuf = "";
4807                         return 0;
4808                 }
4809                 if (diff_populate_filespec(df, 0))
4810                         die("unable to read files to diff");
4811                 *outbuf = df->data;
4812                 return df->size;
4813         }
4814
4815         if (driver->textconv_cache && df->sha1_valid) {
4816                 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
4817                                           &size);
4818                 if (*outbuf)
4819                         return size;
4820         }
4821
4822         *outbuf = run_textconv(driver->textconv, df, &size);
4823         if (!*outbuf)
4824                 die("unable to read files to diff");
4825
4826         if (driver->textconv_cache && df->sha1_valid) {
4827                 /* ignore errors, as we might be in a readonly repository */
4828                 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
4829                                 size);
4830                 /*
4831                  * we could save up changes and flush them all at the end,
4832                  * but we would need an extra call after all diffing is done.
4833                  * Since generating a cache entry is the slow path anyway,
4834                  * this extra overhead probably isn't a big deal.
4835                  */
4836                 notes_cache_write(driver->textconv_cache);
4837         }
4838
4839         return size;
4840 }