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