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