Pileus Git - ~andy/git/blob - 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
  15 #ifdef NO_FAST_WORKING_DIRECTORY
  16 #define FAST_WORKING_DIRECTORY 0
  17 #else
  18 #define FAST_WORKING_DIRECTORY 1
  19 #endif
  20
  21 static int diff_detect_rename_default;
  22 static int diff_rename_limit_default = 200;
  23 static int diff_suppress_blank_empty;
  24 int diff_use_color_default = -1;
  25 static const char *external_diff_cmd_cfg;
  26 int diff_auto_refresh_index = 1;
  27 static int diff_mnemonic_prefix;
  28
  29 static char diff_colors[][COLOR_MAXLEN] = {
  30         "\033[m",       /* reset */
  31         "",             /* PLAIN (normal) */
  32         "\033[1m",      /* METAINFO (bold) */
  33         "\033[36m",     /* FRAGINFO (cyan) */
  34         "\033[31m",     /* OLD (red) */
  35         "\033[32m",     /* NEW (green) */
  36         "\033[33m",     /* COMMIT (yellow) */
  37         "\033[41m",     /* WHITESPACE (red background) */
  38 };
  39
  40 static int parse_diff_color_slot(const char *var, int ofs)
  41 {
  42         if (!strcasecmp(var+ofs, "plain"))
  43                 return DIFF_PLAIN;
  44         if (!strcasecmp(var+ofs, "meta"))
  45                 return DIFF_METAINFO;
  46         if (!strcasecmp(var+ofs, "frag"))
  47                 return DIFF_FRAGINFO;
  48         if (!strcasecmp(var+ofs, "old"))
  49                 return DIFF_FILE_OLD;
  50         if (!strcasecmp(var+ofs, "new"))
  51                 return DIFF_FILE_NEW;
  52         if (!strcasecmp(var+ofs, "commit"))
  53                 return DIFF_COMMIT;
  54         if (!strcasecmp(var+ofs, "whitespace"))
  55                 return DIFF_WHITESPACE;
  56         die("bad config variable '%s'", var);
  57 }
  58
  59 static struct ll_diff_driver {
  60         const char *name;
  61         struct ll_diff_driver *next;
  62         const char *cmd;
  63 } *user_diff, **user_diff_tail;
  64
  65 /*
  66  * Currently there is only "diff.<drivername>.command" variable;
  67  * because there are "diff.color.<slot>" variables, we are parsing
  68  * this in a bit convoluted way to allow low level diff driver
  69  * called "color".
  70  */
  71 static int parse_lldiff_command(const char *var, const char *ep, const char *value)
  72 {
  73         const char *name;
  74         int namelen;
  75         struct ll_diff_driver *drv;
  76
  77         name = var + 5;
  78         namelen = ep - name;
  79         for (drv = user_diff; drv; drv = drv->next)
  80                 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
  81                         break;
  82         if (!drv) {
  83                 drv = xcalloc(1, sizeof(struct ll_diff_driver));
  84                 drv->name = xmemdupz(name, namelen);
  85                 if (!user_diff_tail)
  86                         user_diff_tail = &user_diff;
  87                 *user_diff_tail = drv;
  88                 user_diff_tail = &(drv->next);
  89         }
  90
  91         return git_config_string(&(drv->cmd), var, value);
  92 }
  93
  94 /*
  95  * 'diff.<what>.funcname' attribute can be specified in the configuration
  96  * to define a customized regexp to find the beginning of a function to
  97  * be used for hunk header lines of "diff -p" style output.
  98  */
  99 struct funcname_pattern_entry {
 100         char *name;
 101         char *pattern;
 102         int cflags;
 103 };
 104 static struct funcname_pattern_list {
 105         struct funcname_pattern_list *next;
 106         struct funcname_pattern_entry e;
 107 } *funcname_pattern_list;
 108
 109 static int parse_funcname_pattern(const char *var, const char *ep, const char *value, int cflags)
 110 {
 111         const char *name;
 112         int namelen;
 113         struct funcname_pattern_list *pp;
 114
 115         name = var + 5; /* "diff." */
 116         namelen = ep - name;
 117
 118         for (pp = funcname_pattern_list; pp; pp = pp->next)
 119                 if (!strncmp(pp->e.name, name, namelen) && !pp->e.name[namelen])
 120                         break;
 121         if (!pp) {
 122                 pp = xcalloc(1, sizeof(*pp));
 123                 pp->e.name = xmemdupz(name, namelen);
 124                 pp->next = funcname_pattern_list;
 125                 funcname_pattern_list = pp;
 126         }
 127         free(pp->e.pattern);
 128         pp->e.pattern = xstrdup(value);
 129         pp->e.cflags = cflags;
 130         return 0;
 131 }
 132
 133 /*
 134  * These are to give UI layer defaults.
 135  * The core-level commands such as git-diff-files should
 136  * never be affected by the setting of diff.renames
 137  * the user happens to have in the configuration file.
 138  */
 139 int git_diff_ui_config(const char *var, const char *value, void *cb)
 140 {
 141         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 142                 diff_use_color_default = git_config_colorbool(var, value, -1);
 143                 return 0;
 144         }
 145         if (!strcmp(var, "diff.renames")) {
 146                 if (!value)
 147                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 148                 else if (!strcasecmp(value, "copies") ||
 149                          !strcasecmp(value, "copy"))
 150                         diff_detect_rename_default = DIFF_DETECT_COPY;
 151                 else if (git_config_bool(var,value))
 152                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 153                 return 0;
 154         }
 155         if (!strcmp(var, "diff.autorefreshindex")) {
 156                 diff_auto_refresh_index = git_config_bool(var, value);
 157                 return 0;
 158         }
 159         if (!strcmp(var, "diff.mnemonicprefix")) {
 160                 diff_mnemonic_prefix = git_config_bool(var, value);
 161                 return 0;
 162         }
 163         if (!strcmp(var, "diff.external"))
 164                 return git_config_string(&external_diff_cmd_cfg, var, value);
 165         if (!prefixcmp(var, "diff.")) {
 166                 const char *ep = strrchr(var, '.');
 167
 168                 if (ep != var + 4 && !strcmp(ep, ".command"))
 169                         return parse_lldiff_command(var, ep, value);
 170         }
 171
 172         return git_diff_basic_config(var, value, cb);
 173 }
 174
 175 int git_diff_basic_config(const char *var, const char *value, void *cb)
 176 {
 177         if (!strcmp(var, "diff.renamelimit")) {
 178                 diff_rename_limit_default = git_config_int(var, value);
 179                 return 0;
 180         }
 181
 182         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 183                 int slot = parse_diff_color_slot(var, 11);
 184                 if (!value)
 185                         return config_error_nonbool(var);
 186                 color_parse(value, var, diff_colors[slot]);
 187                 return 0;
 188         }
 189
 190         /* like GNU diff's --suppress-blank-empty option  */
 191         if (!strcmp(var, "diff.suppress-blank-empty")) {
 192                 diff_suppress_blank_empty = git_config_bool(var, value);
 193                 return 0;
 194         }
 195
 196         if (!prefixcmp(var, "diff.")) {
 197                 const char *ep = strrchr(var, '.');
 198                 if (ep != var + 4) {
 199                         if (!strcmp(ep, ".funcname")) {
 200                                 if (!value)
 201                                         return config_error_nonbool(var);
 202                                 return parse_funcname_pattern(var, ep, value,
 203                                         0);
 204                         } else if (!strcmp(ep, ".xfuncname")) {
 205                                 if (!value)
 206                                         return config_error_nonbool(var);
 207                                 return parse_funcname_pattern(var, ep, value,
 208                                         REG_EXTENDED);
 209                         }
 210                 }
 211         }
 212
 213         return git_color_default_config(var, value, cb);
 214 }
 215
 216 static char *quote_two(const char *one, const char *two)
 217 {
 218         int need_one = quote_c_style(one, NULL, NULL, 1);
 219         int need_two = quote_c_style(two, NULL, NULL, 1);
 220         struct strbuf res;
 221
 222         strbuf_init(&res, 0);
 223         if (need_one + need_two) {
 224                 strbuf_addch(&res, '"');
 225                 quote_c_style(one, &res, NULL, 1);
 226                 quote_c_style(two, &res, NULL, 1);
 227                 strbuf_addch(&res, '"');
 228         } else {
 229                 strbuf_addstr(&res, one);
 230                 strbuf_addstr(&res, two);
 231         }
 232         return strbuf_detach(&res, NULL);
 233 }
 234
 235 static const char *external_diff(void)
 236 {
 237         static const char *external_diff_cmd = NULL;
 238         static int done_preparing = 0;
 239
 240         if (done_preparing)
 241                 return external_diff_cmd;
 242         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 243         if (!external_diff_cmd)
 244                 external_diff_cmd = external_diff_cmd_cfg;
 245         done_preparing = 1;
 246         return external_diff_cmd;
 247 }
 248
 249 static struct diff_tempfile {
 250         const char *name; /* filename external diff should read from */
 251         char hex[41];
 252         char mode[10];
 253         char tmp_path[PATH_MAX];
 254 } diff_temp[2];
 255
 256 static int count_lines(const char *data, int size)
 257 {
 258         int count, ch, completely_empty = 1, nl_just_seen = 0;
 259         count = 0;
 260         while (0 < size--) {
 261                 ch = *data++;
 262                 if (ch == '\n') {
 263                         count++;
 264                         nl_just_seen = 1;
 265                         completely_empty = 0;
 266                 }
 267                 else {
 268                         nl_just_seen = 0;
 269                         completely_empty = 0;
 270                 }
 271         }
 272         if (completely_empty)
 273                 return 0;
 274         if (!nl_just_seen)
 275                 count++; /* no trailing newline */
 276         return count;
 277 }
 278
 279 static void print_line_count(FILE *file, int count)
 280 {
 281         switch (count) {
 282         case 0:
 283                 fprintf(file, "0,0");
 284                 break;
 285         case 1:
 286                 fprintf(file, "1");
 287                 break;
 288         default:
 289                 fprintf(file, "1,%d", count);
 290                 break;
 291         }
 292 }
 293
 294 static void copy_file_with_prefix(FILE *file,
 295                                   int prefix, const char *data, int size,
 296                                   const char *set, const char *reset)
 297 {
 298         int ch, nl_just_seen = 1;
 299         while (0 < size--) {
 300                 ch = *data++;
 301                 if (nl_just_seen) {
 302                         fputs(set, file);
 303                         putc(prefix, file);
 304                 }
 305                 if (ch == '\n') {
 306                         nl_just_seen = 1;
 307                         fputs(reset, file);
 308                 } else
 309                         nl_just_seen = 0;
 310                 putc(ch, file);
 311         }
 312         if (!nl_just_seen)
 313                 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
 314 }
 315
 316 static void emit_rewrite_diff(const char *name_a,
 317                               const char *name_b,
 318                               struct diff_filespec *one,
 319                               struct diff_filespec *two,
 320                               struct diff_options *o)
 321 {
 322         int lc_a, lc_b;
 323         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
 324         const char *name_a_tab, *name_b_tab;
 325         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 326         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 327         const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
 328         const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
 329         const char *reset = diff_get_color(color_diff, DIFF_RESET);
 330         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 331         const char *a_prefix, *b_prefix;
 332
 333         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 334                 a_prefix = o->b_prefix;
 335                 b_prefix = o->a_prefix;
 336         } else {
 337                 a_prefix = o->a_prefix;
 338                 b_prefix = o->b_prefix;
 339         }
 340
 341         name_a += (*name_a == '/');
 342         name_b += (*name_b == '/');
 343         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 344         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 345
 346         strbuf_reset(&a_name);
 347         strbuf_reset(&b_name);
 348         quote_two_c_style(&a_name, a_prefix, name_a, 0);
 349         quote_two_c_style(&b_name, b_prefix, name_b, 0);
 350
 351         diff_populate_filespec(one, 0);
 352         diff_populate_filespec(two, 0);
 353         lc_a = count_lines(one->data, one->size);
 354         lc_b = count_lines(two->data, two->size);
 355         fprintf(o->file,
 356                 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
 357                 metainfo, a_name.buf, name_a_tab, reset,
 358                 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
 359         print_line_count(o->file, lc_a);
 360         fprintf(o->file, " +");
 361         print_line_count(o->file, lc_b);
 362         fprintf(o->file, " @@%s\n", reset);
 363         if (lc_a)
 364                 copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
 365         if (lc_b)
 366                 copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
 367 }
 368
 369 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 370 {
 371         if (!DIFF_FILE_VALID(one)) {
 372                 mf->ptr = (char *)""; /* does not matter */
 373                 mf->size = 0;
 374                 return 0;
 375         }
 376         else if (diff_populate_filespec(one, 0))
 377                 return -1;
 378         mf->ptr = one->data;
 379         mf->size = one->size;
 380         return 0;
 381 }
 382
 383 struct diff_words_buffer {
 384         mmfile_t text;
 385         long alloc;
 386         long current; /* output pointer */
 387         int suppressed_newline;
 388 };
 389
 390 static void diff_words_append(char *line, unsigned long len,
 391                 struct diff_words_buffer *buffer)
 392 {
 393         if (buffer->text.size + len > buffer->alloc) {
 394                 buffer->alloc = (buffer->text.size + len) * 3 / 2;
 395                 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
 396         }
 397         line++;
 398         len--;
 399         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 400         buffer->text.size += len;
 401 }
 402
 403 struct diff_words_data {
 404         struct diff_words_buffer minus, plus;
 405         FILE *file;
 406 };
 407
 408 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
 409                 int suppress_newline)
 410 {
 411         const char *ptr;
 412         int eol = 0;
 413
 414         if (len == 0)
 415                 return;
 416
 417         ptr  = buffer->text.ptr + buffer->current;
 418         buffer->current += len;
 419
 420         if (ptr[len - 1] == '\n') {
 421                 eol = 1;
 422                 len--;
 423         }
 424
 425         fputs(diff_get_color(1, color), file);
 426         fwrite(ptr, len, 1, file);
 427         fputs(diff_get_color(1, DIFF_RESET), file);
 428
 429         if (eol) {
 430                 if (suppress_newline)
 431                         buffer->suppressed_newline = 1;
 432                 else
 433                         putc('\n', file);
 434         }
 435 }
 436
 437 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 438 {
 439         struct diff_words_data *diff_words = priv;
 440
 441         if (diff_words->minus.suppressed_newline) {
 442                 if (line[0] != '+')
 443                         putc('\n', diff_words->file);
 444                 diff_words->minus.suppressed_newline = 0;
 445         }
 446
 447         len--;
 448         switch (line[0]) {
 449                 case '-':
 450                         print_word(diff_words->file,
 451                                    &diff_words->minus, len, DIFF_FILE_OLD, 1);
 452                         break;
 453                 case '+':
 454                         print_word(diff_words->file,
 455                                    &diff_words->plus, len, DIFF_FILE_NEW, 0);
 456                         break;
 457                 case ' ':
 458                         print_word(diff_words->file,
 459                                    &diff_words->plus, len, DIFF_PLAIN, 0);
 460                         diff_words->minus.current += len;
 461                         break;
 462         }
 463 }
 464
 465 /* this executes the word diff on the accumulated buffers */
 466 static void diff_words_show(struct diff_words_data *diff_words)
 467 {
 468         xpparam_t xpp;
 469         xdemitconf_t xecfg;
 470         xdemitcb_t ecb;
 471         mmfile_t minus, plus;
 472         int i;
 473
 474         memset(&xecfg, 0, sizeof(xecfg));
 475         minus.size = diff_words->minus.text.size;
 476         minus.ptr = xmalloc(minus.size);
 477         memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
 478         for (i = 0; i < minus.size; i++)
 479                 if (isspace(minus.ptr[i]))
 480                         minus.ptr[i] = '\n';
 481         diff_words->minus.current = 0;
 482
 483         plus.size = diff_words->plus.text.size;
 484         plus.ptr = xmalloc(plus.size);
 485         memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
 486         for (i = 0; i < plus.size; i++)
 487                 if (isspace(plus.ptr[i]))
 488                         plus.ptr[i] = '\n';
 489         diff_words->plus.current = 0;
 490
 491         xpp.flags = XDF_NEED_MINIMAL;
 492         xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
 493         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
 494                       &xpp, &xecfg, &ecb);
 495         free(minus.ptr);
 496         free(plus.ptr);
 497         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 498
 499         if (diff_words->minus.suppressed_newline) {
 500                 putc('\n', diff_words->file);
 501                 diff_words->minus.suppressed_newline = 0;
 502         }
 503 }
 504
 505 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 506
 507 struct emit_callback {
 508         int nparents, color_diff;
 509         unsigned ws_rule;
 510         sane_truncate_fn truncate;
 511         const char **label_path;
 512         struct diff_words_data *diff_words;
 513         int *found_changesp;
 514         FILE *file;
 515 };
 516
 517 static void free_diff_words_data(struct emit_callback *ecbdata)
 518 {
 519         if (ecbdata->diff_words) {
 520                 /* flush buffers */
 521                 if (ecbdata->diff_words->minus.text.size ||
 522                                 ecbdata->diff_words->plus.text.size)
 523                         diff_words_show(ecbdata->diff_words);
 524
 525                 free (ecbdata->diff_words->minus.text.ptr);
 526                 free (ecbdata->diff_words->plus.text.ptr);
 527                 free(ecbdata->diff_words);
 528                 ecbdata->diff_words = NULL;
 529         }
 530 }
 531
 532 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 533 {
 534         if (diff_use_color)
 535                 return diff_colors[ix];
 536         return "";
 537 }
 538
 539 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
 540 {
 541         int has_trailing_newline, has_trailing_carriage_return;
 542
 543         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 544         if (has_trailing_newline)
 545                 len--;
 546         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 547         if (has_trailing_carriage_return)
 548                 len--;
 549
 550         fputs(set, file);
 551         fwrite(line, len, 1, file);
 552         fputs(reset, file);
 553         if (has_trailing_carriage_return)
 554                 fputc('\r', file);
 555         if (has_trailing_newline)
 556                 fputc('\n', file);
 557 }
 558
 559 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
 560 {
 561         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 562         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 563
 564         if (!*ws)
 565                 emit_line(ecbdata->file, set, reset, line, len);
 566         else {
 567                 /* Emit just the prefix, then the rest. */
 568                 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
 569                 ws_check_emit(line + ecbdata->nparents,
 570                               len - ecbdata->nparents, ecbdata->ws_rule,
 571                               ecbdata->file, set, reset, ws);
 572         }
 573 }
 574
 575 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
 576 {
 577         const char *cp;
 578         unsigned long allot;
 579         size_t l = len;
 580
 581         if (ecb->truncate)
 582                 return ecb->truncate(line, len);
 583         cp = line;
 584         allot = l;
 585         while (0 < l) {
 586                 (void) utf8_width(&cp, &l);
 587                 if (!cp)
 588                         break; /* truncated in the middle? */
 589         }
 590         return allot - l;
 591 }
 592
 593 static void fn_out_consume(void *priv, char *line, unsigned long len)
 594 {
 595         int i;
 596         int color;
 597         struct emit_callback *ecbdata = priv;
 598         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 599         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 600         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 601
 602         *(ecbdata->found_changesp) = 1;
 603
 604         if (ecbdata->label_path[0]) {
 605                 const char *name_a_tab, *name_b_tab;
 606
 607                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 608                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 609
 610                 fprintf(ecbdata->file, "%s--- %s%s%s\n",
 611                         meta, ecbdata->label_path[0], reset, name_a_tab);
 612                 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
 613                         meta, ecbdata->label_path[1], reset, name_b_tab);
 614                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 615         }
 616
 617         if (diff_suppress_blank_empty
 618             && len == 2 && line[0] == ' ' && line[1] == '\n') {
 619                 line[0] = '\n';
 620                 len = 1;
 621         }
 622
 623         /* This is not really necessary for now because
 624          * this codepath only deals with two-way diffs.
 625          */
 626         for (i = 0; i < len && line[i] == '@'; i++)
 627                 ;
 628         if (2 <= i && i < len && line[i] == ' ') {
 629                 ecbdata->nparents = i - 1;
 630                 len = sane_truncate_line(ecbdata, line, len);
 631                 emit_line(ecbdata->file,
 632                           diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
 633                           reset, line, len);
 634                 if (line[len-1] != '\n')
 635                         putc('\n', ecbdata->file);
 636                 return;
 637         }
 638
 639         if (len < ecbdata->nparents) {
 640                 emit_line(ecbdata->file, reset, reset, line, len);
 641                 return;
 642         }
 643
 644         color = DIFF_PLAIN;
 645         if (ecbdata->diff_words && ecbdata->nparents != 1)
 646                 /* fall back to normal diff */
 647                 free_diff_words_data(ecbdata);
 648         if (ecbdata->diff_words) {
 649                 if (line[0] == '-') {
 650                         diff_words_append(line, len,
 651                                           &ecbdata->diff_words->minus);
 652                         return;
 653                 } else if (line[0] == '+') {
 654                         diff_words_append(line, len,
 655                                           &ecbdata->diff_words->plus);
 656                         return;
 657                 }
 658                 if (ecbdata->diff_words->minus.text.size ||
 659                     ecbdata->diff_words->plus.text.size)
 660                         diff_words_show(ecbdata->diff_words);
 661                 line++;
 662                 len--;
 663                 emit_line(ecbdata->file, plain, reset, line, len);
 664                 return;
 665         }
 666         for (i = 0; i < ecbdata->nparents && len; i++) {
 667                 if (line[i] == '-')
 668                         color = DIFF_FILE_OLD;
 669                 else if (line[i] == '+')
 670                         color = DIFF_FILE_NEW;
 671         }
 672
 673         if (color != DIFF_FILE_NEW) {
 674                 emit_line(ecbdata->file,
 675                           diff_get_color(ecbdata->color_diff, color),
 676                           reset, line, len);
 677                 return;
 678         }
 679         emit_add_line(reset, ecbdata, line, len);
 680 }
 681
 682 static char *pprint_rename(const char *a, const char *b)
 683 {
 684         const char *old = a;
 685         const char *new = b;
 686         struct strbuf name;
 687         int pfx_length, sfx_length;
 688         int len_a = strlen(a);
 689         int len_b = strlen(b);
 690         int a_midlen, b_midlen;
 691         int qlen_a = quote_c_style(a, NULL, NULL, 0);
 692         int qlen_b = quote_c_style(b, NULL, NULL, 0);
 693
 694         strbuf_init(&name, 0);
 695         if (qlen_a || qlen_b) {
 696                 quote_c_style(a, &name, NULL, 0);
 697                 strbuf_addstr(&name, " => ");
 698                 quote_c_style(b, &name, NULL, 0);
 699                 return strbuf_detach(&name, NULL);
 700         }
 701
 702         /* Find common prefix */
 703         pfx_length = 0;
 704         while (*old && *new && *old == *new) {
 705                 if (*old == '/')
 706                         pfx_length = old - a + 1;
 707                 old++;
 708                 new++;
 709         }
 710
 711         /* Find common suffix */
 712         old = a + len_a;
 713         new = b + len_b;
 714         sfx_length = 0;
 715         while (a <= old && b <= new && *old == *new) {
 716                 if (*old == '/')
 717                         sfx_length = len_a - (old - a);
 718                 old--;
 719                 new--;
 720         }
 721
 722         /*
 723          * pfx{mid-a => mid-b}sfx
 724          * {pfx-a => pfx-b}sfx
 725          * pfx{sfx-a => sfx-b}
 726          * name-a => name-b
 727          */
 728         a_midlen = len_a - pfx_length - sfx_length;
 729         b_midlen = len_b - pfx_length - sfx_length;
 730         if (a_midlen < 0)
 731                 a_midlen = 0;
 732         if (b_midlen < 0)
 733                 b_midlen = 0;
 734
 735         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 736         if (pfx_length + sfx_length) {
 737                 strbuf_add(&name, a, pfx_length);
 738                 strbuf_addch(&name, '{');
 739         }
 740         strbuf_add(&name, a + pfx_length, a_midlen);
 741         strbuf_addstr(&name, " => ");
 742         strbuf_add(&name, b + pfx_length, b_midlen);
 743         if (pfx_length + sfx_length) {
 744                 strbuf_addch(&name, '}');
 745                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 746         }
 747         return strbuf_detach(&name, NULL);
 748 }
 749
 750 struct diffstat_t {
 751         int nr;
 752         int alloc;
 753         struct diffstat_file {
 754                 char *from_name;
 755                 char *name;
 756                 char *print_name;
 757                 unsigned is_unmerged:1;
 758                 unsigned is_binary:1;
 759                 unsigned is_renamed:1;
 760                 unsigned int added, deleted;
 761         } **files;
 762 };
 763
 764 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 765                                           const char *name_a,
 766                                           const char *name_b)
 767 {
 768         struct diffstat_file *x;
 769         x = xcalloc(sizeof (*x), 1);
 770         if (diffstat->nr == diffstat->alloc) {
 771                 diffstat->alloc = alloc_nr(diffstat->alloc);
 772                 diffstat->files = xrealloc(diffstat->files,
 773                                 diffstat->alloc * sizeof(x));
 774         }
 775         diffstat->files[diffstat->nr++] = x;
 776         if (name_b) {
 777                 x->from_name = xstrdup(name_a);
 778                 x->name = xstrdup(name_b);
 779                 x->is_renamed = 1;
 780         }
 781         else {
 782                 x->from_name = NULL;
 783                 x->name = xstrdup(name_a);
 784         }
 785         return x;
 786 }
 787
 788 static void diffstat_consume(void *priv, char *line, unsigned long len)
 789 {
 790         struct diffstat_t *diffstat = priv;
 791         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 792
 793         if (line[0] == '+')
 794                 x->added++;
 795         else if (line[0] == '-')
 796                 x->deleted++;
 797 }
 798
 799 const char mime_boundary_leader[] = "------------";
 800
 801 static int scale_linear(int it, int width, int max_change)
 802 {
 803         /*
 804          * make sure that at least one '-' is printed if there were deletions,
 805          * and likewise for '+'.
 806          */
 807         if (max_change < 2)
 808                 return it;
 809         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
 810 }
 811
 812 static void show_name(FILE *file,
 813                       const char *prefix, const char *name, int len,
 814                       const char *reset, const char *set)
 815 {
 816         fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
 817 }
 818
 819 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
 820 {
 821         if (cnt <= 0)
 822                 return;
 823         fprintf(file, "%s", set);
 824         while (cnt--)
 825                 putc(ch, file);
 826         fprintf(file, "%s", reset);
 827 }
 828
 829 static void fill_print_name(struct diffstat_file *file)
 830 {
 831         char *pname;
 832
 833         if (file->print_name)
 834                 return;
 835
 836         if (!file->is_renamed) {
 837                 struct strbuf buf;
 838                 strbuf_init(&buf, 0);
 839                 if (quote_c_style(file->name, &buf, NULL, 0)) {
 840                         pname = strbuf_detach(&buf, NULL);
 841                 } else {
 842                         pname = file->name;
 843                         strbuf_release(&buf);
 844                 }
 845         } else {
 846                 pname = pprint_rename(file->from_name, file->name);
 847         }
 848         file->print_name = pname;
 849 }
 850
 851 static void show_stats(struct diffstat_t* data, struct diff_options *options)
 852 {
 853         int i, len, add, del, total, adds = 0, dels = 0;
 854         int max_change = 0, max_len = 0;
 855         int total_files = data->nr;
 856         int width, name_width;
 857         const char *reset, *set, *add_c, *del_c;
 858
 859         if (data->nr == 0)
 860                 return;
 861
 862         width = options->stat_width ? options->stat_width : 80;
 863         name_width = options->stat_name_width ? options->stat_name_width : 50;
 864
 865         /* Sanity: give at least 5 columns to the graph,
 866          * but leave at least 10 columns for the name.
 867          */
 868         if (width < 25)
 869                 width = 25;
 870         if (name_width < 10)
 871                 name_width = 10;
 872         else if (width < name_width + 15)
 873                 name_width = width - 15;
 874
 875         /* Find the longest filename and max number of changes */
 876         reset = diff_get_color_opt(options, DIFF_RESET);
 877         set   = diff_get_color_opt(options, DIFF_PLAIN);
 878         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
 879         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
 880
 881         for (i = 0; i < data->nr; i++) {
 882                 struct diffstat_file *file = data->files[i];
 883                 int change = file->added + file->deleted;
 884                 fill_print_name(file);
 885                 len = strlen(file->print_name);
 886                 if (max_len < len)
 887                         max_len = len;
 888
 889                 if (file->is_binary || file->is_unmerged)
 890                         continue;
 891                 if (max_change < change)
 892                         max_change = change;
 893         }
 894
 895         /* Compute the width of the graph part;
 896          * 10 is for one blank at the beginning of the line plus
 897          * " | count " between the name and the graph.
 898          *
 899          * From here on, name_width is the width of the name area,
 900          * and width is the width of the graph area.
 901          */
 902         name_width = (name_width < max_len) ? name_width : max_len;
 903         if (width < (name_width + 10) + max_change)
 904                 width = width - (name_width + 10);
 905         else
 906                 width = max_change;
 907
 908         for (i = 0; i < data->nr; i++) {
 909                 const char *prefix = "";
 910                 char *name = data->files[i]->print_name;
 911                 int added = data->files[i]->added;
 912                 int deleted = data->files[i]->deleted;
 913                 int name_len;
 914
 915                 /*
 916                  * "scale" the filename
 917                  */
 918                 len = name_width;
 919                 name_len = strlen(name);
 920                 if (name_width < name_len) {
 921                         char *slash;
 922                         prefix = "...";
 923                         len -= 3;
 924                         name += name_len - len;
 925                         slash = strchr(name, '/');
 926                         if (slash)
 927                                 name = slash;
 928                 }
 929
 930                 if (data->files[i]->is_binary) {
 931                         show_name(options->file, prefix, name, len, reset, set);
 932                         fprintf(options->file, "  Bin ");
 933                         fprintf(options->file, "%s%d%s", del_c, deleted, reset);
 934                         fprintf(options->file, " -> ");
 935                         fprintf(options->file, "%s%d%s", add_c, added, reset);
 936                         fprintf(options->file, " bytes");
 937                         fprintf(options->file, "\n");
 938                         continue;
 939                 }
 940                 else if (data->files[i]->is_unmerged) {
 941                         show_name(options->file, prefix, name, len, reset, set);
 942                         fprintf(options->file, "  Unmerged\n");
 943                         continue;
 944                 }
 945                 else if (!data->files[i]->is_renamed &&
 946                          (added + deleted == 0)) {
 947                         total_files--;
 948                         continue;
 949                 }
 950
 951                 /*
 952                  * scale the add/delete
 953                  */
 954                 add = added;
 955                 del = deleted;
 956                 total = add + del;
 957                 adds += add;
 958                 dels += del;
 959
 960                 if (width <= max_change) {
 961                         add = scale_linear(add, width, max_change);
 962                         del = scale_linear(del, width, max_change);
 963                         total = add + del;
 964                 }
 965                 show_name(options->file, prefix, name, len, reset, set);
 966                 fprintf(options->file, "%5d%s", added + deleted,
 967                                 added + deleted ? " " : "");
 968                 show_graph(options->file, '+', add, add_c, reset);
 969                 show_graph(options->file, '-', del, del_c, reset);
 970                 fprintf(options->file, "\n");
 971         }
 972         fprintf(options->file,
 973                "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
 974                set, total_files, adds, dels, reset);
 975 }
 976
 977 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
 978 {
 979         int i, adds = 0, dels = 0, total_files = data->nr;
 980
 981         if (data->nr == 0)
 982                 return;
 983
 984         for (i = 0; i < data->nr; i++) {
 985                 if (!data->files[i]->is_binary &&
 986                     !data->files[i]->is_unmerged) {
 987                         int added = data->files[i]->added;
 988                         int deleted= data->files[i]->deleted;
 989                         if (!data->files[i]->is_renamed &&
 990                             (added + deleted == 0)) {
 991                                 total_files--;
 992                         } else {
 993                                 adds += added;
 994                                 dels += deleted;
 995                         }
 996                 }
 997         }
 998         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
 999                total_files, adds, dels);
1000 }
1001
1002 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
1003 {
1004         int i;
1005
1006         if (data->nr == 0)
1007                 return;
1008
1009         for (i = 0; i < data->nr; i++) {
1010                 struct diffstat_file *file = data->files[i];
1011
1012                 if (file->is_binary)
1013                         fprintf(options->file, "-\t-\t");
1014                 else
1015                         fprintf(options->file,
1016                                 "%d\t%d\t", file->added, file->deleted);
1017                 if (options->line_termination) {
1018                         fill_print_name(file);
1019                         if (!file->is_renamed)
1020                                 write_name_quoted(file->name, options->file,
1021                                                   options->line_termination);
1022                         else {
1023                                 fputs(file->print_name, options->file);
1024                                 putc(options->line_termination, options->file);
1025                         }
1026                 } else {
1027                         if (file->is_renamed) {
1028                                 putc('\0', options->file);
1029                                 write_name_quoted(file->from_name, options->file, '\0');
1030                         }
1031                         write_name_quoted(file->name, options->file, '\0');
1032                 }
1033         }
1034 }
1035
1036 struct dirstat_file {
1037         const char *name;
1038         unsigned long changed;
1039 };
1040
1041 struct dirstat_dir {
1042         struct dirstat_file *files;
1043         int alloc, nr, percent, cumulative;
1044 };
1045
1046 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1047 {
1048         unsigned long this_dir = 0;
1049         unsigned int sources = 0;
1050
1051         while (dir->nr) {
1052                 struct dirstat_file *f = dir->files;
1053                 int namelen = strlen(f->name);
1054                 unsigned long this;
1055                 char *slash;
1056
1057                 if (namelen < baselen)
1058                         break;
1059                 if (memcmp(f->name, base, baselen))
1060                         break;
1061                 slash = strchr(f->name + baselen, '/');
1062                 if (slash) {
1063                         int newbaselen = slash + 1 - f->name;
1064                         this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1065                         sources++;
1066                 } else {
1067                         this = f->changed;
1068                         dir->files++;
1069                         dir->nr--;
1070                         sources += 2;
1071                 }
1072                 this_dir += this;
1073         }
1074
1075         /*
1076          * We don't report dirstat's for
1077          *  - the top level
1078          *  - or cases where everything came from a single directory
1079          *    under this directory (sources == 1).
1080          */
1081         if (baselen && sources != 1) {
1082                 int permille = this_dir * 1000 / changed;
1083                 if (permille) {
1084                         int percent = permille / 10;
1085                         if (percent >= dir->percent) {
1086                                 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1087                                 if (!dir->cumulative)
1088                                         return 0;
1089                         }
1090                 }
1091         }
1092         return this_dir;
1093 }
1094
1095 static int dirstat_compare(const void *_a, const void *_b)
1096 {
1097         const struct dirstat_file *a = _a;
1098         const struct dirstat_file *b = _b;
1099         return strcmp(a->name, b->name);
1100 }
1101
1102 static void show_dirstat(struct diff_options *options)
1103 {
1104         int i;
1105         unsigned long changed;
1106         struct dirstat_dir dir;
1107         struct diff_queue_struct *q = &diff_queued_diff;
1108
1109         dir.files = NULL;
1110         dir.alloc = 0;
1111         dir.nr = 0;
1112         dir.percent = options->dirstat_percent;
1113         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1114
1115         changed = 0;
1116         for (i = 0; i < q->nr; i++) {
1117                 struct diff_filepair *p = q->queue[i];
1118                 const char *name;
1119                 unsigned long copied, added, damage;
1120
1121                 name = p->one->path ? p->one->path : p->two->path;
1122
1123                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1124                         diff_populate_filespec(p->one, 0);
1125                         diff_populate_filespec(p->two, 0);
1126                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1127                                                &copied, &added);
1128                         diff_free_filespec_data(p->one);
1129                         diff_free_filespec_data(p->two);
1130                 } else if (DIFF_FILE_VALID(p->one)) {
1131                         diff_populate_filespec(p->one, 1);
1132                         copied = added = 0;
1133                         diff_free_filespec_data(p->one);
1134                 } else if (DIFF_FILE_VALID(p->two)) {
1135                         diff_populate_filespec(p->two, 1);
1136                         copied = 0;
1137                         added = p->two->size;
1138                         diff_free_filespec_data(p->two);
1139                 } else
1140                         continue;
1141
1142                 /*
1143                  * Original minus copied is the removed material,
1144                  * added is the new material.  They are both damages
1145                  * made to the preimage. In --dirstat-by-file mode, count
1146                  * damaged files, not damaged lines. This is done by
1147                  * counting only a single damaged line per file.
1148                  */
1149                 damage = (p->one->size - copied) + added;
1150                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1151                         damage = 1;
1152
1153                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1154                 dir.files[dir.nr].name = name;
1155                 dir.files[dir.nr].changed = damage;
1156                 changed += damage;
1157                 dir.nr++;
1158         }
1159
1160         /* This can happen even with many files, if everything was renames */
1161         if (!changed)
1162                 return;
1163
1164         /* Show all directories with more than x% of the changes */
1165         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1166         gather_dirstat(options->file, &dir, changed, "", 0);
1167 }
1168
1169 static void free_diffstat_info(struct diffstat_t *diffstat)
1170 {
1171         int i;
1172         for (i = 0; i < diffstat->nr; i++) {
1173                 struct diffstat_file *f = diffstat->files[i];
1174                 if (f->name != f->print_name)
1175                         free(f->print_name);
1176                 free(f->name);
1177                 free(f->from_name);
1178                 free(f);
1179         }
1180         free(diffstat->files);
1181 }
1182
1183 struct checkdiff_t {
1184         const char *filename;
1185         int lineno;
1186         struct diff_options *o;
1187         unsigned ws_rule;
1188         unsigned status;
1189         int trailing_blanks_start;
1190 };
1191
1192 static int is_conflict_marker(const char *line, unsigned long len)
1193 {
1194         char firstchar;
1195         int cnt;
1196
1197         if (len < 8)
1198                 return 0;
1199         firstchar = line[0];
1200         switch (firstchar) {
1201         case '=': case '>': case '<':
1202                 break;
1203         default:
1204                 return 0;
1205         }
1206         for (cnt = 1; cnt < 7; cnt++)
1207                 if (line[cnt] != firstchar)
1208                         return 0;
1209         /* line[0] thru line[6] are same as firstchar */
1210         if (firstchar == '=') {
1211                 /* divider between ours and theirs? */
1212                 if (len != 8 || line[7] != '\n')
1213                         return 0;
1214         } else if (len < 8 || !isspace(line[7])) {
1215                 /* not divider before ours nor after theirs */
1216                 return 0;
1217         }
1218         return 1;
1219 }
1220
1221 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1222 {
1223         struct checkdiff_t *data = priv;
1224         int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1225         const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1226         const char *reset = diff_get_color(color_diff, DIFF_RESET);
1227         const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1228         char *err;
1229
1230         if (line[0] == '+') {
1231                 unsigned bad;
1232                 data->lineno++;
1233                 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1234                         data->trailing_blanks_start = 0;
1235                 else if (!data->trailing_blanks_start)
1236                         data->trailing_blanks_start = data->lineno;
1237                 if (is_conflict_marker(line + 1, len - 1)) {
1238                         data->status |= 1;
1239                         fprintf(data->o->file,
1240                                 "%s:%d: leftover conflict marker\n",
1241                                 data->filename, data->lineno);
1242                 }
1243                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1244                 if (!bad)
1245                         return;
1246                 data->status |= bad;
1247                 err = whitespace_error_string(bad);
1248                 fprintf(data->o->file, "%s:%d: %s.\n",
1249                         data->filename, data->lineno, err);
1250                 free(err);
1251                 emit_line(data->o->file, set, reset, line, 1);
1252                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1253                               data->o->file, set, reset, ws);
1254         } else if (line[0] == ' ') {
1255                 data->lineno++;
1256                 data->trailing_blanks_start = 0;
1257         } else if (line[0] == '@') {
1258                 char *plus = strchr(line, '+');
1259                 if (plus)
1260                         data->lineno = strtol(plus, NULL, 10) - 1;
1261                 else
1262                         die("invalid diff");
1263                 data->trailing_blanks_start = 0;
1264         }
1265 }
1266
1267 static unsigned char *deflate_it(char *data,
1268                                  unsigned long size,
1269                                  unsigned long *result_size)
1270 {
1271         int bound;
1272         unsigned char *deflated;
1273         z_stream stream;
1274
1275         memset(&stream, 0, sizeof(stream));
1276         deflateInit(&stream, zlib_compression_level);
1277         bound = deflateBound(&stream, size);
1278         deflated = xmalloc(bound);
1279         stream.next_out = deflated;
1280         stream.avail_out = bound;
1281
1282         stream.next_in = (unsigned char *)data;
1283         stream.avail_in = size;
1284         while (deflate(&stream, Z_FINISH) == Z_OK)
1285                 ; /* nothing */
1286         deflateEnd(&stream);
1287         *result_size = stream.total_out;
1288         return deflated;
1289 }
1290
1291 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1292 {
1293         void *cp;
1294         void *delta;
1295         void *deflated;
1296         void *data;
1297         unsigned long orig_size;
1298         unsigned long delta_size;
1299         unsigned long deflate_size;
1300         unsigned long data_size;
1301
1302         /* We could do deflated delta, or we could do just deflated two,
1303          * whichever is smaller.
1304          */
1305         delta = NULL;
1306         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1307         if (one->size && two->size) {
1308                 delta = diff_delta(one->ptr, one->size,
1309                                    two->ptr, two->size,
1310                                    &delta_size, deflate_size);
1311                 if (delta) {
1312                         void *to_free = delta;
1313                         orig_size = delta_size;
1314                         delta = deflate_it(delta, delta_size, &delta_size);
1315                         free(to_free);
1316                 }
1317         }
1318
1319         if (delta && delta_size < deflate_size) {
1320                 fprintf(file, "delta %lu\n", orig_size);
1321                 free(deflated);
1322                 data = delta;
1323                 data_size = delta_size;
1324         }
1325         else {
1326                 fprintf(file, "literal %lu\n", two->size);
1327                 free(delta);
1328                 data = deflated;
1329                 data_size = deflate_size;
1330         }
1331
1332         /* emit data encoded in base85 */
1333         cp = data;
1334         while (data_size) {
1335                 int bytes = (52 < data_size) ? 52 : data_size;
1336                 char line[70];
1337                 data_size -= bytes;
1338                 if (bytes <= 26)
1339                         line[0] = bytes + 'A' - 1;
1340                 else
1341                         line[0] = bytes - 26 + 'a' - 1;
1342                 encode_85(line + 1, cp, bytes);
1343                 cp = (char *) cp + bytes;
1344                 fputs(line, file);
1345                 fputc('\n', file);
1346         }
1347         fprintf(file, "\n");
1348         free(data);
1349 }
1350
1351 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1352 {
1353         fprintf(file, "GIT binary patch\n");
1354         emit_binary_diff_body(file, one, two);
1355         emit_binary_diff_body(file, two, one);
1356 }
1357
1358 static void setup_diff_attr_check(struct git_attr_check *check)
1359 {
1360         static struct git_attr *attr_diff;
1361
1362         if (!attr_diff) {
1363                 attr_diff = git_attr("diff", 4);
1364         }
1365         check[0].attr = attr_diff;
1366 }
1367
1368 static void diff_filespec_check_attr(struct diff_filespec *one)
1369 {
1370         struct git_attr_check attr_diff_check;
1371         int check_from_data = 0;
1372
1373         if (one->checked_attr)
1374                 return;
1375
1376         setup_diff_attr_check(&attr_diff_check);
1377         one->is_binary = 0;
1378         one->funcname_pattern_ident = NULL;
1379
1380         if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1381                 const char *value;
1382
1383                 /* binaryness */
1384                 value = attr_diff_check.value;
1385                 if (ATTR_TRUE(value))
1386                         ;
1387                 else if (ATTR_FALSE(value))
1388                         one->is_binary = 1;
1389                 else
1390                         check_from_data = 1;
1391
1392                 /* funcname pattern ident */
1393                 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1394                         ;
1395                 else
1396                         one->funcname_pattern_ident = value;
1397         }
1398
1399         if (check_from_data) {
1400                 if (!one->data && DIFF_FILE_VALID(one))
1401                         diff_populate_filespec(one, 0);
1402
1403                 if (one->data)
1404                         one->is_binary = buffer_is_binary(one->data, one->size);
1405         }
1406 }
1407
1408 int diff_filespec_is_binary(struct diff_filespec *one)
1409 {
1410         diff_filespec_check_attr(one);
1411         return one->is_binary;
1412 }
1413
1414 static const struct funcname_pattern_entry *funcname_pattern(const char *ident)
1415 {
1416         struct funcname_pattern_list *pp;
1417
1418         for (pp = funcname_pattern_list; pp; pp = pp->next)
1419                 if (!strcmp(ident, pp->e.name))
1420                         return &pp->e;
1421         return NULL;
1422 }
1423
1424 static const struct funcname_pattern_entry builtin_funcname_pattern[] = {
1425         { "bibtex", "(@[a-zA-Z]{1,}[ \t]*\\{{0,1}[ \t]*[^ \t\"@',\\#}{~%]*).*$",
1426           REG_EXTENDED },
1427         { "html", "^[ \t]*(<[Hh][1-6][ \t].*>.*)$", REG_EXTENDED },
1428         { "java",
1429           "!^[ \t]*(catch|do|for|if|instanceof|new|return|switch|throw|while)\n"
1430           "^[ \t]*(([ \t]*[A-Za-z_][A-Za-z_0-9]*){2,}[ \t]*\\([^;]*)$",
1431           REG_EXTENDED },
1432         { "pascal",
1433           "^((procedure|function|constructor|destructor|interface|"
1434                 "implementation|initialization|finalization)[ \t]*.*)$"
1435           "\n"
1436           "^(.*=[ \t]*(class|record).*)$",
1437           REG_EXTENDED },
1438         { "php", "^[\t ]*((function|class).*)", REG_EXTENDED },
1439         { "python", "^[ \t]*((class|def)[ \t].*)$", REG_EXTENDED },
1440         { "ruby", "^[ \t]*((class|module|def)[ \t].*)$",
1441           REG_EXTENDED },
1442         { "tex",
1443           "^(\\\\((sub)*section|chapter|part)\\*{0,1}\\{.*)$",
1444           REG_EXTENDED },
1445 };
1446
1447 static const struct funcname_pattern_entry *diff_funcname_pattern(struct diff_filespec *one)
1448 {
1449         const char *ident;
1450         const struct funcname_pattern_entry *pe;
1451         int i;
1452
1453         diff_filespec_check_attr(one);
1454         ident = one->funcname_pattern_ident;
1455
1456         if (!ident)
1457                 /*
1458                  * If the config file has "funcname.default" defined, that
1459                  * regexp is used; otherwise NULL is returned and xemit uses
1460                  * the built-in default.
1461                  */
1462                 return funcname_pattern("default");
1463
1464         /* Look up custom "funcname.$ident" regexp from config. */
1465         pe = funcname_pattern(ident);
1466         if (pe)
1467                 return pe;
1468
1469         /*
1470          * And define built-in fallback patterns here.  Note that
1471          * these can be overridden by the user's config settings.
1472          */
1473         for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1474                 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1475                         return &builtin_funcname_pattern[i];
1476
1477         return NULL;
1478 }
1479
1480 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1481 {
1482         if (!options->a_prefix)
1483                 options->a_prefix = a;
1484         if (!options->b_prefix)
1485                 options->b_prefix = b;
1486 }
1487
1488 static void builtin_diff(const char *name_a,
1489                          const char *name_b,
1490                          struct diff_filespec *one,
1491                          struct diff_filespec *two,
1492                          const char *xfrm_msg,
1493                          struct diff_options *o,
1494                          int complete_rewrite)
1495 {
1496         mmfile_t mf1, mf2;
1497         const char *lbl[2];
1498         char *a_one, *b_two;
1499         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1500         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1501         const char *a_prefix, *b_prefix;
1502
1503         diff_set_mnemonic_prefix(o, "a/", "b/");
1504         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1505                 a_prefix = o->b_prefix;
1506                 b_prefix = o->a_prefix;
1507         } else {
1508                 a_prefix = o->a_prefix;
1509                 b_prefix = o->b_prefix;
1510         }
1511
1512         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1513         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1514         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1515         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1516         fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1517         if (lbl[0][0] == '/') {
1518                 /* /dev/null */
1519                 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1520                 if (xfrm_msg && xfrm_msg[0])
1521                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1522         }
1523         else if (lbl[1][0] == '/') {
1524                 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1525                 if (xfrm_msg && xfrm_msg[0])
1526                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1527         }
1528         else {
1529                 if (one->mode != two->mode) {
1530                         fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1531                         fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1532                 }
1533                 if (xfrm_msg && xfrm_msg[0])
1534                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1535                 /*
1536                  * we do not run diff between different kind
1537                  * of objects.
1538                  */
1539                 if ((one->mode ^ two->mode) & S_IFMT)
1540                         goto free_ab_and_return;
1541                 if (complete_rewrite) {
1542                         emit_rewrite_diff(name_a, name_b, one, two, o);
1543                         o->found_changes = 1;
1544                         goto free_ab_and_return;
1545                 }
1546         }
1547
1548         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1549                 die("unable to read files to diff");
1550
1551         if (!DIFF_OPT_TST(o, TEXT) &&
1552             (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1553                 /* Quite common confusing case */
1554                 if (mf1.size == mf2.size &&
1555                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1556                         goto free_ab_and_return;
1557                 if (DIFF_OPT_TST(o, BINARY))
1558                         emit_binary_diff(o->file, &mf1, &mf2);
1559                 else
1560                         fprintf(o->file, "Binary files %s and %s differ\n",
1561                                 lbl[0], lbl[1]);
1562                 o->found_changes = 1;
1563         }
1564         else {
1565                 /* Crazy xdl interfaces.. */
1566                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1567                 xpparam_t xpp;
1568                 xdemitconf_t xecfg;
1569                 xdemitcb_t ecb;
1570                 struct emit_callback ecbdata;
1571                 const struct funcname_pattern_entry *pe;
1572
1573                 pe = diff_funcname_pattern(one);
1574                 if (!pe)
1575                         pe = diff_funcname_pattern(two);
1576
1577                 memset(&xecfg, 0, sizeof(xecfg));
1578                 memset(&ecbdata, 0, sizeof(ecbdata));
1579                 ecbdata.label_path = lbl;
1580                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1581                 ecbdata.found_changesp = &o->found_changes;
1582                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1583                 ecbdata.file = o->file;
1584                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1585                 xecfg.ctxlen = o->context;
1586                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1587                 if (pe)
1588                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1589                 if (!diffopts)
1590                         ;
1591                 else if (!prefixcmp(diffopts, "--unified="))
1592                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1593                 else if (!prefixcmp(diffopts, "-u"))
1594                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1595                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1596                         ecbdata.diff_words =
1597                                 xcalloc(1, sizeof(struct diff_words_data));
1598                         ecbdata.diff_words->file = o->file;
1599                 }
1600                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1601                               &xpp, &xecfg, &ecb);
1602                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1603                         free_diff_words_data(&ecbdata);
1604         }
1605
1606  free_ab_and_return:
1607         diff_free_filespec_data(one);
1608         diff_free_filespec_data(two);
1609         free(a_one);
1610         free(b_two);
1611         return;
1612 }
1613
1614 static void builtin_diffstat(const char *name_a, const char *name_b,
1615                              struct diff_filespec *one,
1616                              struct diff_filespec *two,
1617                              struct diffstat_t *diffstat,
1618                              struct diff_options *o,
1619                              int complete_rewrite)
1620 {
1621         mmfile_t mf1, mf2;
1622         struct diffstat_file *data;
1623
1624         data = diffstat_add(diffstat, name_a, name_b);
1625
1626         if (!one || !two) {
1627                 data->is_unmerged = 1;
1628                 return;
1629         }
1630         if (complete_rewrite) {
1631                 diff_populate_filespec(one, 0);
1632                 diff_populate_filespec(two, 0);
1633                 data->deleted = count_lines(one->data, one->size);
1634                 data->added = count_lines(two->data, two->size);
1635                 goto free_and_return;
1636         }
1637         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1638                 die("unable to read files to diff");
1639
1640         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1641                 data->is_binary = 1;
1642                 data->added = mf2.size;
1643                 data->deleted = mf1.size;
1644         } else {
1645                 /* Crazy xdl interfaces.. */
1646                 xpparam_t xpp;
1647                 xdemitconf_t xecfg;
1648                 xdemitcb_t ecb;
1649
1650                 memset(&xecfg, 0, sizeof(xecfg));
1651                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1652                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1653                               &xpp, &xecfg, &ecb);
1654         }
1655
1656  free_and_return:
1657         diff_free_filespec_data(one);
1658         diff_free_filespec_data(two);
1659 }
1660
1661 static void builtin_checkdiff(const char *name_a, const char *name_b,
1662                               const char *attr_path,
1663                               struct diff_filespec *one,
1664                               struct diff_filespec *two,
1665                               struct diff_options *o)
1666 {
1667         mmfile_t mf1, mf2;
1668         struct checkdiff_t data;
1669
1670         if (!two)
1671                 return;
1672
1673         memset(&data, 0, sizeof(data));
1674         data.filename = name_b ? name_b : name_a;
1675         data.lineno = 0;
1676         data.o = o;
1677         data.ws_rule = whitespace_rule(attr_path);
1678
1679         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1680                 die("unable to read files to diff");
1681
1682         /*
1683          * All the other codepaths check both sides, but not checking
1684          * the "old" side here is deliberate.  We are checking the newly
1685          * introduced changes, and as long as the "new" side is text, we
1686          * can and should check what it introduces.
1687          */
1688         if (diff_filespec_is_binary(two))
1689                 goto free_and_return;
1690         else {
1691                 /* Crazy xdl interfaces.. */
1692                 xpparam_t xpp;
1693                 xdemitconf_t xecfg;
1694                 xdemitcb_t ecb;
1695
1696                 memset(&xecfg, 0, sizeof(xecfg));
1697                 xecfg.ctxlen = 1; /* at least one context line */
1698                 xpp.flags = XDF_NEED_MINIMAL;
1699                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1700                               &xpp, &xecfg, &ecb);
1701
1702                 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1703                     data.trailing_blanks_start) {
1704                         fprintf(o->file, "%s:%d: ends with blank lines.\n",
1705                                 data.filename, data.trailing_blanks_start);
1706                         data.status = 1; /* report errors */
1707                 }
1708         }
1709  free_and_return:
1710         diff_free_filespec_data(one);
1711         diff_free_filespec_data(two);
1712         if (data.status)
1713                 DIFF_OPT_SET(o, CHECK_FAILED);
1714 }
1715
1716 struct diff_filespec *alloc_filespec(const char *path)
1717 {
1718         int namelen = strlen(path);
1719         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1720
1721         memset(spec, 0, sizeof(*spec));
1722         spec->path = (char *)(spec + 1);
1723         memcpy(spec->path, path, namelen+1);
1724         spec->count = 1;
1725         return spec;
1726 }
1727
1728 void free_filespec(struct diff_filespec *spec)
1729 {
1730         if (!--spec->count) {
1731                 diff_free_filespec_data(spec);
1732                 free(spec);
1733         }
1734 }
1735
1736 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1737                    unsigned short mode)
1738 {
1739         if (mode) {
1740                 spec->mode = canon_mode(mode);
1741                 hashcpy(spec->sha1, sha1);
1742                 spec->sha1_valid = !is_null_sha1(sha1);
1743         }
1744 }
1745
1746 /*
1747  * Given a name and sha1 pair, if the index tells us the file in
1748  * the work tree has that object contents, return true, so that
1749  * prepare_temp_file() does not have to inflate and extract.
1750  */
1751 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1752 {
1753         struct cache_entry *ce;
1754         struct stat st;
1755         int pos, len;
1756
1757         /* We do not read the cache ourselves here, because the
1758          * benchmark with my previous version that always reads cache
1759          * shows that it makes things worse for diff-tree comparing
1760          * two linux-2.6 kernel trees in an already checked out work
1761          * tree.  This is because most diff-tree comparisons deal with
1762          * only a small number of files, while reading the cache is
1763          * expensive for a large project, and its cost outweighs the
1764          * savings we get by not inflating the object to a temporary
1765          * file.  Practically, this code only helps when we are used
1766          * by diff-cache --cached, which does read the cache before
1767          * calling us.
1768          */
1769         if (!active_cache)
1770                 return 0;
1771
1772         /* We want to avoid the working directory if our caller
1773          * doesn't need the data in a normal file, this system
1774          * is rather slow with its stat/open/mmap/close syscalls,
1775          * and the object is contained in a pack file.  The pack
1776          * is probably already open and will be faster to obtain
1777          * the data through than the working directory.  Loose
1778          * objects however would tend to be slower as they need
1779          * to be individually opened and inflated.
1780          */
1781         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1782                 return 0;
1783
1784         len = strlen(name);
1785         pos = cache_name_pos(name, len);
1786         if (pos < 0)
1787                 return 0;
1788         ce = active_cache[pos];
1789
1790         /*
1791          * This is not the sha1 we are looking for, or
1792          * unreusable because it is not a regular file.
1793          */
1794         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1795                 return 0;
1796
1797         /*
1798          * If ce matches the file in the work tree, we can reuse it.
1799          */
1800         if (ce_uptodate(ce) ||
1801             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1802                 return 1;
1803
1804         return 0;
1805 }
1806
1807 static int populate_from_stdin(struct diff_filespec *s)
1808 {
1809         struct strbuf buf;
1810         size_t size = 0;
1811
1812         strbuf_init(&buf, 0);
1813         if (strbuf_read(&buf, 0, 0) < 0)
1814                 return error("error while reading from stdin %s",
1815                                      strerror(errno));
1816
1817         s->should_munmap = 0;
1818         s->data = strbuf_detach(&buf, &size);
1819         s->size = size;
1820         s->should_free = 1;
1821         return 0;
1822 }
1823
1824 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1825 {
1826         int len;
1827         char *data = xmalloc(100);
1828         len = snprintf(data, 100,
1829                 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1830         s->data = data;
1831         s->size = len;
1832         s->should_free = 1;
1833         if (size_only) {
1834                 s->data = NULL;
1835                 free(data);
1836         }
1837         return 0;
1838 }
1839
1840 /*
1841  * While doing rename detection and pickaxe operation, we may need to
1842  * grab the data for the blob (or file) for our own in-core comparison.
1843  * diff_filespec has data and size fields for this purpose.
1844  */
1845 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1846 {
1847         int err = 0;
1848         if (!DIFF_FILE_VALID(s))
1849                 die("internal error: asking to populate invalid file.");
1850         if (S_ISDIR(s->mode))
1851                 return -1;
1852
1853         if (s->data)
1854                 return 0;
1855
1856         if (size_only && 0 < s->size)
1857                 return 0;
1858
1859         if (S_ISGITLINK(s->mode))
1860                 return diff_populate_gitlink(s, size_only);
1861
1862         if (!s->sha1_valid ||
1863             reuse_worktree_file(s->path, s->sha1, 0)) {
1864                 struct strbuf buf;
1865                 struct stat st;
1866                 int fd;
1867
1868                 if (!strcmp(s->path, "-"))
1869                         return populate_from_stdin(s);
1870
1871                 if (lstat(s->path, &st) < 0) {
1872                         if (errno == ENOENT) {
1873                         err_empty:
1874                                 err = -1;
1875                         empty:
1876                                 s->data = (char *)"";
1877                                 s->size = 0;
1878                                 return err;
1879                         }
1880                 }
1881                 s->size = xsize_t(st.st_size);
1882                 if (!s->size)
1883                         goto empty;
1884                 if (size_only)
1885                         return 0;
1886                 if (S_ISLNK(st.st_mode)) {
1887                         int ret;
1888                         s->data = xmalloc(s->size);
1889                         s->should_free = 1;
1890                         ret = readlink(s->path, s->data, s->size);
1891                         if (ret < 0) {
1892                                 free(s->data);
1893                                 goto err_empty;
1894                         }
1895                         return 0;
1896                 }
1897                 fd = open(s->path, O_RDONLY);
1898                 if (fd < 0)
1899                         goto err_empty;
1900                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1901                 close(fd);
1902                 s->should_munmap = 1;
1903
1904                 /*
1905                  * Convert from working tree format to canonical git format
1906                  */
1907                 strbuf_init(&buf, 0);
1908                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1909                         size_t size = 0;
1910                         munmap(s->data, s->size);
1911                         s->should_munmap = 0;
1912                         s->data = strbuf_detach(&buf, &size);
1913                         s->size = size;
1914                         s->should_free = 1;
1915                 }
1916         }
1917         else {
1918                 enum object_type type;
1919                 if (size_only)
1920                         type = sha1_object_info(s->sha1, &s->size);
1921                 else {
1922                         s->data = read_sha1_file(s->sha1, &type, &s->size);
1923                         s->should_free = 1;
1924                 }
1925         }
1926         return 0;
1927 }
1928
1929 void diff_free_filespec_blob(struct diff_filespec *s)
1930 {
1931         if (s->should_free)
1932                 free(s->data);
1933         else if (s->should_munmap)
1934                 munmap(s->data, s->size);
1935
1936         if (s->should_free || s->should_munmap) {
1937                 s->should_free = s->should_munmap = 0;
1938                 s->data = NULL;
1939         }
1940 }
1941
1942 void diff_free_filespec_data(struct diff_filespec *s)
1943 {
1944         diff_free_filespec_blob(s);
1945         free(s->cnt_data);
1946         s->cnt_data = NULL;
1947 }
1948
1949 static void prep_temp_blob(struct diff_tempfile *temp,
1950                            void *blob,
1951                            unsigned long size,
1952                            const unsigned char *sha1,
1953                            int mode)
1954 {
1955         int fd;
1956
1957         fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1958         if (fd < 0)
1959                 die("unable to create temp-file: %s", strerror(errno));
1960         if (write_in_full(fd, blob, size) != size)
1961                 die("unable to write temp-file");
1962         close(fd);
1963         temp->name = temp->tmp_path;
1964         strcpy(temp->hex, sha1_to_hex(sha1));
1965         temp->hex[40] = 0;
1966         sprintf(temp->mode, "%06o", mode);
1967 }
1968
1969 static void prepare_temp_file(const char *name,
1970                               struct diff_tempfile *temp,
1971                               struct diff_filespec *one)
1972 {
1973         if (!DIFF_FILE_VALID(one)) {
1974         not_a_valid_file:
1975                 /* A '-' entry produces this for file-2, and
1976                  * a '+' entry produces this for file-1.
1977                  */
1978                 temp->name = "/dev/null";
1979                 strcpy(temp->hex, ".");
1980                 strcpy(temp->mode, ".");
1981                 return;
1982         }
1983
1984         if (!one->sha1_valid ||
1985             reuse_worktree_file(name, one->sha1, 1)) {
1986                 struct stat st;
1987                 if (lstat(name, &st) < 0) {
1988                         if (errno == ENOENT)
1989                                 goto not_a_valid_file;
1990                         die("stat(%s): %s", name, strerror(errno));
1991                 }
1992                 if (S_ISLNK(st.st_mode)) {
1993                         int ret;
1994                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1995                         size_t sz = xsize_t(st.st_size);
1996                         if (sizeof(buf) <= st.st_size)
1997                                 die("symlink too long: %s", name);
1998                         ret = readlink(name, buf, sz);
1999                         if (ret < 0)
2000                                 die("readlink(%s)", name);
2001                         prep_temp_blob(temp, buf, sz,
2002                                        (one->sha1_valid ?
2003                                         one->sha1 : null_sha1),
2004                                        (one->sha1_valid ?
2005                                         one->mode : S_IFLNK));
2006                 }
2007                 else {
2008                         /* we can borrow from the file in the work tree */
2009                         temp->name = name;
2010                         if (!one->sha1_valid)
2011                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2012                         else
2013                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2014                         /* Even though we may sometimes borrow the
2015                          * contents from the work tree, we always want
2016                          * one->mode.  mode is trustworthy even when
2017                          * !(one->sha1_valid), as long as
2018                          * DIFF_FILE_VALID(one).
2019                          */
2020                         sprintf(temp->mode, "%06o", one->mode);
2021                 }
2022                 return;
2023         }
2024         else {
2025                 if (diff_populate_filespec(one, 0))
2026                         die("cannot read data blob for %s", one->path);
2027                 prep_temp_blob(temp, one->data, one->size,
2028                                one->sha1, one->mode);
2029         }
2030 }
2031
2032 static void remove_tempfile(void)
2033 {
2034         int i;
2035
2036         for (i = 0; i < 2; i++)
2037                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
2038                         unlink(diff_temp[i].name);
2039                         diff_temp[i].name = NULL;
2040                 }
2041 }
2042
2043 static void remove_tempfile_on_signal(int signo)
2044 {
2045         remove_tempfile();
2046         signal(SIGINT, SIG_DFL);
2047         raise(signo);
2048 }
2049
2050 /* An external diff command takes:
2051  *
2052  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2053  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2054  *
2055  */
2056 static void run_external_diff(const char *pgm,
2057                               const char *name,
2058                               const char *other,
2059                               struct diff_filespec *one,
2060                               struct diff_filespec *two,
2061                               const char *xfrm_msg,
2062                               int complete_rewrite)
2063 {
2064         const char *spawn_arg[10];
2065         struct diff_tempfile *temp = diff_temp;
2066         int retval;
2067         static int atexit_asked = 0;
2068         const char *othername;
2069         const char **arg = &spawn_arg[0];
2070
2071         othername = (other? other : name);
2072         if (one && two) {
2073                 prepare_temp_file(name, &temp[0], one);
2074                 prepare_temp_file(othername, &temp[1], two);
2075                 if (! atexit_asked &&
2076                     (temp[0].name == temp[0].tmp_path ||
2077                      temp[1].name == temp[1].tmp_path)) {
2078                         atexit_asked = 1;
2079                         atexit(remove_tempfile);
2080                 }
2081                 signal(SIGINT, remove_tempfile_on_signal);
2082         }
2083
2084         if (one && two) {
2085                 *arg++ = pgm;
2086                 *arg++ = name;
2087                 *arg++ = temp[0].name;
2088                 *arg++ = temp[0].hex;
2089                 *arg++ = temp[0].mode;
2090                 *arg++ = temp[1].name;
2091                 *arg++ = temp[1].hex;
2092                 *arg++ = temp[1].mode;
2093                 if (other) {
2094                         *arg++ = other;
2095                         *arg++ = xfrm_msg;
2096                 }
2097         } else {
2098                 *arg++ = pgm;
2099                 *arg++ = name;
2100         }
2101         *arg = NULL;
2102         fflush(NULL);
2103         retval = run_command_v_opt(spawn_arg, 0);
2104         remove_tempfile();
2105         if (retval) {
2106                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2107                 exit(1);
2108         }
2109 }
2110
2111 static const char *external_diff_attr(const char *name)
2112 {
2113         struct git_attr_check attr_diff_check;
2114
2115         if (!name)
2116                 return NULL;
2117
2118         setup_diff_attr_check(&attr_diff_check);
2119         if (!git_checkattr(name, 1, &attr_diff_check)) {
2120                 const char *value = attr_diff_check.value;
2121                 if (!ATTR_TRUE(value) &&
2122                     !ATTR_FALSE(value) &&
2123                     !ATTR_UNSET(value)) {
2124                         struct ll_diff_driver *drv;
2125
2126                         for (drv = user_diff; drv; drv = drv->next)
2127                                 if (!strcmp(drv->name, value))
2128                                         return drv->cmd;
2129                 }
2130         }
2131         return NULL;
2132 }
2133
2134 static void run_diff_cmd(const char *pgm,
2135                          const char *name,
2136                          const char *other,
2137                          const char *attr_path,
2138                          struct diff_filespec *one,
2139                          struct diff_filespec *two,
2140                          const char *xfrm_msg,
2141                          struct diff_options *o,
2142                          int complete_rewrite)
2143 {
2144         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2145                 pgm = NULL;
2146         else {
2147                 const char *cmd = external_diff_attr(attr_path);
2148                 if (cmd)
2149                         pgm = cmd;
2150         }
2151
2152         if (pgm) {
2153                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2154                                   complete_rewrite);
2155                 return;
2156         }
2157         if (one && two)
2158                 builtin_diff(name, other ? other : name,
2159                              one, two, xfrm_msg, o, complete_rewrite);
2160         else
2161                 fprintf(o->file, "* Unmerged path %s\n", name);
2162 }
2163
2164 static void diff_fill_sha1_info(struct diff_filespec *one)
2165 {
2166         if (DIFF_FILE_VALID(one)) {
2167                 if (!one->sha1_valid) {
2168                         struct stat st;
2169                         if (!strcmp(one->path, "-")) {
2170                                 hashcpy(one->sha1, null_sha1);
2171                                 return;
2172                         }
2173                         if (lstat(one->path, &st) < 0)
2174                                 die("stat %s", one->path);
2175                         if (index_path(one->sha1, one->path, &st, 0))
2176                                 die("cannot hash %s\n", one->path);
2177                 }
2178         }
2179         else
2180                 hashclr(one->sha1);
2181 }
2182
2183 static int similarity_index(struct diff_filepair *p)
2184 {
2185         return p->score * 100 / MAX_SCORE;
2186 }
2187
2188 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2189 {
2190         /* Strip the prefix but do not molest /dev/null and absolute paths */
2191         if (*namep && **namep != '/')
2192                 *namep += prefix_length;
2193         if (*otherp && **otherp != '/')
2194                 *otherp += prefix_length;
2195 }
2196
2197 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2198 {
2199         const char *pgm = external_diff();
2200         struct strbuf msg;
2201         char *xfrm_msg;
2202         struct diff_filespec *one = p->one;
2203         struct diff_filespec *two = p->two;
2204         const char *name;
2205         const char *other;
2206         const char *attr_path;
2207         int complete_rewrite = 0;
2208
2209         name  = p->one->path;
2210         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2211         attr_path = name;
2212         if (o->prefix_length)
2213                 strip_prefix(o->prefix_length, &name, &other);
2214
2215         if (DIFF_PAIR_UNMERGED(p)) {
2216                 run_diff_cmd(pgm, name, NULL, attr_path,
2217                              NULL, NULL, NULL, o, 0);
2218                 return;
2219         }
2220
2221         diff_fill_sha1_info(one);
2222         diff_fill_sha1_info(two);
2223
2224         strbuf_init(&msg, PATH_MAX * 2 + 300);
2225         switch (p->status) {
2226         case DIFF_STATUS_COPIED:
2227                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2228                 strbuf_addstr(&msg, "\ncopy from ");
2229                 quote_c_style(name, &msg, NULL, 0);
2230                 strbuf_addstr(&msg, "\ncopy to ");
2231                 quote_c_style(other, &msg, NULL, 0);
2232                 strbuf_addch(&msg, '\n');
2233                 break;
2234         case DIFF_STATUS_RENAMED:
2235                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2236                 strbuf_addstr(&msg, "\nrename from ");
2237                 quote_c_style(name, &msg, NULL, 0);
2238                 strbuf_addstr(&msg, "\nrename to ");
2239                 quote_c_style(other, &msg, NULL, 0);
2240                 strbuf_addch(&msg, '\n');
2241                 break;
2242         case DIFF_STATUS_MODIFIED:
2243                 if (p->score) {
2244                         strbuf_addf(&msg, "dissimilarity index %d%%\n",
2245                                         similarity_index(p));
2246                         complete_rewrite = 1;
2247                         break;
2248                 }
2249                 /* fallthru */
2250         default:
2251                 /* nothing */
2252                 ;
2253         }
2254
2255         if (hashcmp(one->sha1, two->sha1)) {
2256                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2257
2258                 if (DIFF_OPT_TST(o, BINARY)) {
2259                         mmfile_t mf;
2260                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2261                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2262                                 abbrev = 40;
2263                 }
2264                 strbuf_addf(&msg, "index %.*s..%.*s",
2265                                 abbrev, sha1_to_hex(one->sha1),
2266                                 abbrev, sha1_to_hex(two->sha1));
2267                 if (one->mode == two->mode)
2268                         strbuf_addf(&msg, " %06o", one->mode);
2269                 strbuf_addch(&msg, '\n');
2270         }
2271
2272         if (msg.len)
2273                 strbuf_setlen(&msg, msg.len - 1);
2274         xfrm_msg = msg.len ? msg.buf : NULL;
2275
2276         if (!pgm &&
2277             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2278             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2279                 /* a filepair that changes between file and symlink
2280                  * needs to be split into deletion and creation.
2281                  */
2282                 struct diff_filespec *null = alloc_filespec(two->path);
2283                 run_diff_cmd(NULL, name, other, attr_path,
2284                              one, null, xfrm_msg, o, 0);
2285                 free(null);
2286                 null = alloc_filespec(one->path);
2287                 run_diff_cmd(NULL, name, other, attr_path,
2288                              null, two, xfrm_msg, o, 0);
2289                 free(null);
2290         }
2291         else
2292                 run_diff_cmd(pgm, name, other, attr_path,
2293                              one, two, xfrm_msg, o, complete_rewrite);
2294
2295         strbuf_release(&msg);
2296 }
2297
2298 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2299                          struct diffstat_t *diffstat)
2300 {
2301         const char *name;
2302         const char *other;
2303         int complete_rewrite = 0;
2304
2305         if (DIFF_PAIR_UNMERGED(p)) {
2306                 /* unmerged */
2307                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2308                 return;
2309         }
2310
2311         name = p->one->path;
2312         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2313
2314         if (o->prefix_length)
2315                 strip_prefix(o->prefix_length, &name, &other);
2316
2317         diff_fill_sha1_info(p->one);
2318         diff_fill_sha1_info(p->two);
2319
2320         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2321                 complete_rewrite = 1;
2322         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2323 }
2324
2325 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2326 {
2327         const char *name;
2328         const char *other;
2329         const char *attr_path;
2330
2331         if (DIFF_PAIR_UNMERGED(p)) {
2332                 /* unmerged */
2333                 return;
2334         }
2335
2336         name = p->one->path;
2337         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2338         attr_path = other ? other : name;
2339
2340         if (o->prefix_length)
2341                 strip_prefix(o->prefix_length, &name, &other);
2342
2343         diff_fill_sha1_info(p->one);
2344         diff_fill_sha1_info(p->two);
2345
2346         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2347 }
2348
2349 void diff_setup(struct diff_options *options)
2350 {
2351         memset(options, 0, sizeof(*options));
2352
2353         options->file = stdout;
2354
2355         options->line_termination = '\n';
2356         options->break_opt = -1;
2357         options->rename_limit = -1;
2358         options->dirstat_percent = 3;
2359         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
2360         options->context = 3;
2361
2362         options->change = diff_change;
2363         options->add_remove = diff_addremove;
2364         if (diff_use_color_default > 0)
2365                 DIFF_OPT_SET(options, COLOR_DIFF);
2366         else
2367                 DIFF_OPT_CLR(options, COLOR_DIFF);
2368         options->detect_rename = diff_detect_rename_default;
2369
2370         if (!diff_mnemonic_prefix) {
2371                 options->a_prefix = "a/";
2372                 options->b_prefix = "b/";
2373         }
2374 }
2375
2376 int diff_setup_done(struct diff_options *options)
2377 {
2378         int count = 0;
2379
2380         if (options->output_format & DIFF_FORMAT_NAME)
2381                 count++;
2382         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2383                 count++;
2384         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2385                 count++;
2386         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2387                 count++;
2388         if (count > 1)
2389                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2390
2391         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2392                 options->detect_rename = DIFF_DETECT_COPY;
2393
2394         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2395                 options->prefix = NULL;
2396         if (options->prefix)
2397                 options->prefix_length = strlen(options->prefix);
2398         else
2399                 options->prefix_length = 0;
2400
2401         if (options->output_format & (DIFF_FORMAT_NAME |
2402                                       DIFF_FORMAT_NAME_STATUS |
2403                                       DIFF_FORMAT_CHECKDIFF |
2404                                       DIFF_FORMAT_NO_OUTPUT))
2405                 options->output_format &= ~(DIFF_FORMAT_RAW |
2406                                             DIFF_FORMAT_NUMSTAT |
2407                                             DIFF_FORMAT_DIFFSTAT |
2408                                             DIFF_FORMAT_SHORTSTAT |
2409                                             DIFF_FORMAT_DIRSTAT |
2410                                             DIFF_FORMAT_SUMMARY |
2411                                             DIFF_FORMAT_PATCH);
2412
2413         /*
2414          * These cases always need recursive; we do not drop caller-supplied
2415          * recursive bits for other formats here.
2416          */
2417         if (options->output_format & (DIFF_FORMAT_PATCH |
2418                                       DIFF_FORMAT_NUMSTAT |
2419                                       DIFF_FORMAT_DIFFSTAT |
2420                                       DIFF_FORMAT_SHORTSTAT |
2421                                       DIFF_FORMAT_DIRSTAT |
2422                                       DIFF_FORMAT_SUMMARY |
2423                                       DIFF_FORMAT_CHECKDIFF))
2424                 DIFF_OPT_SET(options, RECURSIVE);
2425         /*
2426          * Also pickaxe would not work very well if you do not say recursive
2427          */
2428         if (options->pickaxe)
2429                 DIFF_OPT_SET(options, RECURSIVE);
2430
2431         if (options->detect_rename && options->rename_limit < 0)
2432                 options->rename_limit = diff_rename_limit_default;
2433         if (options->setup & DIFF_SETUP_USE_CACHE) {
2434                 if (!active_cache)
2435                         /* read-cache does not die even when it fails
2436                          * so it is safe for us to do this here.  Also
2437                          * it does not smudge active_cache or active_nr
2438                          * when it fails, so we do not have to worry about
2439                          * cleaning it up ourselves either.
2440                          */
2441                         read_cache();
2442         }
2443         if (options->abbrev <= 0 || 40 < options->abbrev)
2444                 options->abbrev = 40; /* full */
2445
2446         /*
2447          * It does not make sense to show the first hit we happened
2448          * to have found.  It does not make sense not to return with
2449          * exit code in such a case either.
2450          */
2451         if (DIFF_OPT_TST(options, QUIET)) {
2452                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2453                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2454         }
2455
2456         return 0;
2457 }
2458
2459 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2460 {
2461         char c, *eq;
2462         int len;
2463
2464         if (*arg != '-')
2465                 return 0;
2466         c = *++arg;
2467         if (!c)
2468                 return 0;
2469         if (c == arg_short) {
2470                 c = *++arg;
2471                 if (!c)
2472                         return 1;
2473                 if (val && isdigit(c)) {
2474                         char *end;
2475                         int n = strtoul(arg, &end, 10);
2476                         if (*end)
2477                                 return 0;
2478                         *val = n;
2479                         return 1;
2480                 }
2481                 return 0;
2482         }
2483         if (c != '-')
2484                 return 0;
2485         arg++;
2486         eq = strchr(arg, '=');
2487         if (eq)
2488                 len = eq - arg;
2489         else
2490                 len = strlen(arg);
2491         if (!len || strncmp(arg, arg_long, len))
2492                 return 0;
2493         if (eq) {
2494                 int n;
2495                 char *end;
2496                 if (!isdigit(*++eq))
2497                         return 0;
2498                 n = strtoul(eq, &end, 10);
2499                 if (*end)
2500                         return 0;
2501                 *val = n;
2502         }
2503         return 1;
2504 }
2505
2506 static int diff_scoreopt_parse(const char *opt);
2507
2508 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2509 {
2510         const char *arg = av[0];
2511
2512         /* Output format options */
2513         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2514                 options->output_format |= DIFF_FORMAT_PATCH;
2515         else if (opt_arg(arg, 'U', "unified", &options->context))
2516                 options->output_format |= DIFF_FORMAT_PATCH;
2517         else if (!strcmp(arg, "--raw"))
2518                 options->output_format |= DIFF_FORMAT_RAW;
2519         else if (!strcmp(arg, "--patch-with-raw"))
2520                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2521         else if (!strcmp(arg, "--numstat"))
2522                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2523         else if (!strcmp(arg, "--shortstat"))
2524                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2525         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2526                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2527         else if (!strcmp(arg, "--cumulative")) {
2528                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2529                 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2530         } else if (opt_arg(arg, 0, "dirstat-by-file",
2531                            &options->dirstat_percent)) {
2532                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2533                 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
2534         }
2535         else if (!strcmp(arg, "--check"))
2536                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2537         else if (!strcmp(arg, "--summary"))
2538                 options->output_format |= DIFF_FORMAT_SUMMARY;
2539         else if (!strcmp(arg, "--patch-with-stat"))
2540                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2541         else if (!strcmp(arg, "--name-only"))
2542                 options->output_format |= DIFF_FORMAT_NAME;
2543         else if (!strcmp(arg, "--name-status"))
2544                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2545         else if (!strcmp(arg, "-s"))
2546                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2547         else if (!prefixcmp(arg, "--stat")) {
2548                 char *end;
2549                 int width = options->stat_width;
2550                 int name_width = options->stat_name_width;
2551                 arg += 6;
2552                 end = (char *)arg;
2553
2554                 switch (*arg) {
2555                 case '-':
2556                         if (!prefixcmp(arg, "-width="))
2557                                 width = strtoul(arg + 7, &end, 10);
2558                         else if (!prefixcmp(arg, "-name-width="))
2559                                 name_width = strtoul(arg + 12, &end, 10);
2560                         break;
2561                 case '=':
2562                         width = strtoul(arg+1, &end, 10);
2563                         if (*end == ',')
2564                                 name_width = strtoul(end+1, &end, 10);
2565                 }
2566
2567                 /* Important! This checks all the error cases! */
2568                 if (*end)
2569                         return 0;
2570                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2571                 options->stat_name_width = name_width;
2572                 options->stat_width = width;
2573         }
2574
2575         /* renames options */
2576         else if (!prefixcmp(arg, "-B")) {
2577                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2578                         return -1;
2579         }
2580         else if (!prefixcmp(arg, "-M")) {
2581                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2582                         return -1;
2583                 options->detect_rename = DIFF_DETECT_RENAME;
2584         }
2585         else if (!prefixcmp(arg, "-C")) {
2586                 if (options->detect_rename == DIFF_DETECT_COPY)
2587                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2588                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2589                         return -1;
2590                 options->detect_rename = DIFF_DETECT_COPY;
2591         }
2592         else if (!strcmp(arg, "--no-renames"))
2593                 options->detect_rename = 0;
2594         else if (!strcmp(arg, "--relative"))
2595                 DIFF_OPT_SET(options, RELATIVE_NAME);
2596         else if (!prefixcmp(arg, "--relative=")) {
2597                 DIFF_OPT_SET(options, RELATIVE_NAME);
2598                 options->prefix = arg + 11;
2599         }
2600
2601         /* xdiff options */
2602         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2603                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2604         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2605                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2606         else if (!strcmp(arg, "--ignore-space-at-eol"))
2607                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2608
2609         /* flags options */
2610         else if (!strcmp(arg, "--binary")) {
2611                 options->output_format |= DIFF_FORMAT_PATCH;
2612                 DIFF_OPT_SET(options, BINARY);
2613         }
2614         else if (!strcmp(arg, "--full-index"))
2615                 DIFF_OPT_SET(options, FULL_INDEX);
2616         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2617                 DIFF_OPT_SET(options, TEXT);
2618         else if (!strcmp(arg, "-R"))
2619                 DIFF_OPT_SET(options, REVERSE_DIFF);
2620         else if (!strcmp(arg, "--find-copies-harder"))
2621                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2622         else if (!strcmp(arg, "--follow"))
2623                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2624         else if (!strcmp(arg, "--color"))
2625                 DIFF_OPT_SET(options, COLOR_DIFF);
2626         else if (!strcmp(arg, "--no-color"))
2627                 DIFF_OPT_CLR(options, COLOR_DIFF);
2628         else if (!strcmp(arg, "--color-words"))
2629                 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2630         else if (!strcmp(arg, "--exit-code"))
2631                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2632         else if (!strcmp(arg, "--quiet"))
2633                 DIFF_OPT_SET(options, QUIET);
2634         else if (!strcmp(arg, "--ext-diff"))
2635                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2636         else if (!strcmp(arg, "--no-ext-diff"))
2637                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2638         else if (!strcmp(arg, "--ignore-submodules"))
2639                 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2640
2641         /* misc options */
2642         else if (!strcmp(arg, "-z"))
2643                 options->line_termination = 0;
2644         else if (!prefixcmp(arg, "-l"))
2645                 options->rename_limit = strtoul(arg+2, NULL, 10);
2646         else if (!prefixcmp(arg, "-S"))
2647                 options->pickaxe = arg + 2;
2648         else if (!strcmp(arg, "--pickaxe-all"))
2649                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2650         else if (!strcmp(arg, "--pickaxe-regex"))
2651                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2652         else if (!prefixcmp(arg, "-O"))
2653                 options->orderfile = arg + 2;
2654         else if (!prefixcmp(arg, "--diff-filter="))
2655                 options->filter = arg + 14;
2656         else if (!strcmp(arg, "--abbrev"))
2657                 options->abbrev = DEFAULT_ABBREV;
2658         else if (!prefixcmp(arg, "--abbrev=")) {
2659                 options->abbrev = strtoul(arg + 9, NULL, 10);
2660                 if (options->abbrev < MINIMUM_ABBREV)
2661                         options->abbrev = MINIMUM_ABBREV;
2662                 else if (40 < options->abbrev)
2663                         options->abbrev = 40;
2664         }
2665         else if (!prefixcmp(arg, "--src-prefix="))
2666                 options->a_prefix = arg + 13;
2667         else if (!prefixcmp(arg, "--dst-prefix="))
2668                 options->b_prefix = arg + 13;
2669         else if (!strcmp(arg, "--no-prefix"))
2670                 options->a_prefix = options->b_prefix = "";
2671         else if (!prefixcmp(arg, "--output=")) {
2672                 options->file = fopen(arg + strlen("--output="), "w");
2673                 options->close_file = 1;
2674         } else
2675                 return 0;
2676         return 1;
2677 }
2678
2679 static int parse_num(const char **cp_p)
2680 {
2681         unsigned long num, scale;
2682         int ch, dot;
2683         const char *cp = *cp_p;
2684
2685         num = 0;
2686         scale = 1;
2687         dot = 0;
2688         for(;;) {
2689                 ch = *cp;
2690                 if ( !dot && ch == '.' ) {
2691                         scale = 1;
2692                         dot = 1;
2693                 } else if ( ch == '%' ) {
2694                         scale = dot ? scale*100 : 100;
2695                         cp++;   /* % is always at the end */
2696                         break;
2697                 } else if ( ch >= '0' && ch <= '9' ) {
2698                         if ( scale < 100000 ) {
2699                                 scale *= 10;
2700                                 num = (num*10) + (ch-'0');
2701                         }
2702                 } else {
2703                         break;
2704                 }
2705                 cp++;
2706         }
2707         *cp_p = cp;
2708
2709         /* user says num divided by scale and we say internally that
2710          * is MAX_SCORE * num / scale.
2711          */
2712         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2713 }
2714
2715 static int diff_scoreopt_parse(const char *opt)
2716 {
2717         int opt1, opt2, cmd;
2718
2719         if (*opt++ != '-')
2720                 return -1;
2721         cmd = *opt++;
2722         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2723                 return -1; /* that is not a -M, -C nor -B option */
2724
2725         opt1 = parse_num(&opt);
2726         if (cmd != 'B')
2727                 opt2 = 0;
2728         else {
2729                 if (*opt == 0)
2730                         opt2 = 0;
2731                 else if (*opt != '/')
2732                         return -1; /* we expect -B80/99 or -B80 */
2733                 else {
2734                         opt++;
2735                         opt2 = parse_num(&opt);
2736                 }
2737         }
2738         if (*opt != 0)
2739                 return -1;
2740         return opt1 | (opt2 << 16);
2741 }
2742
2743 struct diff_queue_struct diff_queued_diff;
2744
2745 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2746 {
2747         if (queue->alloc <= queue->nr) {
2748                 queue->alloc = alloc_nr(queue->alloc);
2749                 queue->queue = xrealloc(queue->queue,
2750                                         sizeof(dp) * queue->alloc);
2751         }
2752         queue->queue[queue->nr++] = dp;
2753 }
2754
2755 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2756                                  struct diff_filespec *one,
2757                                  struct diff_filespec *two)
2758 {
2759         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2760         dp->one = one;
2761         dp->two = two;
2762         if (queue)
2763                 diff_q(queue, dp);
2764         return dp;
2765 }
2766
2767 void diff_free_filepair(struct diff_filepair *p)
2768 {
2769         free_filespec(p->one);
2770         free_filespec(p->two);
2771         free(p);
2772 }
2773
2774 /* This is different from find_unique_abbrev() in that
2775  * it stuffs the result with dots for alignment.
2776  */
2777 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2778 {
2779         int abblen;
2780         const char *abbrev;
2781         if (len == 40)
2782                 return sha1_to_hex(sha1);
2783
2784         abbrev = find_unique_abbrev(sha1, len);
2785         abblen = strlen(abbrev);
2786         if (abblen < 37) {
2787                 static char hex[41];
2788                 if (len < abblen && abblen <= len + 2)
2789                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2790                 else
2791                         sprintf(hex, "%s...", abbrev);
2792                 return hex;
2793         }
2794         return sha1_to_hex(sha1);
2795 }
2796
2797 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2798 {
2799         int line_termination = opt->line_termination;
2800         int inter_name_termination = line_termination ? '\t' : '\0';
2801
2802         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2803                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2804                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
2805                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2806         }
2807         if (p->score) {
2808                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2809                         inter_name_termination);
2810         } else {
2811                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2812         }
2813
2814         if (p->status == DIFF_STATUS_COPIED ||
2815             p->status == DIFF_STATUS_RENAMED) {
2816                 const char *name_a, *name_b;
2817                 name_a = p->one->path;
2818                 name_b = p->two->path;
2819                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2820                 write_name_quoted(name_a, opt->file, inter_name_termination);
2821                 write_name_quoted(name_b, opt->file, line_termination);
2822         } else {
2823                 const char *name_a, *name_b;
2824                 name_a = p->one->mode ? p->one->path : p->two->path;
2825                 name_b = NULL;
2826                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2827                 write_name_quoted(name_a, opt->file, line_termination);
2828         }
2829 }
2830
2831 int diff_unmodified_pair(struct diff_filepair *p)
2832 {
2833         /* This function is written stricter than necessary to support
2834          * the currently implemented transformers, but the idea is to
2835          * let transformers to produce diff_filepairs any way they want,
2836          * and filter and clean them up here before producing the output.
2837          */
2838         struct diff_filespec *one = p->one, *two = p->two;
2839
2840         if (DIFF_PAIR_UNMERGED(p))
2841                 return 0; /* unmerged is interesting */
2842
2843         /* deletion, addition, mode or type change
2844          * and rename are all interesting.
2845          */
2846         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2847             DIFF_PAIR_MODE_CHANGED(p) ||
2848             strcmp(one->path, two->path))
2849                 return 0;
2850
2851         /* both are valid and point at the same path.  that is, we are
2852          * dealing with a change.
2853          */
2854         if (one->sha1_valid && two->sha1_valid &&
2855             !hashcmp(one->sha1, two->sha1))
2856                 return 1; /* no change */
2857         if (!one->sha1_valid && !two->sha1_valid)
2858                 return 1; /* both look at the same file on the filesystem. */
2859         return 0;
2860 }
2861
2862 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2863 {
2864         if (diff_unmodified_pair(p))
2865                 return;
2866
2867         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2868             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2869                 return; /* no tree diffs in patch format */
2870
2871         run_diff(p, o);
2872 }
2873
2874 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2875                             struct diffstat_t *diffstat)
2876 {
2877         if (diff_unmodified_pair(p))
2878                 return;
2879
2880         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2881             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2882                 return; /* no tree diffs in patch format */
2883
2884         run_diffstat(p, o, diffstat);
2885 }
2886
2887 static void diff_flush_checkdiff(struct diff_filepair *p,
2888                 struct diff_options *o)
2889 {
2890         if (diff_unmodified_pair(p))
2891                 return;
2892
2893         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2894             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2895                 return; /* no tree diffs in patch format */
2896
2897         run_checkdiff(p, o);
2898 }
2899
2900 int diff_queue_is_empty(void)
2901 {
2902         struct diff_queue_struct *q = &diff_queued_diff;
2903         int i;
2904         for (i = 0; i < q->nr; i++)
2905                 if (!diff_unmodified_pair(q->queue[i]))
2906                         return 0;
2907         return 1;
2908 }
2909
2910 #if DIFF_DEBUG
2911 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2912 {
2913         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2914                 x, one ? one : "",
2915                 s->path,
2916                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2917                 s->mode,
2918                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2919         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2920                 x, one ? one : "",
2921                 s->size, s->xfrm_flags);
2922 }
2923
2924 void diff_debug_filepair(const struct diff_filepair *p, int i)
2925 {
2926         diff_debug_filespec(p->one, i, "one");
2927         diff_debug_filespec(p->two, i, "two");
2928         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2929                 p->score, p->status ? p->status : '?',
2930                 p->one->rename_used, p->broken_pair);
2931 }
2932
2933 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2934 {
2935         int i;
2936         if (msg)
2937                 fprintf(stderr, "%s\n", msg);
2938         fprintf(stderr, "q->nr = %d\n", q->nr);
2939         for (i = 0; i < q->nr; i++) {
2940                 struct diff_filepair *p = q->queue[i];
2941                 diff_debug_filepair(p, i);
2942         }
2943 }
2944 #endif
2945
2946 static void diff_resolve_rename_copy(void)
2947 {
2948         int i;
2949         struct diff_filepair *p;
2950         struct diff_queue_struct *q = &diff_queued_diff;
2951
2952         diff_debug_queue("resolve-rename-copy", q);
2953
2954         for (i = 0; i < q->nr; i++) {
2955                 p = q->queue[i];
2956                 p->status = 0; /* undecided */
2957                 if (DIFF_PAIR_UNMERGED(p))
2958                         p->status = DIFF_STATUS_UNMERGED;
2959                 else if (!DIFF_FILE_VALID(p->one))
2960                         p->status = DIFF_STATUS_ADDED;
2961                 else if (!DIFF_FILE_VALID(p->two))
2962                         p->status = DIFF_STATUS_DELETED;
2963                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2964                         p->status = DIFF_STATUS_TYPE_CHANGED;
2965
2966                 /* from this point on, we are dealing with a pair
2967                  * whose both sides are valid and of the same type, i.e.
2968                  * either in-place edit or rename/copy edit.
2969                  */
2970                 else if (DIFF_PAIR_RENAME(p)) {
2971                         /*
2972                          * A rename might have re-connected a broken
2973                          * pair up, causing the pathnames to be the
2974                          * same again. If so, that's not a rename at
2975                          * all, just a modification..
2976                          *
2977                          * Otherwise, see if this source was used for
2978                          * multiple renames, in which case we decrement
2979                          * the count, and call it a copy.
2980                          */
2981                         if (!strcmp(p->one->path, p->two->path))
2982                                 p->status = DIFF_STATUS_MODIFIED;
2983                         else if (--p->one->rename_used > 0)
2984                                 p->status = DIFF_STATUS_COPIED;
2985                         else
2986                                 p->status = DIFF_STATUS_RENAMED;
2987                 }
2988                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2989                          p->one->mode != p->two->mode ||
2990                          is_null_sha1(p->one->sha1))
2991                         p->status = DIFF_STATUS_MODIFIED;
2992                 else {
2993                         /* This is a "no-change" entry and should not
2994                          * happen anymore, but prepare for broken callers.
2995                          */
2996                         error("feeding unmodified %s to diffcore",
2997                               p->one->path);
2998                         p->status = DIFF_STATUS_UNKNOWN;
2999                 }
3000         }
3001         diff_debug_queue("resolve-rename-copy done", q);
3002 }
3003
3004 static int check_pair_status(struct diff_filepair *p)
3005 {
3006         switch (p->status) {
3007         case DIFF_STATUS_UNKNOWN:
3008                 return 0;
3009         case 0:
3010                 die("internal error in diff-resolve-rename-copy");
3011         default:
3012                 return 1;
3013         }
3014 }
3015
3016 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3017 {
3018         int fmt = opt->output_format;
3019
3020         if (fmt & DIFF_FORMAT_CHECKDIFF)
3021                 diff_flush_checkdiff(p, opt);
3022         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3023                 diff_flush_raw(p, opt);
3024         else if (fmt & DIFF_FORMAT_NAME) {
3025                 const char *name_a, *name_b;
3026                 name_a = p->two->path;
3027                 name_b = NULL;
3028                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3029                 write_name_quoted(name_a, opt->file, opt->line_termination);
3030         }
3031 }
3032
3033 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3034 {
3035         if (fs->mode)
3036                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3037         else
3038                 fprintf(file, " %s ", newdelete);
3039         write_name_quoted(fs->path, file, '\n');
3040 }
3041
3042
3043 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
3044 {
3045         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3046                 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
3047                         show_name ? ' ' : '\n');
3048                 if (show_name) {
3049                         write_name_quoted(p->two->path, file, '\n');
3050                 }
3051         }
3052 }
3053
3054 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3055 {
3056         char *names = pprint_rename(p->one->path, p->two->path);
3057
3058         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3059         free(names);
3060         show_mode_change(file, p, 0);
3061 }
3062
3063 static void diff_summary(FILE *file, struct diff_filepair *p)
3064 {
3065         switch(p->status) {
3066         case DIFF_STATUS_DELETED:
3067                 show_file_mode_name(file, "delete", p->one);
3068                 break;
3069         case DIFF_STATUS_ADDED:
3070                 show_file_mode_name(file, "create", p->two);
3071                 break;
3072         case DIFF_STATUS_COPIED:
3073                 show_rename_copy(file, "copy", p);
3074                 break;
3075         case DIFF_STATUS_RENAMED:
3076                 show_rename_copy(file, "rename", p);
3077                 break;
3078         default:
3079                 if (p->score) {
3080                         fputs(" rewrite ", file);
3081                         write_name_quoted(p->two->path, file, ' ');
3082                         fprintf(file, "(%d%%)\n", similarity_index(p));
3083                 }
3084                 show_mode_change(file, p, !p->score);
3085                 break;
3086         }
3087 }
3088
3089 struct patch_id_t {
3090         git_SHA_CTX *ctx;
3091         int patchlen;
3092 };
3093
3094 static int remove_space(char *line, int len)
3095 {
3096         int i;
3097         char *dst = line;
3098         unsigned char c;
3099
3100         for (i = 0; i < len; i++)
3101                 if (!isspace((c = line[i])))
3102                         *dst++ = c;
3103
3104         return dst - line;
3105 }
3106
3107 static void patch_id_consume(void *priv, char *line, unsigned long len)
3108 {
3109         struct patch_id_t *data = priv;
3110         int new_len;
3111
3112         /* Ignore line numbers when computing the SHA1 of the patch */
3113         if (!prefixcmp(line, "@@ -"))
3114                 return;
3115
3116         new_len = remove_space(line, len);
3117
3118         git_SHA1_Update(data->ctx, line, new_len);
3119         data->patchlen += new_len;
3120 }
3121
3122 /* returns 0 upon success, and writes result into sha1 */
3123 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3124 {
3125         struct diff_queue_struct *q = &diff_queued_diff;
3126         int i;
3127         git_SHA_CTX ctx;
3128         struct patch_id_t data;
3129         char buffer[PATH_MAX * 4 + 20];
3130
3131         git_SHA1_Init(&ctx);
3132         memset(&data, 0, sizeof(struct patch_id_t));
3133         data.ctx = &ctx;
3134
3135         for (i = 0; i < q->nr; i++) {
3136                 xpparam_t xpp;
3137                 xdemitconf_t xecfg;
3138                 xdemitcb_t ecb;
3139                 mmfile_t mf1, mf2;
3140                 struct diff_filepair *p = q->queue[i];
3141                 int len1, len2;
3142
3143                 memset(&xecfg, 0, sizeof(xecfg));
3144                 if (p->status == 0)
3145                         return error("internal diff status error");
3146                 if (p->status == DIFF_STATUS_UNKNOWN)
3147                         continue;
3148                 if (diff_unmodified_pair(p))
3149                         continue;
3150                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3151                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3152                         continue;
3153                 if (DIFF_PAIR_UNMERGED(p))
3154                         continue;
3155
3156                 diff_fill_sha1_info(p->one);
3157                 diff_fill_sha1_info(p->two);
3158                 if (fill_mmfile(&mf1, p->one) < 0 ||
3159                                 fill_mmfile(&mf2, p->two) < 0)
3160                         return error("unable to read files to diff");
3161
3162                 len1 = remove_space(p->one->path, strlen(p->one->path));
3163                 len2 = remove_space(p->two->path, strlen(p->two->path));
3164                 if (p->one->mode == 0)
3165                         len1 = snprintf(buffer, sizeof(buffer),
3166                                         "diff--gita/%.*sb/%.*s"
3167                                         "newfilemode%06o"
3168                                         "---/dev/null"
3169                                         "+++b/%.*s",
3170                                         len1, p->one->path,
3171                                         len2, p->two->path,
3172                                         p->two->mode,
3173                                         len2, p->two->path);
3174                 else if (p->two->mode == 0)
3175                         len1 = snprintf(buffer, sizeof(buffer),
3176                                         "diff--gita/%.*sb/%.*s"
3177                                         "deletedfilemode%06o"
3178                                         "---a/%.*s"
3179                                         "+++/dev/null",
3180                                         len1, p->one->path,
3181                                         len2, p->two->path,
3182                                         p->one->mode,
3183                                         len1, p->one->path);
3184                 else
3185                         len1 = snprintf(buffer, sizeof(buffer),
3186                                         "diff--gita/%.*sb/%.*s"
3187                                         "---a/%.*s"
3188                                         "+++b/%.*s",
3189                                         len1, p->one->path,
3190                                         len2, p->two->path,
3191                                         len1, p->one->path,
3192                                         len2, p->two->path);
3193                 git_SHA1_Update(&ctx, buffer, len1);
3194
3195                 xpp.flags = XDF_NEED_MINIMAL;
3196                 xecfg.ctxlen = 3;
3197                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3198                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3199                               &xpp, &xecfg, &ecb);
3200         }
3201
3202         git_SHA1_Final(sha1, &ctx);
3203         return 0;
3204 }
3205
3206 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3207 {
3208         struct diff_queue_struct *q = &diff_queued_diff;
3209         int i;
3210         int result = diff_get_patch_id(options, sha1);
3211
3212         for (i = 0; i < q->nr; i++)
3213                 diff_free_filepair(q->queue[i]);
3214
3215         free(q->queue);
3216         q->queue = NULL;
3217         q->nr = q->alloc = 0;
3218
3219         return result;
3220 }
3221
3222 static int is_summary_empty(const struct diff_queue_struct *q)
3223 {
3224         int i;
3225
3226         for (i = 0; i < q->nr; i++) {
3227                 const struct diff_filepair *p = q->queue[i];
3228
3229                 switch (p->status) {
3230                 case DIFF_STATUS_DELETED:
3231                 case DIFF_STATUS_ADDED:
3232                 case DIFF_STATUS_COPIED:
3233                 case DIFF_STATUS_RENAMED:
3234                         return 0;
3235                 default:
3236                         if (p->score)
3237                                 return 0;
3238                         if (p->one->mode && p->two->mode &&
3239                             p->one->mode != p->two->mode)
3240                                 return 0;
3241                         break;
3242                 }
3243         }
3244         return 1;
3245 }
3246
3247 void diff_flush(struct diff_options *options)
3248 {
3249         struct diff_queue_struct *q = &diff_queued_diff;
3250         int i, output_format = options->output_format;
3251         int separator = 0;
3252
3253         /*
3254          * Order: raw, stat, summary, patch
3255          * or:    name/name-status/checkdiff (other bits clear)
3256          */
3257         if (!q->nr)
3258                 goto free_queue;
3259
3260         if (output_format & (DIFF_FORMAT_RAW |
3261                              DIFF_FORMAT_NAME |
3262                              DIFF_FORMAT_NAME_STATUS |
3263                              DIFF_FORMAT_CHECKDIFF)) {
3264                 for (i = 0; i < q->nr; i++) {
3265                         struct diff_filepair *p = q->queue[i];
3266                         if (check_pair_status(p))
3267                                 flush_one_pair(p, options);
3268                 }
3269                 separator++;
3270         }
3271
3272         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3273                 struct diffstat_t diffstat;
3274
3275                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3276                 for (i = 0; i < q->nr; i++) {
3277                         struct diff_filepair *p = q->queue[i];
3278                         if (check_pair_status(p))
3279                                 diff_flush_stat(p, options, &diffstat);
3280                 }
3281                 if (output_format & DIFF_FORMAT_NUMSTAT)
3282                         show_numstat(&diffstat, options);
3283                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3284                         show_stats(&diffstat, options);
3285                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3286                         show_shortstats(&diffstat, options);
3287                 free_diffstat_info(&diffstat);
3288                 separator++;
3289         }
3290         if (output_format & DIFF_FORMAT_DIRSTAT)
3291                 show_dirstat(options);
3292
3293         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3294                 for (i = 0; i < q->nr; i++)
3295                         diff_summary(options->file, q->queue[i]);
3296                 separator++;
3297         }
3298
3299         if (output_format & DIFF_FORMAT_PATCH) {
3300                 if (separator) {
3301                         putc(options->line_termination, options->file);
3302                         if (options->stat_sep) {
3303                                 /* attach patch instead of inline */
3304                                 fputs(options->stat_sep, options->file);
3305                         }
3306                 }
3307
3308                 for (i = 0; i < q->nr; i++) {
3309                         struct diff_filepair *p = q->queue[i];
3310                         if (check_pair_status(p))
3311                                 diff_flush_patch(p, options);
3312                 }
3313         }
3314
3315         if (output_format & DIFF_FORMAT_CALLBACK)
3316                 options->format_callback(q, options, options->format_callback_data);
3317
3318         for (i = 0; i < q->nr; i++)
3319                 diff_free_filepair(q->queue[i]);
3320 free_queue:
3321         free(q->queue);
3322         q->queue = NULL;
3323         q->nr = q->alloc = 0;
3324         if (options->close_file)
3325                 fclose(options->file);
3326 }
3327
3328 static void diffcore_apply_filter(const char *filter)
3329 {
3330         int i;
3331         struct diff_queue_struct *q = &diff_queued_diff;
3332         struct diff_queue_struct outq;
3333         outq.queue = NULL;
3334         outq.nr = outq.alloc = 0;
3335
3336         if (!filter)
3337                 return;
3338
3339         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3340                 int found;
3341                 for (i = found = 0; !found && i < q->nr; i++) {
3342                         struct diff_filepair *p = q->queue[i];
3343                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3344                              ((p->score &&
3345                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3346                               (!p->score &&
3347                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3348                             ((p->status != DIFF_STATUS_MODIFIED) &&
3349                              strchr(filter, p->status)))
3350                                 found++;
3351                 }
3352                 if (found)
3353                         return;
3354
3355                 /* otherwise we will clear the whole queue
3356                  * by copying the empty outq at the end of this
3357                  * function, but first clear the current entries
3358                  * in the queue.
3359                  */
3360                 for (i = 0; i < q->nr; i++)
3361                         diff_free_filepair(q->queue[i]);
3362         }
3363         else {
3364                 /* Only the matching ones */
3365                 for (i = 0; i < q->nr; i++) {
3366                         struct diff_filepair *p = q->queue[i];
3367
3368                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3369                              ((p->score &&
3370                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3371                               (!p->score &&
3372                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3373                             ((p->status != DIFF_STATUS_MODIFIED) &&
3374                              strchr(filter, p->status)))
3375                                 diff_q(&outq, p);
3376                         else
3377                                 diff_free_filepair(p);
3378                 }
3379         }
3380         free(q->queue);
3381         *q = outq;
3382 }
3383
3384 /* Check whether two filespecs with the same mode and size are identical */
3385 static int diff_filespec_is_identical(struct diff_filespec *one,
3386                                       struct diff_filespec *two)
3387 {
3388         if (S_ISGITLINK(one->mode))
3389                 return 0;
3390         if (diff_populate_filespec(one, 0))
3391                 return 0;
3392         if (diff_populate_filespec(two, 0))
3393                 return 0;
3394         return !memcmp(one->data, two->data, one->size);
3395 }
3396
3397 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3398 {
3399         int i;
3400         struct diff_queue_struct *q = &diff_queued_diff;
3401         struct diff_queue_struct outq;
3402         outq.queue = NULL;
3403         outq.nr = outq.alloc = 0;
3404
3405         for (i = 0; i < q->nr; i++) {
3406                 struct diff_filepair *p = q->queue[i];
3407
3408                 /*
3409                  * 1. Entries that come from stat info dirtyness
3410                  *    always have both sides (iow, not create/delete),
3411                  *    one side of the object name is unknown, with
3412                  *    the same mode and size.  Keep the ones that
3413                  *    do not match these criteria.  They have real
3414                  *    differences.
3415                  *
3416                  * 2. At this point, the file is known to be modified,
3417                  *    with the same mode and size, and the object
3418                  *    name of one side is unknown.  Need to inspect
3419                  *    the identical contents.
3420                  */
3421                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3422                     !DIFF_FILE_VALID(p->two) ||
3423                     (p->one->sha1_valid && p->two->sha1_valid) ||
3424                     (p->one->mode != p->two->mode) ||
3425                     diff_populate_filespec(p->one, 1) ||
3426                     diff_populate_filespec(p->two, 1) ||
3427                     (p->one->size != p->two->size) ||
3428                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3429                         diff_q(&outq, p);
3430                 else {
3431                         /*
3432                          * The caller can subtract 1 from skip_stat_unmatch
3433                          * to determine how many paths were dirty only
3434                          * due to stat info mismatch.
3435                          */
3436                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3437                                 diffopt->skip_stat_unmatch++;
3438                         diff_free_filepair(p);
3439                 }
3440         }
3441         free(q->queue);
3442         *q = outq;
3443 }
3444
3445 void diffcore_std(struct diff_options *options)
3446 {
3447         if (options->skip_stat_unmatch)
3448                 diffcore_skip_stat_unmatch(options);
3449         if (options->break_opt != -1)
3450                 diffcore_break(options->break_opt);
3451         if (options->detect_rename)
3452                 diffcore_rename(options);
3453         if (options->break_opt != -1)
3454                 diffcore_merge_broken();
3455         if (options->pickaxe)
3456                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3457         if (options->orderfile)
3458                 diffcore_order(options->orderfile);
3459         diff_resolve_rename_copy();
3460         diffcore_apply_filter(options->filter);
3461
3462         if (diff_queued_diff.nr)
3463                 DIFF_OPT_SET(options, HAS_CHANGES);
3464         else
3465                 DIFF_OPT_CLR(options, HAS_CHANGES);
3466 }
3467
3468 int diff_result_code(struct diff_options *opt, int status)
3469 {
3470         int result = 0;
3471         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3472             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3473                 return status;
3474         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3475             DIFF_OPT_TST(opt, HAS_CHANGES))
3476                 result |= 01;
3477         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3478             DIFF_OPT_TST(opt, CHECK_FAILED))
3479                 result |= 02;
3480         return result;
3481 }
3482
3483 void diff_addremove(struct diff_options *options,
3484                     int addremove, unsigned mode,
3485                     const unsigned char *sha1,
3486                     const char *concatpath)
3487 {
3488         struct diff_filespec *one, *two;
3489
3490         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3491                 return;
3492
3493         /* This may look odd, but it is a preparation for
3494          * feeding "there are unchanged files which should
3495          * not produce diffs, but when you are doing copy
3496          * detection you would need them, so here they are"
3497          * entries to the diff-core.  They will be prefixed
3498          * with something like '=' or '*' (I haven't decided
3499          * which but should not make any difference).
3500          * Feeding the same new and old to diff_change()
3501          * also has the same effect.
3502          * Before the final output happens, they are pruned after
3503          * merged into rename/copy pairs as appropriate.
3504          */
3505         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3506                 addremove = (addremove == '+' ? '-' :
3507                              addremove == '-' ? '+' : addremove);
3508
3509         if (options->prefix &&
3510             strncmp(concatpath, options->prefix, options->prefix_length))
3511                 return;
3512
3513         one = alloc_filespec(concatpath);
3514         two = alloc_filespec(concatpath);
3515
3516         if (addremove != '+')
3517                 fill_filespec(one, sha1, mode);
3518         if (addremove != '-')
3519                 fill_filespec(two, sha1, mode);
3520
3521         diff_queue(&diff_queued_diff, one, two);
3522         DIFF_OPT_SET(options, HAS_CHANGES);
3523 }
3524
3525 void diff_change(struct diff_options *options,
3526                  unsigned old_mode, unsigned new_mode,
3527                  const unsigned char *old_sha1,
3528                  const unsigned char *new_sha1,
3529                  const char *concatpath)
3530 {
3531         struct diff_filespec *one, *two;
3532
3533         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3534                         && S_ISGITLINK(new_mode))
3535                 return;
3536
3537         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3538                 unsigned tmp;
3539                 const unsigned char *tmp_c;
3540                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3541                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3542         }
3543
3544         if (options->prefix &&
3545             strncmp(concatpath, options->prefix, options->prefix_length))
3546                 return;
3547
3548         one = alloc_filespec(concatpath);
3549         two = alloc_filespec(concatpath);
3550         fill_filespec(one, old_sha1, old_mode);
3551         fill_filespec(two, new_sha1, new_mode);
3552
3553         diff_queue(&diff_queued_diff, one, two);
3554         DIFF_OPT_SET(options, HAS_CHANGES);
3555 }
3556
3557 void diff_unmerge(struct diff_options *options,
3558                   const char *path,
3559                   unsigned mode, const unsigned char *sha1)
3560 {
3561         struct diff_filespec *one, *two;
3562
3563         if (options->prefix &&
3564             strncmp(path, options->prefix, options->prefix_length))
3565                 return;
3566
3567         one = alloc_filespec(path);
3568         two = alloc_filespec(path);
3569         fill_filespec(one, sha1, mode);
3570         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3571 }