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 int diff_use_color_default = -1;
  24 static const char *external_diff_cmd_cfg;
  25 int diff_auto_refresh_index = 1;
  26
  27 static char diff_colors[][COLOR_MAXLEN] = {
  28         "\033[m",       /* reset */
  29         "",             /* PLAIN (normal) */
  30         "\033[1m",      /* METAINFO (bold) */
  31         "\033[36m",     /* FRAGINFO (cyan) */
  32         "\033[31m",     /* OLD (red) */
  33         "\033[32m",     /* NEW (green) */
  34         "\033[33m",     /* COMMIT (yellow) */
  35         "\033[41m",     /* WHITESPACE (red background) */
  36 };
  37
  38 static int parse_diff_color_slot(const char *var, int ofs)
  39 {
  40         if (!strcasecmp(var+ofs, "plain"))
  41                 return DIFF_PLAIN;
  42         if (!strcasecmp(var+ofs, "meta"))
  43                 return DIFF_METAINFO;
  44         if (!strcasecmp(var+ofs, "frag"))
  45                 return DIFF_FRAGINFO;
  46         if (!strcasecmp(var+ofs, "old"))
  47                 return DIFF_FILE_OLD;
  48         if (!strcasecmp(var+ofs, "new"))
  49                 return DIFF_FILE_NEW;
  50         if (!strcasecmp(var+ofs, "commit"))
  51                 return DIFF_COMMIT;
  52         if (!strcasecmp(var+ofs, "whitespace"))
  53                 return DIFF_WHITESPACE;
  54         die("bad config variable '%s'", var);
  55 }
  56
  57 static struct ll_diff_driver {
  58         const char *name;
  59         struct ll_diff_driver *next;
  60         const char *cmd;
  61 } *user_diff, **user_diff_tail;
  62
  63 /*
  64  * Currently there is only "diff.<drivername>.command" variable;
  65  * because there are "diff.color.<slot>" variables, we are parsing
  66  * this in a bit convoluted way to allow low level diff driver
  67  * called "color".
  68  */
  69 static int parse_lldiff_command(const char *var, const char *ep, const char *value)
  70 {
  71         const char *name;
  72         int namelen;
  73         struct ll_diff_driver *drv;
  74
  75         name = var + 5;
  76         namelen = ep - name;
  77         for (drv = user_diff; drv; drv = drv->next)
  78                 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
  79                         break;
  80         if (!drv) {
  81                 drv = xcalloc(1, sizeof(struct ll_diff_driver));
  82                 drv->name = xmemdupz(name, namelen);
  83                 if (!user_diff_tail)
  84                         user_diff_tail = &user_diff;
  85                 *user_diff_tail = drv;
  86                 user_diff_tail = &(drv->next);
  87         }
  88
  89         return git_config_string(&(drv->cmd), var, value);
  90 }
  91
  92 /*
  93  * 'diff.<what>.funcname' attribute can be specified in the configuration
  94  * to define a customized regexp to find the beginning of a function to
  95  * be used for hunk header lines of "diff -p" style output.
  96  */
  97 static struct funcname_pattern {
  98         char *name;
  99         char *pattern;
 100         struct funcname_pattern *next;
 101 } *funcname_pattern_list;
 102
 103 static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
 104 {
 105         const char *name;
 106         int namelen;
 107         struct funcname_pattern *pp;
 108
 109         name = var + 5; /* "diff." */
 110         namelen = ep - name;
 111
 112         for (pp = funcname_pattern_list; pp; pp = pp->next)
 113                 if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
 114                         break;
 115         if (!pp) {
 116                 pp = xcalloc(1, sizeof(*pp));
 117                 pp->name = xmemdupz(name, namelen);
 118                 pp->next = funcname_pattern_list;
 119                 funcname_pattern_list = pp;
 120         }
 121         free(pp->pattern);
 122         pp->pattern = xstrdup(value);
 123         return 0;
 124 }
 125
 126 /*
 127  * These are to give UI layer defaults.
 128  * The core-level commands such as git-diff-files should
 129  * never be affected by the setting of diff.renames
 130  * the user happens to have in the configuration file.
 131  */
 132 int git_diff_ui_config(const char *var, const char *value, void *cb)
 133 {
 134         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 135                 diff_use_color_default = git_config_colorbool(var, value, -1);
 136                 return 0;
 137         }
 138         if (!strcmp(var, "diff.renames")) {
 139                 if (!value)
 140                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 141                 else if (!strcasecmp(value, "copies") ||
 142                          !strcasecmp(value, "copy"))
 143                         diff_detect_rename_default = DIFF_DETECT_COPY;
 144                 else if (git_config_bool(var,value))
 145                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 146                 return 0;
 147         }
 148         if (!strcmp(var, "diff.autorefreshindex")) {
 149                 diff_auto_refresh_index = git_config_bool(var, value);
 150                 return 0;
 151         }
 152         if (!strcmp(var, "diff.external"))
 153                 return git_config_string(&external_diff_cmd_cfg, var, value);
 154         if (!prefixcmp(var, "diff.")) {
 155                 const char *ep = strrchr(var, '.');
 156
 157                 if (ep != var + 4 && !strcmp(ep, ".command"))
 158                         return parse_lldiff_command(var, ep, value);
 159         }
 160
 161         return git_diff_basic_config(var, value, cb);
 162 }
 163
 164 int git_diff_basic_config(const char *var, const char *value, void *cb)
 165 {
 166         if (!strcmp(var, "diff.renamelimit")) {
 167                 diff_rename_limit_default = git_config_int(var, value);
 168                 return 0;
 169         }
 170
 171         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 172                 int slot = parse_diff_color_slot(var, 11);
 173                 if (!value)
 174                         return config_error_nonbool(var);
 175                 color_parse(value, var, diff_colors[slot]);
 176                 return 0;
 177         }
 178
 179         if (!prefixcmp(var, "diff.")) {
 180                 const char *ep = strrchr(var, '.');
 181                 if (ep != var + 4) {
 182                         if (!strcmp(ep, ".funcname")) {
 183                                 if (!value)
 184                                         return config_error_nonbool(var);
 185                                 return parse_funcname_pattern(var, ep, value);
 186                         }
 187                 }
 188         }
 189
 190         return git_color_default_config(var, value, cb);
 191 }
 192
 193 static char *quote_two(const char *one, const char *two)
 194 {
 195         int need_one = quote_c_style(one, NULL, NULL, 1);
 196         int need_two = quote_c_style(two, NULL, NULL, 1);
 197         struct strbuf res;
 198
 199         strbuf_init(&res, 0);
 200         if (need_one + need_two) {
 201                 strbuf_addch(&res, '"');
 202                 quote_c_style(one, &res, NULL, 1);
 203                 quote_c_style(two, &res, NULL, 1);
 204                 strbuf_addch(&res, '"');
 205         } else {
 206                 strbuf_addstr(&res, one);
 207                 strbuf_addstr(&res, two);
 208         }
 209         return strbuf_detach(&res, NULL);
 210 }
 211
 212 static const char *external_diff(void)
 213 {
 214         static const char *external_diff_cmd = NULL;
 215         static int done_preparing = 0;
 216
 217         if (done_preparing)
 218                 return external_diff_cmd;
 219         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 220         if (!external_diff_cmd)
 221                 external_diff_cmd = external_diff_cmd_cfg;
 222         done_preparing = 1;
 223         return external_diff_cmd;
 224 }
 225
 226 static struct diff_tempfile {
 227         const char *name; /* filename external diff should read from */
 228         char hex[41];
 229         char mode[10];
 230         char tmp_path[PATH_MAX];
 231 } diff_temp[2];
 232
 233 static int count_lines(const char *data, int size)
 234 {
 235         int count, ch, completely_empty = 1, nl_just_seen = 0;
 236         count = 0;
 237         while (0 < size--) {
 238                 ch = *data++;
 239                 if (ch == '\n') {
 240                         count++;
 241                         nl_just_seen = 1;
 242                         completely_empty = 0;
 243                 }
 244                 else {
 245                         nl_just_seen = 0;
 246                         completely_empty = 0;
 247                 }
 248         }
 249         if (completely_empty)
 250                 return 0;
 251         if (!nl_just_seen)
 252                 count++; /* no trailing newline */
 253         return count;
 254 }
 255
 256 static void print_line_count(FILE *file, int count)
 257 {
 258         switch (count) {
 259         case 0:
 260                 fprintf(file, "0,0");
 261                 break;
 262         case 1:
 263                 fprintf(file, "1");
 264                 break;
 265         default:
 266                 fprintf(file, "1,%d", count);
 267                 break;
 268         }
 269 }
 270
 271 static void copy_file_with_prefix(FILE *file,
 272                                   int prefix, const char *data, int size,
 273                                   const char *set, const char *reset)
 274 {
 275         int ch, nl_just_seen = 1;
 276         while (0 < size--) {
 277                 ch = *data++;
 278                 if (nl_just_seen) {
 279                         fputs(set, file);
 280                         putc(prefix, file);
 281                 }
 282                 if (ch == '\n') {
 283                         nl_just_seen = 1;
 284                         fputs(reset, file);
 285                 } else
 286                         nl_just_seen = 0;
 287                 putc(ch, file);
 288         }
 289         if (!nl_just_seen)
 290                 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
 291 }
 292
 293 static void emit_rewrite_diff(const char *name_a,
 294                               const char *name_b,
 295                               struct diff_filespec *one,
 296                               struct diff_filespec *two,
 297                               struct diff_options *o)
 298 {
 299         int lc_a, lc_b;
 300         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
 301         const char *name_a_tab, *name_b_tab;
 302         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 303         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 304         const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
 305         const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
 306         const char *reset = diff_get_color(color_diff, DIFF_RESET);
 307         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 308
 309         name_a += (*name_a == '/');
 310         name_b += (*name_b == '/');
 311         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 312         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 313
 314         strbuf_reset(&a_name);
 315         strbuf_reset(&b_name);
 316         quote_two_c_style(&a_name, o->a_prefix, name_a, 0);
 317         quote_two_c_style(&b_name, o->b_prefix, name_b, 0);
 318
 319         diff_populate_filespec(one, 0);
 320         diff_populate_filespec(two, 0);
 321         lc_a = count_lines(one->data, one->size);
 322         lc_b = count_lines(two->data, two->size);
 323         fprintf(o->file,
 324                 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
 325                 metainfo, a_name.buf, name_a_tab, reset,
 326                 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
 327         print_line_count(o->file, lc_a);
 328         fprintf(o->file, " +");
 329         print_line_count(o->file, lc_b);
 330         fprintf(o->file, " @@%s\n", reset);
 331         if (lc_a)
 332                 copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
 333         if (lc_b)
 334                 copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
 335 }
 336
 337 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 338 {
 339         if (!DIFF_FILE_VALID(one)) {
 340                 mf->ptr = (char *)""; /* does not matter */
 341                 mf->size = 0;
 342                 return 0;
 343         }
 344         else if (diff_populate_filespec(one, 0))
 345                 return -1;
 346         mf->ptr = one->data;
 347         mf->size = one->size;
 348         return 0;
 349 }
 350
 351 struct diff_words_buffer {
 352         mmfile_t text;
 353         long alloc;
 354         long current; /* output pointer */
 355         int suppressed_newline;
 356 };
 357
 358 static void diff_words_append(char *line, unsigned long len,
 359                 struct diff_words_buffer *buffer)
 360 {
 361         if (buffer->text.size + len > buffer->alloc) {
 362                 buffer->alloc = (buffer->text.size + len) * 3 / 2;
 363                 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
 364         }
 365         line++;
 366         len--;
 367         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 368         buffer->text.size += len;
 369 }
 370
 371 struct diff_words_data {
 372         struct xdiff_emit_state xm;
 373         struct diff_words_buffer minus, plus;
 374         FILE *file;
 375 };
 376
 377 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
 378                 int suppress_newline)
 379 {
 380         const char *ptr;
 381         int eol = 0;
 382
 383         if (len == 0)
 384                 return;
 385
 386         ptr  = buffer->text.ptr + buffer->current;
 387         buffer->current += len;
 388
 389         if (ptr[len - 1] == '\n') {
 390                 eol = 1;
 391                 len--;
 392         }
 393
 394         fputs(diff_get_color(1, color), file);
 395         fwrite(ptr, len, 1, file);
 396         fputs(diff_get_color(1, DIFF_RESET), file);
 397
 398         if (eol) {
 399                 if (suppress_newline)
 400                         buffer->suppressed_newline = 1;
 401                 else
 402                         putc('\n', file);
 403         }
 404 }
 405
 406 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 407 {
 408         struct diff_words_data *diff_words = priv;
 409
 410         if (diff_words->minus.suppressed_newline) {
 411                 if (line[0] != '+')
 412                         putc('\n', diff_words->file);
 413                 diff_words->minus.suppressed_newline = 0;
 414         }
 415
 416         len--;
 417         switch (line[0]) {
 418                 case '-':
 419                         print_word(diff_words->file,
 420                                    &diff_words->minus, len, DIFF_FILE_OLD, 1);
 421                         break;
 422                 case '+':
 423                         print_word(diff_words->file,
 424                                    &diff_words->plus, len, DIFF_FILE_NEW, 0);
 425                         break;
 426                 case ' ':
 427                         print_word(diff_words->file,
 428                                    &diff_words->plus, len, DIFF_PLAIN, 0);
 429                         diff_words->minus.current += len;
 430                         break;
 431         }
 432 }
 433
 434 /* this executes the word diff on the accumulated buffers */
 435 static void diff_words_show(struct diff_words_data *diff_words)
 436 {
 437         xpparam_t xpp;
 438         xdemitconf_t xecfg;
 439         xdemitcb_t ecb;
 440         mmfile_t minus, plus;
 441         int i;
 442
 443         memset(&xecfg, 0, sizeof(xecfg));
 444         minus.size = diff_words->minus.text.size;
 445         minus.ptr = xmalloc(minus.size);
 446         memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
 447         for (i = 0; i < minus.size; i++)
 448                 if (isspace(minus.ptr[i]))
 449                         minus.ptr[i] = '\n';
 450         diff_words->minus.current = 0;
 451
 452         plus.size = diff_words->plus.text.size;
 453         plus.ptr = xmalloc(plus.size);
 454         memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
 455         for (i = 0; i < plus.size; i++)
 456                 if (isspace(plus.ptr[i]))
 457                         plus.ptr[i] = '\n';
 458         diff_words->plus.current = 0;
 459
 460         xpp.flags = XDF_NEED_MINIMAL;
 461         xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
 462         ecb.outf = xdiff_outf;
 463         ecb.priv = diff_words;
 464         diff_words->xm.consume = fn_out_diff_words_aux;
 465         xdi_diff(&minus, &plus, &xpp, &xecfg, &ecb);
 466
 467         free(minus.ptr);
 468         free(plus.ptr);
 469         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 470
 471         if (diff_words->minus.suppressed_newline) {
 472                 putc('\n', diff_words->file);
 473                 diff_words->minus.suppressed_newline = 0;
 474         }
 475 }
 476
 477 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 478
 479 struct emit_callback {
 480         struct xdiff_emit_state xm;
 481         int nparents, color_diff;
 482         unsigned ws_rule;
 483         sane_truncate_fn truncate;
 484         const char **label_path;
 485         struct diff_words_data *diff_words;
 486         int *found_changesp;
 487         FILE *file;
 488 };
 489
 490 static void free_diff_words_data(struct emit_callback *ecbdata)
 491 {
 492         if (ecbdata->diff_words) {
 493                 /* flush buffers */
 494                 if (ecbdata->diff_words->minus.text.size ||
 495                                 ecbdata->diff_words->plus.text.size)
 496                         diff_words_show(ecbdata->diff_words);
 497
 498                 free (ecbdata->diff_words->minus.text.ptr);
 499                 free (ecbdata->diff_words->plus.text.ptr);
 500                 free(ecbdata->diff_words);
 501                 ecbdata->diff_words = NULL;
 502         }
 503 }
 504
 505 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 506 {
 507         if (diff_use_color)
 508                 return diff_colors[ix];
 509         return "";
 510 }
 511
 512 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
 513 {
 514         int has_trailing_newline = (len > 0 && line[len-1] == '\n');
 515         if (has_trailing_newline)
 516                 len--;
 517
 518         fputs(set, file);
 519         fwrite(line, len, 1, file);
 520         fputs(reset, file);
 521         if (has_trailing_newline)
 522                 fputc('\n', file);
 523 }
 524
 525 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
 526 {
 527         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 528         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 529
 530         if (!*ws)
 531                 emit_line(ecbdata->file, set, reset, line, len);
 532         else {
 533                 /* Emit just the prefix, then the rest. */
 534                 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
 535                 ws_check_emit(line + ecbdata->nparents,
 536                               len - ecbdata->nparents, ecbdata->ws_rule,
 537                               ecbdata->file, set, reset, ws);
 538         }
 539 }
 540
 541 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
 542 {
 543         const char *cp;
 544         unsigned long allot;
 545         size_t l = len;
 546
 547         if (ecb->truncate)
 548                 return ecb->truncate(line, len);
 549         cp = line;
 550         allot = l;
 551         while (0 < l) {
 552                 (void) utf8_width(&cp, &l);
 553                 if (!cp)
 554                         break; /* truncated in the middle? */
 555         }
 556         return allot - l;
 557 }
 558
 559 static void fn_out_consume(void *priv, char *line, unsigned long len)
 560 {
 561         int i;
 562         int color;
 563         struct emit_callback *ecbdata = priv;
 564         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 565         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 566         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 567
 568         *(ecbdata->found_changesp) = 1;
 569
 570         if (ecbdata->label_path[0]) {
 571                 const char *name_a_tab, *name_b_tab;
 572
 573                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 574                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 575
 576                 fprintf(ecbdata->file, "%s--- %s%s%s\n",
 577                         meta, ecbdata->label_path[0], reset, name_a_tab);
 578                 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
 579                         meta, ecbdata->label_path[1], reset, name_b_tab);
 580                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 581         }
 582
 583         /* This is not really necessary for now because
 584          * this codepath only deals with two-way diffs.
 585          */
 586         for (i = 0; i < len && line[i] == '@'; i++)
 587                 ;
 588         if (2 <= i && i < len && line[i] == ' ') {
 589                 ecbdata->nparents = i - 1;
 590                 len = sane_truncate_line(ecbdata, line, len);
 591                 emit_line(ecbdata->file,
 592                           diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
 593                           reset, line, len);
 594                 if (line[len-1] != '\n')
 595                         putc('\n', ecbdata->file);
 596                 return;
 597         }
 598
 599         if (len < ecbdata->nparents) {
 600                 emit_line(ecbdata->file, reset, reset, line, len);
 601                 return;
 602         }
 603
 604         color = DIFF_PLAIN;
 605         if (ecbdata->diff_words && ecbdata->nparents != 1)
 606                 /* fall back to normal diff */
 607                 free_diff_words_data(ecbdata);
 608         if (ecbdata->diff_words) {
 609                 if (line[0] == '-') {
 610                         diff_words_append(line, len,
 611                                           &ecbdata->diff_words->minus);
 612                         return;
 613                 } else if (line[0] == '+') {
 614                         diff_words_append(line, len,
 615                                           &ecbdata->diff_words->plus);
 616                         return;
 617                 }
 618                 if (ecbdata->diff_words->minus.text.size ||
 619                     ecbdata->diff_words->plus.text.size)
 620                         diff_words_show(ecbdata->diff_words);
 621                 line++;
 622                 len--;
 623                 emit_line(ecbdata->file, plain, reset, line, len);
 624                 return;
 625         }
 626         for (i = 0; i < ecbdata->nparents && len; i++) {
 627                 if (line[i] == '-')
 628                         color = DIFF_FILE_OLD;
 629                 else if (line[i] == '+')
 630                         color = DIFF_FILE_NEW;
 631         }
 632
 633         if (color != DIFF_FILE_NEW) {
 634                 emit_line(ecbdata->file,
 635                           diff_get_color(ecbdata->color_diff, color),
 636                           reset, line, len);
 637                 return;
 638         }
 639         emit_add_line(reset, ecbdata, line, len);
 640 }
 641
 642 static char *pprint_rename(const char *a, const char *b)
 643 {
 644         const char *old = a;
 645         const char *new = b;
 646         struct strbuf name;
 647         int pfx_length, sfx_length;
 648         int len_a = strlen(a);
 649         int len_b = strlen(b);
 650         int a_midlen, b_midlen;
 651         int qlen_a = quote_c_style(a, NULL, NULL, 0);
 652         int qlen_b = quote_c_style(b, NULL, NULL, 0);
 653
 654         strbuf_init(&name, 0);
 655         if (qlen_a || qlen_b) {
 656                 quote_c_style(a, &name, NULL, 0);
 657                 strbuf_addstr(&name, " => ");
 658                 quote_c_style(b, &name, NULL, 0);
 659                 return strbuf_detach(&name, NULL);
 660         }
 661
 662         /* Find common prefix */
 663         pfx_length = 0;
 664         while (*old && *new && *old == *new) {
 665                 if (*old == '/')
 666                         pfx_length = old - a + 1;
 667                 old++;
 668                 new++;
 669         }
 670
 671         /* Find common suffix */
 672         old = a + len_a;
 673         new = b + len_b;
 674         sfx_length = 0;
 675         while (a <= old && b <= new && *old == *new) {
 676                 if (*old == '/')
 677                         sfx_length = len_a - (old - a);
 678                 old--;
 679                 new--;
 680         }
 681
 682         /*
 683          * pfx{mid-a => mid-b}sfx
 684          * {pfx-a => pfx-b}sfx
 685          * pfx{sfx-a => sfx-b}
 686          * name-a => name-b
 687          */
 688         a_midlen = len_a - pfx_length - sfx_length;
 689         b_midlen = len_b - pfx_length - sfx_length;
 690         if (a_midlen < 0)
 691                 a_midlen = 0;
 692         if (b_midlen < 0)
 693                 b_midlen = 0;
 694
 695         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 696         if (pfx_length + sfx_length) {
 697                 strbuf_add(&name, a, pfx_length);
 698                 strbuf_addch(&name, '{');
 699         }
 700         strbuf_add(&name, a + pfx_length, a_midlen);
 701         strbuf_addstr(&name, " => ");
 702         strbuf_add(&name, b + pfx_length, b_midlen);
 703         if (pfx_length + sfx_length) {
 704                 strbuf_addch(&name, '}');
 705                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 706         }
 707         return strbuf_detach(&name, NULL);
 708 }
 709
 710 struct diffstat_t {
 711         struct xdiff_emit_state xm;
 712
 713         int nr;
 714         int alloc;
 715         struct diffstat_file {
 716                 char *from_name;
 717                 char *name;
 718                 char *print_name;
 719                 unsigned is_unmerged:1;
 720                 unsigned is_binary:1;
 721                 unsigned is_renamed:1;
 722                 unsigned int added, deleted;
 723         } **files;
 724 };
 725
 726 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 727                                           const char *name_a,
 728                                           const char *name_b)
 729 {
 730         struct diffstat_file *x;
 731         x = xcalloc(sizeof (*x), 1);
 732         if (diffstat->nr == diffstat->alloc) {
 733                 diffstat->alloc = alloc_nr(diffstat->alloc);
 734                 diffstat->files = xrealloc(diffstat->files,
 735                                 diffstat->alloc * sizeof(x));
 736         }
 737         diffstat->files[diffstat->nr++] = x;
 738         if (name_b) {
 739                 x->from_name = xstrdup(name_a);
 740                 x->name = xstrdup(name_b);
 741                 x->is_renamed = 1;
 742         }
 743         else {
 744                 x->from_name = NULL;
 745                 x->name = xstrdup(name_a);
 746         }
 747         return x;
 748 }
 749
 750 static void diffstat_consume(void *priv, char *line, unsigned long len)
 751 {
 752         struct diffstat_t *diffstat = priv;
 753         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 754
 755         if (line[0] == '+')
 756                 x->added++;
 757         else if (line[0] == '-')
 758                 x->deleted++;
 759 }
 760
 761 const char mime_boundary_leader[] = "------------";
 762
 763 static int scale_linear(int it, int width, int max_change)
 764 {
 765         /*
 766          * make sure that at least one '-' is printed if there were deletions,
 767          * and likewise for '+'.
 768          */
 769         if (max_change < 2)
 770                 return it;
 771         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
 772 }
 773
 774 static void show_name(FILE *file,
 775                       const char *prefix, const char *name, int len,
 776                       const char *reset, const char *set)
 777 {
 778         fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
 779 }
 780
 781 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
 782 {
 783         if (cnt <= 0)
 784                 return;
 785         fprintf(file, "%s", set);
 786         while (cnt--)
 787                 putc(ch, file);
 788         fprintf(file, "%s", reset);
 789 }
 790
 791 static void fill_print_name(struct diffstat_file *file)
 792 {
 793         char *pname;
 794
 795         if (file->print_name)
 796                 return;
 797
 798         if (!file->is_renamed) {
 799                 struct strbuf buf;
 800                 strbuf_init(&buf, 0);
 801                 if (quote_c_style(file->name, &buf, NULL, 0)) {
 802                         pname = strbuf_detach(&buf, NULL);
 803                 } else {
 804                         pname = file->name;
 805                         strbuf_release(&buf);
 806                 }
 807         } else {
 808                 pname = pprint_rename(file->from_name, file->name);
 809         }
 810         file->print_name = pname;
 811 }
 812
 813 static void show_stats(struct diffstat_t* data, struct diff_options *options)
 814 {
 815         int i, len, add, del, total, adds = 0, dels = 0;
 816         int max_change = 0, max_len = 0;
 817         int total_files = data->nr;
 818         int width, name_width;
 819         const char *reset, *set, *add_c, *del_c;
 820
 821         if (data->nr == 0)
 822                 return;
 823
 824         width = options->stat_width ? options->stat_width : 80;
 825         name_width = options->stat_name_width ? options->stat_name_width : 50;
 826
 827         /* Sanity: give at least 5 columns to the graph,
 828          * but leave at least 10 columns for the name.
 829          */
 830         if (width < 25)
 831                 width = 25;
 832         if (name_width < 10)
 833                 name_width = 10;
 834         else if (width < name_width + 15)
 835                 name_width = width - 15;
 836
 837         /* Find the longest filename and max number of changes */
 838         reset = diff_get_color_opt(options, DIFF_RESET);
 839         set   = diff_get_color_opt(options, DIFF_PLAIN);
 840         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
 841         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
 842
 843         for (i = 0; i < data->nr; i++) {
 844                 struct diffstat_file *file = data->files[i];
 845                 int change = file->added + file->deleted;
 846                 fill_print_name(file);
 847                 len = strlen(file->print_name);
 848                 if (max_len < len)
 849                         max_len = len;
 850
 851                 if (file->is_binary || file->is_unmerged)
 852                         continue;
 853                 if (max_change < change)
 854                         max_change = change;
 855         }
 856
 857         /* Compute the width of the graph part;
 858          * 10 is for one blank at the beginning of the line plus
 859          * " | count " between the name and the graph.
 860          *
 861          * From here on, name_width is the width of the name area,
 862          * and width is the width of the graph area.
 863          */
 864         name_width = (name_width < max_len) ? name_width : max_len;
 865         if (width < (name_width + 10) + max_change)
 866                 width = width - (name_width + 10);
 867         else
 868                 width = max_change;
 869
 870         for (i = 0; i < data->nr; i++) {
 871                 const char *prefix = "";
 872                 char *name = data->files[i]->print_name;
 873                 int added = data->files[i]->added;
 874                 int deleted = data->files[i]->deleted;
 875                 int name_len;
 876
 877                 /*
 878                  * "scale" the filename
 879                  */
 880                 len = name_width;
 881                 name_len = strlen(name);
 882                 if (name_width < name_len) {
 883                         char *slash;
 884                         prefix = "...";
 885                         len -= 3;
 886                         name += name_len - len;
 887                         slash = strchr(name, '/');
 888                         if (slash)
 889                                 name = slash;
 890                 }
 891
 892                 if (data->files[i]->is_binary) {
 893                         show_name(options->file, prefix, name, len, reset, set);
 894                         fprintf(options->file, "  Bin ");
 895                         fprintf(options->file, "%s%d%s", del_c, deleted, reset);
 896                         fprintf(options->file, " -> ");
 897                         fprintf(options->file, "%s%d%s", add_c, added, reset);
 898                         fprintf(options->file, " bytes");
 899                         fprintf(options->file, "\n");
 900                         continue;
 901                 }
 902                 else if (data->files[i]->is_unmerged) {
 903                         show_name(options->file, prefix, name, len, reset, set);
 904                         fprintf(options->file, "  Unmerged\n");
 905                         continue;
 906                 }
 907                 else if (!data->files[i]->is_renamed &&
 908                          (added + deleted == 0)) {
 909                         total_files--;
 910                         continue;
 911                 }
 912
 913                 /*
 914                  * scale the add/delete
 915                  */
 916                 add = added;
 917                 del = deleted;
 918                 total = add + del;
 919                 adds += add;
 920                 dels += del;
 921
 922                 if (width <= max_change) {
 923                         add = scale_linear(add, width, max_change);
 924                         del = scale_linear(del, width, max_change);
 925                         total = add + del;
 926                 }
 927                 show_name(options->file, prefix, name, len, reset, set);
 928                 fprintf(options->file, "%5d%s", added + deleted,
 929                                 added + deleted ? " " : "");
 930                 show_graph(options->file, '+', add, add_c, reset);
 931                 show_graph(options->file, '-', del, del_c, reset);
 932                 fprintf(options->file, "\n");
 933         }
 934         fprintf(options->file,
 935                "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
 936                set, total_files, adds, dels, reset);
 937 }
 938
 939 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
 940 {
 941         int i, adds = 0, dels = 0, total_files = data->nr;
 942
 943         if (data->nr == 0)
 944                 return;
 945
 946         for (i = 0; i < data->nr; i++) {
 947                 if (!data->files[i]->is_binary &&
 948                     !data->files[i]->is_unmerged) {
 949                         int added = data->files[i]->added;
 950                         int deleted= data->files[i]->deleted;
 951                         if (!data->files[i]->is_renamed &&
 952                             (added + deleted == 0)) {
 953                                 total_files--;
 954                         } else {
 955                                 adds += added;
 956                                 dels += deleted;
 957                         }
 958                 }
 959         }
 960         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
 961                total_files, adds, dels);
 962 }
 963
 964 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
 965 {
 966         int i;
 967
 968         if (data->nr == 0)
 969                 return;
 970
 971         for (i = 0; i < data->nr; i++) {
 972                 struct diffstat_file *file = data->files[i];
 973
 974                 if (file->is_binary)
 975                         fprintf(options->file, "-\t-\t");
 976                 else
 977                         fprintf(options->file,
 978                                 "%d\t%d\t", file->added, file->deleted);
 979                 if (options->line_termination) {
 980                         fill_print_name(file);
 981                         if (!file->is_renamed)
 982                                 write_name_quoted(file->name, options->file,
 983                                                   options->line_termination);
 984                         else {
 985                                 fputs(file->print_name, options->file);
 986                                 putc(options->line_termination, options->file);
 987                         }
 988                 } else {
 989                         if (file->is_renamed) {
 990                                 putc('\0', options->file);
 991                                 write_name_quoted(file->from_name, options->file, '\0');
 992                         }
 993                         write_name_quoted(file->name, options->file, '\0');
 994                 }
 995         }
 996 }
 997
 998 struct dirstat_file {
 999         const char *name;
1000         unsigned long changed;
1001 };
1002
1003 struct dirstat_dir {
1004         struct dirstat_file *files;
1005         int alloc, nr, percent, cumulative;
1006 };
1007
1008 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1009 {
1010         unsigned long this_dir = 0;
1011         unsigned int sources = 0;
1012
1013         while (dir->nr) {
1014                 struct dirstat_file *f = dir->files;
1015                 int namelen = strlen(f->name);
1016                 unsigned long this;
1017                 char *slash;
1018
1019                 if (namelen < baselen)
1020                         break;
1021                 if (memcmp(f->name, base, baselen))
1022                         break;
1023                 slash = strchr(f->name + baselen, '/');
1024                 if (slash) {
1025                         int newbaselen = slash + 1 - f->name;
1026                         this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1027                         sources++;
1028                 } else {
1029                         this = f->changed;
1030                         dir->files++;
1031                         dir->nr--;
1032                         sources += 2;
1033                 }
1034                 this_dir += this;
1035         }
1036
1037         /*
1038          * We don't report dirstat's for
1039          *  - the top level
1040          *  - or cases where everything came from a single directory
1041          *    under this directory (sources == 1).
1042          */
1043         if (baselen && sources != 1) {
1044                 int permille = this_dir * 1000 / changed;
1045                 if (permille) {
1046                         int percent = permille / 10;
1047                         if (percent >= dir->percent) {
1048                                 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1049                                 if (!dir->cumulative)
1050                                         return 0;
1051                         }
1052                 }
1053         }
1054         return this_dir;
1055 }
1056
1057 static int dirstat_compare(const void *_a, const void *_b)
1058 {
1059         const struct dirstat_file *a = _a;
1060         const struct dirstat_file *b = _b;
1061         return strcmp(a->name, b->name);
1062 }
1063
1064 static void show_dirstat(struct diff_options *options)
1065 {
1066         int i;
1067         unsigned long changed;
1068         struct dirstat_dir dir;
1069         struct diff_queue_struct *q = &diff_queued_diff;
1070
1071         dir.files = NULL;
1072         dir.alloc = 0;
1073         dir.nr = 0;
1074         dir.percent = options->dirstat_percent;
1075         dir.cumulative = options->output_format & DIFF_FORMAT_CUMULATIVE;
1076
1077         changed = 0;
1078         for (i = 0; i < q->nr; i++) {
1079                 struct diff_filepair *p = q->queue[i];
1080                 const char *name;
1081                 unsigned long copied, added, damage;
1082
1083                 name = p->one->path ? p->one->path : p->two->path;
1084
1085                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1086                         diff_populate_filespec(p->one, 0);
1087                         diff_populate_filespec(p->two, 0);
1088                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1089                                                &copied, &added);
1090                         diff_free_filespec_data(p->one);
1091                         diff_free_filespec_data(p->two);
1092                 } else if (DIFF_FILE_VALID(p->one)) {
1093                         diff_populate_filespec(p->one, 1);
1094                         copied = added = 0;
1095                         diff_free_filespec_data(p->one);
1096                 } else if (DIFF_FILE_VALID(p->two)) {
1097                         diff_populate_filespec(p->two, 1);
1098                         copied = 0;
1099                         added = p->two->size;
1100                         diff_free_filespec_data(p->two);
1101                 } else
1102                         continue;
1103
1104                 /*
1105                  * Original minus copied is the removed material,
1106                  * added is the new material.  They are both damages
1107                  * made to the preimage.
1108                  */
1109                 damage = (p->one->size - copied) + added;
1110
1111                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1112                 dir.files[dir.nr].name = name;
1113                 dir.files[dir.nr].changed = damage;
1114                 changed += damage;
1115                 dir.nr++;
1116         }
1117
1118         /* This can happen even with many files, if everything was renames */
1119         if (!changed)
1120                 return;
1121
1122         /* Show all directories with more than x% of the changes */
1123         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1124         gather_dirstat(options->file, &dir, changed, "", 0);
1125 }
1126
1127 static void free_diffstat_info(struct diffstat_t *diffstat)
1128 {
1129         int i;
1130         for (i = 0; i < diffstat->nr; i++) {
1131                 struct diffstat_file *f = diffstat->files[i];
1132                 if (f->name != f->print_name)
1133                         free(f->print_name);
1134                 free(f->name);
1135                 free(f->from_name);
1136                 free(f);
1137         }
1138         free(diffstat->files);
1139 }
1140
1141 struct checkdiff_t {
1142         struct xdiff_emit_state xm;
1143         const char *filename;
1144         int lineno;
1145         struct diff_options *o;
1146         unsigned ws_rule;
1147         unsigned status;
1148         int trailing_blanks_start;
1149 };
1150
1151 static int is_conflict_marker(const char *line, unsigned long len)
1152 {
1153         char firstchar;
1154         int cnt;
1155
1156         if (len < 8)
1157                 return 0;
1158         firstchar = line[0];
1159         switch (firstchar) {
1160         case '=': case '>': case '<':
1161                 break;
1162         default:
1163                 return 0;
1164         }
1165         for (cnt = 1; cnt < 7; cnt++)
1166                 if (line[cnt] != firstchar)
1167                         return 0;
1168         /* line[0] thru line[6] are same as firstchar */
1169         if (firstchar == '=') {
1170                 /* divider between ours and theirs? */
1171                 if (len != 8 || line[7] != '\n')
1172                         return 0;
1173         } else if (len < 8 || !isspace(line[7])) {
1174                 /* not divider before ours nor after theirs */
1175                 return 0;
1176         }
1177         return 1;
1178 }
1179
1180 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1181 {
1182         struct checkdiff_t *data = priv;
1183         int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1184         const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1185         const char *reset = diff_get_color(color_diff, DIFF_RESET);
1186         const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1187         char *err;
1188
1189         if (line[0] == '+') {
1190                 unsigned bad;
1191                 data->lineno++;
1192                 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1193                         data->trailing_blanks_start = 0;
1194                 else if (!data->trailing_blanks_start)
1195                         data->trailing_blanks_start = data->lineno;
1196                 if (is_conflict_marker(line + 1, len - 1)) {
1197                         data->status |= 1;
1198                         fprintf(data->o->file,
1199                                 "%s:%d: leftover conflict marker\n",
1200                                 data->filename, data->lineno);
1201                 }
1202                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1203                 if (!bad)
1204                         return;
1205                 data->status |= bad;
1206                 err = whitespace_error_string(bad);
1207                 fprintf(data->o->file, "%s:%d: %s.\n",
1208                         data->filename, data->lineno, err);
1209                 free(err);
1210                 emit_line(data->o->file, set, reset, line, 1);
1211                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1212                               data->o->file, set, reset, ws);
1213         } else if (line[0] == ' ') {
1214                 data->lineno++;
1215                 data->trailing_blanks_start = 0;
1216         } else if (line[0] == '@') {
1217                 char *plus = strchr(line, '+');
1218                 if (plus)
1219                         data->lineno = strtol(plus, NULL, 10) - 1;
1220                 else
1221                         die("invalid diff");
1222                 data->trailing_blanks_start = 0;
1223         }
1224 }
1225
1226 static unsigned char *deflate_it(char *data,
1227                                  unsigned long size,
1228                                  unsigned long *result_size)
1229 {
1230         int bound;
1231         unsigned char *deflated;
1232         z_stream stream;
1233
1234         memset(&stream, 0, sizeof(stream));
1235         deflateInit(&stream, zlib_compression_level);
1236         bound = deflateBound(&stream, size);
1237         deflated = xmalloc(bound);
1238         stream.next_out = deflated;
1239         stream.avail_out = bound;
1240
1241         stream.next_in = (unsigned char *)data;
1242         stream.avail_in = size;
1243         while (deflate(&stream, Z_FINISH) == Z_OK)
1244                 ; /* nothing */
1245         deflateEnd(&stream);
1246         *result_size = stream.total_out;
1247         return deflated;
1248 }
1249
1250 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1251 {
1252         void *cp;
1253         void *delta;
1254         void *deflated;
1255         void *data;
1256         unsigned long orig_size;
1257         unsigned long delta_size;
1258         unsigned long deflate_size;
1259         unsigned long data_size;
1260
1261         /* We could do deflated delta, or we could do just deflated two,
1262          * whichever is smaller.
1263          */
1264         delta = NULL;
1265         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1266         if (one->size && two->size) {
1267                 delta = diff_delta(one->ptr, one->size,
1268                                    two->ptr, two->size,
1269                                    &delta_size, deflate_size);
1270                 if (delta) {
1271                         void *to_free = delta;
1272                         orig_size = delta_size;
1273                         delta = deflate_it(delta, delta_size, &delta_size);
1274                         free(to_free);
1275                 }
1276         }
1277
1278         if (delta && delta_size < deflate_size) {
1279                 fprintf(file, "delta %lu\n", orig_size);
1280                 free(deflated);
1281                 data = delta;
1282                 data_size = delta_size;
1283         }
1284         else {
1285                 fprintf(file, "literal %lu\n", two->size);
1286                 free(delta);
1287                 data = deflated;
1288                 data_size = deflate_size;
1289         }
1290
1291         /* emit data encoded in base85 */
1292         cp = data;
1293         while (data_size) {
1294                 int bytes = (52 < data_size) ? 52 : data_size;
1295                 char line[70];
1296                 data_size -= bytes;
1297                 if (bytes <= 26)
1298                         line[0] = bytes + 'A' - 1;
1299                 else
1300                         line[0] = bytes - 26 + 'a' - 1;
1301                 encode_85(line + 1, cp, bytes);
1302                 cp = (char *) cp + bytes;
1303                 fputs(line, file);
1304                 fputc('\n', file);
1305         }
1306         fprintf(file, "\n");
1307         free(data);
1308 }
1309
1310 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1311 {
1312         fprintf(file, "GIT binary patch\n");
1313         emit_binary_diff_body(file, one, two);
1314         emit_binary_diff_body(file, two, one);
1315 }
1316
1317 static void setup_diff_attr_check(struct git_attr_check *check)
1318 {
1319         static struct git_attr *attr_diff;
1320
1321         if (!attr_diff) {
1322                 attr_diff = git_attr("diff", 4);
1323         }
1324         check[0].attr = attr_diff;
1325 }
1326
1327 static void diff_filespec_check_attr(struct diff_filespec *one)
1328 {
1329         struct git_attr_check attr_diff_check;
1330         int check_from_data = 0;
1331
1332         if (one->checked_attr)
1333                 return;
1334
1335         setup_diff_attr_check(&attr_diff_check);
1336         one->is_binary = 0;
1337         one->funcname_pattern_ident = NULL;
1338
1339         if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1340                 const char *value;
1341
1342                 /* binaryness */
1343                 value = attr_diff_check.value;
1344                 if (ATTR_TRUE(value))
1345                         ;
1346                 else if (ATTR_FALSE(value))
1347                         one->is_binary = 1;
1348                 else
1349                         check_from_data = 1;
1350
1351                 /* funcname pattern ident */
1352                 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1353                         ;
1354                 else
1355                         one->funcname_pattern_ident = value;
1356         }
1357
1358         if (check_from_data) {
1359                 if (!one->data && DIFF_FILE_VALID(one))
1360                         diff_populate_filespec(one, 0);
1361
1362                 if (one->data)
1363                         one->is_binary = buffer_is_binary(one->data, one->size);
1364         }
1365 }
1366
1367 int diff_filespec_is_binary(struct diff_filespec *one)
1368 {
1369         diff_filespec_check_attr(one);
1370         return one->is_binary;
1371 }
1372
1373 static const char *funcname_pattern(const char *ident)
1374 {
1375         struct funcname_pattern *pp;
1376
1377         for (pp = funcname_pattern_list; pp; pp = pp->next)
1378                 if (!strcmp(ident, pp->name))
1379                         return pp->pattern;
1380         return NULL;
1381 }
1382
1383 static struct builtin_funcname_pattern {
1384         const char *name;
1385         const char *pattern;
1386 } builtin_funcname_pattern[] = {
1387         { "java", "!^[  ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1388                         "new\\|return\\|switch\\|throw\\|while\\)\n"
1389                         "^[     ]*\\(\\([       ]*"
1390                         "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1391                         "[      ]*([^;]*\\)$" },
1392         { "pascal", "^\\(\\(procedure\\|function\\|constructor\\|"
1393                         "destructor\\|interface\\|implementation\\|"
1394                         "initialization\\|finalization\\)[ \t]*.*\\)$"
1395                         "\\|"
1396                         "^\\(.*=[ \t]*\\(class\\|record\\).*\\)$"
1397                         },
1398         { "bibtex", "\\(@[a-zA-Z]\\{1,\\}[ \t]*{\\{0,1\\}[ \t]*[^ \t\"@',\\#}{~%]*\\).*$" },
1399         { "tex", "^\\(\\\\\\(\\(sub\\)*section\\|chapter\\|part\\)\\*\\{0,1\\}{.*\\)$" },
1400         { "ruby", "^\\s*\\(\\(class\\|module\\|def\\)\\s.*\\)$" },
1401 };
1402
1403 static const char *diff_funcname_pattern(struct diff_filespec *one)
1404 {
1405         const char *ident, *pattern;
1406         int i;
1407
1408         diff_filespec_check_attr(one);
1409         ident = one->funcname_pattern_ident;
1410
1411         if (!ident)
1412                 /*
1413                  * If the config file has "funcname.default" defined, that
1414                  * regexp is used; otherwise NULL is returned and xemit uses
1415                  * the built-in default.
1416                  */
1417                 return funcname_pattern("default");
1418
1419         /* Look up custom "funcname.$ident" regexp from config. */
1420         pattern = funcname_pattern(ident);
1421         if (pattern)
1422                 return pattern;
1423
1424         /*
1425          * And define built-in fallback patterns here.  Note that
1426          * these can be overridden by the user's config settings.
1427          */
1428         for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1429                 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1430                         return builtin_funcname_pattern[i].pattern;
1431
1432         return NULL;
1433 }
1434
1435 static void builtin_diff(const char *name_a,
1436                          const char *name_b,
1437                          struct diff_filespec *one,
1438                          struct diff_filespec *two,
1439                          const char *xfrm_msg,
1440                          struct diff_options *o,
1441                          int complete_rewrite)
1442 {
1443         mmfile_t mf1, mf2;
1444         const char *lbl[2];
1445         char *a_one, *b_two;
1446         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1447         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1448
1449         a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1450         b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1451         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1452         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1453         fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1454         if (lbl[0][0] == '/') {
1455                 /* /dev/null */
1456                 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1457                 if (xfrm_msg && xfrm_msg[0])
1458                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1459         }
1460         else if (lbl[1][0] == '/') {
1461                 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1462                 if (xfrm_msg && xfrm_msg[0])
1463                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1464         }
1465         else {
1466                 if (one->mode != two->mode) {
1467                         fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1468                         fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1469                 }
1470                 if (xfrm_msg && xfrm_msg[0])
1471                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1472                 /*
1473                  * we do not run diff between different kind
1474                  * of objects.
1475                  */
1476                 if ((one->mode ^ two->mode) & S_IFMT)
1477                         goto free_ab_and_return;
1478                 if (complete_rewrite) {
1479                         emit_rewrite_diff(name_a, name_b, one, two, o);
1480                         o->found_changes = 1;
1481                         goto free_ab_and_return;
1482                 }
1483         }
1484
1485         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1486                 die("unable to read files to diff");
1487
1488         if (!DIFF_OPT_TST(o, TEXT) &&
1489             (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1490                 /* Quite common confusing case */
1491                 if (mf1.size == mf2.size &&
1492                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1493                         goto free_ab_and_return;
1494                 if (DIFF_OPT_TST(o, BINARY))
1495                         emit_binary_diff(o->file, &mf1, &mf2);
1496                 else
1497                         fprintf(o->file, "Binary files %s and %s differ\n",
1498                                 lbl[0], lbl[1]);
1499                 o->found_changes = 1;
1500         }
1501         else {
1502                 /* Crazy xdl interfaces.. */
1503                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1504                 xpparam_t xpp;
1505                 xdemitconf_t xecfg;
1506                 xdemitcb_t ecb;
1507                 struct emit_callback ecbdata;
1508                 const char *funcname_pattern;
1509
1510                 funcname_pattern = diff_funcname_pattern(one);
1511                 if (!funcname_pattern)
1512                         funcname_pattern = diff_funcname_pattern(two);
1513
1514                 memset(&xecfg, 0, sizeof(xecfg));
1515                 memset(&ecbdata, 0, sizeof(ecbdata));
1516                 ecbdata.label_path = lbl;
1517                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1518                 ecbdata.found_changesp = &o->found_changes;
1519                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1520                 ecbdata.file = o->file;
1521                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1522                 xecfg.ctxlen = o->context;
1523                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1524                 if (funcname_pattern)
1525                         xdiff_set_find_func(&xecfg, funcname_pattern);
1526                 if (!diffopts)
1527                         ;
1528                 else if (!prefixcmp(diffopts, "--unified="))
1529                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1530                 else if (!prefixcmp(diffopts, "-u"))
1531                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1532                 ecb.outf = xdiff_outf;
1533                 ecb.priv = &ecbdata;
1534                 ecbdata.xm.consume = fn_out_consume;
1535                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1536                         ecbdata.diff_words =
1537                                 xcalloc(1, sizeof(struct diff_words_data));
1538                         ecbdata.diff_words->file = o->file;
1539                 }
1540                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1541                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1542                         free_diff_words_data(&ecbdata);
1543         }
1544
1545  free_ab_and_return:
1546         diff_free_filespec_data(one);
1547         diff_free_filespec_data(two);
1548         free(a_one);
1549         free(b_two);
1550         return;
1551 }
1552
1553 static void builtin_diffstat(const char *name_a, const char *name_b,
1554                              struct diff_filespec *one,
1555                              struct diff_filespec *two,
1556                              struct diffstat_t *diffstat,
1557                              struct diff_options *o,
1558                              int complete_rewrite)
1559 {
1560         mmfile_t mf1, mf2;
1561         struct diffstat_file *data;
1562
1563         data = diffstat_add(diffstat, name_a, name_b);
1564
1565         if (!one || !two) {
1566                 data->is_unmerged = 1;
1567                 return;
1568         }
1569         if (complete_rewrite) {
1570                 diff_populate_filespec(one, 0);
1571                 diff_populate_filespec(two, 0);
1572                 data->deleted = count_lines(one->data, one->size);
1573                 data->added = count_lines(two->data, two->size);
1574                 goto free_and_return;
1575         }
1576         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1577                 die("unable to read files to diff");
1578
1579         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1580                 data->is_binary = 1;
1581                 data->added = mf2.size;
1582                 data->deleted = mf1.size;
1583         } else {
1584                 /* Crazy xdl interfaces.. */
1585                 xpparam_t xpp;
1586                 xdemitconf_t xecfg;
1587                 xdemitcb_t ecb;
1588
1589                 memset(&xecfg, 0, sizeof(xecfg));
1590                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1591                 ecb.outf = xdiff_outf;
1592                 ecb.priv = diffstat;
1593                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1594         }
1595
1596  free_and_return:
1597         diff_free_filespec_data(one);
1598         diff_free_filespec_data(two);
1599 }
1600
1601 static void builtin_checkdiff(const char *name_a, const char *name_b,
1602                               const char *attr_path,
1603                               struct diff_filespec *one,
1604                               struct diff_filespec *two,
1605                               struct diff_options *o)
1606 {
1607         mmfile_t mf1, mf2;
1608         struct checkdiff_t data;
1609
1610         if (!two)
1611                 return;
1612
1613         memset(&data, 0, sizeof(data));
1614         data.xm.consume = checkdiff_consume;
1615         data.filename = name_b ? name_b : name_a;
1616         data.lineno = 0;
1617         data.o = o;
1618         data.ws_rule = whitespace_rule(attr_path);
1619
1620         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1621                 die("unable to read files to diff");
1622
1623         /*
1624          * All the other codepaths check both sides, but not checking
1625          * the "old" side here is deliberate.  We are checking the newly
1626          * introduced changes, and as long as the "new" side is text, we
1627          * can and should check what it introduces.
1628          */
1629         if (diff_filespec_is_binary(two))
1630                 goto free_and_return;
1631         else {
1632                 /* Crazy xdl interfaces.. */
1633                 xpparam_t xpp;
1634                 xdemitconf_t xecfg;
1635                 xdemitcb_t ecb;
1636
1637                 memset(&xecfg, 0, sizeof(xecfg));
1638                 xecfg.ctxlen = 1; /* at least one context line */
1639                 xpp.flags = XDF_NEED_MINIMAL;
1640                 ecb.outf = xdiff_outf;
1641                 ecb.priv = &data;
1642                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1643
1644                 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1645                     data.trailing_blanks_start) {
1646                         fprintf(o->file, "%s:%d: ends with blank lines.\n",
1647                                 data.filename, data.trailing_blanks_start);
1648                         data.status = 1; /* report errors */
1649                 }
1650         }
1651  free_and_return:
1652         diff_free_filespec_data(one);
1653         diff_free_filespec_data(two);
1654         if (data.status)
1655                 DIFF_OPT_SET(o, CHECK_FAILED);
1656 }
1657
1658 struct diff_filespec *alloc_filespec(const char *path)
1659 {
1660         int namelen = strlen(path);
1661         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1662
1663         memset(spec, 0, sizeof(*spec));
1664         spec->path = (char *)(spec + 1);
1665         memcpy(spec->path, path, namelen+1);
1666         spec->count = 1;
1667         return spec;
1668 }
1669
1670 void free_filespec(struct diff_filespec *spec)
1671 {
1672         if (!--spec->count) {
1673                 diff_free_filespec_data(spec);
1674                 free(spec);
1675         }
1676 }
1677
1678 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1679                    unsigned short mode)
1680 {
1681         if (mode) {
1682                 spec->mode = canon_mode(mode);
1683                 hashcpy(spec->sha1, sha1);
1684                 spec->sha1_valid = !is_null_sha1(sha1);
1685         }
1686 }
1687
1688 /*
1689  * Given a name and sha1 pair, if the index tells us the file in
1690  * the work tree has that object contents, return true, so that
1691  * prepare_temp_file() does not have to inflate and extract.
1692  */
1693 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1694 {
1695         struct cache_entry *ce;
1696         struct stat st;
1697         int pos, len;
1698
1699         /* We do not read the cache ourselves here, because the
1700          * benchmark with my previous version that always reads cache
1701          * shows that it makes things worse for diff-tree comparing
1702          * two linux-2.6 kernel trees in an already checked out work
1703          * tree.  This is because most diff-tree comparisons deal with
1704          * only a small number of files, while reading the cache is
1705          * expensive for a large project, and its cost outweighs the
1706          * savings we get by not inflating the object to a temporary
1707          * file.  Practically, this code only helps when we are used
1708          * by diff-cache --cached, which does read the cache before
1709          * calling us.
1710          */
1711         if (!active_cache)
1712                 return 0;
1713
1714         /* We want to avoid the working directory if our caller
1715          * doesn't need the data in a normal file, this system
1716          * is rather slow with its stat/open/mmap/close syscalls,
1717          * and the object is contained in a pack file.  The pack
1718          * is probably already open and will be faster to obtain
1719          * the data through than the working directory.  Loose
1720          * objects however would tend to be slower as they need
1721          * to be individually opened and inflated.
1722          */
1723         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1724                 return 0;
1725
1726         len = strlen(name);
1727         pos = cache_name_pos(name, len);
1728         if (pos < 0)
1729                 return 0;
1730         ce = active_cache[pos];
1731
1732         /*
1733          * This is not the sha1 we are looking for, or
1734          * unreusable because it is not a regular file.
1735          */
1736         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1737                 return 0;
1738
1739         /*
1740          * If ce matches the file in the work tree, we can reuse it.
1741          */
1742         if (ce_uptodate(ce) ||
1743             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1744                 return 1;
1745
1746         return 0;
1747 }
1748
1749 static int populate_from_stdin(struct diff_filespec *s)
1750 {
1751         struct strbuf buf;
1752         size_t size = 0;
1753
1754         strbuf_init(&buf, 0);
1755         if (strbuf_read(&buf, 0, 0) < 0)
1756                 return error("error while reading from stdin %s",
1757                                      strerror(errno));
1758
1759         s->should_munmap = 0;
1760         s->data = strbuf_detach(&buf, &size);
1761         s->size = size;
1762         s->should_free = 1;
1763         return 0;
1764 }
1765
1766 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1767 {
1768         int len;
1769         char *data = xmalloc(100);
1770         len = snprintf(data, 100,
1771                 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1772         s->data = data;
1773         s->size = len;
1774         s->should_free = 1;
1775         if (size_only) {
1776                 s->data = NULL;
1777                 free(data);
1778         }
1779         return 0;
1780 }
1781
1782 /*
1783  * While doing rename detection and pickaxe operation, we may need to
1784  * grab the data for the blob (or file) for our own in-core comparison.
1785  * diff_filespec has data and size fields for this purpose.
1786  */
1787 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1788 {
1789         int err = 0;
1790         if (!DIFF_FILE_VALID(s))
1791                 die("internal error: asking to populate invalid file.");
1792         if (S_ISDIR(s->mode))
1793                 return -1;
1794
1795         if (s->data)
1796                 return 0;
1797
1798         if (size_only && 0 < s->size)
1799                 return 0;
1800
1801         if (S_ISGITLINK(s->mode))
1802                 return diff_populate_gitlink(s, size_only);
1803
1804         if (!s->sha1_valid ||
1805             reuse_worktree_file(s->path, s->sha1, 0)) {
1806                 struct strbuf buf;
1807                 struct stat st;
1808                 int fd;
1809
1810                 if (!strcmp(s->path, "-"))
1811                         return populate_from_stdin(s);
1812
1813                 if (lstat(s->path, &st) < 0) {
1814                         if (errno == ENOENT) {
1815                         err_empty:
1816                                 err = -1;
1817                         empty:
1818                                 s->data = (char *)"";
1819                                 s->size = 0;
1820                                 return err;
1821                         }
1822                 }
1823                 s->size = xsize_t(st.st_size);
1824                 if (!s->size)
1825                         goto empty;
1826                 if (size_only)
1827                         return 0;
1828                 if (S_ISLNK(st.st_mode)) {
1829                         int ret;
1830                         s->data = xmalloc(s->size);
1831                         s->should_free = 1;
1832                         ret = readlink(s->path, s->data, s->size);
1833                         if (ret < 0) {
1834                                 free(s->data);
1835                                 goto err_empty;
1836                         }
1837                         return 0;
1838                 }
1839                 fd = open(s->path, O_RDONLY);
1840                 if (fd < 0)
1841                         goto err_empty;
1842                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1843                 close(fd);
1844                 s->should_munmap = 1;
1845
1846                 /*
1847                  * Convert from working tree format to canonical git format
1848                  */
1849                 strbuf_init(&buf, 0);
1850                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1851                         size_t size = 0;
1852                         munmap(s->data, s->size);
1853                         s->should_munmap = 0;
1854                         s->data = strbuf_detach(&buf, &size);
1855                         s->size = size;
1856                         s->should_free = 1;
1857                 }
1858         }
1859         else {
1860                 enum object_type type;
1861                 if (size_only)
1862                         type = sha1_object_info(s->sha1, &s->size);
1863                 else {
1864                         s->data = read_sha1_file(s->sha1, &type, &s->size);
1865                         s->should_free = 1;
1866                 }
1867         }
1868         return 0;
1869 }
1870
1871 void diff_free_filespec_blob(struct diff_filespec *s)
1872 {
1873         if (s->should_free)
1874                 free(s->data);
1875         else if (s->should_munmap)
1876                 munmap(s->data, s->size);
1877
1878         if (s->should_free || s->should_munmap) {
1879                 s->should_free = s->should_munmap = 0;
1880                 s->data = NULL;
1881         }
1882 }
1883
1884 void diff_free_filespec_data(struct diff_filespec *s)
1885 {
1886         diff_free_filespec_blob(s);
1887         free(s->cnt_data);
1888         s->cnt_data = NULL;
1889 }
1890
1891 static void prep_temp_blob(struct diff_tempfile *temp,
1892                            void *blob,
1893                            unsigned long size,
1894                            const unsigned char *sha1,
1895                            int mode)
1896 {
1897         int fd;
1898
1899         fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1900         if (fd < 0)
1901                 die("unable to create temp-file: %s", strerror(errno));
1902         if (write_in_full(fd, blob, size) != size)
1903                 die("unable to write temp-file");
1904         close(fd);
1905         temp->name = temp->tmp_path;
1906         strcpy(temp->hex, sha1_to_hex(sha1));
1907         temp->hex[40] = 0;
1908         sprintf(temp->mode, "%06o", mode);
1909 }
1910
1911 static void prepare_temp_file(const char *name,
1912                               struct diff_tempfile *temp,
1913                               struct diff_filespec *one)
1914 {
1915         if (!DIFF_FILE_VALID(one)) {
1916         not_a_valid_file:
1917                 /* A '-' entry produces this for file-2, and
1918                  * a '+' entry produces this for file-1.
1919                  */
1920                 temp->name = "/dev/null";
1921                 strcpy(temp->hex, ".");
1922                 strcpy(temp->mode, ".");
1923                 return;
1924         }
1925
1926         if (!one->sha1_valid ||
1927             reuse_worktree_file(name, one->sha1, 1)) {
1928                 struct stat st;
1929                 if (lstat(name, &st) < 0) {
1930                         if (errno == ENOENT)
1931                                 goto not_a_valid_file;
1932                         die("stat(%s): %s", name, strerror(errno));
1933                 }
1934                 if (S_ISLNK(st.st_mode)) {
1935                         int ret;
1936                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1937                         size_t sz = xsize_t(st.st_size);
1938                         if (sizeof(buf) <= st.st_size)
1939                                 die("symlink too long: %s", name);
1940                         ret = readlink(name, buf, sz);
1941                         if (ret < 0)
1942                                 die("readlink(%s)", name);
1943                         prep_temp_blob(temp, buf, sz,
1944                                        (one->sha1_valid ?
1945                                         one->sha1 : null_sha1),
1946                                        (one->sha1_valid ?
1947                                         one->mode : S_IFLNK));
1948                 }
1949                 else {
1950                         /* we can borrow from the file in the work tree */
1951                         temp->name = name;
1952                         if (!one->sha1_valid)
1953                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
1954                         else
1955                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
1956                         /* Even though we may sometimes borrow the
1957                          * contents from the work tree, we always want
1958                          * one->mode.  mode is trustworthy even when
1959                          * !(one->sha1_valid), as long as
1960                          * DIFF_FILE_VALID(one).
1961                          */
1962                         sprintf(temp->mode, "%06o", one->mode);
1963                 }
1964                 return;
1965         }
1966         else {
1967                 if (diff_populate_filespec(one, 0))
1968                         die("cannot read data blob for %s", one->path);
1969                 prep_temp_blob(temp, one->data, one->size,
1970                                one->sha1, one->mode);
1971         }
1972 }
1973
1974 static void remove_tempfile(void)
1975 {
1976         int i;
1977
1978         for (i = 0; i < 2; i++)
1979                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1980                         unlink(diff_temp[i].name);
1981                         diff_temp[i].name = NULL;
1982                 }
1983 }
1984
1985 static void remove_tempfile_on_signal(int signo)
1986 {
1987         remove_tempfile();
1988         signal(SIGINT, SIG_DFL);
1989         raise(signo);
1990 }
1991
1992 /* An external diff command takes:
1993  *
1994  * diff-cmd name infile1 infile1-sha1 infile1-mode \
1995  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
1996  *
1997  */
1998 static void run_external_diff(const char *pgm,
1999                               const char *name,
2000                               const char *other,
2001                               struct diff_filespec *one,
2002                               struct diff_filespec *two,
2003                               const char *xfrm_msg,
2004                               int complete_rewrite)
2005 {
2006         const char *spawn_arg[10];
2007         struct diff_tempfile *temp = diff_temp;
2008         int retval;
2009         static int atexit_asked = 0;
2010         const char *othername;
2011         const char **arg = &spawn_arg[0];
2012
2013         othername = (other? other : name);
2014         if (one && two) {
2015                 prepare_temp_file(name, &temp[0], one);
2016                 prepare_temp_file(othername, &temp[1], two);
2017                 if (! atexit_asked &&
2018                     (temp[0].name == temp[0].tmp_path ||
2019                      temp[1].name == temp[1].tmp_path)) {
2020                         atexit_asked = 1;
2021                         atexit(remove_tempfile);
2022                 }
2023                 signal(SIGINT, remove_tempfile_on_signal);
2024         }
2025
2026         if (one && two) {
2027                 *arg++ = pgm;
2028                 *arg++ = name;
2029                 *arg++ = temp[0].name;
2030                 *arg++ = temp[0].hex;
2031                 *arg++ = temp[0].mode;
2032                 *arg++ = temp[1].name;
2033                 *arg++ = temp[1].hex;
2034                 *arg++ = temp[1].mode;
2035                 if (other) {
2036                         *arg++ = other;
2037                         *arg++ = xfrm_msg;
2038                 }
2039         } else {
2040                 *arg++ = pgm;
2041                 *arg++ = name;
2042         }
2043         *arg = NULL;
2044         fflush(NULL);
2045         retval = run_command_v_opt(spawn_arg, 0);
2046         remove_tempfile();
2047         if (retval) {
2048                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2049                 exit(1);
2050         }
2051 }
2052
2053 static const char *external_diff_attr(const char *name)
2054 {
2055         struct git_attr_check attr_diff_check;
2056
2057         if (!name)
2058                 return NULL;
2059
2060         setup_diff_attr_check(&attr_diff_check);
2061         if (!git_checkattr(name, 1, &attr_diff_check)) {
2062                 const char *value = attr_diff_check.value;
2063                 if (!ATTR_TRUE(value) &&
2064                     !ATTR_FALSE(value) &&
2065                     !ATTR_UNSET(value)) {
2066                         struct ll_diff_driver *drv;
2067
2068                         for (drv = user_diff; drv; drv = drv->next)
2069                                 if (!strcmp(drv->name, value))
2070                                         return drv->cmd;
2071                 }
2072         }
2073         return NULL;
2074 }
2075
2076 static void run_diff_cmd(const char *pgm,
2077                          const char *name,
2078                          const char *other,
2079                          const char *attr_path,
2080                          struct diff_filespec *one,
2081                          struct diff_filespec *two,
2082                          const char *xfrm_msg,
2083                          struct diff_options *o,
2084                          int complete_rewrite)
2085 {
2086         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2087                 pgm = NULL;
2088         else {
2089                 const char *cmd = external_diff_attr(attr_path);
2090                 if (cmd)
2091                         pgm = cmd;
2092         }
2093
2094         if (pgm) {
2095                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2096                                   complete_rewrite);
2097                 return;
2098         }
2099         if (one && two)
2100                 builtin_diff(name, other ? other : name,
2101                              one, two, xfrm_msg, o, complete_rewrite);
2102         else
2103                 fprintf(o->file, "* Unmerged path %s\n", name);
2104 }
2105
2106 static void diff_fill_sha1_info(struct diff_filespec *one)
2107 {
2108         if (DIFF_FILE_VALID(one)) {
2109                 if (!one->sha1_valid) {
2110                         struct stat st;
2111                         if (!strcmp(one->path, "-")) {
2112                                 hashcpy(one->sha1, null_sha1);
2113                                 return;
2114                         }
2115                         if (lstat(one->path, &st) < 0)
2116                                 die("stat %s", one->path);
2117                         if (index_path(one->sha1, one->path, &st, 0))
2118                                 die("cannot hash %s\n", one->path);
2119                 }
2120         }
2121         else
2122                 hashclr(one->sha1);
2123 }
2124
2125 static int similarity_index(struct diff_filepair *p)
2126 {
2127         return p->score * 100 / MAX_SCORE;
2128 }
2129
2130 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2131 {
2132         /* Strip the prefix but do not molest /dev/null and absolute paths */
2133         if (*namep && **namep != '/')
2134                 *namep += prefix_length;
2135         if (*otherp && **otherp != '/')
2136                 *otherp += prefix_length;
2137 }
2138
2139 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2140 {
2141         const char *pgm = external_diff();
2142         struct strbuf msg;
2143         char *xfrm_msg;
2144         struct diff_filespec *one = p->one;
2145         struct diff_filespec *two = p->two;
2146         const char *name;
2147         const char *other;
2148         const char *attr_path;
2149         int complete_rewrite = 0;
2150
2151         name  = p->one->path;
2152         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2153         attr_path = name;
2154         if (o->prefix_length)
2155                 strip_prefix(o->prefix_length, &name, &other);
2156
2157         if (DIFF_PAIR_UNMERGED(p)) {
2158                 run_diff_cmd(pgm, name, NULL, attr_path,
2159                              NULL, NULL, NULL, o, 0);
2160                 return;
2161         }
2162
2163         diff_fill_sha1_info(one);
2164         diff_fill_sha1_info(two);
2165
2166         strbuf_init(&msg, PATH_MAX * 2 + 300);
2167         switch (p->status) {
2168         case DIFF_STATUS_COPIED:
2169                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2170                 strbuf_addstr(&msg, "\ncopy from ");
2171                 quote_c_style(name, &msg, NULL, 0);
2172                 strbuf_addstr(&msg, "\ncopy to ");
2173                 quote_c_style(other, &msg, NULL, 0);
2174                 strbuf_addch(&msg, '\n');
2175                 break;
2176         case DIFF_STATUS_RENAMED:
2177                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2178                 strbuf_addstr(&msg, "\nrename from ");
2179                 quote_c_style(name, &msg, NULL, 0);
2180                 strbuf_addstr(&msg, "\nrename to ");
2181                 quote_c_style(other, &msg, NULL, 0);
2182                 strbuf_addch(&msg, '\n');
2183                 break;
2184         case DIFF_STATUS_MODIFIED:
2185                 if (p->score) {
2186                         strbuf_addf(&msg, "dissimilarity index %d%%\n",
2187                                         similarity_index(p));
2188                         complete_rewrite = 1;
2189                         break;
2190                 }
2191                 /* fallthru */
2192         default:
2193                 /* nothing */
2194                 ;
2195         }
2196
2197         if (hashcmp(one->sha1, two->sha1)) {
2198                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2199
2200                 if (DIFF_OPT_TST(o, BINARY)) {
2201                         mmfile_t mf;
2202                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2203                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2204                                 abbrev = 40;
2205                 }
2206                 strbuf_addf(&msg, "index %.*s..%.*s",
2207                                 abbrev, sha1_to_hex(one->sha1),
2208                                 abbrev, sha1_to_hex(two->sha1));
2209                 if (one->mode == two->mode)
2210                         strbuf_addf(&msg, " %06o", one->mode);
2211                 strbuf_addch(&msg, '\n');
2212         }
2213
2214         if (msg.len)
2215                 strbuf_setlen(&msg, msg.len - 1);
2216         xfrm_msg = msg.len ? msg.buf : NULL;
2217
2218         if (!pgm &&
2219             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2220             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2221                 /* a filepair that changes between file and symlink
2222                  * needs to be split into deletion and creation.
2223                  */
2224                 struct diff_filespec *null = alloc_filespec(two->path);
2225                 run_diff_cmd(NULL, name, other, attr_path,
2226                              one, null, xfrm_msg, o, 0);
2227                 free(null);
2228                 null = alloc_filespec(one->path);
2229                 run_diff_cmd(NULL, name, other, attr_path,
2230                              null, two, xfrm_msg, o, 0);
2231                 free(null);
2232         }
2233         else
2234                 run_diff_cmd(pgm, name, other, attr_path,
2235                              one, two, xfrm_msg, o, complete_rewrite);
2236
2237         strbuf_release(&msg);
2238 }
2239
2240 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2241                          struct diffstat_t *diffstat)
2242 {
2243         const char *name;
2244         const char *other;
2245         int complete_rewrite = 0;
2246
2247         if (DIFF_PAIR_UNMERGED(p)) {
2248                 /* unmerged */
2249                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2250                 return;
2251         }
2252
2253         name = p->one->path;
2254         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2255
2256         if (o->prefix_length)
2257                 strip_prefix(o->prefix_length, &name, &other);
2258
2259         diff_fill_sha1_info(p->one);
2260         diff_fill_sha1_info(p->two);
2261
2262         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2263                 complete_rewrite = 1;
2264         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2265 }
2266
2267 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2268 {
2269         const char *name;
2270         const char *other;
2271         const char *attr_path;
2272
2273         if (DIFF_PAIR_UNMERGED(p)) {
2274                 /* unmerged */
2275                 return;
2276         }
2277
2278         name = p->one->path;
2279         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2280         attr_path = other ? other : name;
2281
2282         if (o->prefix_length)
2283                 strip_prefix(o->prefix_length, &name, &other);
2284
2285         diff_fill_sha1_info(p->one);
2286         diff_fill_sha1_info(p->two);
2287
2288         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2289 }
2290
2291 void diff_setup(struct diff_options *options)
2292 {
2293         memset(options, 0, sizeof(*options));
2294
2295         options->file = stdout;
2296
2297         options->line_termination = '\n';
2298         options->break_opt = -1;
2299         options->rename_limit = -1;
2300         options->dirstat_percent = 3;
2301         options->context = 3;
2302
2303         options->change = diff_change;
2304         options->add_remove = diff_addremove;
2305         if (diff_use_color_default > 0)
2306                 DIFF_OPT_SET(options, COLOR_DIFF);
2307         else
2308                 DIFF_OPT_CLR(options, COLOR_DIFF);
2309         options->detect_rename = diff_detect_rename_default;
2310
2311         options->a_prefix = "a/";
2312         options->b_prefix = "b/";
2313 }
2314
2315 int diff_setup_done(struct diff_options *options)
2316 {
2317         int count = 0;
2318
2319         if (options->output_format & DIFF_FORMAT_NAME)
2320                 count++;
2321         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2322                 count++;
2323         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2324                 count++;
2325         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2326                 count++;
2327         if (count > 1)
2328                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2329
2330         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2331                 options->detect_rename = DIFF_DETECT_COPY;
2332
2333         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2334                 options->prefix = NULL;
2335         if (options->prefix)
2336                 options->prefix_length = strlen(options->prefix);
2337         else
2338                 options->prefix_length = 0;
2339
2340         if (options->output_format & (DIFF_FORMAT_NAME |
2341                                       DIFF_FORMAT_NAME_STATUS |
2342                                       DIFF_FORMAT_CHECKDIFF |
2343                                       DIFF_FORMAT_NO_OUTPUT))
2344                 options->output_format &= ~(DIFF_FORMAT_RAW |
2345                                             DIFF_FORMAT_NUMSTAT |
2346                                             DIFF_FORMAT_DIFFSTAT |
2347                                             DIFF_FORMAT_SHORTSTAT |
2348                                             DIFF_FORMAT_DIRSTAT |
2349                                             DIFF_FORMAT_SUMMARY |
2350                                             DIFF_FORMAT_PATCH);
2351
2352         /*
2353          * These cases always need recursive; we do not drop caller-supplied
2354          * recursive bits for other formats here.
2355          */
2356         if (options->output_format & (DIFF_FORMAT_PATCH |
2357                                       DIFF_FORMAT_NUMSTAT |
2358                                       DIFF_FORMAT_DIFFSTAT |
2359                                       DIFF_FORMAT_SHORTSTAT |
2360                                       DIFF_FORMAT_DIRSTAT |
2361                                       DIFF_FORMAT_SUMMARY |
2362                                       DIFF_FORMAT_CHECKDIFF))
2363                 DIFF_OPT_SET(options, RECURSIVE);
2364         /*
2365          * Also pickaxe would not work very well if you do not say recursive
2366          */
2367         if (options->pickaxe)
2368                 DIFF_OPT_SET(options, RECURSIVE);
2369
2370         if (options->detect_rename && options->rename_limit < 0)
2371                 options->rename_limit = diff_rename_limit_default;
2372         if (options->setup & DIFF_SETUP_USE_CACHE) {
2373                 if (!active_cache)
2374                         /* read-cache does not die even when it fails
2375                          * so it is safe for us to do this here.  Also
2376                          * it does not smudge active_cache or active_nr
2377                          * when it fails, so we do not have to worry about
2378                          * cleaning it up ourselves either.
2379                          */
2380                         read_cache();
2381         }
2382         if (options->abbrev <= 0 || 40 < options->abbrev)
2383                 options->abbrev = 40; /* full */
2384
2385         /*
2386          * It does not make sense to show the first hit we happened
2387          * to have found.  It does not make sense not to return with
2388          * exit code in such a case either.
2389          */
2390         if (DIFF_OPT_TST(options, QUIET)) {
2391                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2392                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2393         }
2394
2395         /*
2396          * If we postprocess in diffcore, we cannot simply return
2397          * upon the first hit.  We need to run diff as usual.
2398          */
2399         if (options->pickaxe || options->filter)
2400                 DIFF_OPT_CLR(options, QUIET);
2401
2402         return 0;
2403 }
2404
2405 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2406 {
2407         char c, *eq;
2408         int len;
2409
2410         if (*arg != '-')
2411                 return 0;
2412         c = *++arg;
2413         if (!c)
2414                 return 0;
2415         if (c == arg_short) {
2416                 c = *++arg;
2417                 if (!c)
2418                         return 1;
2419                 if (val && isdigit(c)) {
2420                         char *end;
2421                         int n = strtoul(arg, &end, 10);
2422                         if (*end)
2423                                 return 0;
2424                         *val = n;
2425                         return 1;
2426                 }
2427                 return 0;
2428         }
2429         if (c != '-')
2430                 return 0;
2431         arg++;
2432         eq = strchr(arg, '=');
2433         if (eq)
2434                 len = eq - arg;
2435         else
2436                 len = strlen(arg);
2437         if (!len || strncmp(arg, arg_long, len))
2438                 return 0;
2439         if (eq) {
2440                 int n;
2441                 char *end;
2442                 if (!isdigit(*++eq))
2443                         return 0;
2444                 n = strtoul(eq, &end, 10);
2445                 if (*end)
2446                         return 0;
2447                 *val = n;
2448         }
2449         return 1;
2450 }
2451
2452 static int diff_scoreopt_parse(const char *opt);
2453
2454 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2455 {
2456         const char *arg = av[0];
2457
2458         /* Output format options */
2459         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2460                 options->output_format |= DIFF_FORMAT_PATCH;
2461         else if (opt_arg(arg, 'U', "unified", &options->context))
2462                 options->output_format |= DIFF_FORMAT_PATCH;
2463         else if (!strcmp(arg, "--raw"))
2464                 options->output_format |= DIFF_FORMAT_RAW;
2465         else if (!strcmp(arg, "--patch-with-raw"))
2466                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2467         else if (!strcmp(arg, "--numstat"))
2468                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2469         else if (!strcmp(arg, "--shortstat"))
2470                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2471         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2472                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2473         else if (!strcmp(arg, "--cumulative"))
2474                 options->output_format |= DIFF_FORMAT_CUMULATIVE;
2475         else if (!strcmp(arg, "--check"))
2476                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2477         else if (!strcmp(arg, "--summary"))
2478                 options->output_format |= DIFF_FORMAT_SUMMARY;
2479         else if (!strcmp(arg, "--patch-with-stat"))
2480                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2481         else if (!strcmp(arg, "--name-only"))
2482                 options->output_format |= DIFF_FORMAT_NAME;
2483         else if (!strcmp(arg, "--name-status"))
2484                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2485         else if (!strcmp(arg, "-s"))
2486                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2487         else if (!prefixcmp(arg, "--stat")) {
2488                 char *end;
2489                 int width = options->stat_width;
2490                 int name_width = options->stat_name_width;
2491                 arg += 6;
2492                 end = (char *)arg;
2493
2494                 switch (*arg) {
2495                 case '-':
2496                         if (!prefixcmp(arg, "-width="))
2497                                 width = strtoul(arg + 7, &end, 10);
2498                         else if (!prefixcmp(arg, "-name-width="))
2499                                 name_width = strtoul(arg + 12, &end, 10);
2500                         break;
2501                 case '=':
2502                         width = strtoul(arg+1, &end, 10);
2503                         if (*end == ',')
2504                                 name_width = strtoul(end+1, &end, 10);
2505                 }
2506
2507                 /* Important! This checks all the error cases! */
2508                 if (*end)
2509                         return 0;
2510                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2511                 options->stat_name_width = name_width;
2512                 options->stat_width = width;
2513         }
2514
2515         /* renames options */
2516         else if (!prefixcmp(arg, "-B")) {
2517                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2518                         return -1;
2519         }
2520         else if (!prefixcmp(arg, "-M")) {
2521                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2522                         return -1;
2523                 options->detect_rename = DIFF_DETECT_RENAME;
2524         }
2525         else if (!prefixcmp(arg, "-C")) {
2526                 if (options->detect_rename == DIFF_DETECT_COPY)
2527                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2528                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2529                         return -1;
2530                 options->detect_rename = DIFF_DETECT_COPY;
2531         }
2532         else if (!strcmp(arg, "--no-renames"))
2533                 options->detect_rename = 0;
2534         else if (!strcmp(arg, "--relative"))
2535                 DIFF_OPT_SET(options, RELATIVE_NAME);
2536         else if (!prefixcmp(arg, "--relative=")) {
2537                 DIFF_OPT_SET(options, RELATIVE_NAME);
2538                 options->prefix = arg + 11;
2539         }
2540
2541         /* xdiff options */
2542         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2543                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2544         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2545                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2546         else if (!strcmp(arg, "--ignore-space-at-eol"))
2547                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2548
2549         /* flags options */
2550         else if (!strcmp(arg, "--binary")) {
2551                 options->output_format |= DIFF_FORMAT_PATCH;
2552                 DIFF_OPT_SET(options, BINARY);
2553         }
2554         else if (!strcmp(arg, "--full-index"))
2555                 DIFF_OPT_SET(options, FULL_INDEX);
2556         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2557                 DIFF_OPT_SET(options, TEXT);
2558         else if (!strcmp(arg, "-R"))
2559                 DIFF_OPT_SET(options, REVERSE_DIFF);
2560         else if (!strcmp(arg, "--find-copies-harder"))
2561                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2562         else if (!strcmp(arg, "--follow"))
2563                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2564         else if (!strcmp(arg, "--color"))
2565                 DIFF_OPT_SET(options, COLOR_DIFF);
2566         else if (!strcmp(arg, "--no-color"))
2567                 DIFF_OPT_CLR(options, COLOR_DIFF);
2568         else if (!strcmp(arg, "--color-words"))
2569                 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2570         else if (!strcmp(arg, "--exit-code"))
2571                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2572         else if (!strcmp(arg, "--quiet"))
2573                 DIFF_OPT_SET(options, QUIET);
2574         else if (!strcmp(arg, "--ext-diff"))
2575                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2576         else if (!strcmp(arg, "--no-ext-diff"))
2577                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2578         else if (!strcmp(arg, "--ignore-submodules"))
2579                 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2580
2581         /* misc options */
2582         else if (!strcmp(arg, "-z"))
2583                 options->line_termination = 0;
2584         else if (!prefixcmp(arg, "-l"))
2585                 options->rename_limit = strtoul(arg+2, NULL, 10);
2586         else if (!prefixcmp(arg, "-S"))
2587                 options->pickaxe = arg + 2;
2588         else if (!strcmp(arg, "--pickaxe-all"))
2589                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2590         else if (!strcmp(arg, "--pickaxe-regex"))
2591                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2592         else if (!prefixcmp(arg, "-O"))
2593                 options->orderfile = arg + 2;
2594         else if (!prefixcmp(arg, "--diff-filter="))
2595                 options->filter = arg + 14;
2596         else if (!strcmp(arg, "--abbrev"))
2597                 options->abbrev = DEFAULT_ABBREV;
2598         else if (!prefixcmp(arg, "--abbrev=")) {
2599                 options->abbrev = strtoul(arg + 9, NULL, 10);
2600                 if (options->abbrev < MINIMUM_ABBREV)
2601                         options->abbrev = MINIMUM_ABBREV;
2602                 else if (40 < options->abbrev)
2603                         options->abbrev = 40;
2604         }
2605         else if (!prefixcmp(arg, "--src-prefix="))
2606                 options->a_prefix = arg + 13;
2607         else if (!prefixcmp(arg, "--dst-prefix="))
2608                 options->b_prefix = arg + 13;
2609         else if (!strcmp(arg, "--no-prefix"))
2610                 options->a_prefix = options->b_prefix = "";
2611         else if (!prefixcmp(arg, "--output=")) {
2612                 options->file = fopen(arg + strlen("--output="), "w");
2613                 options->close_file = 1;
2614         } else
2615                 return 0;
2616         return 1;
2617 }
2618
2619 static int parse_num(const char **cp_p)
2620 {
2621         unsigned long num, scale;
2622         int ch, dot;
2623         const char *cp = *cp_p;
2624
2625         num = 0;
2626         scale = 1;
2627         dot = 0;
2628         for(;;) {
2629                 ch = *cp;
2630                 if ( !dot && ch == '.' ) {
2631                         scale = 1;
2632                         dot = 1;
2633                 } else if ( ch == '%' ) {
2634                         scale = dot ? scale*100 : 100;
2635                         cp++;   /* % is always at the end */
2636                         break;
2637                 } else if ( ch >= '0' && ch <= '9' ) {
2638                         if ( scale < 100000 ) {
2639                                 scale *= 10;
2640                                 num = (num*10) + (ch-'0');
2641                         }
2642                 } else {
2643                         break;
2644                 }
2645                 cp++;
2646         }
2647         *cp_p = cp;
2648
2649         /* user says num divided by scale and we say internally that
2650          * is MAX_SCORE * num / scale.
2651          */
2652         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2653 }
2654
2655 static int diff_scoreopt_parse(const char *opt)
2656 {
2657         int opt1, opt2, cmd;
2658
2659         if (*opt++ != '-')
2660                 return -1;
2661         cmd = *opt++;
2662         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2663                 return -1; /* that is not a -M, -C nor -B option */
2664
2665         opt1 = parse_num(&opt);
2666         if (cmd != 'B')
2667                 opt2 = 0;
2668         else {
2669                 if (*opt == 0)
2670                         opt2 = 0;
2671                 else if (*opt != '/')
2672                         return -1; /* we expect -B80/99 or -B80 */
2673                 else {
2674                         opt++;
2675                         opt2 = parse_num(&opt);
2676                 }
2677         }
2678         if (*opt != 0)
2679                 return -1;
2680         return opt1 | (opt2 << 16);
2681 }
2682
2683 struct diff_queue_struct diff_queued_diff;
2684
2685 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2686 {
2687         if (queue->alloc <= queue->nr) {
2688                 queue->alloc = alloc_nr(queue->alloc);
2689                 queue->queue = xrealloc(queue->queue,
2690                                         sizeof(dp) * queue->alloc);
2691         }
2692         queue->queue[queue->nr++] = dp;
2693 }
2694
2695 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2696                                  struct diff_filespec *one,
2697                                  struct diff_filespec *two)
2698 {
2699         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2700         dp->one = one;
2701         dp->two = two;
2702         if (queue)
2703                 diff_q(queue, dp);
2704         return dp;
2705 }
2706
2707 void diff_free_filepair(struct diff_filepair *p)
2708 {
2709         free_filespec(p->one);
2710         free_filespec(p->two);
2711         free(p);
2712 }
2713
2714 /* This is different from find_unique_abbrev() in that
2715  * it stuffs the result with dots for alignment.
2716  */
2717 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2718 {
2719         int abblen;
2720         const char *abbrev;
2721         if (len == 40)
2722                 return sha1_to_hex(sha1);
2723
2724         abbrev = find_unique_abbrev(sha1, len);
2725         abblen = strlen(abbrev);
2726         if (abblen < 37) {
2727                 static char hex[41];
2728                 if (len < abblen && abblen <= len + 2)
2729                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2730                 else
2731                         sprintf(hex, "%s...", abbrev);
2732                 return hex;
2733         }
2734         return sha1_to_hex(sha1);
2735 }
2736
2737 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2738 {
2739         int line_termination = opt->line_termination;
2740         int inter_name_termination = line_termination ? '\t' : '\0';
2741
2742         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2743                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2744                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
2745                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2746         }
2747         if (p->score) {
2748                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2749                         inter_name_termination);
2750         } else {
2751                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2752         }
2753
2754         if (p->status == DIFF_STATUS_COPIED ||
2755             p->status == DIFF_STATUS_RENAMED) {
2756                 const char *name_a, *name_b;
2757                 name_a = p->one->path;
2758                 name_b = p->two->path;
2759                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2760                 write_name_quoted(name_a, opt->file, inter_name_termination);
2761                 write_name_quoted(name_b, opt->file, line_termination);
2762         } else {
2763                 const char *name_a, *name_b;
2764                 name_a = p->one->mode ? p->one->path : p->two->path;
2765                 name_b = NULL;
2766                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2767                 write_name_quoted(name_a, opt->file, line_termination);
2768         }
2769 }
2770
2771 int diff_unmodified_pair(struct diff_filepair *p)
2772 {
2773         /* This function is written stricter than necessary to support
2774          * the currently implemented transformers, but the idea is to
2775          * let transformers to produce diff_filepairs any way they want,
2776          * and filter and clean them up here before producing the output.
2777          */
2778         struct diff_filespec *one = p->one, *two = p->two;
2779
2780         if (DIFF_PAIR_UNMERGED(p))
2781                 return 0; /* unmerged is interesting */
2782
2783         /* deletion, addition, mode or type change
2784          * and rename are all interesting.
2785          */
2786         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2787             DIFF_PAIR_MODE_CHANGED(p) ||
2788             strcmp(one->path, two->path))
2789                 return 0;
2790
2791         /* both are valid and point at the same path.  that is, we are
2792          * dealing with a change.
2793          */
2794         if (one->sha1_valid && two->sha1_valid &&
2795             !hashcmp(one->sha1, two->sha1))
2796                 return 1; /* no change */
2797         if (!one->sha1_valid && !two->sha1_valid)
2798                 return 1; /* both look at the same file on the filesystem. */
2799         return 0;
2800 }
2801
2802 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2803 {
2804         if (diff_unmodified_pair(p))
2805                 return;
2806
2807         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2808             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2809                 return; /* no tree diffs in patch format */
2810
2811         run_diff(p, o);
2812 }
2813
2814 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2815                             struct diffstat_t *diffstat)
2816 {
2817         if (diff_unmodified_pair(p))
2818                 return;
2819
2820         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2821             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2822                 return; /* no tree diffs in patch format */
2823
2824         run_diffstat(p, o, diffstat);
2825 }
2826
2827 static void diff_flush_checkdiff(struct diff_filepair *p,
2828                 struct diff_options *o)
2829 {
2830         if (diff_unmodified_pair(p))
2831                 return;
2832
2833         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2834             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2835                 return; /* no tree diffs in patch format */
2836
2837         run_checkdiff(p, o);
2838 }
2839
2840 int diff_queue_is_empty(void)
2841 {
2842         struct diff_queue_struct *q = &diff_queued_diff;
2843         int i;
2844         for (i = 0; i < q->nr; i++)
2845                 if (!diff_unmodified_pair(q->queue[i]))
2846                         return 0;
2847         return 1;
2848 }
2849
2850 #if DIFF_DEBUG
2851 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2852 {
2853         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2854                 x, one ? one : "",
2855                 s->path,
2856                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2857                 s->mode,
2858                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2859         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2860                 x, one ? one : "",
2861                 s->size, s->xfrm_flags);
2862 }
2863
2864 void diff_debug_filepair(const struct diff_filepair *p, int i)
2865 {
2866         diff_debug_filespec(p->one, i, "one");
2867         diff_debug_filespec(p->two, i, "two");
2868         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2869                 p->score, p->status ? p->status : '?',
2870                 p->one->rename_used, p->broken_pair);
2871 }
2872
2873 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2874 {
2875         int i;
2876         if (msg)
2877                 fprintf(stderr, "%s\n", msg);
2878         fprintf(stderr, "q->nr = %d\n", q->nr);
2879         for (i = 0; i < q->nr; i++) {
2880                 struct diff_filepair *p = q->queue[i];
2881                 diff_debug_filepair(p, i);
2882         }
2883 }
2884 #endif
2885
2886 static void diff_resolve_rename_copy(void)
2887 {
2888         int i;
2889         struct diff_filepair *p;
2890         struct diff_queue_struct *q = &diff_queued_diff;
2891
2892         diff_debug_queue("resolve-rename-copy", q);
2893
2894         for (i = 0; i < q->nr; i++) {
2895                 p = q->queue[i];
2896                 p->status = 0; /* undecided */
2897                 if (DIFF_PAIR_UNMERGED(p))
2898                         p->status = DIFF_STATUS_UNMERGED;
2899                 else if (!DIFF_FILE_VALID(p->one))
2900                         p->status = DIFF_STATUS_ADDED;
2901                 else if (!DIFF_FILE_VALID(p->two))
2902                         p->status = DIFF_STATUS_DELETED;
2903                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2904                         p->status = DIFF_STATUS_TYPE_CHANGED;
2905
2906                 /* from this point on, we are dealing with a pair
2907                  * whose both sides are valid and of the same type, i.e.
2908                  * either in-place edit or rename/copy edit.
2909                  */
2910                 else if (DIFF_PAIR_RENAME(p)) {
2911                         /*
2912                          * A rename might have re-connected a broken
2913                          * pair up, causing the pathnames to be the
2914                          * same again. If so, that's not a rename at
2915                          * all, just a modification..
2916                          *
2917                          * Otherwise, see if this source was used for
2918                          * multiple renames, in which case we decrement
2919                          * the count, and call it a copy.
2920                          */
2921                         if (!strcmp(p->one->path, p->two->path))
2922                                 p->status = DIFF_STATUS_MODIFIED;
2923                         else if (--p->one->rename_used > 0)
2924                                 p->status = DIFF_STATUS_COPIED;
2925                         else
2926                                 p->status = DIFF_STATUS_RENAMED;
2927                 }
2928                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2929                          p->one->mode != p->two->mode ||
2930                          is_null_sha1(p->one->sha1))
2931                         p->status = DIFF_STATUS_MODIFIED;
2932                 else {
2933                         /* This is a "no-change" entry and should not
2934                          * happen anymore, but prepare for broken callers.
2935                          */
2936                         error("feeding unmodified %s to diffcore",
2937                               p->one->path);
2938                         p->status = DIFF_STATUS_UNKNOWN;
2939                 }
2940         }
2941         diff_debug_queue("resolve-rename-copy done", q);
2942 }
2943
2944 static int check_pair_status(struct diff_filepair *p)
2945 {
2946         switch (p->status) {
2947         case DIFF_STATUS_UNKNOWN:
2948                 return 0;
2949         case 0:
2950                 die("internal error in diff-resolve-rename-copy");
2951         default:
2952                 return 1;
2953         }
2954 }
2955
2956 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2957 {
2958         int fmt = opt->output_format;
2959
2960         if (fmt & DIFF_FORMAT_CHECKDIFF)
2961                 diff_flush_checkdiff(p, opt);
2962         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2963                 diff_flush_raw(p, opt);
2964         else if (fmt & DIFF_FORMAT_NAME) {
2965                 const char *name_a, *name_b;
2966                 name_a = p->two->path;
2967                 name_b = NULL;
2968                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2969                 write_name_quoted(name_a, opt->file, opt->line_termination);
2970         }
2971 }
2972
2973 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2974 {
2975         if (fs->mode)
2976                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2977         else
2978                 fprintf(file, " %s ", newdelete);
2979         write_name_quoted(fs->path, file, '\n');
2980 }
2981
2982
2983 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
2984 {
2985         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2986                 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
2987                         show_name ? ' ' : '\n');
2988                 if (show_name) {
2989                         write_name_quoted(p->two->path, file, '\n');
2990                 }
2991         }
2992 }
2993
2994 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
2995 {
2996         char *names = pprint_rename(p->one->path, p->two->path);
2997
2998         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2999         free(names);
3000         show_mode_change(file, p, 0);
3001 }
3002
3003 static void diff_summary(FILE *file, struct diff_filepair *p)
3004 {
3005         switch(p->status) {
3006         case DIFF_STATUS_DELETED:
3007                 show_file_mode_name(file, "delete", p->one);
3008                 break;
3009         case DIFF_STATUS_ADDED:
3010                 show_file_mode_name(file, "create", p->two);
3011                 break;
3012         case DIFF_STATUS_COPIED:
3013                 show_rename_copy(file, "copy", p);
3014                 break;
3015         case DIFF_STATUS_RENAMED:
3016                 show_rename_copy(file, "rename", p);
3017                 break;
3018         default:
3019                 if (p->score) {
3020                         fputs(" rewrite ", file);
3021                         write_name_quoted(p->two->path, file, ' ');
3022                         fprintf(file, "(%d%%)\n", similarity_index(p));
3023                 }
3024                 show_mode_change(file, p, !p->score);
3025                 break;
3026         }
3027 }
3028
3029 struct patch_id_t {
3030         struct xdiff_emit_state xm;
3031         SHA_CTX *ctx;
3032         int patchlen;
3033 };
3034
3035 static int remove_space(char *line, int len)
3036 {
3037         int i;
3038         char *dst = line;
3039         unsigned char c;
3040
3041         for (i = 0; i < len; i++)
3042                 if (!isspace((c = line[i])))
3043                         *dst++ = c;
3044
3045         return dst - line;
3046 }
3047
3048 static void patch_id_consume(void *priv, char *line, unsigned long len)
3049 {
3050         struct patch_id_t *data = priv;
3051         int new_len;
3052
3053         /* Ignore line numbers when computing the SHA1 of the patch */
3054         if (!prefixcmp(line, "@@ -"))
3055                 return;
3056
3057         new_len = remove_space(line, len);
3058
3059         SHA1_Update(data->ctx, line, new_len);
3060         data->patchlen += new_len;
3061 }
3062
3063 /* returns 0 upon success, and writes result into sha1 */
3064 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3065 {
3066         struct diff_queue_struct *q = &diff_queued_diff;
3067         int i;
3068         SHA_CTX ctx;
3069         struct patch_id_t data;
3070         char buffer[PATH_MAX * 4 + 20];
3071
3072         SHA1_Init(&ctx);
3073         memset(&data, 0, sizeof(struct patch_id_t));
3074         data.ctx = &ctx;
3075         data.xm.consume = patch_id_consume;
3076
3077         for (i = 0; i < q->nr; i++) {
3078                 xpparam_t xpp;
3079                 xdemitconf_t xecfg;
3080                 xdemitcb_t ecb;
3081                 mmfile_t mf1, mf2;
3082                 struct diff_filepair *p = q->queue[i];
3083                 int len1, len2;
3084
3085                 memset(&xecfg, 0, sizeof(xecfg));
3086                 if (p->status == 0)
3087                         return error("internal diff status error");
3088                 if (p->status == DIFF_STATUS_UNKNOWN)
3089                         continue;
3090                 if (diff_unmodified_pair(p))
3091                         continue;
3092                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3093                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3094                         continue;
3095                 if (DIFF_PAIR_UNMERGED(p))
3096                         continue;
3097
3098                 diff_fill_sha1_info(p->one);
3099                 diff_fill_sha1_info(p->two);
3100                 if (fill_mmfile(&mf1, p->one) < 0 ||
3101                                 fill_mmfile(&mf2, p->two) < 0)
3102                         return error("unable to read files to diff");
3103
3104                 len1 = remove_space(p->one->path, strlen(p->one->path));
3105                 len2 = remove_space(p->two->path, strlen(p->two->path));
3106                 if (p->one->mode == 0)
3107                         len1 = snprintf(buffer, sizeof(buffer),
3108                                         "diff--gita/%.*sb/%.*s"
3109                                         "newfilemode%06o"
3110                                         "---/dev/null"
3111                                         "+++b/%.*s",
3112                                         len1, p->one->path,
3113                                         len2, p->two->path,
3114                                         p->two->mode,
3115                                         len2, p->two->path);
3116                 else if (p->two->mode == 0)
3117                         len1 = snprintf(buffer, sizeof(buffer),
3118                                         "diff--gita/%.*sb/%.*s"
3119                                         "deletedfilemode%06o"
3120                                         "---a/%.*s"
3121                                         "+++/dev/null",
3122                                         len1, p->one->path,
3123                                         len2, p->two->path,
3124                                         p->one->mode,
3125                                         len1, p->one->path);
3126                 else
3127                         len1 = snprintf(buffer, sizeof(buffer),
3128                                         "diff--gita/%.*sb/%.*s"
3129                                         "---a/%.*s"
3130                                         "+++b/%.*s",
3131                                         len1, p->one->path,
3132                                         len2, p->two->path,
3133                                         len1, p->one->path,
3134                                         len2, p->two->path);
3135                 SHA1_Update(&ctx, buffer, len1);
3136
3137                 xpp.flags = XDF_NEED_MINIMAL;
3138                 xecfg.ctxlen = 3;
3139                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3140                 ecb.outf = xdiff_outf;
3141                 ecb.priv = &data;
3142                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
3143         }
3144
3145         SHA1_Final(sha1, &ctx);
3146         return 0;
3147 }
3148
3149 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3150 {
3151         struct diff_queue_struct *q = &diff_queued_diff;
3152         int i;
3153         int result = diff_get_patch_id(options, sha1);
3154
3155         for (i = 0; i < q->nr; i++)
3156                 diff_free_filepair(q->queue[i]);
3157
3158         free(q->queue);
3159         q->queue = NULL;
3160         q->nr = q->alloc = 0;
3161
3162         return result;
3163 }
3164
3165 static int is_summary_empty(const struct diff_queue_struct *q)
3166 {
3167         int i;
3168
3169         for (i = 0; i < q->nr; i++) {
3170                 const struct diff_filepair *p = q->queue[i];
3171
3172                 switch (p->status) {
3173                 case DIFF_STATUS_DELETED:
3174                 case DIFF_STATUS_ADDED:
3175                 case DIFF_STATUS_COPIED:
3176                 case DIFF_STATUS_RENAMED:
3177                         return 0;
3178                 default:
3179                         if (p->score)
3180                                 return 0;
3181                         if (p->one->mode && p->two->mode &&
3182                             p->one->mode != p->two->mode)
3183                                 return 0;
3184                         break;
3185                 }
3186         }
3187         return 1;
3188 }
3189
3190 void diff_flush(struct diff_options *options)
3191 {
3192         struct diff_queue_struct *q = &diff_queued_diff;
3193         int i, output_format = options->output_format;
3194         int separator = 0;
3195
3196         /*
3197          * Order: raw, stat, summary, patch
3198          * or:    name/name-status/checkdiff (other bits clear)
3199          */
3200         if (!q->nr)
3201                 goto free_queue;
3202
3203         if (output_format & (DIFF_FORMAT_RAW |
3204                              DIFF_FORMAT_NAME |
3205                              DIFF_FORMAT_NAME_STATUS |
3206                              DIFF_FORMAT_CHECKDIFF)) {
3207                 for (i = 0; i < q->nr; i++) {
3208                         struct diff_filepair *p = q->queue[i];
3209                         if (check_pair_status(p))
3210                                 flush_one_pair(p, options);
3211                 }
3212                 separator++;
3213         }
3214
3215         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3216                 struct diffstat_t diffstat;
3217
3218                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3219                 diffstat.xm.consume = diffstat_consume;
3220                 for (i = 0; i < q->nr; i++) {
3221                         struct diff_filepair *p = q->queue[i];
3222                         if (check_pair_status(p))
3223                                 diff_flush_stat(p, options, &diffstat);
3224                 }
3225                 if (output_format & DIFF_FORMAT_NUMSTAT)
3226                         show_numstat(&diffstat, options);
3227                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3228                         show_stats(&diffstat, options);
3229                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3230                         show_shortstats(&diffstat, options);
3231                 free_diffstat_info(&diffstat);
3232                 separator++;
3233         }
3234         if (output_format & DIFF_FORMAT_DIRSTAT)
3235                 show_dirstat(options);
3236
3237         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3238                 for (i = 0; i < q->nr; i++)
3239                         diff_summary(options->file, q->queue[i]);
3240                 separator++;
3241         }
3242
3243         if (output_format & DIFF_FORMAT_PATCH) {
3244                 if (separator) {
3245                         putc(options->line_termination, options->file);
3246                         if (options->stat_sep) {
3247                                 /* attach patch instead of inline */
3248                                 fputs(options->stat_sep, options->file);
3249                         }
3250                 }
3251
3252                 for (i = 0; i < q->nr; i++) {
3253                         struct diff_filepair *p = q->queue[i];
3254                         if (check_pair_status(p))
3255                                 diff_flush_patch(p, options);
3256                 }
3257         }
3258
3259         if (output_format & DIFF_FORMAT_CALLBACK)
3260                 options->format_callback(q, options, options->format_callback_data);
3261
3262         for (i = 0; i < q->nr; i++)
3263                 diff_free_filepair(q->queue[i]);
3264 free_queue:
3265         free(q->queue);
3266         q->queue = NULL;
3267         q->nr = q->alloc = 0;
3268         if (options->close_file)
3269                 fclose(options->file);
3270 }
3271
3272 static void diffcore_apply_filter(const char *filter)
3273 {
3274         int i;
3275         struct diff_queue_struct *q = &diff_queued_diff;
3276         struct diff_queue_struct outq;
3277         outq.queue = NULL;
3278         outq.nr = outq.alloc = 0;
3279
3280         if (!filter)
3281                 return;
3282
3283         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3284                 int found;
3285                 for (i = found = 0; !found && i < q->nr; i++) {
3286                         struct diff_filepair *p = q->queue[i];
3287                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3288                              ((p->score &&
3289                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3290                               (!p->score &&
3291                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3292                             ((p->status != DIFF_STATUS_MODIFIED) &&
3293                              strchr(filter, p->status)))
3294                                 found++;
3295                 }
3296                 if (found)
3297                         return;
3298
3299                 /* otherwise we will clear the whole queue
3300                  * by copying the empty outq at the end of this
3301                  * function, but first clear the current entries
3302                  * in the queue.
3303                  */
3304                 for (i = 0; i < q->nr; i++)
3305                         diff_free_filepair(q->queue[i]);
3306         }
3307         else {
3308                 /* Only the matching ones */
3309                 for (i = 0; i < q->nr; i++) {
3310                         struct diff_filepair *p = q->queue[i];
3311
3312                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3313                              ((p->score &&
3314                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3315                               (!p->score &&
3316                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3317                             ((p->status != DIFF_STATUS_MODIFIED) &&
3318                              strchr(filter, p->status)))
3319                                 diff_q(&outq, p);
3320                         else
3321                                 diff_free_filepair(p);
3322                 }
3323         }
3324         free(q->queue);
3325         *q = outq;
3326 }
3327
3328 /* Check whether two filespecs with the same mode and size are identical */
3329 static int diff_filespec_is_identical(struct diff_filespec *one,
3330                                       struct diff_filespec *two)
3331 {
3332         if (S_ISGITLINK(one->mode))
3333                 return 0;
3334         if (diff_populate_filespec(one, 0))
3335                 return 0;
3336         if (diff_populate_filespec(two, 0))
3337                 return 0;
3338         return !memcmp(one->data, two->data, one->size);
3339 }
3340
3341 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3342 {
3343         int i;
3344         struct diff_queue_struct *q = &diff_queued_diff;
3345         struct diff_queue_struct outq;
3346         outq.queue = NULL;
3347         outq.nr = outq.alloc = 0;
3348
3349         for (i = 0; i < q->nr; i++) {
3350                 struct diff_filepair *p = q->queue[i];
3351
3352                 /*
3353                  * 1. Entries that come from stat info dirtyness
3354                  *    always have both sides (iow, not create/delete),
3355                  *    one side of the object name is unknown, with
3356                  *    the same mode and size.  Keep the ones that
3357                  *    do not match these criteria.  They have real
3358                  *    differences.
3359                  *
3360                  * 2. At this point, the file is known to be modified,
3361                  *    with the same mode and size, and the object
3362                  *    name of one side is unknown.  Need to inspect
3363                  *    the identical contents.
3364                  */
3365                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3366                     !DIFF_FILE_VALID(p->two) ||
3367                     (p->one->sha1_valid && p->two->sha1_valid) ||
3368                     (p->one->mode != p->two->mode) ||
3369                     diff_populate_filespec(p->one, 1) ||
3370                     diff_populate_filespec(p->two, 1) ||
3371                     (p->one->size != p->two->size) ||
3372                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3373                         diff_q(&outq, p);
3374                 else {
3375                         /*
3376                          * The caller can subtract 1 from skip_stat_unmatch
3377                          * to determine how many paths were dirty only
3378                          * due to stat info mismatch.
3379                          */
3380                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3381                                 diffopt->skip_stat_unmatch++;
3382                         diff_free_filepair(p);
3383                 }
3384         }
3385         free(q->queue);
3386         *q = outq;
3387 }
3388
3389 void diffcore_std(struct diff_options *options)
3390 {
3391         if (DIFF_OPT_TST(options, QUIET))
3392                 return;
3393
3394         if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3395                 diffcore_skip_stat_unmatch(options);
3396         if (options->break_opt != -1)
3397                 diffcore_break(options->break_opt);
3398         if (options->detect_rename)
3399                 diffcore_rename(options);
3400         if (options->break_opt != -1)
3401                 diffcore_merge_broken();
3402         if (options->pickaxe)
3403                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3404         if (options->orderfile)
3405                 diffcore_order(options->orderfile);
3406         diff_resolve_rename_copy();
3407         diffcore_apply_filter(options->filter);
3408
3409         if (diff_queued_diff.nr)
3410                 DIFF_OPT_SET(options, HAS_CHANGES);
3411         else
3412                 DIFF_OPT_CLR(options, HAS_CHANGES);
3413 }
3414
3415 int diff_result_code(struct diff_options *opt, int status)
3416 {
3417         int result = 0;
3418         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3419             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3420                 return status;
3421         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3422             DIFF_OPT_TST(opt, HAS_CHANGES))
3423                 result |= 01;
3424         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3425             DIFF_OPT_TST(opt, CHECK_FAILED))
3426                 result |= 02;
3427         return result;
3428 }
3429
3430 void diff_addremove(struct diff_options *options,
3431                     int addremove, unsigned mode,
3432                     const unsigned char *sha1,
3433                     const char *concatpath)
3434 {
3435         struct diff_filespec *one, *two;
3436
3437         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3438                 return;
3439
3440         /* This may look odd, but it is a preparation for
3441          * feeding "there are unchanged files which should
3442          * not produce diffs, but when you are doing copy
3443          * detection you would need them, so here they are"
3444          * entries to the diff-core.  They will be prefixed
3445          * with something like '=' or '*' (I haven't decided
3446          * which but should not make any difference).
3447          * Feeding the same new and old to diff_change()
3448          * also has the same effect.
3449          * Before the final output happens, they are pruned after
3450          * merged into rename/copy pairs as appropriate.
3451          */
3452         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3453                 addremove = (addremove == '+' ? '-' :
3454                              addremove == '-' ? '+' : addremove);
3455
3456         if (options->prefix &&
3457             strncmp(concatpath, options->prefix, options->prefix_length))
3458                 return;
3459
3460         one = alloc_filespec(concatpath);
3461         two = alloc_filespec(concatpath);
3462
3463         if (addremove != '+')
3464                 fill_filespec(one, sha1, mode);
3465         if (addremove != '-')
3466                 fill_filespec(two, sha1, mode);
3467
3468         diff_queue(&diff_queued_diff, one, two);
3469         DIFF_OPT_SET(options, HAS_CHANGES);
3470 }
3471
3472 void diff_change(struct diff_options *options,
3473                  unsigned old_mode, unsigned new_mode,
3474                  const unsigned char *old_sha1,
3475                  const unsigned char *new_sha1,
3476                  const char *concatpath)
3477 {
3478         struct diff_filespec *one, *two;
3479
3480         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3481                         && S_ISGITLINK(new_mode))
3482                 return;
3483
3484         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3485                 unsigned tmp;
3486                 const unsigned char *tmp_c;
3487                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3488                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3489         }
3490
3491         if (options->prefix &&
3492             strncmp(concatpath, options->prefix, options->prefix_length))
3493                 return;
3494
3495         one = alloc_filespec(concatpath);
3496         two = alloc_filespec(concatpath);
3497         fill_filespec(one, old_sha1, old_mode);
3498         fill_filespec(two, new_sha1, new_mode);
3499
3500         diff_queue(&diff_queued_diff, one, two);
3501         DIFF_OPT_SET(options, HAS_CHANGES);
3502 }
3503
3504 void diff_unmerge(struct diff_options *options,
3505                   const char *path,
3506                   unsigned mode, const unsigned char *sha1)
3507 {
3508         struct diff_filespec *one, *two;
3509
3510         if (options->prefix &&
3511             strncmp(path, options->prefix, options->prefix_length))
3512                 return;
3513
3514         one = alloc_filespec(path);
3515         two = alloc_filespec(path);
3516         fill_filespec(one, sha1, mode);
3517         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3518 }