git.oblomov.eu Git - git/blob - diff.c

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "quote.h"
   6 #include "diff.h"
   7 #include "diffcore.h"
   8 #include "delta.h"
   9 #include "xdiff-interface.h"
  10 #include "color.h"
  11 #include "attr.h"
  12 #include "run-command.h"
  13 #include "utf8.h"
  14
  15 #ifdef NO_FAST_WORKING_DIRECTORY
  16 #define FAST_WORKING_DIRECTORY 0
  17 #else
  18 #define FAST_WORKING_DIRECTORY 1
  19 #endif
  20
  21 static int diff_detect_rename_default;
  22 static int diff_rename_limit_default = 200;
  23 static int diff_suppress_blank_empty;
  24 int diff_use_color_default = -1;
  25 static const char *external_diff_cmd_cfg;
  26 int diff_auto_refresh_index = 1;
  27
  28 static char diff_colors[][COLOR_MAXLEN] = {
  29         "\033[m",       /* reset */
  30         "",             /* PLAIN (normal) */
  31         "\033[1m",      /* METAINFO (bold) */
  32         "\033[36m",     /* FRAGINFO (cyan) */
  33         "\033[31m",     /* OLD (red) */
  34         "\033[32m",     /* NEW (green) */
  35         "\033[33m",     /* COMMIT (yellow) */
  36         "\033[41m",     /* WHITESPACE (red background) */
  37 };
  38
  39 static int parse_diff_color_slot(const char *var, int ofs)
  40 {
  41         if (!strcasecmp(var+ofs, "plain"))
  42                 return DIFF_PLAIN;
  43         if (!strcasecmp(var+ofs, "meta"))
  44                 return DIFF_METAINFO;
  45         if (!strcasecmp(var+ofs, "frag"))
  46                 return DIFF_FRAGINFO;
  47         if (!strcasecmp(var+ofs, "old"))
  48                 return DIFF_FILE_OLD;
  49         if (!strcasecmp(var+ofs, "new"))
  50                 return DIFF_FILE_NEW;
  51         if (!strcasecmp(var+ofs, "commit"))
  52                 return DIFF_COMMIT;
  53         if (!strcasecmp(var+ofs, "whitespace"))
  54                 return DIFF_WHITESPACE;
  55         die("bad config variable '%s'", var);
  56 }
  57
  58 static struct ll_diff_driver {
  59         const char *name;
  60         struct ll_diff_driver *next;
  61         const char *cmd;
  62 } *user_diff, **user_diff_tail;
  63
  64 /*
  65  * Currently there is only "diff.<drivername>.command" variable;
  66  * because there are "diff.color.<slot>" variables, we are parsing
  67  * this in a bit convoluted way to allow low level diff driver
  68  * called "color".
  69  */
  70 static int parse_lldiff_command(const char *var, const char *ep, const char *value)
  71 {
  72         const char *name;
  73         int namelen;
  74         struct ll_diff_driver *drv;
  75
  76         name = var + 5;
  77         namelen = ep - name;
  78         for (drv = user_diff; drv; drv = drv->next)
  79                 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
  80                         break;
  81         if (!drv) {
  82                 drv = xcalloc(1, sizeof(struct ll_diff_driver));
  83                 drv->name = xmemdupz(name, namelen);
  84                 if (!user_diff_tail)
  85                         user_diff_tail = &user_diff;
  86                 *user_diff_tail = drv;
  87                 user_diff_tail = &(drv->next);
  88         }
  89
  90         return git_config_string(&(drv->cmd), var, value);
  91 }
  92
  93 /*
  94  * 'diff.<what>.funcname' attribute can be specified in the configuration
  95  * to define a customized regexp to find the beginning of a function to
  96  * be used for hunk header lines of "diff -p" style output.
  97  */
  98 static struct funcname_pattern {
  99         char *name;
 100         char *pattern;
 101         struct funcname_pattern *next;
 102 } *funcname_pattern_list;
 103
 104 static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
 105 {
 106         const char *name;
 107         int namelen;
 108         struct funcname_pattern *pp;
 109
 110         name = var + 5; /* "diff." */
 111         namelen = ep - name;
 112
 113         for (pp = funcname_pattern_list; pp; pp = pp->next)
 114                 if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
 115                         break;
 116         if (!pp) {
 117                 pp = xcalloc(1, sizeof(*pp));
 118                 pp->name = xmemdupz(name, namelen);
 119                 pp->next = funcname_pattern_list;
 120                 funcname_pattern_list = pp;
 121         }
 122         free(pp->pattern);
 123         pp->pattern = xstrdup(value);
 124         return 0;
 125 }
 126
 127 /*
 128  * These are to give UI layer defaults.
 129  * The core-level commands such as git-diff-files should
 130  * never be affected by the setting of diff.renames
 131  * the user happens to have in the configuration file.
 132  */
 133 int git_diff_ui_config(const char *var, const char *value, void *cb)
 134 {
 135         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 136                 diff_use_color_default = git_config_colorbool(var, value, -1);
 137                 return 0;
 138         }
 139         if (!strcmp(var, "diff.renames")) {
 140                 if (!value)
 141                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 142                 else if (!strcasecmp(value, "copies") ||
 143                          !strcasecmp(value, "copy"))
 144                         diff_detect_rename_default = DIFF_DETECT_COPY;
 145                 else if (git_config_bool(var,value))
 146                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 147                 return 0;
 148         }
 149         if (!strcmp(var, "diff.autorefreshindex")) {
 150                 diff_auto_refresh_index = git_config_bool(var, value);
 151                 return 0;
 152         }
 153         if (!strcmp(var, "diff.external"))
 154                 return git_config_string(&external_diff_cmd_cfg, var, value);
 155         if (!prefixcmp(var, "diff.")) {
 156                 const char *ep = strrchr(var, '.');
 157
 158                 if (ep != var + 4 && !strcmp(ep, ".command"))
 159                         return parse_lldiff_command(var, ep, value);
 160         }
 161
 162         return git_diff_basic_config(var, value, cb);
 163 }
 164
 165 int git_diff_basic_config(const char *var, const char *value, void *cb)
 166 {
 167         if (!strcmp(var, "diff.renamelimit")) {
 168                 diff_rename_limit_default = git_config_int(var, value);
 169                 return 0;
 170         }
 171
 172         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 173                 int slot = parse_diff_color_slot(var, 11);
 174                 if (!value)
 175                         return config_error_nonbool(var);
 176                 color_parse(value, var, diff_colors[slot]);
 177                 return 0;
 178         }
 179
 180         /* like GNU diff's --suppress-blank-empty option  */
 181         if (!strcmp(var, "diff.suppress-blank-empty")) {
 182                 diff_suppress_blank_empty = git_config_bool(var, value);
 183                 return 0;
 184         }
 185
 186         if (!prefixcmp(var, "diff.")) {
 187                 const char *ep = strrchr(var, '.');
 188                 if (ep != var + 4) {
 189                         if (!strcmp(ep, ".funcname")) {
 190                                 if (!value)
 191                                         return config_error_nonbool(var);
 192                                 return parse_funcname_pattern(var, ep, value);
 193                         }
 194                 }
 195         }
 196
 197         return git_color_default_config(var, value, cb);
 198 }
 199
 200 static char *quote_two(const char *one, const char *two)
 201 {
 202         int need_one = quote_c_style(one, NULL, NULL, 1);
 203         int need_two = quote_c_style(two, NULL, NULL, 1);
 204         struct strbuf res;
 205
 206         strbuf_init(&res, 0);
 207         if (need_one + need_two) {
 208                 strbuf_addch(&res, '"');
 209                 quote_c_style(one, &res, NULL, 1);
 210                 quote_c_style(two, &res, NULL, 1);
 211                 strbuf_addch(&res, '"');
 212         } else {
 213                 strbuf_addstr(&res, one);
 214                 strbuf_addstr(&res, two);
 215         }
 216         return strbuf_detach(&res, NULL);
 217 }
 218
 219 static const char *external_diff(void)
 220 {
 221         static const char *external_diff_cmd = NULL;
 222         static int done_preparing = 0;
 223
 224         if (done_preparing)
 225                 return external_diff_cmd;
 226         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 227         if (!external_diff_cmd)
 228                 external_diff_cmd = external_diff_cmd_cfg;
 229         done_preparing = 1;
 230         return external_diff_cmd;
 231 }
 232
 233 static struct diff_tempfile {
 234         const char *name; /* filename external diff should read from */
 235         char hex[41];
 236         char mode[10];
 237         char tmp_path[PATH_MAX];
 238 } diff_temp[2];
 239
 240 static int count_lines(const char *data, int size)
 241 {
 242         int count, ch, completely_empty = 1, nl_just_seen = 0;
 243         count = 0;
 244         while (0 < size--) {
 245                 ch = *data++;
 246                 if (ch == '\n') {
 247                         count++;
 248                         nl_just_seen = 1;
 249                         completely_empty = 0;
 250                 }
 251                 else {
 252                         nl_just_seen = 0;
 253                         completely_empty = 0;
 254                 }
 255         }
 256         if (completely_empty)
 257                 return 0;
 258         if (!nl_just_seen)
 259                 count++; /* no trailing newline */
 260         return count;
 261 }
 262
 263 static void print_line_count(FILE *file, int count)
 264 {
 265         switch (count) {
 266         case 0:
 267                 fprintf(file, "0,0");
 268                 break;
 269         case 1:
 270                 fprintf(file, "1");
 271                 break;
 272         default:
 273                 fprintf(file, "1,%d", count);
 274                 break;
 275         }
 276 }
 277
 278 static void copy_file_with_prefix(FILE *file,
 279                                   int prefix, const char *data, int size,
 280                                   const char *set, const char *reset)
 281 {
 282         int ch, nl_just_seen = 1;
 283         while (0 < size--) {
 284                 ch = *data++;
 285                 if (nl_just_seen) {
 286                         fputs(set, file);
 287                         putc(prefix, file);
 288                 }
 289                 if (ch == '\n') {
 290                         nl_just_seen = 1;
 291                         fputs(reset, file);
 292                 } else
 293                         nl_just_seen = 0;
 294                 putc(ch, file);
 295         }
 296         if (!nl_just_seen)
 297                 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
 298 }
 299
 300 static void emit_rewrite_diff(const char *name_a,
 301                               const char *name_b,
 302                               struct diff_filespec *one,
 303                               struct diff_filespec *two,
 304                               struct diff_options *o)
 305 {
 306         int lc_a, lc_b;
 307         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
 308         const char *name_a_tab, *name_b_tab;
 309         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 310         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 311         const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
 312         const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
 313         const char *reset = diff_get_color(color_diff, DIFF_RESET);
 314         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 315
 316         name_a += (*name_a == '/');
 317         name_b += (*name_b == '/');
 318         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 319         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 320
 321         strbuf_reset(&a_name);
 322         strbuf_reset(&b_name);
 323         quote_two_c_style(&a_name, o->a_prefix, name_a, 0);
 324         quote_two_c_style(&b_name, o->b_prefix, name_b, 0);
 325
 326         diff_populate_filespec(one, 0);
 327         diff_populate_filespec(two, 0);
 328         lc_a = count_lines(one->data, one->size);
 329         lc_b = count_lines(two->data, two->size);
 330         fprintf(o->file,
 331                 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
 332                 metainfo, a_name.buf, name_a_tab, reset,
 333                 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
 334         print_line_count(o->file, lc_a);
 335         fprintf(o->file, " +");
 336         print_line_count(o->file, lc_b);
 337         fprintf(o->file, " @@%s\n", reset);
 338         if (lc_a)
 339                 copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
 340         if (lc_b)
 341                 copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
 342 }
 343
 344 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 345 {
 346         if (!DIFF_FILE_VALID(one)) {
 347                 mf->ptr = (char *)""; /* does not matter */
 348                 mf->size = 0;
 349                 return 0;
 350         }
 351         else if (diff_populate_filespec(one, 0))
 352                 return -1;
 353         mf->ptr = one->data;
 354         mf->size = one->size;
 355         return 0;
 356 }
 357
 358 struct diff_words_buffer {
 359         mmfile_t text;
 360         long alloc;
 361         long current; /* output pointer */
 362         int suppressed_newline;
 363 };
 364
 365 static void diff_words_append(char *line, unsigned long len,
 366                 struct diff_words_buffer *buffer)
 367 {
 368         if (buffer->text.size + len > buffer->alloc) {
 369                 buffer->alloc = (buffer->text.size + len) * 3 / 2;
 370                 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
 371         }
 372         line++;
 373         len--;
 374         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 375         buffer->text.size += len;
 376 }
 377
 378 struct diff_words_data {
 379         struct diff_words_buffer minus, plus;
 380         FILE *file;
 381 };
 382
 383 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
 384                 int suppress_newline)
 385 {
 386         const char *ptr;
 387         int eol = 0;
 388
 389         if (len == 0)
 390                 return;
 391
 392         ptr  = buffer->text.ptr + buffer->current;
 393         buffer->current += len;
 394
 395         if (ptr[len - 1] == '\n') {
 396                 eol = 1;
 397                 len--;
 398         }
 399
 400         fputs(diff_get_color(1, color), file);
 401         fwrite(ptr, len, 1, file);
 402         fputs(diff_get_color(1, DIFF_RESET), file);
 403
 404         if (eol) {
 405                 if (suppress_newline)
 406                         buffer->suppressed_newline = 1;
 407                 else
 408                         putc('\n', file);
 409         }
 410 }
 411
 412 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 413 {
 414         struct diff_words_data *diff_words = priv;
 415
 416         if (diff_words->minus.suppressed_newline) {
 417                 if (line[0] != '+')
 418                         putc('\n', diff_words->file);
 419                 diff_words->minus.suppressed_newline = 0;
 420         }
 421
 422         len--;
 423         switch (line[0]) {
 424                 case '-':
 425                         print_word(diff_words->file,
 426                                    &diff_words->minus, len, DIFF_FILE_OLD, 1);
 427                         break;
 428                 case '+':
 429                         print_word(diff_words->file,
 430                                    &diff_words->plus, len, DIFF_FILE_NEW, 0);
 431                         break;
 432                 case ' ':
 433                         print_word(diff_words->file,
 434                                    &diff_words->plus, len, DIFF_PLAIN, 0);
 435                         diff_words->minus.current += len;
 436                         break;
 437         }
 438 }
 439
 440 /* this executes the word diff on the accumulated buffers */
 441 static void diff_words_show(struct diff_words_data *diff_words)
 442 {
 443         xpparam_t xpp;
 444         xdemitconf_t xecfg;
 445         xdemitcb_t ecb;
 446         mmfile_t minus, plus;
 447         int i;
 448
 449         memset(&xecfg, 0, sizeof(xecfg));
 450         minus.size = diff_words->minus.text.size;
 451         minus.ptr = xmalloc(minus.size);
 452         memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
 453         for (i = 0; i < minus.size; i++)
 454                 if (isspace(minus.ptr[i]))
 455                         minus.ptr[i] = '\n';
 456         diff_words->minus.current = 0;
 457
 458         plus.size = diff_words->plus.text.size;
 459         plus.ptr = xmalloc(plus.size);
 460         memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
 461         for (i = 0; i < plus.size; i++)
 462                 if (isspace(plus.ptr[i]))
 463                         plus.ptr[i] = '\n';
 464         diff_words->plus.current = 0;
 465
 466         xpp.flags = XDF_NEED_MINIMAL;
 467         xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
 468         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
 469                       &xpp, &xecfg, &ecb);
 470         free(minus.ptr);
 471         free(plus.ptr);
 472         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 473
 474         if (diff_words->minus.suppressed_newline) {
 475                 putc('\n', diff_words->file);
 476                 diff_words->minus.suppressed_newline = 0;
 477         }
 478 }
 479
 480 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 481
 482 struct emit_callback {
 483         int nparents, color_diff;
 484         unsigned ws_rule;
 485         sane_truncate_fn truncate;
 486         const char **label_path;
 487         struct diff_words_data *diff_words;
 488         int *found_changesp;
 489         FILE *file;
 490 };
 491
 492 static void free_diff_words_data(struct emit_callback *ecbdata)
 493 {
 494         if (ecbdata->diff_words) {
 495                 /* flush buffers */
 496                 if (ecbdata->diff_words->minus.text.size ||
 497                                 ecbdata->diff_words->plus.text.size)
 498                         diff_words_show(ecbdata->diff_words);
 499
 500                 free (ecbdata->diff_words->minus.text.ptr);
 501                 free (ecbdata->diff_words->plus.text.ptr);
 502                 free(ecbdata->diff_words);
 503                 ecbdata->diff_words = NULL;
 504         }
 505 }
 506
 507 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 508 {
 509         if (diff_use_color)
 510                 return diff_colors[ix];
 511         return "";
 512 }
 513
 514 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
 515 {
 516         int has_trailing_newline, has_trailing_carriage_return;
 517
 518         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 519         if (has_trailing_newline)
 520                 len--;
 521         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 522         if (has_trailing_carriage_return)
 523                 len--;
 524
 525         fputs(set, file);
 526         fwrite(line, len, 1, file);
 527         fputs(reset, file);
 528         if (has_trailing_carriage_return)
 529                 fputc('\r', file);
 530         if (has_trailing_newline)
 531                 fputc('\n', file);
 532 }
 533
 534 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
 535 {
 536         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 537         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 538
 539         if (!*ws)
 540                 emit_line(ecbdata->file, set, reset, line, len);
 541         else {
 542                 /* Emit just the prefix, then the rest. */
 543                 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
 544                 ws_check_emit(line + ecbdata->nparents,
 545                               len - ecbdata->nparents, ecbdata->ws_rule,
 546                               ecbdata->file, set, reset, ws);
 547         }
 548 }
 549
 550 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
 551 {
 552         const char *cp;
 553         unsigned long allot;
 554         size_t l = len;
 555
 556         if (ecb->truncate)
 557                 return ecb->truncate(line, len);
 558         cp = line;
 559         allot = l;
 560         while (0 < l) {
 561                 (void) utf8_width(&cp, &l);
 562                 if (!cp)
 563                         break; /* truncated in the middle? */
 564         }
 565         return allot - l;
 566 }
 567
 568 static void fn_out_consume(void *priv, char *line, unsigned long len)
 569 {
 570         int i;
 571         int color;
 572         struct emit_callback *ecbdata = priv;
 573         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 574         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 575         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 576
 577         *(ecbdata->found_changesp) = 1;
 578
 579         if (ecbdata->label_path[0]) {
 580                 const char *name_a_tab, *name_b_tab;
 581
 582                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 583                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 584
 585                 fprintf(ecbdata->file, "%s--- %s%s%s\n",
 586                         meta, ecbdata->label_path[0], reset, name_a_tab);
 587                 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
 588                         meta, ecbdata->label_path[1], reset, name_b_tab);
 589                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 590         }
 591
 592         if (diff_suppress_blank_empty
 593             && len == 2 && line[0] == ' ' && line[1] == '\n') {
 594                 line[0] = '\n';
 595                 len = 1;
 596         }
 597
 598         /* This is not really necessary for now because
 599          * this codepath only deals with two-way diffs.
 600          */
 601         for (i = 0; i < len && line[i] == '@'; i++)
 602                 ;
 603         if (2 <= i && i < len && line[i] == ' ') {
 604                 ecbdata->nparents = i - 1;
 605                 len = sane_truncate_line(ecbdata, line, len);
 606                 emit_line(ecbdata->file,
 607                           diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
 608                           reset, line, len);
 609                 if (line[len-1] != '\n')
 610                         putc('\n', ecbdata->file);
 611                 return;
 612         }
 613
 614         if (len < ecbdata->nparents) {
 615                 emit_line(ecbdata->file, reset, reset, line, len);
 616                 return;
 617         }
 618
 619         color = DIFF_PLAIN;
 620         if (ecbdata->diff_words && ecbdata->nparents != 1)
 621                 /* fall back to normal diff */
 622                 free_diff_words_data(ecbdata);
 623         if (ecbdata->diff_words) {
 624                 if (line[0] == '-') {
 625                         diff_words_append(line, len,
 626                                           &ecbdata->diff_words->minus);
 627                         return;
 628                 } else if (line[0] == '+') {
 629                         diff_words_append(line, len,
 630                                           &ecbdata->diff_words->plus);
 631                         return;
 632                 }
 633                 if (ecbdata->diff_words->minus.text.size ||
 634                     ecbdata->diff_words->plus.text.size)
 635                         diff_words_show(ecbdata->diff_words);
 636                 line++;
 637                 len--;
 638                 emit_line(ecbdata->file, plain, reset, line, len);
 639                 return;
 640         }
 641         for (i = 0; i < ecbdata->nparents && len; i++) {
 642                 if (line[i] == '-')
 643                         color = DIFF_FILE_OLD;
 644                 else if (line[i] == '+')
 645                         color = DIFF_FILE_NEW;
 646         }
 647
 648         if (color != DIFF_FILE_NEW) {
 649                 emit_line(ecbdata->file,
 650                           diff_get_color(ecbdata->color_diff, color),
 651                           reset, line, len);
 652                 return;
 653         }
 654         emit_add_line(reset, ecbdata, line, len);
 655 }
 656
 657 static char *pprint_rename(const char *a, const char *b)
 658 {
 659         const char *old = a;
 660         const char *new = b;
 661         struct strbuf name;
 662         int pfx_length, sfx_length;
 663         int len_a = strlen(a);
 664         int len_b = strlen(b);
 665         int a_midlen, b_midlen;
 666         int qlen_a = quote_c_style(a, NULL, NULL, 0);
 667         int qlen_b = quote_c_style(b, NULL, NULL, 0);
 668
 669         strbuf_init(&name, 0);
 670         if (qlen_a || qlen_b) {
 671                 quote_c_style(a, &name, NULL, 0);
 672                 strbuf_addstr(&name, " => ");
 673                 quote_c_style(b, &name, NULL, 0);
 674                 return strbuf_detach(&name, NULL);
 675         }
 676
 677         /* Find common prefix */
 678         pfx_length = 0;
 679         while (*old && *new && *old == *new) {
 680                 if (*old == '/')
 681                         pfx_length = old - a + 1;
 682                 old++;
 683                 new++;
 684         }
 685
 686         /* Find common suffix */
 687         old = a + len_a;
 688         new = b + len_b;
 689         sfx_length = 0;
 690         while (a <= old && b <= new && *old == *new) {
 691                 if (*old == '/')
 692                         sfx_length = len_a - (old - a);
 693                 old--;
 694                 new--;
 695         }
 696
 697         /*
 698          * pfx{mid-a => mid-b}sfx
 699          * {pfx-a => pfx-b}sfx
 700          * pfx{sfx-a => sfx-b}
 701          * name-a => name-b
 702          */
 703         a_midlen = len_a - pfx_length - sfx_length;
 704         b_midlen = len_b - pfx_length - sfx_length;
 705         if (a_midlen < 0)
 706                 a_midlen = 0;
 707         if (b_midlen < 0)
 708                 b_midlen = 0;
 709
 710         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 711         if (pfx_length + sfx_length) {
 712                 strbuf_add(&name, a, pfx_length);
 713                 strbuf_addch(&name, '{');
 714         }
 715         strbuf_add(&name, a + pfx_length, a_midlen);
 716         strbuf_addstr(&name, " => ");
 717         strbuf_add(&name, b + pfx_length, b_midlen);
 718         if (pfx_length + sfx_length) {
 719                 strbuf_addch(&name, '}');
 720                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 721         }
 722         return strbuf_detach(&name, NULL);
 723 }
 724
 725 struct diffstat_t {
 726         int nr;
 727         int alloc;
 728         struct diffstat_file {
 729                 char *from_name;
 730                 char *name;
 731                 char *print_name;
 732                 unsigned is_unmerged:1;
 733                 unsigned is_binary:1;
 734                 unsigned is_renamed:1;
 735                 unsigned int added, deleted;
 736         } **files;
 737 };
 738
 739 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 740                                           const char *name_a,
 741                                           const char *name_b)
 742 {
 743         struct diffstat_file *x;
 744         x = xcalloc(sizeof (*x), 1);
 745         if (diffstat->nr == diffstat->alloc) {
 746                 diffstat->alloc = alloc_nr(diffstat->alloc);
 747                 diffstat->files = xrealloc(diffstat->files,
 748                                 diffstat->alloc * sizeof(x));
 749         }
 750         diffstat->files[diffstat->nr++] = x;
 751         if (name_b) {
 752                 x->from_name = xstrdup(name_a);
 753                 x->name = xstrdup(name_b);
 754                 x->is_renamed = 1;
 755         }
 756         else {
 757                 x->from_name = NULL;
 758                 x->name = xstrdup(name_a);
 759         }
 760         return x;
 761 }
 762
 763 static void diffstat_consume(void *priv, char *line, unsigned long len)
 764 {
 765         struct diffstat_t *diffstat = priv;
 766         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 767
 768         if (line[0] == '+')
 769                 x->added++;
 770         else if (line[0] == '-')
 771                 x->deleted++;
 772 }
 773
 774 const char mime_boundary_leader[] = "------------";
 775
 776 static int scale_linear(int it, int width, int max_change)
 777 {
 778         /*
 779          * make sure that at least one '-' is printed if there were deletions,
 780          * and likewise for '+'.
 781          */
 782         if (max_change < 2)
 783                 return it;
 784         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
 785 }
 786
 787 static void show_name(FILE *file,
 788                       const char *prefix, const char *name, int len,
 789                       const char *reset, const char *set)
 790 {
 791         fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
 792 }
 793
 794 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
 795 {
 796         if (cnt <= 0)
 797                 return;
 798         fprintf(file, "%s", set);
 799         while (cnt--)
 800                 putc(ch, file);
 801         fprintf(file, "%s", reset);
 802 }
 803
 804 static void fill_print_name(struct diffstat_file *file)
 805 {
 806         char *pname;
 807
 808         if (file->print_name)
 809                 return;
 810
 811         if (!file->is_renamed) {
 812                 struct strbuf buf;
 813                 strbuf_init(&buf, 0);
 814                 if (quote_c_style(file->name, &buf, NULL, 0)) {
 815                         pname = strbuf_detach(&buf, NULL);
 816                 } else {
 817                         pname = file->name;
 818                         strbuf_release(&buf);
 819                 }
 820         } else {
 821                 pname = pprint_rename(file->from_name, file->name);
 822         }
 823         file->print_name = pname;
 824 }
 825
 826 static void show_stats(struct diffstat_t* data, struct diff_options *options)
 827 {
 828         int i, len, add, del, total, adds = 0, dels = 0;
 829         int max_change = 0, max_len = 0;
 830         int total_files = data->nr;
 831         int width, name_width;
 832         const char *reset, *set, *add_c, *del_c;
 833
 834         if (data->nr == 0)
 835                 return;
 836
 837         width = options->stat_width ? options->stat_width : 80;
 838         name_width = options->stat_name_width ? options->stat_name_width : 50;
 839
 840         /* Sanity: give at least 5 columns to the graph,
 841          * but leave at least 10 columns for the name.
 842          */
 843         if (width < 25)
 844                 width = 25;
 845         if (name_width < 10)
 846                 name_width = 10;
 847         else if (width < name_width + 15)
 848                 name_width = width - 15;
 849
 850         /* Find the longest filename and max number of changes */
 851         reset = diff_get_color_opt(options, DIFF_RESET);
 852         set   = diff_get_color_opt(options, DIFF_PLAIN);
 853         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
 854         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
 855
 856         for (i = 0; i < data->nr; i++) {
 857                 struct diffstat_file *file = data->files[i];
 858                 int change = file->added + file->deleted;
 859                 fill_print_name(file);
 860                 len = strlen(file->print_name);
 861                 if (max_len < len)
 862                         max_len = len;
 863
 864                 if (file->is_binary || file->is_unmerged)
 865                         continue;
 866                 if (max_change < change)
 867                         max_change = change;
 868         }
 869
 870         /* Compute the width of the graph part;
 871          * 10 is for one blank at the beginning of the line plus
 872          * " | count " between the name and the graph.
 873          *
 874          * From here on, name_width is the width of the name area,
 875          * and width is the width of the graph area.
 876          */
 877         name_width = (name_width < max_len) ? name_width : max_len;
 878         if (width < (name_width + 10) + max_change)
 879                 width = width - (name_width + 10);
 880         else
 881                 width = max_change;
 882
 883         for (i = 0; i < data->nr; i++) {
 884                 const char *prefix = "";
 885                 char *name = data->files[i]->print_name;
 886                 int added = data->files[i]->added;
 887                 int deleted = data->files[i]->deleted;
 888                 int name_len;
 889
 890                 /*
 891                  * "scale" the filename
 892                  */
 893                 len = name_width;
 894                 name_len = strlen(name);
 895                 if (name_width < name_len) {
 896                         char *slash;
 897                         prefix = "...";
 898                         len -= 3;
 899                         name += name_len - len;
 900                         slash = strchr(name, '/');
 901                         if (slash)
 902                                 name = slash;
 903                 }
 904
 905                 if (data->files[i]->is_binary) {
 906                         show_name(options->file, prefix, name, len, reset, set);
 907                         fprintf(options->file, "  Bin ");
 908                         fprintf(options->file, "%s%d%s", del_c, deleted, reset);
 909                         fprintf(options->file, " -> ");
 910                         fprintf(options->file, "%s%d%s", add_c, added, reset);
 911                         fprintf(options->file, " bytes");
 912                         fprintf(options->file, "\n");
 913                         continue;
 914                 }
 915                 else if (data->files[i]->is_unmerged) {
 916                         show_name(options->file, prefix, name, len, reset, set);
 917                         fprintf(options->file, "  Unmerged\n");
 918                         continue;
 919                 }
 920                 else if (!data->files[i]->is_renamed &&
 921                          (added + deleted == 0)) {
 922                         total_files--;
 923                         continue;
 924                 }
 925
 926                 /*
 927                  * scale the add/delete
 928                  */
 929                 add = added;
 930                 del = deleted;
 931                 total = add + del;
 932                 adds += add;
 933                 dels += del;
 934
 935                 if (width <= max_change) {
 936                         add = scale_linear(add, width, max_change);
 937                         del = scale_linear(del, width, max_change);
 938                         total = add + del;
 939                 }
 940                 show_name(options->file, prefix, name, len, reset, set);
 941                 fprintf(options->file, "%5d%s", added + deleted,
 942                                 added + deleted ? " " : "");
 943                 show_graph(options->file, '+', add, add_c, reset);
 944                 show_graph(options->file, '-', del, del_c, reset);
 945                 fprintf(options->file, "\n");
 946         }
 947         fprintf(options->file,
 948                "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
 949                set, total_files, adds, dels, reset);
 950 }
 951
 952 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
 953 {
 954         int i, adds = 0, dels = 0, total_files = data->nr;
 955
 956         if (data->nr == 0)
 957                 return;
 958
 959         for (i = 0; i < data->nr; i++) {
 960                 if (!data->files[i]->is_binary &&
 961                     !data->files[i]->is_unmerged) {
 962                         int added = data->files[i]->added;
 963                         int deleted= data->files[i]->deleted;
 964                         if (!data->files[i]->is_renamed &&
 965                             (added + deleted == 0)) {
 966                                 total_files--;
 967                         } else {
 968                                 adds += added;
 969                                 dels += deleted;
 970                         }
 971                 }
 972         }
 973         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
 974                total_files, adds, dels);
 975 }
 976
 977 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
 978 {
 979         int i;
 980
 981         if (data->nr == 0)
 982                 return;
 983
 984         for (i = 0; i < data->nr; i++) {
 985                 struct diffstat_file *file = data->files[i];
 986
 987                 if (file->is_binary)
 988                         fprintf(options->file, "-\t-\t");
 989                 else
 990                         fprintf(options->file,
 991                                 "%d\t%d\t", file->added, file->deleted);
 992                 if (options->line_termination) {
 993                         fill_print_name(file);
 994                         if (!file->is_renamed)
 995                                 write_name_quoted(file->name, options->file,
 996                                                   options->line_termination);
 997                         else {
 998                                 fputs(file->print_name, options->file);
 999                                 putc(options->line_termination, options->file);
1000                         }
1001                 } else {
1002                         if (file->is_renamed) {
1003                                 putc('\0', options->file);
1004                                 write_name_quoted(file->from_name, options->file, '\0');
1005                         }
1006                         write_name_quoted(file->name, options->file, '\0');
1007                 }
1008         }
1009 }
1010
1011 struct dirstat_file {
1012         const char *name;
1013         unsigned long changed;
1014 };
1015
1016 struct dirstat_dir {
1017         struct dirstat_file *files;
1018         int alloc, nr, percent, cumulative;
1019 };
1020
1021 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1022 {
1023         unsigned long this_dir = 0;
1024         unsigned int sources = 0;
1025
1026         while (dir->nr) {
1027                 struct dirstat_file *f = dir->files;
1028                 int namelen = strlen(f->name);
1029                 unsigned long this;
1030                 char *slash;
1031
1032                 if (namelen < baselen)
1033                         break;
1034                 if (memcmp(f->name, base, baselen))
1035                         break;
1036                 slash = strchr(f->name + baselen, '/');
1037                 if (slash) {
1038                         int newbaselen = slash + 1 - f->name;
1039                         this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1040                         sources++;
1041                 } else {
1042                         this = f->changed;
1043                         dir->files++;
1044                         dir->nr--;
1045                         sources += 2;
1046                 }
1047                 this_dir += this;
1048         }
1049
1050         /*
1051          * We don't report dirstat's for
1052          *  - the top level
1053          *  - or cases where everything came from a single directory
1054          *    under this directory (sources == 1).
1055          */
1056         if (baselen && sources != 1) {
1057                 int permille = this_dir * 1000 / changed;
1058                 if (permille) {
1059                         int percent = permille / 10;
1060                         if (percent >= dir->percent) {
1061                                 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1062                                 if (!dir->cumulative)
1063                                         return 0;
1064                         }
1065                 }
1066         }
1067         return this_dir;
1068 }
1069
1070 static int dirstat_compare(const void *_a, const void *_b)
1071 {
1072         const struct dirstat_file *a = _a;
1073         const struct dirstat_file *b = _b;
1074         return strcmp(a->name, b->name);
1075 }
1076
1077 static void show_dirstat(struct diff_options *options)
1078 {
1079         int i;
1080         unsigned long changed;
1081         struct dirstat_dir dir;
1082         struct diff_queue_struct *q = &diff_queued_diff;
1083
1084         dir.files = NULL;
1085         dir.alloc = 0;
1086         dir.nr = 0;
1087         dir.percent = options->dirstat_percent;
1088         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1089
1090         changed = 0;
1091         for (i = 0; i < q->nr; i++) {
1092                 struct diff_filepair *p = q->queue[i];
1093                 const char *name;
1094                 unsigned long copied, added, damage;
1095
1096                 name = p->one->path ? p->one->path : p->two->path;
1097
1098                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1099                         diff_populate_filespec(p->one, 0);
1100                         diff_populate_filespec(p->two, 0);
1101                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1102                                                &copied, &added);
1103                         diff_free_filespec_data(p->one);
1104                         diff_free_filespec_data(p->two);
1105                 } else if (DIFF_FILE_VALID(p->one)) {
1106                         diff_populate_filespec(p->one, 1);
1107                         copied = added = 0;
1108                         diff_free_filespec_data(p->one);
1109                 } else if (DIFF_FILE_VALID(p->two)) {
1110                         diff_populate_filespec(p->two, 1);
1111                         copied = 0;
1112                         added = p->two->size;
1113                         diff_free_filespec_data(p->two);
1114                 } else
1115                         continue;
1116
1117                 /*
1118                  * Original minus copied is the removed material,
1119                  * added is the new material.  They are both damages
1120                  * made to the preimage.
1121                  */
1122                 damage = (p->one->size - copied) + added;
1123
1124                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1125                 dir.files[dir.nr].name = name;
1126                 dir.files[dir.nr].changed = damage;
1127                 changed += damage;
1128                 dir.nr++;
1129         }
1130
1131         /* This can happen even with many files, if everything was renames */
1132         if (!changed)
1133                 return;
1134
1135         /* Show all directories with more than x% of the changes */
1136         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1137         gather_dirstat(options->file, &dir, changed, "", 0);
1138 }
1139
1140 static void free_diffstat_info(struct diffstat_t *diffstat)
1141 {
1142         int i;
1143         for (i = 0; i < diffstat->nr; i++) {
1144                 struct diffstat_file *f = diffstat->files[i];
1145                 if (f->name != f->print_name)
1146                         free(f->print_name);
1147                 free(f->name);
1148                 free(f->from_name);
1149                 free(f);
1150         }
1151         free(diffstat->files);
1152 }
1153
1154 struct checkdiff_t {
1155         const char *filename;
1156         int lineno;
1157         struct diff_options *o;
1158         unsigned ws_rule;
1159         unsigned status;
1160         int trailing_blanks_start;
1161 };
1162
1163 static int is_conflict_marker(const char *line, unsigned long len)
1164 {
1165         char firstchar;
1166         int cnt;
1167
1168         if (len < 8)
1169                 return 0;
1170         firstchar = line[0];
1171         switch (firstchar) {
1172         case '=': case '>': case '<':
1173                 break;
1174         default:
1175                 return 0;
1176         }
1177         for (cnt = 1; cnt < 7; cnt++)
1178                 if (line[cnt] != firstchar)
1179                         return 0;
1180         /* line[0] thru line[6] are same as firstchar */
1181         if (firstchar == '=') {
1182                 /* divider between ours and theirs? */
1183                 if (len != 8 || line[7] != '\n')
1184                         return 0;
1185         } else if (len < 8 || !isspace(line[7])) {
1186                 /* not divider before ours nor after theirs */
1187                 return 0;
1188         }
1189         return 1;
1190 }
1191
1192 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1193 {
1194         struct checkdiff_t *data = priv;
1195         int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1196         const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1197         const char *reset = diff_get_color(color_diff, DIFF_RESET);
1198         const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1199         char *err;
1200
1201         if (line[0] == '+') {
1202                 unsigned bad;
1203                 data->lineno++;
1204                 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1205                         data->trailing_blanks_start = 0;
1206                 else if (!data->trailing_blanks_start)
1207                         data->trailing_blanks_start = data->lineno;
1208                 if (is_conflict_marker(line + 1, len - 1)) {
1209                         data->status |= 1;
1210                         fprintf(data->o->file,
1211                                 "%s:%d: leftover conflict marker\n",
1212                                 data->filename, data->lineno);
1213                 }
1214                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1215                 if (!bad)
1216                         return;
1217                 data->status |= bad;
1218                 err = whitespace_error_string(bad);
1219                 fprintf(data->o->file, "%s:%d: %s.\n",
1220                         data->filename, data->lineno, err);
1221                 free(err);
1222                 emit_line(data->o->file, set, reset, line, 1);
1223                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1224                               data->o->file, set, reset, ws);
1225         } else if (line[0] == ' ') {
1226                 data->lineno++;
1227                 data->trailing_blanks_start = 0;
1228         } else if (line[0] == '@') {
1229                 char *plus = strchr(line, '+');
1230                 if (plus)
1231                         data->lineno = strtol(plus, NULL, 10) - 1;
1232                 else
1233                         die("invalid diff");
1234                 data->trailing_blanks_start = 0;
1235         }
1236 }
1237
1238 static unsigned char *deflate_it(char *data,
1239                                  unsigned long size,
1240                                  unsigned long *result_size)
1241 {
1242         int bound;
1243         unsigned char *deflated;
1244         z_stream stream;
1245
1246         memset(&stream, 0, sizeof(stream));
1247         deflateInit(&stream, zlib_compression_level);
1248         bound = deflateBound(&stream, size);
1249         deflated = xmalloc(bound);
1250         stream.next_out = deflated;
1251         stream.avail_out = bound;
1252
1253         stream.next_in = (unsigned char *)data;
1254         stream.avail_in = size;
1255         while (deflate(&stream, Z_FINISH) == Z_OK)
1256                 ; /* nothing */
1257         deflateEnd(&stream);
1258         *result_size = stream.total_out;
1259         return deflated;
1260 }
1261
1262 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1263 {
1264         void *cp;
1265         void *delta;
1266         void *deflated;
1267         void *data;
1268         unsigned long orig_size;
1269         unsigned long delta_size;
1270         unsigned long deflate_size;
1271         unsigned long data_size;
1272
1273         /* We could do deflated delta, or we could do just deflated two,
1274          * whichever is smaller.
1275          */
1276         delta = NULL;
1277         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1278         if (one->size && two->size) {
1279                 delta = diff_delta(one->ptr, one->size,
1280                                    two->ptr, two->size,
1281                                    &delta_size, deflate_size);
1282                 if (delta) {
1283                         void *to_free = delta;
1284                         orig_size = delta_size;
1285                         delta = deflate_it(delta, delta_size, &delta_size);
1286                         free(to_free);
1287                 }
1288         }
1289
1290         if (delta && delta_size < deflate_size) {
1291                 fprintf(file, "delta %lu\n", orig_size);
1292                 free(deflated);
1293                 data = delta;
1294                 data_size = delta_size;
1295         }
1296         else {
1297                 fprintf(file, "literal %lu\n", two->size);
1298                 free(delta);
1299                 data = deflated;
1300                 data_size = deflate_size;
1301         }
1302
1303         /* emit data encoded in base85 */
1304         cp = data;
1305         while (data_size) {
1306                 int bytes = (52 < data_size) ? 52 : data_size;
1307                 char line[70];
1308                 data_size -= bytes;
1309                 if (bytes <= 26)
1310                         line[0] = bytes + 'A' - 1;
1311                 else
1312                         line[0] = bytes - 26 + 'a' - 1;
1313                 encode_85(line + 1, cp, bytes);
1314                 cp = (char *) cp + bytes;
1315                 fputs(line, file);
1316                 fputc('\n', file);
1317         }
1318         fprintf(file, "\n");
1319         free(data);
1320 }
1321
1322 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1323 {
1324         fprintf(file, "GIT binary patch\n");
1325         emit_binary_diff_body(file, one, two);
1326         emit_binary_diff_body(file, two, one);
1327 }
1328
1329 static void setup_diff_attr_check(struct git_attr_check *check)
1330 {
1331         static struct git_attr *attr_diff;
1332
1333         if (!attr_diff) {
1334                 attr_diff = git_attr("diff", 4);
1335         }
1336         check[0].attr = attr_diff;
1337 }
1338
1339 static void diff_filespec_check_attr(struct diff_filespec *one)
1340 {
1341         struct git_attr_check attr_diff_check;
1342         int check_from_data = 0;
1343
1344         if (one->checked_attr)
1345                 return;
1346
1347         setup_diff_attr_check(&attr_diff_check);
1348         one->is_binary = 0;
1349         one->funcname_pattern_ident = NULL;
1350
1351         if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1352                 const char *value;
1353
1354                 /* binaryness */
1355                 value = attr_diff_check.value;
1356                 if (ATTR_TRUE(value))
1357                         ;
1358                 else if (ATTR_FALSE(value))
1359                         one->is_binary = 1;
1360                 else
1361                         check_from_data = 1;
1362
1363                 /* funcname pattern ident */
1364                 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1365                         ;
1366                 else
1367                         one->funcname_pattern_ident = value;
1368         }
1369
1370         if (check_from_data) {
1371                 if (!one->data && DIFF_FILE_VALID(one))
1372                         diff_populate_filespec(one, 0);
1373
1374                 if (one->data)
1375                         one->is_binary = buffer_is_binary(one->data, one->size);
1376         }
1377 }
1378
1379 int diff_filespec_is_binary(struct diff_filespec *one)
1380 {
1381         diff_filespec_check_attr(one);
1382         return one->is_binary;
1383 }
1384
1385 static const char *funcname_pattern(const char *ident)
1386 {
1387         struct funcname_pattern *pp;
1388
1389         for (pp = funcname_pattern_list; pp; pp = pp->next)
1390                 if (!strcmp(ident, pp->name))
1391                         return pp->pattern;
1392         return NULL;
1393 }
1394
1395 static struct builtin_funcname_pattern {
1396         const char *name;
1397         const char *pattern;
1398 } builtin_funcname_pattern[] = {
1399         { "bibtex", "\\(@[a-zA-Z]\\{1,\\}[ \t]*{\\{0,1\\}[ \t]*[^ \t\"@',\\#}{~%]*\\).*$" },
1400         { "html", "^\\s*\\(<[Hh][1-6]\\s.*>.*\\)$" },
1401         { "java", "!^[  ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1402                         "new\\|return\\|switch\\|throw\\|while\\)\n"
1403                         "^[     ]*\\(\\([       ]*"
1404                         "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1405                         "[      ]*([^;]*\\)$" },
1406         { "pascal", "^\\(\\(procedure\\|function\\|constructor\\|"
1407                         "destructor\\|interface\\|implementation\\|"
1408                         "initialization\\|finalization\\)[ \t]*.*\\)$"
1409                         "\\|"
1410                         "^\\(.*=[ \t]*\\(class\\|record\\).*\\)$"
1411                         },
1412         { "php", "^[\t ]*\\(\\(function\\|class\\).*\\)" },
1413         { "python", "^\\s*\\(\\(class\\|def\\)\\s.*\\)$" },
1414         { "ruby", "^\\s*\\(\\(class\\|module\\|def\\)\\s.*\\)$" },
1415         { "tex", "^\\(\\\\\\(\\(sub\\)*section\\|chapter\\|part\\)\\*\\{0,1\\}{.*\\)$" },
1416 };
1417
1418 static const char *diff_funcname_pattern(struct diff_filespec *one)
1419 {
1420         const char *ident, *pattern;
1421         int i;
1422
1423         diff_filespec_check_attr(one);
1424         ident = one->funcname_pattern_ident;
1425
1426         if (!ident)
1427                 /*
1428                  * If the config file has "funcname.default" defined, that
1429                  * regexp is used; otherwise NULL is returned and xemit uses
1430                  * the built-in default.
1431                  */
1432                 return funcname_pattern("default");
1433
1434         /* Look up custom "funcname.$ident" regexp from config. */
1435         pattern = funcname_pattern(ident);
1436         if (pattern)
1437                 return pattern;
1438
1439         /*
1440          * And define built-in fallback patterns here.  Note that
1441          * these can be overridden by the user's config settings.
1442          */
1443         for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1444                 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1445                         return builtin_funcname_pattern[i].pattern;
1446
1447         return NULL;
1448 }
1449
1450 static void builtin_diff(const char *name_a,
1451                          const char *name_b,
1452                          struct diff_filespec *one,
1453                          struct diff_filespec *two,
1454                          const char *xfrm_msg,
1455                          struct diff_options *o,
1456                          int complete_rewrite)
1457 {
1458         mmfile_t mf1, mf2;
1459         const char *lbl[2];
1460         char *a_one, *b_two;
1461         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1462         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1463
1464         a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1465         b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1466         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1467         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1468         fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1469         if (lbl[0][0] == '/') {
1470                 /* /dev/null */
1471                 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1472                 if (xfrm_msg && xfrm_msg[0])
1473                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1474         }
1475         else if (lbl[1][0] == '/') {
1476                 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1477                 if (xfrm_msg && xfrm_msg[0])
1478                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1479         }
1480         else {
1481                 if (one->mode != two->mode) {
1482                         fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1483                         fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1484                 }
1485                 if (xfrm_msg && xfrm_msg[0])
1486                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1487                 /*
1488                  * we do not run diff between different kind
1489                  * of objects.
1490                  */
1491                 if ((one->mode ^ two->mode) & S_IFMT)
1492                         goto free_ab_and_return;
1493                 if (complete_rewrite) {
1494                         emit_rewrite_diff(name_a, name_b, one, two, o);
1495                         o->found_changes = 1;
1496                         goto free_ab_and_return;
1497                 }
1498         }
1499
1500         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1501                 die("unable to read files to diff");
1502
1503         if (!DIFF_OPT_TST(o, TEXT) &&
1504             (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1505                 /* Quite common confusing case */
1506                 if (mf1.size == mf2.size &&
1507                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1508                         goto free_ab_and_return;
1509                 if (DIFF_OPT_TST(o, BINARY))
1510                         emit_binary_diff(o->file, &mf1, &mf2);
1511                 else
1512                         fprintf(o->file, "Binary files %s and %s differ\n",
1513                                 lbl[0], lbl[1]);
1514                 o->found_changes = 1;
1515         }
1516         else {
1517                 /* Crazy xdl interfaces.. */
1518                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1519                 xpparam_t xpp;
1520                 xdemitconf_t xecfg;
1521                 xdemitcb_t ecb;
1522                 struct emit_callback ecbdata;
1523                 const char *funcname_pattern;
1524
1525                 funcname_pattern = diff_funcname_pattern(one);
1526                 if (!funcname_pattern)
1527                         funcname_pattern = diff_funcname_pattern(two);
1528
1529                 memset(&xecfg, 0, sizeof(xecfg));
1530                 memset(&ecbdata, 0, sizeof(ecbdata));
1531                 ecbdata.label_path = lbl;
1532                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1533                 ecbdata.found_changesp = &o->found_changes;
1534                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1535                 ecbdata.file = o->file;
1536                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1537                 xecfg.ctxlen = o->context;
1538                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1539                 if (funcname_pattern)
1540                         xdiff_set_find_func(&xecfg, funcname_pattern);
1541                 if (!diffopts)
1542                         ;
1543                 else if (!prefixcmp(diffopts, "--unified="))
1544                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1545                 else if (!prefixcmp(diffopts, "-u"))
1546                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1547                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1548                         ecbdata.diff_words =
1549                                 xcalloc(1, sizeof(struct diff_words_data));
1550                         ecbdata.diff_words->file = o->file;
1551                 }
1552                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1553                               &xpp, &xecfg, &ecb);
1554                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1555                         free_diff_words_data(&ecbdata);
1556         }
1557
1558  free_ab_and_return:
1559         diff_free_filespec_data(one);
1560         diff_free_filespec_data(two);
1561         free(a_one);
1562         free(b_two);
1563         return;
1564 }
1565
1566 static void builtin_diffstat(const char *name_a, const char *name_b,
1567                              struct diff_filespec *one,
1568                              struct diff_filespec *two,
1569                              struct diffstat_t *diffstat,
1570                              struct diff_options *o,
1571                              int complete_rewrite)
1572 {
1573         mmfile_t mf1, mf2;
1574         struct diffstat_file *data;
1575
1576         data = diffstat_add(diffstat, name_a, name_b);
1577
1578         if (!one || !two) {
1579                 data->is_unmerged = 1;
1580                 return;
1581         }
1582         if (complete_rewrite) {
1583                 diff_populate_filespec(one, 0);
1584                 diff_populate_filespec(two, 0);
1585                 data->deleted = count_lines(one->data, one->size);
1586                 data->added = count_lines(two->data, two->size);
1587                 goto free_and_return;
1588         }
1589         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1590                 die("unable to read files to diff");
1591
1592         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1593                 data->is_binary = 1;
1594                 data->added = mf2.size;
1595                 data->deleted = mf1.size;
1596         } else {
1597                 /* Crazy xdl interfaces.. */
1598                 xpparam_t xpp;
1599                 xdemitconf_t xecfg;
1600                 xdemitcb_t ecb;
1601
1602                 memset(&xecfg, 0, sizeof(xecfg));
1603                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1604                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1605                               &xpp, &xecfg, &ecb);
1606         }
1607
1608  free_and_return:
1609         diff_free_filespec_data(one);
1610         diff_free_filespec_data(two);
1611 }
1612
1613 static void builtin_checkdiff(const char *name_a, const char *name_b,
1614                               const char *attr_path,
1615                               struct diff_filespec *one,
1616                               struct diff_filespec *two,
1617                               struct diff_options *o)
1618 {
1619         mmfile_t mf1, mf2;
1620         struct checkdiff_t data;
1621
1622         if (!two)
1623                 return;
1624
1625         memset(&data, 0, sizeof(data));
1626         data.filename = name_b ? name_b : name_a;
1627         data.lineno = 0;
1628         data.o = o;
1629         data.ws_rule = whitespace_rule(attr_path);
1630
1631         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1632                 die("unable to read files to diff");
1633
1634         /*
1635          * All the other codepaths check both sides, but not checking
1636          * the "old" side here is deliberate.  We are checking the newly
1637          * introduced changes, and as long as the "new" side is text, we
1638          * can and should check what it introduces.
1639          */
1640         if (diff_filespec_is_binary(two))
1641                 goto free_and_return;
1642         else {
1643                 /* Crazy xdl interfaces.. */
1644                 xpparam_t xpp;
1645                 xdemitconf_t xecfg;
1646                 xdemitcb_t ecb;
1647
1648                 memset(&xecfg, 0, sizeof(xecfg));
1649                 xecfg.ctxlen = 1; /* at least one context line */
1650                 xpp.flags = XDF_NEED_MINIMAL;
1651                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1652                               &xpp, &xecfg, &ecb);
1653
1654                 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1655                     data.trailing_blanks_start) {
1656                         fprintf(o->file, "%s:%d: ends with blank lines.\n",
1657                                 data.filename, data.trailing_blanks_start);
1658                         data.status = 1; /* report errors */
1659                 }
1660         }
1661  free_and_return:
1662         diff_free_filespec_data(one);
1663         diff_free_filespec_data(two);
1664         if (data.status)
1665                 DIFF_OPT_SET(o, CHECK_FAILED);
1666 }
1667
1668 struct diff_filespec *alloc_filespec(const char *path)
1669 {
1670         int namelen = strlen(path);
1671         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1672
1673         memset(spec, 0, sizeof(*spec));
1674         spec->path = (char *)(spec + 1);
1675         memcpy(spec->path, path, namelen+1);
1676         spec->count = 1;
1677         return spec;
1678 }
1679
1680 void free_filespec(struct diff_filespec *spec)
1681 {
1682         if (!--spec->count) {
1683                 diff_free_filespec_data(spec);
1684                 free(spec);
1685         }
1686 }
1687
1688 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1689                    unsigned short mode)
1690 {
1691         if (mode) {
1692                 spec->mode = canon_mode(mode);
1693                 hashcpy(spec->sha1, sha1);
1694                 spec->sha1_valid = !is_null_sha1(sha1);
1695         }
1696 }
1697
1698 /*
1699  * Given a name and sha1 pair, if the index tells us the file in
1700  * the work tree has that object contents, return true, so that
1701  * prepare_temp_file() does not have to inflate and extract.
1702  */
1703 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1704 {
1705         struct cache_entry *ce;
1706         struct stat st;
1707         int pos, len;
1708
1709         /* We do not read the cache ourselves here, because the
1710          * benchmark with my previous version that always reads cache
1711          * shows that it makes things worse for diff-tree comparing
1712          * two linux-2.6 kernel trees in an already checked out work
1713          * tree.  This is because most diff-tree comparisons deal with
1714          * only a small number of files, while reading the cache is
1715          * expensive for a large project, and its cost outweighs the
1716          * savings we get by not inflating the object to a temporary
1717          * file.  Practically, this code only helps when we are used
1718          * by diff-cache --cached, which does read the cache before
1719          * calling us.
1720          */
1721         if (!active_cache)
1722                 return 0;
1723
1724         /* We want to avoid the working directory if our caller
1725          * doesn't need the data in a normal file, this system
1726          * is rather slow with its stat/open/mmap/close syscalls,
1727          * and the object is contained in a pack file.  The pack
1728          * is probably already open and will be faster to obtain
1729          * the data through than the working directory.  Loose
1730          * objects however would tend to be slower as they need
1731          * to be individually opened and inflated.
1732          */
1733         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1734                 return 0;
1735
1736         len = strlen(name);
1737         pos = cache_name_pos(name, len);
1738         if (pos < 0)
1739                 return 0;
1740         ce = active_cache[pos];
1741
1742         /*
1743          * This is not the sha1 we are looking for, or
1744          * unreusable because it is not a regular file.
1745          */
1746         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1747                 return 0;
1748
1749         /*
1750          * If ce matches the file in the work tree, we can reuse it.
1751          */
1752         if (ce_uptodate(ce) ||
1753             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1754                 return 1;
1755
1756         return 0;
1757 }
1758
1759 static int populate_from_stdin(struct diff_filespec *s)
1760 {
1761         struct strbuf buf;
1762         size_t size = 0;
1763
1764         strbuf_init(&buf, 0);
1765         if (strbuf_read(&buf, 0, 0) < 0)
1766                 return error("error while reading from stdin %s",
1767                                      strerror(errno));
1768
1769         s->should_munmap = 0;
1770         s->data = strbuf_detach(&buf, &size);
1771         s->size = size;
1772         s->should_free = 1;
1773         return 0;
1774 }
1775
1776 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1777 {
1778         int len;
1779         char *data = xmalloc(100);
1780         len = snprintf(data, 100,
1781                 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1782         s->data = data;
1783         s->size = len;
1784         s->should_free = 1;
1785         if (size_only) {
1786                 s->data = NULL;
1787                 free(data);
1788         }
1789         return 0;
1790 }
1791
1792 /*
1793  * While doing rename detection and pickaxe operation, we may need to
1794  * grab the data for the blob (or file) for our own in-core comparison.
1795  * diff_filespec has data and size fields for this purpose.
1796  */
1797 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1798 {
1799         int err = 0;
1800         if (!DIFF_FILE_VALID(s))
1801                 die("internal error: asking to populate invalid file.");
1802         if (S_ISDIR(s->mode))
1803                 return -1;
1804
1805         if (s->data)
1806                 return 0;
1807
1808         if (size_only && 0 < s->size)
1809                 return 0;
1810
1811         if (S_ISGITLINK(s->mode))
1812                 return diff_populate_gitlink(s, size_only);
1813
1814         if (!s->sha1_valid ||
1815             reuse_worktree_file(s->path, s->sha1, 0)) {
1816                 struct strbuf buf;
1817                 struct stat st;
1818                 int fd;
1819
1820                 if (!strcmp(s->path, "-"))
1821                         return populate_from_stdin(s);
1822
1823                 if (lstat(s->path, &st) < 0) {
1824                         if (errno == ENOENT) {
1825                         err_empty:
1826                                 err = -1;
1827                         empty:
1828                                 s->data = (char *)"";
1829                                 s->size = 0;
1830                                 return err;
1831                         }
1832                 }
1833                 s->size = xsize_t(st.st_size);
1834                 if (!s->size)
1835                         goto empty;
1836                 if (size_only)
1837                         return 0;
1838                 if (S_ISLNK(st.st_mode)) {
1839                         int ret;
1840                         s->data = xmalloc(s->size);
1841                         s->should_free = 1;
1842                         ret = readlink(s->path, s->data, s->size);
1843                         if (ret < 0) {
1844                                 free(s->data);
1845                                 goto err_empty;
1846                         }
1847                         return 0;
1848                 }
1849                 fd = open(s->path, O_RDONLY);
1850                 if (fd < 0)
1851                         goto err_empty;
1852                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1853                 close(fd);
1854                 s->should_munmap = 1;
1855
1856                 /*
1857                  * Convert from working tree format to canonical git format
1858                  */
1859                 strbuf_init(&buf, 0);
1860                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1861                         size_t size = 0;
1862                         munmap(s->data, s->size);
1863                         s->should_munmap = 0;
1864                         s->data = strbuf_detach(&buf, &size);
1865                         s->size = size;
1866                         s->should_free = 1;
1867                 }
1868         }
1869         else {
1870                 enum object_type type;
1871                 if (size_only)
1872                         type = sha1_object_info(s->sha1, &s->size);
1873                 else {
1874                         s->data = read_sha1_file(s->sha1, &type, &s->size);
1875                         s->should_free = 1;
1876                 }
1877         }
1878         return 0;
1879 }
1880
1881 void diff_free_filespec_blob(struct diff_filespec *s)
1882 {
1883         if (s->should_free)
1884                 free(s->data);
1885         else if (s->should_munmap)
1886                 munmap(s->data, s->size);
1887
1888         if (s->should_free || s->should_munmap) {
1889                 s->should_free = s->should_munmap = 0;
1890                 s->data = NULL;
1891         }
1892 }
1893
1894 void diff_free_filespec_data(struct diff_filespec *s)
1895 {
1896         diff_free_filespec_blob(s);
1897         free(s->cnt_data);
1898         s->cnt_data = NULL;
1899 }
1900
1901 static void prep_temp_blob(struct diff_tempfile *temp,
1902                            void *blob,
1903                            unsigned long size,
1904                            const unsigned char *sha1,
1905                            int mode)
1906 {
1907         int fd;
1908
1909         fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1910         if (fd < 0)
1911                 die("unable to create temp-file: %s", strerror(errno));
1912         if (write_in_full(fd, blob, size) != size)
1913                 die("unable to write temp-file");
1914         close(fd);
1915         temp->name = temp->tmp_path;
1916         strcpy(temp->hex, sha1_to_hex(sha1));
1917         temp->hex[40] = 0;
1918         sprintf(temp->mode, "%06o", mode);
1919 }
1920
1921 static void prepare_temp_file(const char *name,
1922                               struct diff_tempfile *temp,
1923                               struct diff_filespec *one)
1924 {
1925         if (!DIFF_FILE_VALID(one)) {
1926         not_a_valid_file:
1927                 /* A '-' entry produces this for file-2, and
1928                  * a '+' entry produces this for file-1.
1929                  */
1930                 temp->name = "/dev/null";
1931                 strcpy(temp->hex, ".");
1932                 strcpy(temp->mode, ".");
1933                 return;
1934         }
1935
1936         if (!one->sha1_valid ||
1937             reuse_worktree_file(name, one->sha1, 1)) {
1938                 struct stat st;
1939                 if (lstat(name, &st) < 0) {
1940                         if (errno == ENOENT)
1941                                 goto not_a_valid_file;
1942                         die("stat(%s): %s", name, strerror(errno));
1943                 }
1944                 if (S_ISLNK(st.st_mode)) {
1945                         int ret;
1946                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1947                         size_t sz = xsize_t(st.st_size);
1948                         if (sizeof(buf) <= st.st_size)
1949                                 die("symlink too long: %s", name);
1950                         ret = readlink(name, buf, sz);
1951                         if (ret < 0)
1952                                 die("readlink(%s)", name);
1953                         prep_temp_blob(temp, buf, sz,
1954                                        (one->sha1_valid ?
1955                                         one->sha1 : null_sha1),
1956                                        (one->sha1_valid ?
1957                                         one->mode : S_IFLNK));
1958                 }
1959                 else {
1960                         /* we can borrow from the file in the work tree */
1961                         temp->name = name;
1962                         if (!one->sha1_valid)
1963                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
1964                         else
1965                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
1966                         /* Even though we may sometimes borrow the
1967                          * contents from the work tree, we always want
1968                          * one->mode.  mode is trustworthy even when
1969                          * !(one->sha1_valid), as long as
1970                          * DIFF_FILE_VALID(one).
1971                          */
1972                         sprintf(temp->mode, "%06o", one->mode);
1973                 }
1974                 return;
1975         }
1976         else {
1977                 if (diff_populate_filespec(one, 0))
1978                         die("cannot read data blob for %s", one->path);
1979                 prep_temp_blob(temp, one->data, one->size,
1980                                one->sha1, one->mode);
1981         }
1982 }
1983
1984 static void remove_tempfile(void)
1985 {
1986         int i;
1987
1988         for (i = 0; i < 2; i++)
1989                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1990                         unlink(diff_temp[i].name);
1991                         diff_temp[i].name = NULL;
1992                 }
1993 }
1994
1995 static void remove_tempfile_on_signal(int signo)
1996 {
1997         remove_tempfile();
1998         signal(SIGINT, SIG_DFL);
1999         raise(signo);
2000 }
2001
2002 /* An external diff command takes:
2003  *
2004  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2005  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2006  *
2007  */
2008 static void run_external_diff(const char *pgm,
2009                               const char *name,
2010                               const char *other,
2011                               struct diff_filespec *one,
2012                               struct diff_filespec *two,
2013                               const char *xfrm_msg,
2014                               int complete_rewrite)
2015 {
2016         const char *spawn_arg[10];
2017         struct diff_tempfile *temp = diff_temp;
2018         int retval;
2019         static int atexit_asked = 0;
2020         const char *othername;
2021         const char **arg = &spawn_arg[0];
2022
2023         othername = (other? other : name);
2024         if (one && two) {
2025                 prepare_temp_file(name, &temp[0], one);
2026                 prepare_temp_file(othername, &temp[1], two);
2027                 if (! atexit_asked &&
2028                     (temp[0].name == temp[0].tmp_path ||
2029                      temp[1].name == temp[1].tmp_path)) {
2030                         atexit_asked = 1;
2031                         atexit(remove_tempfile);
2032                 }
2033                 signal(SIGINT, remove_tempfile_on_signal);
2034         }
2035
2036         if (one && two) {
2037                 *arg++ = pgm;
2038                 *arg++ = name;
2039                 *arg++ = temp[0].name;
2040                 *arg++ = temp[0].hex;
2041                 *arg++ = temp[0].mode;
2042                 *arg++ = temp[1].name;
2043                 *arg++ = temp[1].hex;
2044                 *arg++ = temp[1].mode;
2045                 if (other) {
2046                         *arg++ = other;
2047                         *arg++ = xfrm_msg;
2048                 }
2049         } else {
2050                 *arg++ = pgm;
2051                 *arg++ = name;
2052         }
2053         *arg = NULL;
2054         fflush(NULL);
2055         retval = run_command_v_opt(spawn_arg, 0);
2056         remove_tempfile();
2057         if (retval) {
2058                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2059                 exit(1);
2060         }
2061 }
2062
2063 static const char *external_diff_attr(const char *name)
2064 {
2065         struct git_attr_check attr_diff_check;
2066
2067         if (!name)
2068                 return NULL;
2069
2070         setup_diff_attr_check(&attr_diff_check);
2071         if (!git_checkattr(name, 1, &attr_diff_check)) {
2072                 const char *value = attr_diff_check.value;
2073                 if (!ATTR_TRUE(value) &&
2074                     !ATTR_FALSE(value) &&
2075                     !ATTR_UNSET(value)) {
2076                         struct ll_diff_driver *drv;
2077
2078                         for (drv = user_diff; drv; drv = drv->next)
2079                                 if (!strcmp(drv->name, value))
2080                                         return drv->cmd;
2081                 }
2082         }
2083         return NULL;
2084 }
2085
2086 static void run_diff_cmd(const char *pgm,
2087                          const char *name,
2088                          const char *other,
2089                          const char *attr_path,
2090                          struct diff_filespec *one,
2091                          struct diff_filespec *two,
2092                          const char *xfrm_msg,
2093                          struct diff_options *o,
2094                          int complete_rewrite)
2095 {
2096         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2097                 pgm = NULL;
2098         else {
2099                 const char *cmd = external_diff_attr(attr_path);
2100                 if (cmd)
2101                         pgm = cmd;
2102         }
2103
2104         if (pgm) {
2105                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2106                                   complete_rewrite);
2107                 return;
2108         }
2109         if (one && two)
2110                 builtin_diff(name, other ? other : name,
2111                              one, two, xfrm_msg, o, complete_rewrite);
2112         else
2113                 fprintf(o->file, "* Unmerged path %s\n", name);
2114 }
2115
2116 static void diff_fill_sha1_info(struct diff_filespec *one)
2117 {
2118         if (DIFF_FILE_VALID(one)) {
2119                 if (!one->sha1_valid) {
2120                         struct stat st;
2121                         if (!strcmp(one->path, "-")) {
2122                                 hashcpy(one->sha1, null_sha1);
2123                                 return;
2124                         }
2125                         if (lstat(one->path, &st) < 0)
2126                                 die("stat %s", one->path);
2127                         if (index_path(one->sha1, one->path, &st, 0))
2128                                 die("cannot hash %s\n", one->path);
2129                 }
2130         }
2131         else
2132                 hashclr(one->sha1);
2133 }
2134
2135 static int similarity_index(struct diff_filepair *p)
2136 {
2137         return p->score * 100 / MAX_SCORE;
2138 }
2139
2140 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2141 {
2142         /* Strip the prefix but do not molest /dev/null and absolute paths */
2143         if (*namep && **namep != '/')
2144                 *namep += prefix_length;
2145         if (*otherp && **otherp != '/')
2146                 *otherp += prefix_length;
2147 }
2148
2149 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2150 {
2151         const char *pgm = external_diff();
2152         struct strbuf msg;
2153         char *xfrm_msg;
2154         struct diff_filespec *one = p->one;
2155         struct diff_filespec *two = p->two;
2156         const char *name;
2157         const char *other;
2158         const char *attr_path;
2159         int complete_rewrite = 0;
2160
2161         name  = p->one->path;
2162         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2163         attr_path = name;
2164         if (o->prefix_length)
2165                 strip_prefix(o->prefix_length, &name, &other);
2166
2167         if (DIFF_PAIR_UNMERGED(p)) {
2168                 run_diff_cmd(pgm, name, NULL, attr_path,
2169                              NULL, NULL, NULL, o, 0);
2170                 return;
2171         }
2172
2173         diff_fill_sha1_info(one);
2174         diff_fill_sha1_info(two);
2175
2176         strbuf_init(&msg, PATH_MAX * 2 + 300);
2177         switch (p->status) {
2178         case DIFF_STATUS_COPIED:
2179                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2180                 strbuf_addstr(&msg, "\ncopy from ");
2181                 quote_c_style(name, &msg, NULL, 0);
2182                 strbuf_addstr(&msg, "\ncopy to ");
2183                 quote_c_style(other, &msg, NULL, 0);
2184                 strbuf_addch(&msg, '\n');
2185                 break;
2186         case DIFF_STATUS_RENAMED:
2187                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2188                 strbuf_addstr(&msg, "\nrename from ");
2189                 quote_c_style(name, &msg, NULL, 0);
2190                 strbuf_addstr(&msg, "\nrename to ");
2191                 quote_c_style(other, &msg, NULL, 0);
2192                 strbuf_addch(&msg, '\n');
2193                 break;
2194         case DIFF_STATUS_MODIFIED:
2195                 if (p->score) {
2196                         strbuf_addf(&msg, "dissimilarity index %d%%\n",
2197                                         similarity_index(p));
2198                         complete_rewrite = 1;
2199                         break;
2200                 }
2201                 /* fallthru */
2202         default:
2203                 /* nothing */
2204                 ;
2205         }
2206
2207         if (hashcmp(one->sha1, two->sha1)) {
2208                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2209
2210                 if (DIFF_OPT_TST(o, BINARY)) {
2211                         mmfile_t mf;
2212                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2213                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2214                                 abbrev = 40;
2215                 }
2216                 strbuf_addf(&msg, "index %.*s..%.*s",
2217                                 abbrev, sha1_to_hex(one->sha1),
2218                                 abbrev, sha1_to_hex(two->sha1));
2219                 if (one->mode == two->mode)
2220                         strbuf_addf(&msg, " %06o", one->mode);
2221                 strbuf_addch(&msg, '\n');
2222         }
2223
2224         if (msg.len)
2225                 strbuf_setlen(&msg, msg.len - 1);
2226         xfrm_msg = msg.len ? msg.buf : NULL;
2227
2228         if (!pgm &&
2229             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2230             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2231                 /* a filepair that changes between file and symlink
2232                  * needs to be split into deletion and creation.
2233                  */
2234                 struct diff_filespec *null = alloc_filespec(two->path);
2235                 run_diff_cmd(NULL, name, other, attr_path,
2236                              one, null, xfrm_msg, o, 0);
2237                 free(null);
2238                 null = alloc_filespec(one->path);
2239                 run_diff_cmd(NULL, name, other, attr_path,
2240                              null, two, xfrm_msg, o, 0);
2241                 free(null);
2242         }
2243         else
2244                 run_diff_cmd(pgm, name, other, attr_path,
2245                              one, two, xfrm_msg, o, complete_rewrite);
2246
2247         strbuf_release(&msg);
2248 }
2249
2250 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2251                          struct diffstat_t *diffstat)
2252 {
2253         const char *name;
2254         const char *other;
2255         int complete_rewrite = 0;
2256
2257         if (DIFF_PAIR_UNMERGED(p)) {
2258                 /* unmerged */
2259                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2260                 return;
2261         }
2262
2263         name = p->one->path;
2264         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2265
2266         if (o->prefix_length)
2267                 strip_prefix(o->prefix_length, &name, &other);
2268
2269         diff_fill_sha1_info(p->one);
2270         diff_fill_sha1_info(p->two);
2271
2272         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2273                 complete_rewrite = 1;
2274         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2275 }
2276
2277 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2278 {
2279         const char *name;
2280         const char *other;
2281         const char *attr_path;
2282
2283         if (DIFF_PAIR_UNMERGED(p)) {
2284                 /* unmerged */
2285                 return;
2286         }
2287
2288         name = p->one->path;
2289         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2290         attr_path = other ? other : name;
2291
2292         if (o->prefix_length)
2293                 strip_prefix(o->prefix_length, &name, &other);
2294
2295         diff_fill_sha1_info(p->one);
2296         diff_fill_sha1_info(p->two);
2297
2298         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2299 }
2300
2301 void diff_setup(struct diff_options *options)
2302 {
2303         memset(options, 0, sizeof(*options));
2304
2305         options->file = stdout;
2306
2307         options->line_termination = '\n';
2308         options->break_opt = -1;
2309         options->rename_limit = -1;
2310         options->dirstat_percent = 3;
2311         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
2312         options->context = 3;
2313
2314         options->change = diff_change;
2315         options->add_remove = diff_addremove;
2316         if (diff_use_color_default > 0)
2317                 DIFF_OPT_SET(options, COLOR_DIFF);
2318         else
2319                 DIFF_OPT_CLR(options, COLOR_DIFF);
2320         options->detect_rename = diff_detect_rename_default;
2321
2322         options->a_prefix = "a/";
2323         options->b_prefix = "b/";
2324 }
2325
2326 int diff_setup_done(struct diff_options *options)
2327 {
2328         int count = 0;
2329
2330         if (options->output_format & DIFF_FORMAT_NAME)
2331                 count++;
2332         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2333                 count++;
2334         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2335                 count++;
2336         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2337                 count++;
2338         if (count > 1)
2339                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2340
2341         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2342                 options->detect_rename = DIFF_DETECT_COPY;
2343
2344         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2345                 options->prefix = NULL;
2346         if (options->prefix)
2347                 options->prefix_length = strlen(options->prefix);
2348         else
2349                 options->prefix_length = 0;
2350
2351         if (options->output_format & (DIFF_FORMAT_NAME |
2352                                       DIFF_FORMAT_NAME_STATUS |
2353                                       DIFF_FORMAT_CHECKDIFF |
2354                                       DIFF_FORMAT_NO_OUTPUT))
2355                 options->output_format &= ~(DIFF_FORMAT_RAW |
2356                                             DIFF_FORMAT_NUMSTAT |
2357                                             DIFF_FORMAT_DIFFSTAT |
2358                                             DIFF_FORMAT_SHORTSTAT |
2359                                             DIFF_FORMAT_DIRSTAT |
2360                                             DIFF_FORMAT_SUMMARY |
2361                                             DIFF_FORMAT_PATCH);
2362
2363         /*
2364          * These cases always need recursive; we do not drop caller-supplied
2365          * recursive bits for other formats here.
2366          */
2367         if (options->output_format & (DIFF_FORMAT_PATCH |
2368                                       DIFF_FORMAT_NUMSTAT |
2369                                       DIFF_FORMAT_DIFFSTAT |
2370                                       DIFF_FORMAT_SHORTSTAT |
2371                                       DIFF_FORMAT_DIRSTAT |
2372                                       DIFF_FORMAT_SUMMARY |
2373                                       DIFF_FORMAT_CHECKDIFF))
2374                 DIFF_OPT_SET(options, RECURSIVE);
2375         /*
2376          * Also pickaxe would not work very well if you do not say recursive
2377          */
2378         if (options->pickaxe)
2379                 DIFF_OPT_SET(options, RECURSIVE);
2380
2381         if (options->detect_rename && options->rename_limit < 0)
2382                 options->rename_limit = diff_rename_limit_default;
2383         if (options->setup & DIFF_SETUP_USE_CACHE) {
2384                 if (!active_cache)
2385                         /* read-cache does not die even when it fails
2386                          * so it is safe for us to do this here.  Also
2387                          * it does not smudge active_cache or active_nr
2388                          * when it fails, so we do not have to worry about
2389                          * cleaning it up ourselves either.
2390                          */
2391                         read_cache();
2392         }
2393         if (options->abbrev <= 0 || 40 < options->abbrev)
2394                 options->abbrev = 40; /* full */
2395
2396         /*
2397          * It does not make sense to show the first hit we happened
2398          * to have found.  It does not make sense not to return with
2399          * exit code in such a case either.
2400          */
2401         if (DIFF_OPT_TST(options, QUIET)) {
2402                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2403                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2404         }
2405
2406         /*
2407          * If we postprocess in diffcore, we cannot simply return
2408          * upon the first hit.  We need to run diff as usual.
2409          */
2410         if (options->pickaxe || options->filter)
2411                 DIFF_OPT_CLR(options, QUIET);
2412
2413         return 0;
2414 }
2415
2416 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2417 {
2418         char c, *eq;
2419         int len;
2420
2421         if (*arg != '-')
2422                 return 0;
2423         c = *++arg;
2424         if (!c)
2425                 return 0;
2426         if (c == arg_short) {
2427                 c = *++arg;
2428                 if (!c)
2429                         return 1;
2430                 if (val && isdigit(c)) {
2431                         char *end;
2432                         int n = strtoul(arg, &end, 10);
2433                         if (*end)
2434                                 return 0;
2435                         *val = n;
2436                         return 1;
2437                 }
2438                 return 0;
2439         }
2440         if (c != '-')
2441                 return 0;
2442         arg++;
2443         eq = strchr(arg, '=');
2444         if (eq)
2445                 len = eq - arg;
2446         else
2447                 len = strlen(arg);
2448         if (!len || strncmp(arg, arg_long, len))
2449                 return 0;
2450         if (eq) {
2451                 int n;
2452                 char *end;
2453                 if (!isdigit(*++eq))
2454                         return 0;
2455                 n = strtoul(eq, &end, 10);
2456                 if (*end)
2457                         return 0;
2458                 *val = n;
2459         }
2460         return 1;
2461 }
2462
2463 static int diff_scoreopt_parse(const char *opt);
2464
2465 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2466 {
2467         const char *arg = av[0];
2468
2469         /* Output format options */
2470         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2471                 options->output_format |= DIFF_FORMAT_PATCH;
2472         else if (opt_arg(arg, 'U', "unified", &options->context))
2473                 options->output_format |= DIFF_FORMAT_PATCH;
2474         else if (!strcmp(arg, "--raw"))
2475                 options->output_format |= DIFF_FORMAT_RAW;
2476         else if (!strcmp(arg, "--patch-with-raw"))
2477                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2478         else if (!strcmp(arg, "--numstat"))
2479                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2480         else if (!strcmp(arg, "--shortstat"))
2481                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2482         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2483                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2484         else if (!strcmp(arg, "--cumulative")) {
2485                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2486                 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2487         }
2488         else if (!strcmp(arg, "--check"))
2489                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2490         else if (!strcmp(arg, "--summary"))
2491                 options->output_format |= DIFF_FORMAT_SUMMARY;
2492         else if (!strcmp(arg, "--patch-with-stat"))
2493                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2494         else if (!strcmp(arg, "--name-only"))
2495                 options->output_format |= DIFF_FORMAT_NAME;
2496         else if (!strcmp(arg, "--name-status"))
2497                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2498         else if (!strcmp(arg, "-s"))
2499                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2500         else if (!prefixcmp(arg, "--stat")) {
2501                 char *end;
2502                 int width = options->stat_width;
2503                 int name_width = options->stat_name_width;
2504                 arg += 6;
2505                 end = (char *)arg;
2506
2507                 switch (*arg) {
2508                 case '-':
2509                         if (!prefixcmp(arg, "-width="))
2510                                 width = strtoul(arg + 7, &end, 10);
2511                         else if (!prefixcmp(arg, "-name-width="))
2512                                 name_width = strtoul(arg + 12, &end, 10);
2513                         break;
2514                 case '=':
2515                         width = strtoul(arg+1, &end, 10);
2516                         if (*end == ',')
2517                                 name_width = strtoul(end+1, &end, 10);
2518                 }
2519
2520                 /* Important! This checks all the error cases! */
2521                 if (*end)
2522                         return 0;
2523                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2524                 options->stat_name_width = name_width;
2525                 options->stat_width = width;
2526         }
2527
2528         /* renames options */
2529         else if (!prefixcmp(arg, "-B")) {
2530                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2531                         return -1;
2532         }
2533         else if (!prefixcmp(arg, "-M")) {
2534                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2535                         return -1;
2536                 options->detect_rename = DIFF_DETECT_RENAME;
2537         }
2538         else if (!prefixcmp(arg, "-C")) {
2539                 if (options->detect_rename == DIFF_DETECT_COPY)
2540                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2541                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2542                         return -1;
2543                 options->detect_rename = DIFF_DETECT_COPY;
2544         }
2545         else if (!strcmp(arg, "--no-renames"))
2546                 options->detect_rename = 0;
2547         else if (!strcmp(arg, "--relative"))
2548                 DIFF_OPT_SET(options, RELATIVE_NAME);
2549         else if (!prefixcmp(arg, "--relative=")) {
2550                 DIFF_OPT_SET(options, RELATIVE_NAME);
2551                 options->prefix = arg + 11;
2552         }
2553
2554         /* xdiff options */
2555         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2556                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2557         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2558                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2559         else if (!strcmp(arg, "--ignore-space-at-eol"))
2560                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2561
2562         /* flags options */
2563         else if (!strcmp(arg, "--binary")) {
2564                 options->output_format |= DIFF_FORMAT_PATCH;
2565                 DIFF_OPT_SET(options, BINARY);
2566         }
2567         else if (!strcmp(arg, "--full-index"))
2568                 DIFF_OPT_SET(options, FULL_INDEX);
2569         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2570                 DIFF_OPT_SET(options, TEXT);
2571         else if (!strcmp(arg, "-R"))
2572                 DIFF_OPT_SET(options, REVERSE_DIFF);
2573         else if (!strcmp(arg, "--find-copies-harder"))
2574                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2575         else if (!strcmp(arg, "--follow"))
2576                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2577         else if (!strcmp(arg, "--color"))
2578                 DIFF_OPT_SET(options, COLOR_DIFF);
2579         else if (!strcmp(arg, "--no-color"))
2580                 DIFF_OPT_CLR(options, COLOR_DIFF);
2581         else if (!strcmp(arg, "--color-words"))
2582                 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2583         else if (!strcmp(arg, "--exit-code"))
2584                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2585         else if (!strcmp(arg, "--quiet"))
2586                 DIFF_OPT_SET(options, QUIET);
2587         else if (!strcmp(arg, "--ext-diff"))
2588                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2589         else if (!strcmp(arg, "--no-ext-diff"))
2590                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2591         else if (!strcmp(arg, "--ignore-submodules"))
2592                 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2593
2594         /* misc options */
2595         else if (!strcmp(arg, "-z"))
2596                 options->line_termination = 0;
2597         else if (!prefixcmp(arg, "-l"))
2598                 options->rename_limit = strtoul(arg+2, NULL, 10);
2599         else if (!prefixcmp(arg, "-S"))
2600                 options->pickaxe = arg + 2;
2601         else if (!strcmp(arg, "--pickaxe-all"))
2602                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2603         else if (!strcmp(arg, "--pickaxe-regex"))
2604                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2605         else if (!prefixcmp(arg, "-O"))
2606                 options->orderfile = arg + 2;
2607         else if (!prefixcmp(arg, "--diff-filter="))
2608                 options->filter = arg + 14;
2609         else if (!strcmp(arg, "--abbrev"))
2610                 options->abbrev = DEFAULT_ABBREV;
2611         else if (!prefixcmp(arg, "--abbrev=")) {
2612                 options->abbrev = strtoul(arg + 9, NULL, 10);
2613                 if (options->abbrev < MINIMUM_ABBREV)
2614                         options->abbrev = MINIMUM_ABBREV;
2615                 else if (40 < options->abbrev)
2616                         options->abbrev = 40;
2617         }
2618         else if (!prefixcmp(arg, "--src-prefix="))
2619                 options->a_prefix = arg + 13;
2620         else if (!prefixcmp(arg, "--dst-prefix="))
2621                 options->b_prefix = arg + 13;
2622         else if (!strcmp(arg, "--no-prefix"))
2623                 options->a_prefix = options->b_prefix = "";
2624         else if (!prefixcmp(arg, "--output=")) {
2625                 options->file = fopen(arg + strlen("--output="), "w");
2626                 options->close_file = 1;
2627         } else
2628                 return 0;
2629         return 1;
2630 }
2631
2632 static int parse_num(const char **cp_p)
2633 {
2634         unsigned long num, scale;
2635         int ch, dot;
2636         const char *cp = *cp_p;
2637
2638         num = 0;
2639         scale = 1;
2640         dot = 0;
2641         for(;;) {
2642                 ch = *cp;
2643                 if ( !dot && ch == '.' ) {
2644                         scale = 1;
2645                         dot = 1;
2646                 } else if ( ch == '%' ) {
2647                         scale = dot ? scale*100 : 100;
2648                         cp++;   /* % is always at the end */
2649                         break;
2650                 } else if ( ch >= '0' && ch <= '9' ) {
2651                         if ( scale < 100000 ) {
2652                                 scale *= 10;
2653                                 num = (num*10) + (ch-'0');
2654                         }
2655                 } else {
2656                         break;
2657                 }
2658                 cp++;
2659         }
2660         *cp_p = cp;
2661
2662         /* user says num divided by scale and we say internally that
2663          * is MAX_SCORE * num / scale.
2664          */
2665         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2666 }
2667
2668 static int diff_scoreopt_parse(const char *opt)
2669 {
2670         int opt1, opt2, cmd;
2671
2672         if (*opt++ != '-')
2673                 return -1;
2674         cmd = *opt++;
2675         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2676                 return -1; /* that is not a -M, -C nor -B option */
2677
2678         opt1 = parse_num(&opt);
2679         if (cmd != 'B')
2680                 opt2 = 0;
2681         else {
2682                 if (*opt == 0)
2683                         opt2 = 0;
2684                 else if (*opt != '/')
2685                         return -1; /* we expect -B80/99 or -B80 */
2686                 else {
2687                         opt++;
2688                         opt2 = parse_num(&opt);
2689                 }
2690         }
2691         if (*opt != 0)
2692                 return -1;
2693         return opt1 | (opt2 << 16);
2694 }
2695
2696 struct diff_queue_struct diff_queued_diff;
2697
2698 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2699 {
2700         if (queue->alloc <= queue->nr) {
2701                 queue->alloc = alloc_nr(queue->alloc);
2702                 queue->queue = xrealloc(queue->queue,
2703                                         sizeof(dp) * queue->alloc);
2704         }
2705         queue->queue[queue->nr++] = dp;
2706 }
2707
2708 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2709                                  struct diff_filespec *one,
2710                                  struct diff_filespec *two)
2711 {
2712         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2713         dp->one = one;
2714         dp->two = two;
2715         if (queue)
2716                 diff_q(queue, dp);
2717         return dp;
2718 }
2719
2720 void diff_free_filepair(struct diff_filepair *p)
2721 {
2722         free_filespec(p->one);
2723         free_filespec(p->two);
2724         free(p);
2725 }
2726
2727 /* This is different from find_unique_abbrev() in that
2728  * it stuffs the result with dots for alignment.
2729  */
2730 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2731 {
2732         int abblen;
2733         const char *abbrev;
2734         if (len == 40)
2735                 return sha1_to_hex(sha1);
2736
2737         abbrev = find_unique_abbrev(sha1, len);
2738         abblen = strlen(abbrev);
2739         if (abblen < 37) {
2740                 static char hex[41];
2741                 if (len < abblen && abblen <= len + 2)
2742                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2743                 else
2744                         sprintf(hex, "%s...", abbrev);
2745                 return hex;
2746         }
2747         return sha1_to_hex(sha1);
2748 }
2749
2750 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2751 {
2752         int line_termination = opt->line_termination;
2753         int inter_name_termination = line_termination ? '\t' : '\0';
2754
2755         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2756                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2757                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
2758                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2759         }
2760         if (p->score) {
2761                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2762                         inter_name_termination);
2763         } else {
2764                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2765         }
2766
2767         if (p->status == DIFF_STATUS_COPIED ||
2768             p->status == DIFF_STATUS_RENAMED) {
2769                 const char *name_a, *name_b;
2770                 name_a = p->one->path;
2771                 name_b = p->two->path;
2772                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2773                 write_name_quoted(name_a, opt->file, inter_name_termination);
2774                 write_name_quoted(name_b, opt->file, line_termination);
2775         } else {
2776                 const char *name_a, *name_b;
2777                 name_a = p->one->mode ? p->one->path : p->two->path;
2778                 name_b = NULL;
2779                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2780                 write_name_quoted(name_a, opt->file, line_termination);
2781         }
2782 }
2783
2784 int diff_unmodified_pair(struct diff_filepair *p)
2785 {
2786         /* This function is written stricter than necessary to support
2787          * the currently implemented transformers, but the idea is to
2788          * let transformers to produce diff_filepairs any way they want,
2789          * and filter and clean them up here before producing the output.
2790          */
2791         struct diff_filespec *one = p->one, *two = p->two;
2792
2793         if (DIFF_PAIR_UNMERGED(p))
2794                 return 0; /* unmerged is interesting */
2795
2796         /* deletion, addition, mode or type change
2797          * and rename are all interesting.
2798          */
2799         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2800             DIFF_PAIR_MODE_CHANGED(p) ||
2801             strcmp(one->path, two->path))
2802                 return 0;
2803
2804         /* both are valid and point at the same path.  that is, we are
2805          * dealing with a change.
2806          */
2807         if (one->sha1_valid && two->sha1_valid &&
2808             !hashcmp(one->sha1, two->sha1))
2809                 return 1; /* no change */
2810         if (!one->sha1_valid && !two->sha1_valid)
2811                 return 1; /* both look at the same file on the filesystem. */
2812         return 0;
2813 }
2814
2815 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
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_diff(p, o);
2825 }
2826
2827 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2828                             struct diffstat_t *diffstat)
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_diffstat(p, o, diffstat);
2838 }
2839
2840 static void diff_flush_checkdiff(struct diff_filepair *p,
2841                 struct diff_options *o)
2842 {
2843         if (diff_unmodified_pair(p))
2844                 return;
2845
2846         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2847             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2848                 return; /* no tree diffs in patch format */
2849
2850         run_checkdiff(p, o);
2851 }
2852
2853 int diff_queue_is_empty(void)
2854 {
2855         struct diff_queue_struct *q = &diff_queued_diff;
2856         int i;
2857         for (i = 0; i < q->nr; i++)
2858                 if (!diff_unmodified_pair(q->queue[i]))
2859                         return 0;
2860         return 1;
2861 }
2862
2863 #if DIFF_DEBUG
2864 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2865 {
2866         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2867                 x, one ? one : "",
2868                 s->path,
2869                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2870                 s->mode,
2871                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2872         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2873                 x, one ? one : "",
2874                 s->size, s->xfrm_flags);
2875 }
2876
2877 void diff_debug_filepair(const struct diff_filepair *p, int i)
2878 {
2879         diff_debug_filespec(p->one, i, "one");
2880         diff_debug_filespec(p->two, i, "two");
2881         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2882                 p->score, p->status ? p->status : '?',
2883                 p->one->rename_used, p->broken_pair);
2884 }
2885
2886 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2887 {
2888         int i;
2889         if (msg)
2890                 fprintf(stderr, "%s\n", msg);
2891         fprintf(stderr, "q->nr = %d\n", q->nr);
2892         for (i = 0; i < q->nr; i++) {
2893                 struct diff_filepair *p = q->queue[i];
2894                 diff_debug_filepair(p, i);
2895         }
2896 }
2897 #endif
2898
2899 static void diff_resolve_rename_copy(void)
2900 {
2901         int i;
2902         struct diff_filepair *p;
2903         struct diff_queue_struct *q = &diff_queued_diff;
2904
2905         diff_debug_queue("resolve-rename-copy", q);
2906
2907         for (i = 0; i < q->nr; i++) {
2908                 p = q->queue[i];
2909                 p->status = 0; /* undecided */
2910                 if (DIFF_PAIR_UNMERGED(p))
2911                         p->status = DIFF_STATUS_UNMERGED;
2912                 else if (!DIFF_FILE_VALID(p->one))
2913                         p->status = DIFF_STATUS_ADDED;
2914                 else if (!DIFF_FILE_VALID(p->two))
2915                         p->status = DIFF_STATUS_DELETED;
2916                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2917                         p->status = DIFF_STATUS_TYPE_CHANGED;
2918
2919                 /* from this point on, we are dealing with a pair
2920                  * whose both sides are valid and of the same type, i.e.
2921                  * either in-place edit or rename/copy edit.
2922                  */
2923                 else if (DIFF_PAIR_RENAME(p)) {
2924                         /*
2925                          * A rename might have re-connected a broken
2926                          * pair up, causing the pathnames to be the
2927                          * same again. If so, that's not a rename at
2928                          * all, just a modification..
2929                          *
2930                          * Otherwise, see if this source was used for
2931                          * multiple renames, in which case we decrement
2932                          * the count, and call it a copy.
2933                          */
2934                         if (!strcmp(p->one->path, p->two->path))
2935                                 p->status = DIFF_STATUS_MODIFIED;
2936                         else if (--p->one->rename_used > 0)
2937                                 p->status = DIFF_STATUS_COPIED;
2938                         else
2939                                 p->status = DIFF_STATUS_RENAMED;
2940                 }
2941                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2942                          p->one->mode != p->two->mode ||
2943                          is_null_sha1(p->one->sha1))
2944                         p->status = DIFF_STATUS_MODIFIED;
2945                 else {
2946                         /* This is a "no-change" entry and should not
2947                          * happen anymore, but prepare for broken callers.
2948                          */
2949                         error("feeding unmodified %s to diffcore",
2950                               p->one->path);
2951                         p->status = DIFF_STATUS_UNKNOWN;
2952                 }
2953         }
2954         diff_debug_queue("resolve-rename-copy done", q);
2955 }
2956
2957 static int check_pair_status(struct diff_filepair *p)
2958 {
2959         switch (p->status) {
2960         case DIFF_STATUS_UNKNOWN:
2961                 return 0;
2962         case 0:
2963                 die("internal error in diff-resolve-rename-copy");
2964         default:
2965                 return 1;
2966         }
2967 }
2968
2969 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2970 {
2971         int fmt = opt->output_format;
2972
2973         if (fmt & DIFF_FORMAT_CHECKDIFF)
2974                 diff_flush_checkdiff(p, opt);
2975         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2976                 diff_flush_raw(p, opt);
2977         else if (fmt & DIFF_FORMAT_NAME) {
2978                 const char *name_a, *name_b;
2979                 name_a = p->two->path;
2980                 name_b = NULL;
2981                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2982                 write_name_quoted(name_a, opt->file, opt->line_termination);
2983         }
2984 }
2985
2986 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2987 {
2988         if (fs->mode)
2989                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2990         else
2991                 fprintf(file, " %s ", newdelete);
2992         write_name_quoted(fs->path, file, '\n');
2993 }
2994
2995
2996 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
2997 {
2998         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2999                 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
3000                         show_name ? ' ' : '\n');
3001                 if (show_name) {
3002                         write_name_quoted(p->two->path, file, '\n');
3003                 }
3004         }
3005 }
3006
3007 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3008 {
3009         char *names = pprint_rename(p->one->path, p->two->path);
3010
3011         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3012         free(names);
3013         show_mode_change(file, p, 0);
3014 }
3015
3016 static void diff_summary(FILE *file, struct diff_filepair *p)
3017 {
3018         switch(p->status) {
3019         case DIFF_STATUS_DELETED:
3020                 show_file_mode_name(file, "delete", p->one);
3021                 break;
3022         case DIFF_STATUS_ADDED:
3023                 show_file_mode_name(file, "create", p->two);
3024                 break;
3025         case DIFF_STATUS_COPIED:
3026                 show_rename_copy(file, "copy", p);
3027                 break;
3028         case DIFF_STATUS_RENAMED:
3029                 show_rename_copy(file, "rename", p);
3030                 break;
3031         default:
3032                 if (p->score) {
3033                         fputs(" rewrite ", file);
3034                         write_name_quoted(p->two->path, file, ' ');
3035                         fprintf(file, "(%d%%)\n", similarity_index(p));
3036                 }
3037                 show_mode_change(file, p, !p->score);
3038                 break;
3039         }
3040 }
3041
3042 struct patch_id_t {
3043         SHA_CTX *ctx;
3044         int patchlen;
3045 };
3046
3047 static int remove_space(char *line, int len)
3048 {
3049         int i;
3050         char *dst = line;
3051         unsigned char c;
3052
3053         for (i = 0; i < len; i++)
3054                 if (!isspace((c = line[i])))
3055                         *dst++ = c;
3056
3057         return dst - line;
3058 }
3059
3060 static void patch_id_consume(void *priv, char *line, unsigned long len)
3061 {
3062         struct patch_id_t *data = priv;
3063         int new_len;
3064
3065         /* Ignore line numbers when computing the SHA1 of the patch */
3066         if (!prefixcmp(line, "@@ -"))
3067                 return;
3068
3069         new_len = remove_space(line, len);
3070
3071         SHA1_Update(data->ctx, line, new_len);
3072         data->patchlen += new_len;
3073 }
3074
3075 /* returns 0 upon success, and writes result into sha1 */
3076 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3077 {
3078         struct diff_queue_struct *q = &diff_queued_diff;
3079         int i;
3080         SHA_CTX ctx;
3081         struct patch_id_t data;
3082         char buffer[PATH_MAX * 4 + 20];
3083
3084         SHA1_Init(&ctx);
3085         memset(&data, 0, sizeof(struct patch_id_t));
3086         data.ctx = &ctx;
3087
3088         for (i = 0; i < q->nr; i++) {
3089                 xpparam_t xpp;
3090                 xdemitconf_t xecfg;
3091                 xdemitcb_t ecb;
3092                 mmfile_t mf1, mf2;
3093                 struct diff_filepair *p = q->queue[i];
3094                 int len1, len2;
3095
3096                 memset(&xecfg, 0, sizeof(xecfg));
3097                 if (p->status == 0)
3098                         return error("internal diff status error");
3099                 if (p->status == DIFF_STATUS_UNKNOWN)
3100                         continue;
3101                 if (diff_unmodified_pair(p))
3102                         continue;
3103                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3104                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3105                         continue;
3106                 if (DIFF_PAIR_UNMERGED(p))
3107                         continue;
3108
3109                 diff_fill_sha1_info(p->one);
3110                 diff_fill_sha1_info(p->two);
3111                 if (fill_mmfile(&mf1, p->one) < 0 ||
3112                                 fill_mmfile(&mf2, p->two) < 0)
3113                         return error("unable to read files to diff");
3114
3115                 len1 = remove_space(p->one->path, strlen(p->one->path));
3116                 len2 = remove_space(p->two->path, strlen(p->two->path));
3117                 if (p->one->mode == 0)
3118                         len1 = snprintf(buffer, sizeof(buffer),
3119                                         "diff--gita/%.*sb/%.*s"
3120                                         "newfilemode%06o"
3121                                         "---/dev/null"
3122                                         "+++b/%.*s",
3123                                         len1, p->one->path,
3124                                         len2, p->two->path,
3125                                         p->two->mode,
3126                                         len2, p->two->path);
3127                 else if (p->two->mode == 0)
3128                         len1 = snprintf(buffer, sizeof(buffer),
3129                                         "diff--gita/%.*sb/%.*s"
3130                                         "deletedfilemode%06o"
3131                                         "---a/%.*s"
3132                                         "+++/dev/null",
3133                                         len1, p->one->path,
3134                                         len2, p->two->path,
3135                                         p->one->mode,
3136                                         len1, p->one->path);
3137                 else
3138                         len1 = snprintf(buffer, sizeof(buffer),
3139                                         "diff--gita/%.*sb/%.*s"
3140                                         "---a/%.*s"
3141                                         "+++b/%.*s",
3142                                         len1, p->one->path,
3143                                         len2, p->two->path,
3144                                         len1, p->one->path,
3145                                         len2, p->two->path);
3146                 SHA1_Update(&ctx, buffer, len1);
3147
3148                 xpp.flags = XDF_NEED_MINIMAL;
3149                 xecfg.ctxlen = 3;
3150                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3151                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3152                               &xpp, &xecfg, &ecb);
3153         }
3154
3155         SHA1_Final(sha1, &ctx);
3156         return 0;
3157 }
3158
3159 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3160 {
3161         struct diff_queue_struct *q = &diff_queued_diff;
3162         int i;
3163         int result = diff_get_patch_id(options, sha1);
3164
3165         for (i = 0; i < q->nr; i++)
3166                 diff_free_filepair(q->queue[i]);
3167
3168         free(q->queue);
3169         q->queue = NULL;
3170         q->nr = q->alloc = 0;
3171
3172         return result;
3173 }
3174
3175 static int is_summary_empty(const struct diff_queue_struct *q)
3176 {
3177         int i;
3178
3179         for (i = 0; i < q->nr; i++) {
3180                 const struct diff_filepair *p = q->queue[i];
3181
3182                 switch (p->status) {
3183                 case DIFF_STATUS_DELETED:
3184                 case DIFF_STATUS_ADDED:
3185                 case DIFF_STATUS_COPIED:
3186                 case DIFF_STATUS_RENAMED:
3187                         return 0;
3188                 default:
3189                         if (p->score)
3190                                 return 0;
3191                         if (p->one->mode && p->two->mode &&
3192                             p->one->mode != p->two->mode)
3193                                 return 0;
3194                         break;
3195                 }
3196         }
3197         return 1;
3198 }
3199
3200 void diff_flush(struct diff_options *options)
3201 {
3202         struct diff_queue_struct *q = &diff_queued_diff;
3203         int i, output_format = options->output_format;
3204         int separator = 0;
3205
3206         /*
3207          * Order: raw, stat, summary, patch
3208          * or:    name/name-status/checkdiff (other bits clear)
3209          */
3210         if (!q->nr)
3211                 goto free_queue;
3212
3213         if (output_format & (DIFF_FORMAT_RAW |
3214                              DIFF_FORMAT_NAME |
3215                              DIFF_FORMAT_NAME_STATUS |
3216                              DIFF_FORMAT_CHECKDIFF)) {
3217                 for (i = 0; i < q->nr; i++) {
3218                         struct diff_filepair *p = q->queue[i];
3219                         if (check_pair_status(p))
3220                                 flush_one_pair(p, options);
3221                 }
3222                 separator++;
3223         }
3224
3225         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3226                 struct diffstat_t diffstat;
3227
3228                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3229                 for (i = 0; i < q->nr; i++) {
3230                         struct diff_filepair *p = q->queue[i];
3231                         if (check_pair_status(p))
3232                                 diff_flush_stat(p, options, &diffstat);
3233                 }
3234                 if (output_format & DIFF_FORMAT_NUMSTAT)
3235                         show_numstat(&diffstat, options);
3236                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3237                         show_stats(&diffstat, options);
3238                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3239                         show_shortstats(&diffstat, options);
3240                 free_diffstat_info(&diffstat);
3241                 separator++;
3242         }
3243         if (output_format & DIFF_FORMAT_DIRSTAT)
3244                 show_dirstat(options);
3245
3246         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3247                 for (i = 0; i < q->nr; i++)
3248                         diff_summary(options->file, q->queue[i]);
3249                 separator++;
3250         }
3251
3252         if (output_format & DIFF_FORMAT_PATCH) {
3253                 if (separator) {
3254                         putc(options->line_termination, options->file);
3255                         if (options->stat_sep) {
3256                                 /* attach patch instead of inline */
3257                                 fputs(options->stat_sep, options->file);
3258                         }
3259                 }
3260
3261                 for (i = 0; i < q->nr; i++) {
3262                         struct diff_filepair *p = q->queue[i];
3263                         if (check_pair_status(p))
3264                                 diff_flush_patch(p, options);
3265                 }
3266         }
3267
3268         if (output_format & DIFF_FORMAT_CALLBACK)
3269                 options->format_callback(q, options, options->format_callback_data);
3270
3271         for (i = 0; i < q->nr; i++)
3272                 diff_free_filepair(q->queue[i]);
3273 free_queue:
3274         free(q->queue);
3275         q->queue = NULL;
3276         q->nr = q->alloc = 0;
3277         if (options->close_file)
3278                 fclose(options->file);
3279 }
3280
3281 static void diffcore_apply_filter(const char *filter)
3282 {
3283         int i;
3284         struct diff_queue_struct *q = &diff_queued_diff;
3285         struct diff_queue_struct outq;
3286         outq.queue = NULL;
3287         outq.nr = outq.alloc = 0;
3288
3289         if (!filter)
3290                 return;
3291
3292         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3293                 int found;
3294                 for (i = found = 0; !found && i < q->nr; i++) {
3295                         struct diff_filepair *p = q->queue[i];
3296                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3297                              ((p->score &&
3298                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3299                               (!p->score &&
3300                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3301                             ((p->status != DIFF_STATUS_MODIFIED) &&
3302                              strchr(filter, p->status)))
3303                                 found++;
3304                 }
3305                 if (found)
3306                         return;
3307
3308                 /* otherwise we will clear the whole queue
3309                  * by copying the empty outq at the end of this
3310                  * function, but first clear the current entries
3311                  * in the queue.
3312                  */
3313                 for (i = 0; i < q->nr; i++)
3314                         diff_free_filepair(q->queue[i]);
3315         }
3316         else {
3317                 /* Only the matching ones */
3318                 for (i = 0; i < q->nr; i++) {
3319                         struct diff_filepair *p = q->queue[i];
3320
3321                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3322                              ((p->score &&
3323                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3324                               (!p->score &&
3325                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3326                             ((p->status != DIFF_STATUS_MODIFIED) &&
3327                              strchr(filter, p->status)))
3328                                 diff_q(&outq, p);
3329                         else
3330                                 diff_free_filepair(p);
3331                 }
3332         }
3333         free(q->queue);
3334         *q = outq;
3335 }
3336
3337 /* Check whether two filespecs with the same mode and size are identical */
3338 static int diff_filespec_is_identical(struct diff_filespec *one,
3339                                       struct diff_filespec *two)
3340 {
3341         if (S_ISGITLINK(one->mode))
3342                 return 0;
3343         if (diff_populate_filespec(one, 0))
3344                 return 0;
3345         if (diff_populate_filespec(two, 0))
3346                 return 0;
3347         return !memcmp(one->data, two->data, one->size);
3348 }
3349
3350 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3351 {
3352         int i;
3353         struct diff_queue_struct *q = &diff_queued_diff;
3354         struct diff_queue_struct outq;
3355         outq.queue = NULL;
3356         outq.nr = outq.alloc = 0;
3357
3358         for (i = 0; i < q->nr; i++) {
3359                 struct diff_filepair *p = q->queue[i];
3360
3361                 /*
3362                  * 1. Entries that come from stat info dirtyness
3363                  *    always have both sides (iow, not create/delete),
3364                  *    one side of the object name is unknown, with
3365                  *    the same mode and size.  Keep the ones that
3366                  *    do not match these criteria.  They have real
3367                  *    differences.
3368                  *
3369                  * 2. At this point, the file is known to be modified,
3370                  *    with the same mode and size, and the object
3371                  *    name of one side is unknown.  Need to inspect
3372                  *    the identical contents.
3373                  */
3374                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3375                     !DIFF_FILE_VALID(p->two) ||
3376                     (p->one->sha1_valid && p->two->sha1_valid) ||
3377                     (p->one->mode != p->two->mode) ||
3378                     diff_populate_filespec(p->one, 1) ||
3379                     diff_populate_filespec(p->two, 1) ||
3380                     (p->one->size != p->two->size) ||
3381                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3382                         diff_q(&outq, p);
3383                 else {
3384                         /*
3385                          * The caller can subtract 1 from skip_stat_unmatch
3386                          * to determine how many paths were dirty only
3387                          * due to stat info mismatch.
3388                          */
3389                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3390                                 diffopt->skip_stat_unmatch++;
3391                         diff_free_filepair(p);
3392                 }
3393         }
3394         free(q->queue);
3395         *q = outq;
3396 }
3397
3398 void diffcore_std(struct diff_options *options)
3399 {
3400         if (DIFF_OPT_TST(options, QUIET))
3401                 return;
3402
3403         if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3404                 diffcore_skip_stat_unmatch(options);
3405         if (options->break_opt != -1)
3406                 diffcore_break(options->break_opt);
3407         if (options->detect_rename)
3408                 diffcore_rename(options);
3409         if (options->break_opt != -1)
3410                 diffcore_merge_broken();
3411         if (options->pickaxe)
3412                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3413         if (options->orderfile)
3414                 diffcore_order(options->orderfile);
3415         diff_resolve_rename_copy();
3416         diffcore_apply_filter(options->filter);
3417
3418         if (diff_queued_diff.nr)
3419                 DIFF_OPT_SET(options, HAS_CHANGES);
3420         else
3421                 DIFF_OPT_CLR(options, HAS_CHANGES);
3422 }
3423
3424 int diff_result_code(struct diff_options *opt, int status)
3425 {
3426         int result = 0;
3427         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3428             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3429                 return status;
3430         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3431             DIFF_OPT_TST(opt, HAS_CHANGES))
3432                 result |= 01;
3433         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3434             DIFF_OPT_TST(opt, CHECK_FAILED))
3435                 result |= 02;
3436         return result;
3437 }
3438
3439 void diff_addremove(struct diff_options *options,
3440                     int addremove, unsigned mode,
3441                     const unsigned char *sha1,
3442                     const char *concatpath)
3443 {
3444         struct diff_filespec *one, *two;
3445
3446         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3447                 return;
3448
3449         /* This may look odd, but it is a preparation for
3450          * feeding "there are unchanged files which should
3451          * not produce diffs, but when you are doing copy
3452          * detection you would need them, so here they are"
3453          * entries to the diff-core.  They will be prefixed
3454          * with something like '=' or '*' (I haven't decided
3455          * which but should not make any difference).
3456          * Feeding the same new and old to diff_change()
3457          * also has the same effect.
3458          * Before the final output happens, they are pruned after
3459          * merged into rename/copy pairs as appropriate.
3460          */
3461         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3462                 addremove = (addremove == '+' ? '-' :
3463                              addremove == '-' ? '+' : addremove);
3464
3465         if (options->prefix &&
3466             strncmp(concatpath, options->prefix, options->prefix_length))
3467                 return;
3468
3469         one = alloc_filespec(concatpath);
3470         two = alloc_filespec(concatpath);
3471
3472         if (addremove != '+')
3473                 fill_filespec(one, sha1, mode);
3474         if (addremove != '-')
3475                 fill_filespec(two, sha1, mode);
3476
3477         diff_queue(&diff_queued_diff, one, two);
3478         DIFF_OPT_SET(options, HAS_CHANGES);
3479 }
3480
3481 void diff_change(struct diff_options *options,
3482                  unsigned old_mode, unsigned new_mode,
3483                  const unsigned char *old_sha1,
3484                  const unsigned char *new_sha1,
3485                  const char *concatpath)
3486 {
3487         struct diff_filespec *one, *two;
3488
3489         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3490                         && S_ISGITLINK(new_mode))
3491                 return;
3492
3493         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3494                 unsigned tmp;
3495                 const unsigned char *tmp_c;
3496                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3497                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3498         }
3499
3500         if (options->prefix &&
3501             strncmp(concatpath, options->prefix, options->prefix_length))
3502                 return;
3503
3504         one = alloc_filespec(concatpath);
3505         two = alloc_filespec(concatpath);
3506         fill_filespec(one, old_sha1, old_mode);
3507         fill_filespec(two, new_sha1, new_mode);
3508
3509         diff_queue(&diff_queued_diff, one, two);
3510         DIFF_OPT_SET(options, HAS_CHANGES);
3511 }
3512
3513 void diff_unmerge(struct diff_options *options,
3514                   const char *path,
3515                   unsigned mode, const unsigned char *sha1)
3516 {
3517         struct diff_filespec *one, *two;
3518
3519         if (options->prefix &&
3520             strncmp(path, options->prefix, options->prefix_length))
3521                 return;
3522
3523         one = alloc_filespec(path);
3524         two = alloc_filespec(path);
3525         fill_filespec(one, sha1, mode);
3526         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3527 }