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 int diff_use_color_default = -1;
  24 static const char *external_diff_cmd_cfg;
  25 int diff_auto_refresh_index = 1;
  26
  27 static char diff_colors[][COLOR_MAXLEN] = {
  28         "\033[m",       /* reset */
  29         "",             /* PLAIN (normal) */
  30         "\033[1m",      /* METAINFO (bold) */
  31         "\033[36m",     /* FRAGINFO (cyan) */
  32         "\033[31m",     /* OLD (red) */
  33         "\033[32m",     /* NEW (green) */
  34         "\033[33m",     /* COMMIT (yellow) */
  35         "\033[41m",     /* WHITESPACE (red background) */
  36 };
  37
  38 static int parse_diff_color_slot(const char *var, int ofs)
  39 {
  40         if (!strcasecmp(var+ofs, "plain"))
  41                 return DIFF_PLAIN;
  42         if (!strcasecmp(var+ofs, "meta"))
  43                 return DIFF_METAINFO;
  44         if (!strcasecmp(var+ofs, "frag"))
  45                 return DIFF_FRAGINFO;
  46         if (!strcasecmp(var+ofs, "old"))
  47                 return DIFF_FILE_OLD;
  48         if (!strcasecmp(var+ofs, "new"))
  49                 return DIFF_FILE_NEW;
  50         if (!strcasecmp(var+ofs, "commit"))
  51                 return DIFF_COMMIT;
  52         if (!strcasecmp(var+ofs, "whitespace"))
  53                 return DIFF_WHITESPACE;
  54         die("bad config variable '%s'", var);
  55 }
  56
  57 static struct ll_diff_driver {
  58         const char *name;
  59         struct ll_diff_driver *next;
  60         const char *cmd;
  61 } *user_diff, **user_diff_tail;
  62
  63 /*
  64  * Currently there is only "diff.<drivername>.command" variable;
  65  * because there are "diff.color.<slot>" variables, we are parsing
  66  * this in a bit convoluted way to allow low level diff driver
  67  * called "color".
  68  */
  69 static int parse_lldiff_command(const char *var, const char *ep, const char *value)
  70 {
  71         const char *name;
  72         int namelen;
  73         struct ll_diff_driver *drv;
  74
  75         name = var + 5;
  76         namelen = ep - name;
  77         for (drv = user_diff; drv; drv = drv->next)
  78                 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
  79                         break;
  80         if (!drv) {
  81                 drv = xcalloc(1, sizeof(struct ll_diff_driver));
  82                 drv->name = xmemdupz(name, namelen);
  83                 if (!user_diff_tail)
  84                         user_diff_tail = &user_diff;
  85                 *user_diff_tail = drv;
  86                 user_diff_tail = &(drv->next);
  87         }
  88
  89         return git_config_string(&(drv->cmd), var, value);
  90 }
  91
  92 /*
  93  * 'diff.<what>.funcname' attribute can be specified in the configuration
  94  * to define a customized regexp to find the beginning of a function to
  95  * be used for hunk header lines of "diff -p" style output.
  96  */
  97 struct funcname_pattern_entry {
  98         char *name;
  99         char *pattern;
 100 };
 101 static struct funcname_pattern_list {
 102         struct funcname_pattern_list *next;
 103         struct funcname_pattern_entry e;
 104 } *funcname_pattern_list;
 105
 106 static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
 107 {
 108         const char *name;
 109         int namelen;
 110         struct funcname_pattern_list *pp;
 111
 112         name = var + 5; /* "diff." */
 113         namelen = ep - name;
 114
 115         for (pp = funcname_pattern_list; pp; pp = pp->next)
 116                 if (!strncmp(pp->e.name, name, namelen) && !pp->e.name[namelen])
 117                         break;
 118         if (!pp) {
 119                 pp = xcalloc(1, sizeof(*pp));
 120                 pp->e.name = xmemdupz(name, namelen);
 121                 pp->next = funcname_pattern_list;
 122                 funcname_pattern_list = pp;
 123         }
 124         free(pp->e.pattern);
 125         pp->e.pattern = xstrdup(value);
 126         return 0;
 127 }
 128
 129 /*
 130  * These are to give UI layer defaults.
 131  * The core-level commands such as git-diff-files should
 132  * never be affected by the setting of diff.renames
 133  * the user happens to have in the configuration file.
 134  */
 135 int git_diff_ui_config(const char *var, const char *value, void *cb)
 136 {
 137         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 138                 diff_use_color_default = git_config_colorbool(var, value, -1);
 139                 return 0;
 140         }
 141         if (!strcmp(var, "diff.renames")) {
 142                 if (!value)
 143                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 144                 else if (!strcasecmp(value, "copies") ||
 145                          !strcasecmp(value, "copy"))
 146                         diff_detect_rename_default = DIFF_DETECT_COPY;
 147                 else if (git_config_bool(var,value))
 148                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 149                 return 0;
 150         }
 151         if (!strcmp(var, "diff.autorefreshindex")) {
 152                 diff_auto_refresh_index = git_config_bool(var, value);
 153                 return 0;
 154         }
 155         if (!strcmp(var, "diff.external"))
 156                 return git_config_string(&external_diff_cmd_cfg, var, value);
 157         if (!prefixcmp(var, "diff.")) {
 158                 const char *ep = strrchr(var, '.');
 159
 160                 if (ep != var + 4 && !strcmp(ep, ".command"))
 161                         return parse_lldiff_command(var, ep, value);
 162         }
 163
 164         return git_diff_basic_config(var, value, cb);
 165 }
 166
 167 int git_diff_basic_config(const char *var, const char *value, void *cb)
 168 {
 169         if (!strcmp(var, "diff.renamelimit")) {
 170                 diff_rename_limit_default = git_config_int(var, value);
 171                 return 0;
 172         }
 173
 174         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 175                 int slot = parse_diff_color_slot(var, 11);
 176                 if (!value)
 177                         return config_error_nonbool(var);
 178                 color_parse(value, var, diff_colors[slot]);
 179                 return 0;
 180         }
 181
 182         if (!prefixcmp(var, "diff.")) {
 183                 const char *ep = strrchr(var, '.');
 184                 if (ep != var + 4) {
 185                         if (!strcmp(ep, ".funcname")) {
 186                                 if (!value)
 187                                         return config_error_nonbool(var);
 188                                 return parse_funcname_pattern(var, ep, value);
 189                         }
 190                 }
 191         }
 192
 193         return git_color_default_config(var, value, cb);
 194 }
 195
 196 static char *quote_two(const char *one, const char *two)
 197 {
 198         int need_one = quote_c_style(one, NULL, NULL, 1);
 199         int need_two = quote_c_style(two, NULL, NULL, 1);
 200         struct strbuf res;
 201
 202         strbuf_init(&res, 0);
 203         if (need_one + need_two) {
 204                 strbuf_addch(&res, '"');
 205                 quote_c_style(one, &res, NULL, 1);
 206                 quote_c_style(two, &res, NULL, 1);
 207                 strbuf_addch(&res, '"');
 208         } else {
 209                 strbuf_addstr(&res, one);
 210                 strbuf_addstr(&res, two);
 211         }
 212         return strbuf_detach(&res, NULL);
 213 }
 214
 215 static const char *external_diff(void)
 216 {
 217         static const char *external_diff_cmd = NULL;
 218         static int done_preparing = 0;
 219
 220         if (done_preparing)
 221                 return external_diff_cmd;
 222         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 223         if (!external_diff_cmd)
 224                 external_diff_cmd = external_diff_cmd_cfg;
 225         done_preparing = 1;
 226         return external_diff_cmd;
 227 }
 228
 229 static struct diff_tempfile {
 230         const char *name; /* filename external diff should read from */
 231         char hex[41];
 232         char mode[10];
 233         char tmp_path[PATH_MAX];
 234 } diff_temp[2];
 235
 236 static int count_lines(const char *data, int size)
 237 {
 238         int count, ch, completely_empty = 1, nl_just_seen = 0;
 239         count = 0;
 240         while (0 < size--) {
 241                 ch = *data++;
 242                 if (ch == '\n') {
 243                         count++;
 244                         nl_just_seen = 1;
 245                         completely_empty = 0;
 246                 }
 247                 else {
 248                         nl_just_seen = 0;
 249                         completely_empty = 0;
 250                 }
 251         }
 252         if (completely_empty)
 253                 return 0;
 254         if (!nl_just_seen)
 255                 count++; /* no trailing newline */
 256         return count;
 257 }
 258
 259 static void print_line_count(FILE *file, int count)
 260 {
 261         switch (count) {
 262         case 0:
 263                 fprintf(file, "0,0");
 264                 break;
 265         case 1:
 266                 fprintf(file, "1");
 267                 break;
 268         default:
 269                 fprintf(file, "1,%d", count);
 270                 break;
 271         }
 272 }
 273
 274 static void copy_file_with_prefix(FILE *file,
 275                                   int prefix, const char *data, int size,
 276                                   const char *set, const char *reset)
 277 {
 278         int ch, nl_just_seen = 1;
 279         while (0 < size--) {
 280                 ch = *data++;
 281                 if (nl_just_seen) {
 282                         fputs(set, file);
 283                         putc(prefix, file);
 284                 }
 285                 if (ch == '\n') {
 286                         nl_just_seen = 1;
 287                         fputs(reset, file);
 288                 } else
 289                         nl_just_seen = 0;
 290                 putc(ch, file);
 291         }
 292         if (!nl_just_seen)
 293                 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
 294 }
 295
 296 static void emit_rewrite_diff(const char *name_a,
 297                               const char *name_b,
 298                               struct diff_filespec *one,
 299                               struct diff_filespec *two,
 300                               struct diff_options *o)
 301 {
 302         int lc_a, lc_b;
 303         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
 304         const char *name_a_tab, *name_b_tab;
 305         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 306         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 307         const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
 308         const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
 309         const char *reset = diff_get_color(color_diff, DIFF_RESET);
 310         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 311
 312         name_a += (*name_a == '/');
 313         name_b += (*name_b == '/');
 314         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 315         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 316
 317         strbuf_reset(&a_name);
 318         strbuf_reset(&b_name);
 319         quote_two_c_style(&a_name, o->a_prefix, name_a, 0);
 320         quote_two_c_style(&b_name, o->b_prefix, name_b, 0);
 321
 322         diff_populate_filespec(one, 0);
 323         diff_populate_filespec(two, 0);
 324         lc_a = count_lines(one->data, one->size);
 325         lc_b = count_lines(two->data, two->size);
 326         fprintf(o->file,
 327                 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
 328                 metainfo, a_name.buf, name_a_tab, reset,
 329                 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
 330         print_line_count(o->file, lc_a);
 331         fprintf(o->file, " +");
 332         print_line_count(o->file, lc_b);
 333         fprintf(o->file, " @@%s\n", reset);
 334         if (lc_a)
 335                 copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
 336         if (lc_b)
 337                 copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
 338 }
 339
 340 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 341 {
 342         if (!DIFF_FILE_VALID(one)) {
 343                 mf->ptr = (char *)""; /* does not matter */
 344                 mf->size = 0;
 345                 return 0;
 346         }
 347         else if (diff_populate_filespec(one, 0))
 348                 return -1;
 349         mf->ptr = one->data;
 350         mf->size = one->size;
 351         return 0;
 352 }
 353
 354 struct diff_words_buffer {
 355         mmfile_t text;
 356         long alloc;
 357         long current; /* output pointer */
 358         int suppressed_newline;
 359 };
 360
 361 static void diff_words_append(char *line, unsigned long len,
 362                 struct diff_words_buffer *buffer)
 363 {
 364         if (buffer->text.size + len > buffer->alloc) {
 365                 buffer->alloc = (buffer->text.size + len) * 3 / 2;
 366                 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
 367         }
 368         line++;
 369         len--;
 370         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 371         buffer->text.size += len;
 372 }
 373
 374 struct diff_words_data {
 375         struct xdiff_emit_state xm;
 376         struct diff_words_buffer minus, plus;
 377         FILE *file;
 378 };
 379
 380 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
 381                 int suppress_newline)
 382 {
 383         const char *ptr;
 384         int eol = 0;
 385
 386         if (len == 0)
 387                 return;
 388
 389         ptr  = buffer->text.ptr + buffer->current;
 390         buffer->current += len;
 391
 392         if (ptr[len - 1] == '\n') {
 393                 eol = 1;
 394                 len--;
 395         }
 396
 397         fputs(diff_get_color(1, color), file);
 398         fwrite(ptr, len, 1, file);
 399         fputs(diff_get_color(1, DIFF_RESET), file);
 400
 401         if (eol) {
 402                 if (suppress_newline)
 403                         buffer->suppressed_newline = 1;
 404                 else
 405                         putc('\n', file);
 406         }
 407 }
 408
 409 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 410 {
 411         struct diff_words_data *diff_words = priv;
 412
 413         if (diff_words->minus.suppressed_newline) {
 414                 if (line[0] != '+')
 415                         putc('\n', diff_words->file);
 416                 diff_words->minus.suppressed_newline = 0;
 417         }
 418
 419         len--;
 420         switch (line[0]) {
 421                 case '-':
 422                         print_word(diff_words->file,
 423                                    &diff_words->minus, len, DIFF_FILE_OLD, 1);
 424                         break;
 425                 case '+':
 426                         print_word(diff_words->file,
 427                                    &diff_words->plus, len, DIFF_FILE_NEW, 0);
 428                         break;
 429                 case ' ':
 430                         print_word(diff_words->file,
 431                                    &diff_words->plus, len, DIFF_PLAIN, 0);
 432                         diff_words->minus.current += len;
 433                         break;
 434         }
 435 }
 436
 437 /* this executes the word diff on the accumulated buffers */
 438 static void diff_words_show(struct diff_words_data *diff_words)
 439 {
 440         xpparam_t xpp;
 441         xdemitconf_t xecfg;
 442         xdemitcb_t ecb;
 443         mmfile_t minus, plus;
 444         int i;
 445
 446         memset(&xecfg, 0, sizeof(xecfg));
 447         minus.size = diff_words->minus.text.size;
 448         minus.ptr = xmalloc(minus.size);
 449         memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
 450         for (i = 0; i < minus.size; i++)
 451                 if (isspace(minus.ptr[i]))
 452                         minus.ptr[i] = '\n';
 453         diff_words->minus.current = 0;
 454
 455         plus.size = diff_words->plus.text.size;
 456         plus.ptr = xmalloc(plus.size);
 457         memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
 458         for (i = 0; i < plus.size; i++)
 459                 if (isspace(plus.ptr[i]))
 460                         plus.ptr[i] = '\n';
 461         diff_words->plus.current = 0;
 462
 463         xpp.flags = XDF_NEED_MINIMAL;
 464         xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
 465         ecb.outf = xdiff_outf;
 466         ecb.priv = diff_words;
 467         diff_words->xm.consume = fn_out_diff_words_aux;
 468         xdi_diff(&minus, &plus, &xpp, &xecfg, &ecb);
 469
 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         struct xdiff_emit_state xm;
 484         int nparents, color_diff;
 485         unsigned ws_rule;
 486         sane_truncate_fn truncate;
 487         const char **label_path;
 488         struct diff_words_data *diff_words;
 489         int *found_changesp;
 490         FILE *file;
 491 };
 492
 493 static void free_diff_words_data(struct emit_callback *ecbdata)
 494 {
 495         if (ecbdata->diff_words) {
 496                 /* flush buffers */
 497                 if (ecbdata->diff_words->minus.text.size ||
 498                                 ecbdata->diff_words->plus.text.size)
 499                         diff_words_show(ecbdata->diff_words);
 500
 501                 free (ecbdata->diff_words->minus.text.ptr);
 502                 free (ecbdata->diff_words->plus.text.ptr);
 503                 free(ecbdata->diff_words);
 504                 ecbdata->diff_words = NULL;
 505         }
 506 }
 507
 508 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 509 {
 510         if (diff_use_color)
 511                 return diff_colors[ix];
 512         return "";
 513 }
 514
 515 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
 516 {
 517         int has_trailing_newline, has_trailing_carriage_return;
 518
 519         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 520         if (has_trailing_newline)
 521                 len--;
 522         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 523         if (has_trailing_carriage_return)
 524                 len--;
 525
 526         fputs(set, file);
 527         fwrite(line, len, 1, file);
 528         fputs(reset, file);
 529         if (has_trailing_carriage_return)
 530                 fputc('\r', file);
 531         if (has_trailing_newline)
 532                 fputc('\n', file);
 533 }
 534
 535 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
 536 {
 537         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 538         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 539
 540         if (!*ws)
 541                 emit_line(ecbdata->file, set, reset, line, len);
 542         else {
 543                 /* Emit just the prefix, then the rest. */
 544                 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
 545                 ws_check_emit(line + ecbdata->nparents,
 546                               len - ecbdata->nparents, ecbdata->ws_rule,
 547                               ecbdata->file, set, reset, ws);
 548         }
 549 }
 550
 551 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
 552 {
 553         const char *cp;
 554         unsigned long allot;
 555         size_t l = len;
 556
 557         if (ecb->truncate)
 558                 return ecb->truncate(line, len);
 559         cp = line;
 560         allot = l;
 561         while (0 < l) {
 562                 (void) utf8_width(&cp, &l);
 563                 if (!cp)
 564                         break; /* truncated in the middle? */
 565         }
 566         return allot - l;
 567 }
 568
 569 static void fn_out_consume(void *priv, char *line, unsigned long len)
 570 {
 571         int i;
 572         int color;
 573         struct emit_callback *ecbdata = priv;
 574         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 575         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 576         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 577
 578         *(ecbdata->found_changesp) = 1;
 579
 580         if (ecbdata->label_path[0]) {
 581                 const char *name_a_tab, *name_b_tab;
 582
 583                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 584                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 585
 586                 fprintf(ecbdata->file, "%s--- %s%s%s\n",
 587                         meta, ecbdata->label_path[0], reset, name_a_tab);
 588                 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
 589                         meta, ecbdata->label_path[1], reset, name_b_tab);
 590                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 591         }
 592
 593         /* This is not really necessary for now because
 594          * this codepath only deals with two-way diffs.
 595          */
 596         for (i = 0; i < len && line[i] == '@'; i++)
 597                 ;
 598         if (2 <= i && i < len && line[i] == ' ') {
 599                 ecbdata->nparents = i - 1;
 600                 len = sane_truncate_line(ecbdata, line, len);
 601                 emit_line(ecbdata->file,
 602                           diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
 603                           reset, line, len);
 604                 if (line[len-1] != '\n')
 605                         putc('\n', ecbdata->file);
 606                 return;
 607         }
 608
 609         if (len < ecbdata->nparents) {
 610                 emit_line(ecbdata->file, reset, reset, line, len);
 611                 return;
 612         }
 613
 614         color = DIFF_PLAIN;
 615         if (ecbdata->diff_words && ecbdata->nparents != 1)
 616                 /* fall back to normal diff */
 617                 free_diff_words_data(ecbdata);
 618         if (ecbdata->diff_words) {
 619                 if (line[0] == '-') {
 620                         diff_words_append(line, len,
 621                                           &ecbdata->diff_words->minus);
 622                         return;
 623                 } else if (line[0] == '+') {
 624                         diff_words_append(line, len,
 625                                           &ecbdata->diff_words->plus);
 626                         return;
 627                 }
 628                 if (ecbdata->diff_words->minus.text.size ||
 629                     ecbdata->diff_words->plus.text.size)
 630                         diff_words_show(ecbdata->diff_words);
 631                 line++;
 632                 len--;
 633                 emit_line(ecbdata->file, plain, reset, line, len);
 634                 return;
 635         }
 636         for (i = 0; i < ecbdata->nparents && len; i++) {
 637                 if (line[i] == '-')
 638                         color = DIFF_FILE_OLD;
 639                 else if (line[i] == '+')
 640                         color = DIFF_FILE_NEW;
 641         }
 642
 643         if (color != DIFF_FILE_NEW) {
 644                 emit_line(ecbdata->file,
 645                           diff_get_color(ecbdata->color_diff, color),
 646                           reset, line, len);
 647                 return;
 648         }
 649         emit_add_line(reset, ecbdata, line, len);
 650 }
 651
 652 static char *pprint_rename(const char *a, const char *b)
 653 {
 654         const char *old = a;
 655         const char *new = b;
 656         struct strbuf name;
 657         int pfx_length, sfx_length;
 658         int len_a = strlen(a);
 659         int len_b = strlen(b);
 660         int a_midlen, b_midlen;
 661         int qlen_a = quote_c_style(a, NULL, NULL, 0);
 662         int qlen_b = quote_c_style(b, NULL, NULL, 0);
 663
 664         strbuf_init(&name, 0);
 665         if (qlen_a || qlen_b) {
 666                 quote_c_style(a, &name, NULL, 0);
 667                 strbuf_addstr(&name, " => ");
 668                 quote_c_style(b, &name, NULL, 0);
 669                 return strbuf_detach(&name, NULL);
 670         }
 671
 672         /* Find common prefix */
 673         pfx_length = 0;
 674         while (*old && *new && *old == *new) {
 675                 if (*old == '/')
 676                         pfx_length = old - a + 1;
 677                 old++;
 678                 new++;
 679         }
 680
 681         /* Find common suffix */
 682         old = a + len_a;
 683         new = b + len_b;
 684         sfx_length = 0;
 685         while (a <= old && b <= new && *old == *new) {
 686                 if (*old == '/')
 687                         sfx_length = len_a - (old - a);
 688                 old--;
 689                 new--;
 690         }
 691
 692         /*
 693          * pfx{mid-a => mid-b}sfx
 694          * {pfx-a => pfx-b}sfx
 695          * pfx{sfx-a => sfx-b}
 696          * name-a => name-b
 697          */
 698         a_midlen = len_a - pfx_length - sfx_length;
 699         b_midlen = len_b - pfx_length - sfx_length;
 700         if (a_midlen < 0)
 701                 a_midlen = 0;
 702         if (b_midlen < 0)
 703                 b_midlen = 0;
 704
 705         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 706         if (pfx_length + sfx_length) {
 707                 strbuf_add(&name, a, pfx_length);
 708                 strbuf_addch(&name, '{');
 709         }
 710         strbuf_add(&name, a + pfx_length, a_midlen);
 711         strbuf_addstr(&name, " => ");
 712         strbuf_add(&name, b + pfx_length, b_midlen);
 713         if (pfx_length + sfx_length) {
 714                 strbuf_addch(&name, '}');
 715                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 716         }
 717         return strbuf_detach(&name, NULL);
 718 }
 719
 720 struct diffstat_t {
 721         struct xdiff_emit_state xm;
 722
 723         int nr;
 724         int alloc;
 725         struct diffstat_file {
 726                 char *from_name;
 727                 char *name;
 728                 char *print_name;
 729                 unsigned is_unmerged:1;
 730                 unsigned is_binary:1;
 731                 unsigned is_renamed:1;
 732                 unsigned int added, deleted;
 733         } **files;
 734 };
 735
 736 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 737                                           const char *name_a,
 738                                           const char *name_b)
 739 {
 740         struct diffstat_file *x;
 741         x = xcalloc(sizeof (*x), 1);
 742         if (diffstat->nr == diffstat->alloc) {
 743                 diffstat->alloc = alloc_nr(diffstat->alloc);
 744                 diffstat->files = xrealloc(diffstat->files,
 745                                 diffstat->alloc * sizeof(x));
 746         }
 747         diffstat->files[diffstat->nr++] = x;
 748         if (name_b) {
 749                 x->from_name = xstrdup(name_a);
 750                 x->name = xstrdup(name_b);
 751                 x->is_renamed = 1;
 752         }
 753         else {
 754                 x->from_name = NULL;
 755                 x->name = xstrdup(name_a);
 756         }
 757         return x;
 758 }
 759
 760 static void diffstat_consume(void *priv, char *line, unsigned long len)
 761 {
 762         struct diffstat_t *diffstat = priv;
 763         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 764
 765         if (line[0] == '+')
 766                 x->added++;
 767         else if (line[0] == '-')
 768                 x->deleted++;
 769 }
 770
 771 const char mime_boundary_leader[] = "------------";
 772
 773 static int scale_linear(int it, int width, int max_change)
 774 {
 775         /*
 776          * make sure that at least one '-' is printed if there were deletions,
 777          * and likewise for '+'.
 778          */
 779         if (max_change < 2)
 780                 return it;
 781         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
 782 }
 783
 784 static void show_name(FILE *file,
 785                       const char *prefix, const char *name, int len,
 786                       const char *reset, const char *set)
 787 {
 788         fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
 789 }
 790
 791 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
 792 {
 793         if (cnt <= 0)
 794                 return;
 795         fprintf(file, "%s", set);
 796         while (cnt--)
 797                 putc(ch, file);
 798         fprintf(file, "%s", reset);
 799 }
 800
 801 static void fill_print_name(struct diffstat_file *file)
 802 {
 803         char *pname;
 804
 805         if (file->print_name)
 806                 return;
 807
 808         if (!file->is_renamed) {
 809                 struct strbuf buf;
 810                 strbuf_init(&buf, 0);
 811                 if (quote_c_style(file->name, &buf, NULL, 0)) {
 812                         pname = strbuf_detach(&buf, NULL);
 813                 } else {
 814                         pname = file->name;
 815                         strbuf_release(&buf);
 816                 }
 817         } else {
 818                 pname = pprint_rename(file->from_name, file->name);
 819         }
 820         file->print_name = pname;
 821 }
 822
 823 static void show_stats(struct diffstat_t* data, struct diff_options *options)
 824 {
 825         int i, len, add, del, total, adds = 0, dels = 0;
 826         int max_change = 0, max_len = 0;
 827         int total_files = data->nr;
 828         int width, name_width;
 829         const char *reset, *set, *add_c, *del_c;
 830
 831         if (data->nr == 0)
 832                 return;
 833
 834         width = options->stat_width ? options->stat_width : 80;
 835         name_width = options->stat_name_width ? options->stat_name_width : 50;
 836
 837         /* Sanity: give at least 5 columns to the graph,
 838          * but leave at least 10 columns for the name.
 839          */
 840         if (width < 25)
 841                 width = 25;
 842         if (name_width < 10)
 843                 name_width = 10;
 844         else if (width < name_width + 15)
 845                 name_width = width - 15;
 846
 847         /* Find the longest filename and max number of changes */
 848         reset = diff_get_color_opt(options, DIFF_RESET);
 849         set   = diff_get_color_opt(options, DIFF_PLAIN);
 850         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
 851         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
 852
 853         for (i = 0; i < data->nr; i++) {
 854                 struct diffstat_file *file = data->files[i];
 855                 int change = file->added + file->deleted;
 856                 fill_print_name(file);
 857                 len = strlen(file->print_name);
 858                 if (max_len < len)
 859                         max_len = len;
 860
 861                 if (file->is_binary || file->is_unmerged)
 862                         continue;
 863                 if (max_change < change)
 864                         max_change = change;
 865         }
 866
 867         /* Compute the width of the graph part;
 868          * 10 is for one blank at the beginning of the line plus
 869          * " | count " between the name and the graph.
 870          *
 871          * From here on, name_width is the width of the name area,
 872          * and width is the width of the graph area.
 873          */
 874         name_width = (name_width < max_len) ? name_width : max_len;
 875         if (width < (name_width + 10) + max_change)
 876                 width = width - (name_width + 10);
 877         else
 878                 width = max_change;
 879
 880         for (i = 0; i < data->nr; i++) {
 881                 const char *prefix = "";
 882                 char *name = data->files[i]->print_name;
 883                 int added = data->files[i]->added;
 884                 int deleted = data->files[i]->deleted;
 885                 int name_len;
 886
 887                 /*
 888                  * "scale" the filename
 889                  */
 890                 len = name_width;
 891                 name_len = strlen(name);
 892                 if (name_width < name_len) {
 893                         char *slash;
 894                         prefix = "...";
 895                         len -= 3;
 896                         name += name_len - len;
 897                         slash = strchr(name, '/');
 898                         if (slash)
 899                                 name = slash;
 900                 }
 901
 902                 if (data->files[i]->is_binary) {
 903                         show_name(options->file, prefix, name, len, reset, set);
 904                         fprintf(options->file, "  Bin ");
 905                         fprintf(options->file, "%s%d%s", del_c, deleted, reset);
 906                         fprintf(options->file, " -> ");
 907                         fprintf(options->file, "%s%d%s", add_c, added, reset);
 908                         fprintf(options->file, " bytes");
 909                         fprintf(options->file, "\n");
 910                         continue;
 911                 }
 912                 else if (data->files[i]->is_unmerged) {
 913                         show_name(options->file, prefix, name, len, reset, set);
 914                         fprintf(options->file, "  Unmerged\n");
 915                         continue;
 916                 }
 917                 else if (!data->files[i]->is_renamed &&
 918                          (added + deleted == 0)) {
 919                         total_files--;
 920                         continue;
 921                 }
 922
 923                 /*
 924                  * scale the add/delete
 925                  */
 926                 add = added;
 927                 del = deleted;
 928                 total = add + del;
 929                 adds += add;
 930                 dels += del;
 931
 932                 if (width <= max_change) {
 933                         add = scale_linear(add, width, max_change);
 934                         del = scale_linear(del, width, max_change);
 935                         total = add + del;
 936                 }
 937                 show_name(options->file, prefix, name, len, reset, set);
 938                 fprintf(options->file, "%5d%s", added + deleted,
 939                                 added + deleted ? " " : "");
 940                 show_graph(options->file, '+', add, add_c, reset);
 941                 show_graph(options->file, '-', del, del_c, reset);
 942                 fprintf(options->file, "\n");
 943         }
 944         fprintf(options->file,
 945                "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
 946                set, total_files, adds, dels, reset);
 947 }
 948
 949 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
 950 {
 951         int i, adds = 0, dels = 0, total_files = data->nr;
 952
 953         if (data->nr == 0)
 954                 return;
 955
 956         for (i = 0; i < data->nr; i++) {
 957                 if (!data->files[i]->is_binary &&
 958                     !data->files[i]->is_unmerged) {
 959                         int added = data->files[i]->added;
 960                         int deleted= data->files[i]->deleted;
 961                         if (!data->files[i]->is_renamed &&
 962                             (added + deleted == 0)) {
 963                                 total_files--;
 964                         } else {
 965                                 adds += added;
 966                                 dels += deleted;
 967                         }
 968                 }
 969         }
 970         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
 971                total_files, adds, dels);
 972 }
 973
 974 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
 975 {
 976         int i;
 977
 978         if (data->nr == 0)
 979                 return;
 980
 981         for (i = 0; i < data->nr; i++) {
 982                 struct diffstat_file *file = data->files[i];
 983
 984                 if (file->is_binary)
 985                         fprintf(options->file, "-\t-\t");
 986                 else
 987                         fprintf(options->file,
 988                                 "%d\t%d\t", file->added, file->deleted);
 989                 if (options->line_termination) {
 990                         fill_print_name(file);
 991                         if (!file->is_renamed)
 992                                 write_name_quoted(file->name, options->file,
 993                                                   options->line_termination);
 994                         else {
 995                                 fputs(file->print_name, options->file);
 996                                 putc(options->line_termination, options->file);
 997                         }
 998                 } else {
 999                         if (file->is_renamed) {
1000                                 putc('\0', options->file);
1001                                 write_name_quoted(file->from_name, options->file, '\0');
1002                         }
1003                         write_name_quoted(file->name, options->file, '\0');
1004                 }
1005         }
1006 }
1007
1008 struct dirstat_file {
1009         const char *name;
1010         unsigned long changed;
1011 };
1012
1013 struct dirstat_dir {
1014         struct dirstat_file *files;
1015         int alloc, nr, percent, cumulative;
1016 };
1017
1018 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1019 {
1020         unsigned long this_dir = 0;
1021         unsigned int sources = 0;
1022
1023         while (dir->nr) {
1024                 struct dirstat_file *f = dir->files;
1025                 int namelen = strlen(f->name);
1026                 unsigned long this;
1027                 char *slash;
1028
1029                 if (namelen < baselen)
1030                         break;
1031                 if (memcmp(f->name, base, baselen))
1032                         break;
1033                 slash = strchr(f->name + baselen, '/');
1034                 if (slash) {
1035                         int newbaselen = slash + 1 - f->name;
1036                         this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1037                         sources++;
1038                 } else {
1039                         this = f->changed;
1040                         dir->files++;
1041                         dir->nr--;
1042                         sources += 2;
1043                 }
1044                 this_dir += this;
1045         }
1046
1047         /*
1048          * We don't report dirstat's for
1049          *  - the top level
1050          *  - or cases where everything came from a single directory
1051          *    under this directory (sources == 1).
1052          */
1053         if (baselen && sources != 1) {
1054                 int permille = this_dir * 1000 / changed;
1055                 if (permille) {
1056                         int percent = permille / 10;
1057                         if (percent >= dir->percent) {
1058                                 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1059                                 if (!dir->cumulative)
1060                                         return 0;
1061                         }
1062                 }
1063         }
1064         return this_dir;
1065 }
1066
1067 static int dirstat_compare(const void *_a, const void *_b)
1068 {
1069         const struct dirstat_file *a = _a;
1070         const struct dirstat_file *b = _b;
1071         return strcmp(a->name, b->name);
1072 }
1073
1074 static void show_dirstat(struct diff_options *options)
1075 {
1076         int i;
1077         unsigned long changed;
1078         struct dirstat_dir dir;
1079         struct diff_queue_struct *q = &diff_queued_diff;
1080
1081         dir.files = NULL;
1082         dir.alloc = 0;
1083         dir.nr = 0;
1084         dir.percent = options->dirstat_percent;
1085         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1086
1087         changed = 0;
1088         for (i = 0; i < q->nr; i++) {
1089                 struct diff_filepair *p = q->queue[i];
1090                 const char *name;
1091                 unsigned long copied, added, damage;
1092
1093                 name = p->one->path ? p->one->path : p->two->path;
1094
1095                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1096                         diff_populate_filespec(p->one, 0);
1097                         diff_populate_filespec(p->two, 0);
1098                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1099                                                &copied, &added);
1100                         diff_free_filespec_data(p->one);
1101                         diff_free_filespec_data(p->two);
1102                 } else if (DIFF_FILE_VALID(p->one)) {
1103                         diff_populate_filespec(p->one, 1);
1104                         copied = added = 0;
1105                         diff_free_filespec_data(p->one);
1106                 } else if (DIFF_FILE_VALID(p->two)) {
1107                         diff_populate_filespec(p->two, 1);
1108                         copied = 0;
1109                         added = p->two->size;
1110                         diff_free_filespec_data(p->two);
1111                 } else
1112                         continue;
1113
1114                 /*
1115                  * Original minus copied is the removed material,
1116                  * added is the new material.  They are both damages
1117                  * made to the preimage.
1118                  */
1119                 damage = (p->one->size - copied) + added;
1120
1121                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1122                 dir.files[dir.nr].name = name;
1123                 dir.files[dir.nr].changed = damage;
1124                 changed += damage;
1125                 dir.nr++;
1126         }
1127
1128         /* This can happen even with many files, if everything was renames */
1129         if (!changed)
1130                 return;
1131
1132         /* Show all directories with more than x% of the changes */
1133         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1134         gather_dirstat(options->file, &dir, changed, "", 0);
1135 }
1136
1137 static void free_diffstat_info(struct diffstat_t *diffstat)
1138 {
1139         int i;
1140         for (i = 0; i < diffstat->nr; i++) {
1141                 struct diffstat_file *f = diffstat->files[i];
1142                 if (f->name != f->print_name)
1143                         free(f->print_name);
1144                 free(f->name);
1145                 free(f->from_name);
1146                 free(f);
1147         }
1148         free(diffstat->files);
1149 }
1150
1151 struct checkdiff_t {
1152         struct xdiff_emit_state xm;
1153         const char *filename;
1154         int lineno;
1155         struct diff_options *o;
1156         unsigned ws_rule;
1157         unsigned status;
1158         int trailing_blanks_start;
1159 };
1160
1161 static int is_conflict_marker(const char *line, unsigned long len)
1162 {
1163         char firstchar;
1164         int cnt;
1165
1166         if (len < 8)
1167                 return 0;
1168         firstchar = line[0];
1169         switch (firstchar) {
1170         case '=': case '>': case '<':
1171                 break;
1172         default:
1173                 return 0;
1174         }
1175         for (cnt = 1; cnt < 7; cnt++)
1176                 if (line[cnt] != firstchar)
1177                         return 0;
1178         /* line[0] thru line[6] are same as firstchar */
1179         if (firstchar == '=') {
1180                 /* divider between ours and theirs? */
1181                 if (len != 8 || line[7] != '\n')
1182                         return 0;
1183         } else if (len < 8 || !isspace(line[7])) {
1184                 /* not divider before ours nor after theirs */
1185                 return 0;
1186         }
1187         return 1;
1188 }
1189
1190 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1191 {
1192         struct checkdiff_t *data = priv;
1193         int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1194         const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1195         const char *reset = diff_get_color(color_diff, DIFF_RESET);
1196         const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1197         char *err;
1198
1199         if (line[0] == '+') {
1200                 unsigned bad;
1201                 data->lineno++;
1202                 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1203                         data->trailing_blanks_start = 0;
1204                 else if (!data->trailing_blanks_start)
1205                         data->trailing_blanks_start = data->lineno;
1206                 if (is_conflict_marker(line + 1, len - 1)) {
1207                         data->status |= 1;
1208                         fprintf(data->o->file,
1209                                 "%s:%d: leftover conflict marker\n",
1210                                 data->filename, data->lineno);
1211                 }
1212                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1213                 if (!bad)
1214                         return;
1215                 data->status |= bad;
1216                 err = whitespace_error_string(bad);
1217                 fprintf(data->o->file, "%s:%d: %s.\n",
1218                         data->filename, data->lineno, err);
1219                 free(err);
1220                 emit_line(data->o->file, set, reset, line, 1);
1221                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1222                               data->o->file, set, reset, ws);
1223         } else if (line[0] == ' ') {
1224                 data->lineno++;
1225                 data->trailing_blanks_start = 0;
1226         } else if (line[0] == '@') {
1227                 char *plus = strchr(line, '+');
1228                 if (plus)
1229                         data->lineno = strtol(plus, NULL, 10) - 1;
1230                 else
1231                         die("invalid diff");
1232                 data->trailing_blanks_start = 0;
1233         }
1234 }
1235
1236 static unsigned char *deflate_it(char *data,
1237                                  unsigned long size,
1238                                  unsigned long *result_size)
1239 {
1240         int bound;
1241         unsigned char *deflated;
1242         z_stream stream;
1243
1244         memset(&stream, 0, sizeof(stream));
1245         deflateInit(&stream, zlib_compression_level);
1246         bound = deflateBound(&stream, size);
1247         deflated = xmalloc(bound);
1248         stream.next_out = deflated;
1249         stream.avail_out = bound;
1250
1251         stream.next_in = (unsigned char *)data;
1252         stream.avail_in = size;
1253         while (deflate(&stream, Z_FINISH) == Z_OK)
1254                 ; /* nothing */
1255         deflateEnd(&stream);
1256         *result_size = stream.total_out;
1257         return deflated;
1258 }
1259
1260 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1261 {
1262         void *cp;
1263         void *delta;
1264         void *deflated;
1265         void *data;
1266         unsigned long orig_size;
1267         unsigned long delta_size;
1268         unsigned long deflate_size;
1269         unsigned long data_size;
1270
1271         /* We could do deflated delta, or we could do just deflated two,
1272          * whichever is smaller.
1273          */
1274         delta = NULL;
1275         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1276         if (one->size && two->size) {
1277                 delta = diff_delta(one->ptr, one->size,
1278                                    two->ptr, two->size,
1279                                    &delta_size, deflate_size);
1280                 if (delta) {
1281                         void *to_free = delta;
1282                         orig_size = delta_size;
1283                         delta = deflate_it(delta, delta_size, &delta_size);
1284                         free(to_free);
1285                 }
1286         }
1287
1288         if (delta && delta_size < deflate_size) {
1289                 fprintf(file, "delta %lu\n", orig_size);
1290                 free(deflated);
1291                 data = delta;
1292                 data_size = delta_size;
1293         }
1294         else {
1295                 fprintf(file, "literal %lu\n", two->size);
1296                 free(delta);
1297                 data = deflated;
1298                 data_size = deflate_size;
1299         }
1300
1301         /* emit data encoded in base85 */
1302         cp = data;
1303         while (data_size) {
1304                 int bytes = (52 < data_size) ? 52 : data_size;
1305                 char line[70];
1306                 data_size -= bytes;
1307                 if (bytes <= 26)
1308                         line[0] = bytes + 'A' - 1;
1309                 else
1310                         line[0] = bytes - 26 + 'a' - 1;
1311                 encode_85(line + 1, cp, bytes);
1312                 cp = (char *) cp + bytes;
1313                 fputs(line, file);
1314                 fputc('\n', file);
1315         }
1316         fprintf(file, "\n");
1317         free(data);
1318 }
1319
1320 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1321 {
1322         fprintf(file, "GIT binary patch\n");
1323         emit_binary_diff_body(file, one, two);
1324         emit_binary_diff_body(file, two, one);
1325 }
1326
1327 static void setup_diff_attr_check(struct git_attr_check *check)
1328 {
1329         static struct git_attr *attr_diff;
1330
1331         if (!attr_diff) {
1332                 attr_diff = git_attr("diff", 4);
1333         }
1334         check[0].attr = attr_diff;
1335 }
1336
1337 static void diff_filespec_check_attr(struct diff_filespec *one)
1338 {
1339         struct git_attr_check attr_diff_check;
1340         int check_from_data = 0;
1341
1342         if (one->checked_attr)
1343                 return;
1344
1345         setup_diff_attr_check(&attr_diff_check);
1346         one->is_binary = 0;
1347         one->funcname_pattern_ident = NULL;
1348
1349         if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1350                 const char *value;
1351
1352                 /* binaryness */
1353                 value = attr_diff_check.value;
1354                 if (ATTR_TRUE(value))
1355                         ;
1356                 else if (ATTR_FALSE(value))
1357                         one->is_binary = 1;
1358                 else
1359                         check_from_data = 1;
1360
1361                 /* funcname pattern ident */
1362                 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1363                         ;
1364                 else
1365                         one->funcname_pattern_ident = value;
1366         }
1367
1368         if (check_from_data) {
1369                 if (!one->data && DIFF_FILE_VALID(one))
1370                         diff_populate_filespec(one, 0);
1371
1372                 if (one->data)
1373                         one->is_binary = buffer_is_binary(one->data, one->size);
1374         }
1375 }
1376
1377 int diff_filespec_is_binary(struct diff_filespec *one)
1378 {
1379         diff_filespec_check_attr(one);
1380         return one->is_binary;
1381 }
1382
1383 static const struct funcname_pattern_entry *funcname_pattern(const char *ident)
1384 {
1385         struct funcname_pattern_list *pp;
1386
1387         for (pp = funcname_pattern_list; pp; pp = pp->next)
1388                 if (!strcmp(ident, pp->e.name))
1389                         return &pp->e;
1390         return NULL;
1391 }
1392
1393 static const struct funcname_pattern_entry builtin_funcname_pattern[] = {
1394         { "java", "!^[  ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1395                         "new\\|return\\|switch\\|throw\\|while\\)\n"
1396                         "^[     ]*\\(\\([       ]*"
1397                         "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1398                         "[      ]*([^;]*\\)$" },
1399         { "pascal", "^\\(\\(procedure\\|function\\|constructor\\|"
1400                         "destructor\\|interface\\|implementation\\|"
1401                         "initialization\\|finalization\\)[ \t]*.*\\)$"
1402                         "\\|"
1403                         "^\\(.*=[ \t]*\\(class\\|record\\).*\\)$"
1404                         },
1405         { "bibtex", "\\(@[a-zA-Z]\\{1,\\}[ \t]*{\\{0,1\\}[ \t]*[^ \t\"@',\\#}{~%]*\\).*$" },
1406         { "tex", "^\\(\\\\\\(\\(sub\\)*section\\|chapter\\|part\\)\\*\\{0,1\\}{.*\\)$" },
1407         { "ruby", "^\\s*\\(\\(class\\|module\\|def\\)\\s.*\\)$" },
1408 };
1409
1410 static const struct funcname_pattern_entry *diff_funcname_pattern(struct diff_filespec *one)
1411 {
1412         const char *ident;
1413         const struct funcname_pattern_entry *pe;
1414         int i;
1415
1416         diff_filespec_check_attr(one);
1417         ident = one->funcname_pattern_ident;
1418
1419         if (!ident)
1420                 /*
1421                  * If the config file has "funcname.default" defined, that
1422                  * regexp is used; otherwise NULL is returned and xemit uses
1423                  * the built-in default.
1424                  */
1425                 return funcname_pattern("default");
1426
1427         /* Look up custom "funcname.$ident" regexp from config. */
1428         pe = funcname_pattern(ident);
1429         if (pe)
1430                 return pe;
1431
1432         /*
1433          * And define built-in fallback patterns here.  Note that
1434          * these can be overridden by the user's config settings.
1435          */
1436         for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1437                 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1438                         return &builtin_funcname_pattern[i];
1439
1440         return NULL;
1441 }
1442
1443 static void builtin_diff(const char *name_a,
1444                          const char *name_b,
1445                          struct diff_filespec *one,
1446                          struct diff_filespec *two,
1447                          const char *xfrm_msg,
1448                          struct diff_options *o,
1449                          int complete_rewrite)
1450 {
1451         mmfile_t mf1, mf2;
1452         const char *lbl[2];
1453         char *a_one, *b_two;
1454         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1455         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1456
1457         a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1458         b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1459         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1460         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1461         fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1462         if (lbl[0][0] == '/') {
1463                 /* /dev/null */
1464                 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1465                 if (xfrm_msg && xfrm_msg[0])
1466                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1467         }
1468         else if (lbl[1][0] == '/') {
1469                 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1470                 if (xfrm_msg && xfrm_msg[0])
1471                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1472         }
1473         else {
1474                 if (one->mode != two->mode) {
1475                         fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1476                         fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1477                 }
1478                 if (xfrm_msg && xfrm_msg[0])
1479                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1480                 /*
1481                  * we do not run diff between different kind
1482                  * of objects.
1483                  */
1484                 if ((one->mode ^ two->mode) & S_IFMT)
1485                         goto free_ab_and_return;
1486                 if (complete_rewrite) {
1487                         emit_rewrite_diff(name_a, name_b, one, two, o);
1488                         o->found_changes = 1;
1489                         goto free_ab_and_return;
1490                 }
1491         }
1492
1493         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1494                 die("unable to read files to diff");
1495
1496         if (!DIFF_OPT_TST(o, TEXT) &&
1497             (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1498                 /* Quite common confusing case */
1499                 if (mf1.size == mf2.size &&
1500                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1501                         goto free_ab_and_return;
1502                 if (DIFF_OPT_TST(o, BINARY))
1503                         emit_binary_diff(o->file, &mf1, &mf2);
1504                 else
1505                         fprintf(o->file, "Binary files %s and %s differ\n",
1506                                 lbl[0], lbl[1]);
1507                 o->found_changes = 1;
1508         }
1509         else {
1510                 /* Crazy xdl interfaces.. */
1511                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1512                 xpparam_t xpp;
1513                 xdemitconf_t xecfg;
1514                 xdemitcb_t ecb;
1515                 struct emit_callback ecbdata;
1516                 const struct funcname_pattern_entry *pe;
1517
1518                 pe = diff_funcname_pattern(one);
1519                 if (!pe)
1520                         pe = diff_funcname_pattern(two);
1521
1522                 memset(&xecfg, 0, sizeof(xecfg));
1523                 memset(&ecbdata, 0, sizeof(ecbdata));
1524                 ecbdata.label_path = lbl;
1525                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1526                 ecbdata.found_changesp = &o->found_changes;
1527                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1528                 ecbdata.file = o->file;
1529                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1530                 xecfg.ctxlen = o->context;
1531                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1532                 if (pe)
1533                         xdiff_set_find_func(&xecfg, pe->pattern);
1534                 if (!diffopts)
1535                         ;
1536                 else if (!prefixcmp(diffopts, "--unified="))
1537                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1538                 else if (!prefixcmp(diffopts, "-u"))
1539                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1540                 ecb.outf = xdiff_outf;
1541                 ecb.priv = &ecbdata;
1542                 ecbdata.xm.consume = fn_out_consume;
1543                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1544                         ecbdata.diff_words =
1545                                 xcalloc(1, sizeof(struct diff_words_data));
1546                         ecbdata.diff_words->file = o->file;
1547                 }
1548                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1549                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1550                         free_diff_words_data(&ecbdata);
1551         }
1552
1553  free_ab_and_return:
1554         diff_free_filespec_data(one);
1555         diff_free_filespec_data(two);
1556         free(a_one);
1557         free(b_two);
1558         return;
1559 }
1560
1561 static void builtin_diffstat(const char *name_a, const char *name_b,
1562                              struct diff_filespec *one,
1563                              struct diff_filespec *two,
1564                              struct diffstat_t *diffstat,
1565                              struct diff_options *o,
1566                              int complete_rewrite)
1567 {
1568         mmfile_t mf1, mf2;
1569         struct diffstat_file *data;
1570
1571         data = diffstat_add(diffstat, name_a, name_b);
1572
1573         if (!one || !two) {
1574                 data->is_unmerged = 1;
1575                 return;
1576         }
1577         if (complete_rewrite) {
1578                 diff_populate_filespec(one, 0);
1579                 diff_populate_filespec(two, 0);
1580                 data->deleted = count_lines(one->data, one->size);
1581                 data->added = count_lines(two->data, two->size);
1582                 goto free_and_return;
1583         }
1584         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1585                 die("unable to read files to diff");
1586
1587         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1588                 data->is_binary = 1;
1589                 data->added = mf2.size;
1590                 data->deleted = mf1.size;
1591         } else {
1592                 /* Crazy xdl interfaces.. */
1593                 xpparam_t xpp;
1594                 xdemitconf_t xecfg;
1595                 xdemitcb_t ecb;
1596
1597                 memset(&xecfg, 0, sizeof(xecfg));
1598                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1599                 ecb.outf = xdiff_outf;
1600                 ecb.priv = diffstat;
1601                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1602         }
1603
1604  free_and_return:
1605         diff_free_filespec_data(one);
1606         diff_free_filespec_data(two);
1607 }
1608
1609 static void builtin_checkdiff(const char *name_a, const char *name_b,
1610                               const char *attr_path,
1611                               struct diff_filespec *one,
1612                               struct diff_filespec *two,
1613                               struct diff_options *o)
1614 {
1615         mmfile_t mf1, mf2;
1616         struct checkdiff_t data;
1617
1618         if (!two)
1619                 return;
1620
1621         memset(&data, 0, sizeof(data));
1622         data.xm.consume = checkdiff_consume;
1623         data.filename = name_b ? name_b : name_a;
1624         data.lineno = 0;
1625         data.o = o;
1626         data.ws_rule = whitespace_rule(attr_path);
1627
1628         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1629                 die("unable to read files to diff");
1630
1631         /*
1632          * All the other codepaths check both sides, but not checking
1633          * the "old" side here is deliberate.  We are checking the newly
1634          * introduced changes, and as long as the "new" side is text, we
1635          * can and should check what it introduces.
1636          */
1637         if (diff_filespec_is_binary(two))
1638                 goto free_and_return;
1639         else {
1640                 /* Crazy xdl interfaces.. */
1641                 xpparam_t xpp;
1642                 xdemitconf_t xecfg;
1643                 xdemitcb_t ecb;
1644
1645                 memset(&xecfg, 0, sizeof(xecfg));
1646                 xecfg.ctxlen = 1; /* at least one context line */
1647                 xpp.flags = XDF_NEED_MINIMAL;
1648                 ecb.outf = xdiff_outf;
1649                 ecb.priv = &data;
1650                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1651
1652                 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1653                     data.trailing_blanks_start) {
1654                         fprintf(o->file, "%s:%d: ends with blank lines.\n",
1655                                 data.filename, data.trailing_blanks_start);
1656                         data.status = 1; /* report errors */
1657                 }
1658         }
1659  free_and_return:
1660         diff_free_filespec_data(one);
1661         diff_free_filespec_data(two);
1662         if (data.status)
1663                 DIFF_OPT_SET(o, CHECK_FAILED);
1664 }
1665
1666 struct diff_filespec *alloc_filespec(const char *path)
1667 {
1668         int namelen = strlen(path);
1669         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1670
1671         memset(spec, 0, sizeof(*spec));
1672         spec->path = (char *)(spec + 1);
1673         memcpy(spec->path, path, namelen+1);
1674         spec->count = 1;
1675         return spec;
1676 }
1677
1678 void free_filespec(struct diff_filespec *spec)
1679 {
1680         if (!--spec->count) {
1681                 diff_free_filespec_data(spec);
1682                 free(spec);
1683         }
1684 }
1685
1686 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1687                    unsigned short mode)
1688 {
1689         if (mode) {
1690                 spec->mode = canon_mode(mode);
1691                 hashcpy(spec->sha1, sha1);
1692                 spec->sha1_valid = !is_null_sha1(sha1);
1693         }
1694 }
1695
1696 /*
1697  * Given a name and sha1 pair, if the index tells us the file in
1698  * the work tree has that object contents, return true, so that
1699  * prepare_temp_file() does not have to inflate and extract.
1700  */
1701 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1702 {
1703         struct cache_entry *ce;
1704         struct stat st;
1705         int pos, len;
1706
1707         /* We do not read the cache ourselves here, because the
1708          * benchmark with my previous version that always reads cache
1709          * shows that it makes things worse for diff-tree comparing
1710          * two linux-2.6 kernel trees in an already checked out work
1711          * tree.  This is because most diff-tree comparisons deal with
1712          * only a small number of files, while reading the cache is
1713          * expensive for a large project, and its cost outweighs the
1714          * savings we get by not inflating the object to a temporary
1715          * file.  Practically, this code only helps when we are used
1716          * by diff-cache --cached, which does read the cache before
1717          * calling us.
1718          */
1719         if (!active_cache)
1720                 return 0;
1721
1722         /* We want to avoid the working directory if our caller
1723          * doesn't need the data in a normal file, this system
1724          * is rather slow with its stat/open/mmap/close syscalls,
1725          * and the object is contained in a pack file.  The pack
1726          * is probably already open and will be faster to obtain
1727          * the data through than the working directory.  Loose
1728          * objects however would tend to be slower as they need
1729          * to be individually opened and inflated.
1730          */
1731         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1732                 return 0;
1733
1734         len = strlen(name);
1735         pos = cache_name_pos(name, len);
1736         if (pos < 0)
1737                 return 0;
1738         ce = active_cache[pos];
1739
1740         /*
1741          * This is not the sha1 we are looking for, or
1742          * unreusable because it is not a regular file.
1743          */
1744         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1745                 return 0;
1746
1747         /*
1748          * If ce matches the file in the work tree, we can reuse it.
1749          */
1750         if (ce_uptodate(ce) ||
1751             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1752                 return 1;
1753
1754         return 0;
1755 }
1756
1757 static int populate_from_stdin(struct diff_filespec *s)
1758 {
1759         struct strbuf buf;
1760         size_t size = 0;
1761
1762         strbuf_init(&buf, 0);
1763         if (strbuf_read(&buf, 0, 0) < 0)
1764                 return error("error while reading from stdin %s",
1765                                      strerror(errno));
1766
1767         s->should_munmap = 0;
1768         s->data = strbuf_detach(&buf, &size);
1769         s->size = size;
1770         s->should_free = 1;
1771         return 0;
1772 }
1773
1774 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1775 {
1776         int len;
1777         char *data = xmalloc(100);
1778         len = snprintf(data, 100,
1779                 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1780         s->data = data;
1781         s->size = len;
1782         s->should_free = 1;
1783         if (size_only) {
1784                 s->data = NULL;
1785                 free(data);
1786         }
1787         return 0;
1788 }
1789
1790 /*
1791  * While doing rename detection and pickaxe operation, we may need to
1792  * grab the data for the blob (or file) for our own in-core comparison.
1793  * diff_filespec has data and size fields for this purpose.
1794  */
1795 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1796 {
1797         int err = 0;
1798         if (!DIFF_FILE_VALID(s))
1799                 die("internal error: asking to populate invalid file.");
1800         if (S_ISDIR(s->mode))
1801                 return -1;
1802
1803         if (s->data)
1804                 return 0;
1805
1806         if (size_only && 0 < s->size)
1807                 return 0;
1808
1809         if (S_ISGITLINK(s->mode))
1810                 return diff_populate_gitlink(s, size_only);
1811
1812         if (!s->sha1_valid ||
1813             reuse_worktree_file(s->path, s->sha1, 0)) {
1814                 struct strbuf buf;
1815                 struct stat st;
1816                 int fd;
1817
1818                 if (!strcmp(s->path, "-"))
1819                         return populate_from_stdin(s);
1820
1821                 if (lstat(s->path, &st) < 0) {
1822                         if (errno == ENOENT) {
1823                         err_empty:
1824                                 err = -1;
1825                         empty:
1826                                 s->data = (char *)"";
1827                                 s->size = 0;
1828                                 return err;
1829                         }
1830                 }
1831                 s->size = xsize_t(st.st_size);
1832                 if (!s->size)
1833                         goto empty;
1834                 if (size_only)
1835                         return 0;
1836                 if (S_ISLNK(st.st_mode)) {
1837                         int ret;
1838                         s->data = xmalloc(s->size);
1839                         s->should_free = 1;
1840                         ret = readlink(s->path, s->data, s->size);
1841                         if (ret < 0) {
1842                                 free(s->data);
1843                                 goto err_empty;
1844                         }
1845                         return 0;
1846                 }
1847                 fd = open(s->path, O_RDONLY);
1848                 if (fd < 0)
1849                         goto err_empty;
1850                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1851                 close(fd);
1852                 s->should_munmap = 1;
1853
1854                 /*
1855                  * Convert from working tree format to canonical git format
1856                  */
1857                 strbuf_init(&buf, 0);
1858                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1859                         size_t size = 0;
1860                         munmap(s->data, s->size);
1861                         s->should_munmap = 0;
1862                         s->data = strbuf_detach(&buf, &size);
1863                         s->size = size;
1864                         s->should_free = 1;
1865                 }
1866         }
1867         else {
1868                 enum object_type type;
1869                 if (size_only)
1870                         type = sha1_object_info(s->sha1, &s->size);
1871                 else {
1872                         s->data = read_sha1_file(s->sha1, &type, &s->size);
1873                         s->should_free = 1;
1874                 }
1875         }
1876         return 0;
1877 }
1878
1879 void diff_free_filespec_blob(struct diff_filespec *s)
1880 {
1881         if (s->should_free)
1882                 free(s->data);
1883         else if (s->should_munmap)
1884                 munmap(s->data, s->size);
1885
1886         if (s->should_free || s->should_munmap) {
1887                 s->should_free = s->should_munmap = 0;
1888                 s->data = NULL;
1889         }
1890 }
1891
1892 void diff_free_filespec_data(struct diff_filespec *s)
1893 {
1894         diff_free_filespec_blob(s);
1895         free(s->cnt_data);
1896         s->cnt_data = NULL;
1897 }
1898
1899 static void prep_temp_blob(struct diff_tempfile *temp,
1900                            void *blob,
1901                            unsigned long size,
1902                            const unsigned char *sha1,
1903                            int mode)
1904 {
1905         int fd;
1906
1907         fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1908         if (fd < 0)
1909                 die("unable to create temp-file: %s", strerror(errno));
1910         if (write_in_full(fd, blob, size) != size)
1911                 die("unable to write temp-file");
1912         close(fd);
1913         temp->name = temp->tmp_path;
1914         strcpy(temp->hex, sha1_to_hex(sha1));
1915         temp->hex[40] = 0;
1916         sprintf(temp->mode, "%06o", mode);
1917 }
1918
1919 static void prepare_temp_file(const char *name,
1920                               struct diff_tempfile *temp,
1921                               struct diff_filespec *one)
1922 {
1923         if (!DIFF_FILE_VALID(one)) {
1924         not_a_valid_file:
1925                 /* A '-' entry produces this for file-2, and
1926                  * a '+' entry produces this for file-1.
1927                  */
1928                 temp->name = "/dev/null";
1929                 strcpy(temp->hex, ".");
1930                 strcpy(temp->mode, ".");
1931                 return;
1932         }
1933
1934         if (!one->sha1_valid ||
1935             reuse_worktree_file(name, one->sha1, 1)) {
1936                 struct stat st;
1937                 if (lstat(name, &st) < 0) {
1938                         if (errno == ENOENT)
1939                                 goto not_a_valid_file;
1940                         die("stat(%s): %s", name, strerror(errno));
1941                 }
1942                 if (S_ISLNK(st.st_mode)) {
1943                         int ret;
1944                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1945                         size_t sz = xsize_t(st.st_size);
1946                         if (sizeof(buf) <= st.st_size)
1947                                 die("symlink too long: %s", name);
1948                         ret = readlink(name, buf, sz);
1949                         if (ret < 0)
1950                                 die("readlink(%s)", name);
1951                         prep_temp_blob(temp, buf, sz,
1952                                        (one->sha1_valid ?
1953                                         one->sha1 : null_sha1),
1954                                        (one->sha1_valid ?
1955                                         one->mode : S_IFLNK));
1956                 }
1957                 else {
1958                         /* we can borrow from the file in the work tree */
1959                         temp->name = name;
1960                         if (!one->sha1_valid)
1961                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
1962                         else
1963                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
1964                         /* Even though we may sometimes borrow the
1965                          * contents from the work tree, we always want
1966                          * one->mode.  mode is trustworthy even when
1967                          * !(one->sha1_valid), as long as
1968                          * DIFF_FILE_VALID(one).
1969                          */
1970                         sprintf(temp->mode, "%06o", one->mode);
1971                 }
1972                 return;
1973         }
1974         else {
1975                 if (diff_populate_filespec(one, 0))
1976                         die("cannot read data blob for %s", one->path);
1977                 prep_temp_blob(temp, one->data, one->size,
1978                                one->sha1, one->mode);
1979         }
1980 }
1981
1982 static void remove_tempfile(void)
1983 {
1984         int i;
1985
1986         for (i = 0; i < 2; i++)
1987                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1988                         unlink(diff_temp[i].name);
1989                         diff_temp[i].name = NULL;
1990                 }
1991 }
1992
1993 static void remove_tempfile_on_signal(int signo)
1994 {
1995         remove_tempfile();
1996         signal(SIGINT, SIG_DFL);
1997         raise(signo);
1998 }
1999
2000 /* An external diff command takes:
2001  *
2002  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2003  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2004  *
2005  */
2006 static void run_external_diff(const char *pgm,
2007                               const char *name,
2008                               const char *other,
2009                               struct diff_filespec *one,
2010                               struct diff_filespec *two,
2011                               const char *xfrm_msg,
2012                               int complete_rewrite)
2013 {
2014         const char *spawn_arg[10];
2015         struct diff_tempfile *temp = diff_temp;
2016         int retval;
2017         static int atexit_asked = 0;
2018         const char *othername;
2019         const char **arg = &spawn_arg[0];
2020
2021         othername = (other? other : name);
2022         if (one && two) {
2023                 prepare_temp_file(name, &temp[0], one);
2024                 prepare_temp_file(othername, &temp[1], two);
2025                 if (! atexit_asked &&
2026                     (temp[0].name == temp[0].tmp_path ||
2027                      temp[1].name == temp[1].tmp_path)) {
2028                         atexit_asked = 1;
2029                         atexit(remove_tempfile);
2030                 }
2031                 signal(SIGINT, remove_tempfile_on_signal);
2032         }
2033
2034         if (one && two) {
2035                 *arg++ = pgm;
2036                 *arg++ = name;
2037                 *arg++ = temp[0].name;
2038                 *arg++ = temp[0].hex;
2039                 *arg++ = temp[0].mode;
2040                 *arg++ = temp[1].name;
2041                 *arg++ = temp[1].hex;
2042                 *arg++ = temp[1].mode;
2043                 if (other) {
2044                         *arg++ = other;
2045                         *arg++ = xfrm_msg;
2046                 }
2047         } else {
2048                 *arg++ = pgm;
2049                 *arg++ = name;
2050         }
2051         *arg = NULL;
2052         fflush(NULL);
2053         retval = run_command_v_opt(spawn_arg, 0);
2054         remove_tempfile();
2055         if (retval) {
2056                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2057                 exit(1);
2058         }
2059 }
2060
2061 static const char *external_diff_attr(const char *name)
2062 {
2063         struct git_attr_check attr_diff_check;
2064
2065         if (!name)
2066                 return NULL;
2067
2068         setup_diff_attr_check(&attr_diff_check);
2069         if (!git_checkattr(name, 1, &attr_diff_check)) {
2070                 const char *value = attr_diff_check.value;
2071                 if (!ATTR_TRUE(value) &&
2072                     !ATTR_FALSE(value) &&
2073                     !ATTR_UNSET(value)) {
2074                         struct ll_diff_driver *drv;
2075
2076                         for (drv = user_diff; drv; drv = drv->next)
2077                                 if (!strcmp(drv->name, value))
2078                                         return drv->cmd;
2079                 }
2080         }
2081         return NULL;
2082 }
2083
2084 static void run_diff_cmd(const char *pgm,
2085                          const char *name,
2086                          const char *other,
2087                          const char *attr_path,
2088                          struct diff_filespec *one,
2089                          struct diff_filespec *two,
2090                          const char *xfrm_msg,
2091                          struct diff_options *o,
2092                          int complete_rewrite)
2093 {
2094         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2095                 pgm = NULL;
2096         else {
2097                 const char *cmd = external_diff_attr(attr_path);
2098                 if (cmd)
2099                         pgm = cmd;
2100         }
2101
2102         if (pgm) {
2103                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2104                                   complete_rewrite);
2105                 return;
2106         }
2107         if (one && two)
2108                 builtin_diff(name, other ? other : name,
2109                              one, two, xfrm_msg, o, complete_rewrite);
2110         else
2111                 fprintf(o->file, "* Unmerged path %s\n", name);
2112 }
2113
2114 static void diff_fill_sha1_info(struct diff_filespec *one)
2115 {
2116         if (DIFF_FILE_VALID(one)) {
2117                 if (!one->sha1_valid) {
2118                         struct stat st;
2119                         if (!strcmp(one->path, "-")) {
2120                                 hashcpy(one->sha1, null_sha1);
2121                                 return;
2122                         }
2123                         if (lstat(one->path, &st) < 0)
2124                                 die("stat %s", one->path);
2125                         if (index_path(one->sha1, one->path, &st, 0))
2126                                 die("cannot hash %s\n", one->path);
2127                 }
2128         }
2129         else
2130                 hashclr(one->sha1);
2131 }
2132
2133 static int similarity_index(struct diff_filepair *p)
2134 {
2135         return p->score * 100 / MAX_SCORE;
2136 }
2137
2138 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2139 {
2140         /* Strip the prefix but do not molest /dev/null and absolute paths */
2141         if (*namep && **namep != '/')
2142                 *namep += prefix_length;
2143         if (*otherp && **otherp != '/')
2144                 *otherp += prefix_length;
2145 }
2146
2147 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2148 {
2149         const char *pgm = external_diff();
2150         struct strbuf msg;
2151         char *xfrm_msg;
2152         struct diff_filespec *one = p->one;
2153         struct diff_filespec *two = p->two;
2154         const char *name;
2155         const char *other;
2156         const char *attr_path;
2157         int complete_rewrite = 0;
2158
2159         name  = p->one->path;
2160         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2161         attr_path = name;
2162         if (o->prefix_length)
2163                 strip_prefix(o->prefix_length, &name, &other);
2164
2165         if (DIFF_PAIR_UNMERGED(p)) {
2166                 run_diff_cmd(pgm, name, NULL, attr_path,
2167                              NULL, NULL, NULL, o, 0);
2168                 return;
2169         }
2170
2171         diff_fill_sha1_info(one);
2172         diff_fill_sha1_info(two);
2173
2174         strbuf_init(&msg, PATH_MAX * 2 + 300);
2175         switch (p->status) {
2176         case DIFF_STATUS_COPIED:
2177                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2178                 strbuf_addstr(&msg, "\ncopy from ");
2179                 quote_c_style(name, &msg, NULL, 0);
2180                 strbuf_addstr(&msg, "\ncopy to ");
2181                 quote_c_style(other, &msg, NULL, 0);
2182                 strbuf_addch(&msg, '\n');
2183                 break;
2184         case DIFF_STATUS_RENAMED:
2185                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2186                 strbuf_addstr(&msg, "\nrename from ");
2187                 quote_c_style(name, &msg, NULL, 0);
2188                 strbuf_addstr(&msg, "\nrename to ");
2189                 quote_c_style(other, &msg, NULL, 0);
2190                 strbuf_addch(&msg, '\n');
2191                 break;
2192         case DIFF_STATUS_MODIFIED:
2193                 if (p->score) {
2194                         strbuf_addf(&msg, "dissimilarity index %d%%\n",
2195                                         similarity_index(p));
2196                         complete_rewrite = 1;
2197                         break;
2198                 }
2199                 /* fallthru */
2200         default:
2201                 /* nothing */
2202                 ;
2203         }
2204
2205         if (hashcmp(one->sha1, two->sha1)) {
2206                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2207
2208                 if (DIFF_OPT_TST(o, BINARY)) {
2209                         mmfile_t mf;
2210                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2211                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2212                                 abbrev = 40;
2213                 }
2214                 strbuf_addf(&msg, "index %.*s..%.*s",
2215                                 abbrev, sha1_to_hex(one->sha1),
2216                                 abbrev, sha1_to_hex(two->sha1));
2217                 if (one->mode == two->mode)
2218                         strbuf_addf(&msg, " %06o", one->mode);
2219                 strbuf_addch(&msg, '\n');
2220         }
2221
2222         if (msg.len)
2223                 strbuf_setlen(&msg, msg.len - 1);
2224         xfrm_msg = msg.len ? msg.buf : NULL;
2225
2226         if (!pgm &&
2227             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2228             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2229                 /* a filepair that changes between file and symlink
2230                  * needs to be split into deletion and creation.
2231                  */
2232                 struct diff_filespec *null = alloc_filespec(two->path);
2233                 run_diff_cmd(NULL, name, other, attr_path,
2234                              one, null, xfrm_msg, o, 0);
2235                 free(null);
2236                 null = alloc_filespec(one->path);
2237                 run_diff_cmd(NULL, name, other, attr_path,
2238                              null, two, xfrm_msg, o, 0);
2239                 free(null);
2240         }
2241         else
2242                 run_diff_cmd(pgm, name, other, attr_path,
2243                              one, two, xfrm_msg, o, complete_rewrite);
2244
2245         strbuf_release(&msg);
2246 }
2247
2248 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2249                          struct diffstat_t *diffstat)
2250 {
2251         const char *name;
2252         const char *other;
2253         int complete_rewrite = 0;
2254
2255         if (DIFF_PAIR_UNMERGED(p)) {
2256                 /* unmerged */
2257                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2258                 return;
2259         }
2260
2261         name = p->one->path;
2262         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2263
2264         if (o->prefix_length)
2265                 strip_prefix(o->prefix_length, &name, &other);
2266
2267         diff_fill_sha1_info(p->one);
2268         diff_fill_sha1_info(p->two);
2269
2270         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2271                 complete_rewrite = 1;
2272         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2273 }
2274
2275 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2276 {
2277         const char *name;
2278         const char *other;
2279         const char *attr_path;
2280
2281         if (DIFF_PAIR_UNMERGED(p)) {
2282                 /* unmerged */
2283                 return;
2284         }
2285
2286         name = p->one->path;
2287         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2288         attr_path = other ? other : name;
2289
2290         if (o->prefix_length)
2291                 strip_prefix(o->prefix_length, &name, &other);
2292
2293         diff_fill_sha1_info(p->one);
2294         diff_fill_sha1_info(p->two);
2295
2296         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2297 }
2298
2299 void diff_setup(struct diff_options *options)
2300 {
2301         memset(options, 0, sizeof(*options));
2302
2303         options->file = stdout;
2304
2305         options->line_termination = '\n';
2306         options->break_opt = -1;
2307         options->rename_limit = -1;
2308         options->dirstat_percent = 3;
2309         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
2310         options->context = 3;
2311
2312         options->change = diff_change;
2313         options->add_remove = diff_addremove;
2314         if (diff_use_color_default > 0)
2315                 DIFF_OPT_SET(options, COLOR_DIFF);
2316         else
2317                 DIFF_OPT_CLR(options, COLOR_DIFF);
2318         options->detect_rename = diff_detect_rename_default;
2319
2320         options->a_prefix = "a/";
2321         options->b_prefix = "b/";
2322 }
2323
2324 int diff_setup_done(struct diff_options *options)
2325 {
2326         int count = 0;
2327
2328         if (options->output_format & DIFF_FORMAT_NAME)
2329                 count++;
2330         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2331                 count++;
2332         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2333                 count++;
2334         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2335                 count++;
2336         if (count > 1)
2337                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2338
2339         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2340                 options->detect_rename = DIFF_DETECT_COPY;
2341
2342         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2343                 options->prefix = NULL;
2344         if (options->prefix)
2345                 options->prefix_length = strlen(options->prefix);
2346         else
2347                 options->prefix_length = 0;
2348
2349         if (options->output_format & (DIFF_FORMAT_NAME |
2350                                       DIFF_FORMAT_NAME_STATUS |
2351                                       DIFF_FORMAT_CHECKDIFF |
2352                                       DIFF_FORMAT_NO_OUTPUT))
2353                 options->output_format &= ~(DIFF_FORMAT_RAW |
2354                                             DIFF_FORMAT_NUMSTAT |
2355                                             DIFF_FORMAT_DIFFSTAT |
2356                                             DIFF_FORMAT_SHORTSTAT |
2357                                             DIFF_FORMAT_DIRSTAT |
2358                                             DIFF_FORMAT_SUMMARY |
2359                                             DIFF_FORMAT_PATCH);
2360
2361         /*
2362          * These cases always need recursive; we do not drop caller-supplied
2363          * recursive bits for other formats here.
2364          */
2365         if (options->output_format & (DIFF_FORMAT_PATCH |
2366                                       DIFF_FORMAT_NUMSTAT |
2367                                       DIFF_FORMAT_DIFFSTAT |
2368                                       DIFF_FORMAT_SHORTSTAT |
2369                                       DIFF_FORMAT_DIRSTAT |
2370                                       DIFF_FORMAT_SUMMARY |
2371                                       DIFF_FORMAT_CHECKDIFF))
2372                 DIFF_OPT_SET(options, RECURSIVE);
2373         /*
2374          * Also pickaxe would not work very well if you do not say recursive
2375          */
2376         if (options->pickaxe)
2377                 DIFF_OPT_SET(options, RECURSIVE);
2378
2379         if (options->detect_rename && options->rename_limit < 0)
2380                 options->rename_limit = diff_rename_limit_default;
2381         if (options->setup & DIFF_SETUP_USE_CACHE) {
2382                 if (!active_cache)
2383                         /* read-cache does not die even when it fails
2384                          * so it is safe for us to do this here.  Also
2385                          * it does not smudge active_cache or active_nr
2386                          * when it fails, so we do not have to worry about
2387                          * cleaning it up ourselves either.
2388                          */
2389                         read_cache();
2390         }
2391         if (options->abbrev <= 0 || 40 < options->abbrev)
2392                 options->abbrev = 40; /* full */
2393
2394         /*
2395          * It does not make sense to show the first hit we happened
2396          * to have found.  It does not make sense not to return with
2397          * exit code in such a case either.
2398          */
2399         if (DIFF_OPT_TST(options, QUIET)) {
2400                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2401                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2402         }
2403
2404         /*
2405          * If we postprocess in diffcore, we cannot simply return
2406          * upon the first hit.  We need to run diff as usual.
2407          */
2408         if (options->pickaxe || options->filter)
2409                 DIFF_OPT_CLR(options, QUIET);
2410
2411         return 0;
2412 }
2413
2414 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2415 {
2416         char c, *eq;
2417         int len;
2418
2419         if (*arg != '-')
2420                 return 0;
2421         c = *++arg;
2422         if (!c)
2423                 return 0;
2424         if (c == arg_short) {
2425                 c = *++arg;
2426                 if (!c)
2427                         return 1;
2428                 if (val && isdigit(c)) {
2429                         char *end;
2430                         int n = strtoul(arg, &end, 10);
2431                         if (*end)
2432                                 return 0;
2433                         *val = n;
2434                         return 1;
2435                 }
2436                 return 0;
2437         }
2438         if (c != '-')
2439                 return 0;
2440         arg++;
2441         eq = strchr(arg, '=');
2442         if (eq)
2443                 len = eq - arg;
2444         else
2445                 len = strlen(arg);
2446         if (!len || strncmp(arg, arg_long, len))
2447                 return 0;
2448         if (eq) {
2449                 int n;
2450                 char *end;
2451                 if (!isdigit(*++eq))
2452                         return 0;
2453                 n = strtoul(eq, &end, 10);
2454                 if (*end)
2455                         return 0;
2456                 *val = n;
2457         }
2458         return 1;
2459 }
2460
2461 static int diff_scoreopt_parse(const char *opt);
2462
2463 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2464 {
2465         const char *arg = av[0];
2466
2467         /* Output format options */
2468         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2469                 options->output_format |= DIFF_FORMAT_PATCH;
2470         else if (opt_arg(arg, 'U', "unified", &options->context))
2471                 options->output_format |= DIFF_FORMAT_PATCH;
2472         else if (!strcmp(arg, "--raw"))
2473                 options->output_format |= DIFF_FORMAT_RAW;
2474         else if (!strcmp(arg, "--patch-with-raw"))
2475                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2476         else if (!strcmp(arg, "--numstat"))
2477                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2478         else if (!strcmp(arg, "--shortstat"))
2479                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2480         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2481                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2482         else if (!strcmp(arg, "--cumulative")) {
2483                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2484                 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2485         }
2486         else if (!strcmp(arg, "--check"))
2487                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2488         else if (!strcmp(arg, "--summary"))
2489                 options->output_format |= DIFF_FORMAT_SUMMARY;
2490         else if (!strcmp(arg, "--patch-with-stat"))
2491                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2492         else if (!strcmp(arg, "--name-only"))
2493                 options->output_format |= DIFF_FORMAT_NAME;
2494         else if (!strcmp(arg, "--name-status"))
2495                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2496         else if (!strcmp(arg, "-s"))
2497                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2498         else if (!prefixcmp(arg, "--stat")) {
2499                 char *end;
2500                 int width = options->stat_width;
2501                 int name_width = options->stat_name_width;
2502                 arg += 6;
2503                 end = (char *)arg;
2504
2505                 switch (*arg) {
2506                 case '-':
2507                         if (!prefixcmp(arg, "-width="))
2508                                 width = strtoul(arg + 7, &end, 10);
2509                         else if (!prefixcmp(arg, "-name-width="))
2510                                 name_width = strtoul(arg + 12, &end, 10);
2511                         break;
2512                 case '=':
2513                         width = strtoul(arg+1, &end, 10);
2514                         if (*end == ',')
2515                                 name_width = strtoul(end+1, &end, 10);
2516                 }
2517
2518                 /* Important! This checks all the error cases! */
2519                 if (*end)
2520                         return 0;
2521                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2522                 options->stat_name_width = name_width;
2523                 options->stat_width = width;
2524         }
2525
2526         /* renames options */
2527         else if (!prefixcmp(arg, "-B")) {
2528                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2529                         return -1;
2530         }
2531         else if (!prefixcmp(arg, "-M")) {
2532                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2533                         return -1;
2534                 options->detect_rename = DIFF_DETECT_RENAME;
2535         }
2536         else if (!prefixcmp(arg, "-C")) {
2537                 if (options->detect_rename == DIFF_DETECT_COPY)
2538                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2539                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2540                         return -1;
2541                 options->detect_rename = DIFF_DETECT_COPY;
2542         }
2543         else if (!strcmp(arg, "--no-renames"))
2544                 options->detect_rename = 0;
2545         else if (!strcmp(arg, "--relative"))
2546                 DIFF_OPT_SET(options, RELATIVE_NAME);
2547         else if (!prefixcmp(arg, "--relative=")) {
2548                 DIFF_OPT_SET(options, RELATIVE_NAME);
2549                 options->prefix = arg + 11;
2550         }
2551
2552         /* xdiff options */
2553         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2554                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2555         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2556                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2557         else if (!strcmp(arg, "--ignore-space-at-eol"))
2558                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2559
2560         /* flags options */
2561         else if (!strcmp(arg, "--binary")) {
2562                 options->output_format |= DIFF_FORMAT_PATCH;
2563                 DIFF_OPT_SET(options, BINARY);
2564         }
2565         else if (!strcmp(arg, "--full-index"))
2566                 DIFF_OPT_SET(options, FULL_INDEX);
2567         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2568                 DIFF_OPT_SET(options, TEXT);
2569         else if (!strcmp(arg, "-R"))
2570                 DIFF_OPT_SET(options, REVERSE_DIFF);
2571         else if (!strcmp(arg, "--find-copies-harder"))
2572                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2573         else if (!strcmp(arg, "--follow"))
2574                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2575         else if (!strcmp(arg, "--color"))
2576                 DIFF_OPT_SET(options, COLOR_DIFF);
2577         else if (!strcmp(arg, "--no-color"))
2578                 DIFF_OPT_CLR(options, COLOR_DIFF);
2579         else if (!strcmp(arg, "--color-words"))
2580                 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2581         else if (!strcmp(arg, "--exit-code"))
2582                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2583         else if (!strcmp(arg, "--quiet"))
2584                 DIFF_OPT_SET(options, QUIET);
2585         else if (!strcmp(arg, "--ext-diff"))
2586                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2587         else if (!strcmp(arg, "--no-ext-diff"))
2588                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2589         else if (!strcmp(arg, "--ignore-submodules"))
2590                 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2591
2592         /* misc options */
2593         else if (!strcmp(arg, "-z"))
2594                 options->line_termination = 0;
2595         else if (!prefixcmp(arg, "-l"))
2596                 options->rename_limit = strtoul(arg+2, NULL, 10);
2597         else if (!prefixcmp(arg, "-S"))
2598                 options->pickaxe = arg + 2;
2599         else if (!strcmp(arg, "--pickaxe-all"))
2600                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2601         else if (!strcmp(arg, "--pickaxe-regex"))
2602                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2603         else if (!prefixcmp(arg, "-O"))
2604                 options->orderfile = arg + 2;
2605         else if (!prefixcmp(arg, "--diff-filter="))
2606                 options->filter = arg + 14;
2607         else if (!strcmp(arg, "--abbrev"))
2608                 options->abbrev = DEFAULT_ABBREV;
2609         else if (!prefixcmp(arg, "--abbrev=")) {
2610                 options->abbrev = strtoul(arg + 9, NULL, 10);
2611                 if (options->abbrev < MINIMUM_ABBREV)
2612                         options->abbrev = MINIMUM_ABBREV;
2613                 else if (40 < options->abbrev)
2614                         options->abbrev = 40;
2615         }
2616         else if (!prefixcmp(arg, "--src-prefix="))
2617                 options->a_prefix = arg + 13;
2618         else if (!prefixcmp(arg, "--dst-prefix="))
2619                 options->b_prefix = arg + 13;
2620         else if (!strcmp(arg, "--no-prefix"))
2621                 options->a_prefix = options->b_prefix = "";
2622         else if (!prefixcmp(arg, "--output=")) {
2623                 options->file = fopen(arg + strlen("--output="), "w");
2624                 options->close_file = 1;
2625         } else
2626                 return 0;
2627         return 1;
2628 }
2629
2630 static int parse_num(const char **cp_p)
2631 {
2632         unsigned long num, scale;
2633         int ch, dot;
2634         const char *cp = *cp_p;
2635
2636         num = 0;
2637         scale = 1;
2638         dot = 0;
2639         for(;;) {
2640                 ch = *cp;
2641                 if ( !dot && ch == '.' ) {
2642                         scale = 1;
2643                         dot = 1;
2644                 } else if ( ch == '%' ) {
2645                         scale = dot ? scale*100 : 100;
2646                         cp++;   /* % is always at the end */
2647                         break;
2648                 } else if ( ch >= '0' && ch <= '9' ) {
2649                         if ( scale < 100000 ) {
2650                                 scale *= 10;
2651                                 num = (num*10) + (ch-'0');
2652                         }
2653                 } else {
2654                         break;
2655                 }
2656                 cp++;
2657         }
2658         *cp_p = cp;
2659
2660         /* user says num divided by scale and we say internally that
2661          * is MAX_SCORE * num / scale.
2662          */
2663         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2664 }
2665
2666 static int diff_scoreopt_parse(const char *opt)
2667 {
2668         int opt1, opt2, cmd;
2669
2670         if (*opt++ != '-')
2671                 return -1;
2672         cmd = *opt++;
2673         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2674                 return -1; /* that is not a -M, -C nor -B option */
2675
2676         opt1 = parse_num(&opt);
2677         if (cmd != 'B')
2678                 opt2 = 0;
2679         else {
2680                 if (*opt == 0)
2681                         opt2 = 0;
2682                 else if (*opt != '/')
2683                         return -1; /* we expect -B80/99 or -B80 */
2684                 else {
2685                         opt++;
2686                         opt2 = parse_num(&opt);
2687                 }
2688         }
2689         if (*opt != 0)
2690                 return -1;
2691         return opt1 | (opt2 << 16);
2692 }
2693
2694 struct diff_queue_struct diff_queued_diff;
2695
2696 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2697 {
2698         if (queue->alloc <= queue->nr) {
2699                 queue->alloc = alloc_nr(queue->alloc);
2700                 queue->queue = xrealloc(queue->queue,
2701                                         sizeof(dp) * queue->alloc);
2702         }
2703         queue->queue[queue->nr++] = dp;
2704 }
2705
2706 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2707                                  struct diff_filespec *one,
2708                                  struct diff_filespec *two)
2709 {
2710         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2711         dp->one = one;
2712         dp->two = two;
2713         if (queue)
2714                 diff_q(queue, dp);
2715         return dp;
2716 }
2717
2718 void diff_free_filepair(struct diff_filepair *p)
2719 {
2720         free_filespec(p->one);
2721         free_filespec(p->two);
2722         free(p);
2723 }
2724
2725 /* This is different from find_unique_abbrev() in that
2726  * it stuffs the result with dots for alignment.
2727  */
2728 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2729 {
2730         int abblen;
2731         const char *abbrev;
2732         if (len == 40)
2733                 return sha1_to_hex(sha1);
2734
2735         abbrev = find_unique_abbrev(sha1, len);
2736         abblen = strlen(abbrev);
2737         if (abblen < 37) {
2738                 static char hex[41];
2739                 if (len < abblen && abblen <= len + 2)
2740                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2741                 else
2742                         sprintf(hex, "%s...", abbrev);
2743                 return hex;
2744         }
2745         return sha1_to_hex(sha1);
2746 }
2747
2748 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2749 {
2750         int line_termination = opt->line_termination;
2751         int inter_name_termination = line_termination ? '\t' : '\0';
2752
2753         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2754                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2755                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
2756                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2757         }
2758         if (p->score) {
2759                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2760                         inter_name_termination);
2761         } else {
2762                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2763         }
2764
2765         if (p->status == DIFF_STATUS_COPIED ||
2766             p->status == DIFF_STATUS_RENAMED) {
2767                 const char *name_a, *name_b;
2768                 name_a = p->one->path;
2769                 name_b = p->two->path;
2770                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2771                 write_name_quoted(name_a, opt->file, inter_name_termination);
2772                 write_name_quoted(name_b, opt->file, line_termination);
2773         } else {
2774                 const char *name_a, *name_b;
2775                 name_a = p->one->mode ? p->one->path : p->two->path;
2776                 name_b = NULL;
2777                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2778                 write_name_quoted(name_a, opt->file, line_termination);
2779         }
2780 }
2781
2782 int diff_unmodified_pair(struct diff_filepair *p)
2783 {
2784         /* This function is written stricter than necessary to support
2785          * the currently implemented transformers, but the idea is to
2786          * let transformers to produce diff_filepairs any way they want,
2787          * and filter and clean them up here before producing the output.
2788          */
2789         struct diff_filespec *one = p->one, *two = p->two;
2790
2791         if (DIFF_PAIR_UNMERGED(p))
2792                 return 0; /* unmerged is interesting */
2793
2794         /* deletion, addition, mode or type change
2795          * and rename are all interesting.
2796          */
2797         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2798             DIFF_PAIR_MODE_CHANGED(p) ||
2799             strcmp(one->path, two->path))
2800                 return 0;
2801
2802         /* both are valid and point at the same path.  that is, we are
2803          * dealing with a change.
2804          */
2805         if (one->sha1_valid && two->sha1_valid &&
2806             !hashcmp(one->sha1, two->sha1))
2807                 return 1; /* no change */
2808         if (!one->sha1_valid && !two->sha1_valid)
2809                 return 1; /* both look at the same file on the filesystem. */
2810         return 0;
2811 }
2812
2813 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2814 {
2815         if (diff_unmodified_pair(p))
2816                 return;
2817
2818         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2819             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2820                 return; /* no tree diffs in patch format */
2821
2822         run_diff(p, o);
2823 }
2824
2825 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2826                             struct diffstat_t *diffstat)
2827 {
2828         if (diff_unmodified_pair(p))
2829                 return;
2830
2831         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2832             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2833                 return; /* no tree diffs in patch format */
2834
2835         run_diffstat(p, o, diffstat);
2836 }
2837
2838 static void diff_flush_checkdiff(struct diff_filepair *p,
2839                 struct diff_options *o)
2840 {
2841         if (diff_unmodified_pair(p))
2842                 return;
2843
2844         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2845             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2846                 return; /* no tree diffs in patch format */
2847
2848         run_checkdiff(p, o);
2849 }
2850
2851 int diff_queue_is_empty(void)
2852 {
2853         struct diff_queue_struct *q = &diff_queued_diff;
2854         int i;
2855         for (i = 0; i < q->nr; i++)
2856                 if (!diff_unmodified_pair(q->queue[i]))
2857                         return 0;
2858         return 1;
2859 }
2860
2861 #if DIFF_DEBUG
2862 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2863 {
2864         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2865                 x, one ? one : "",
2866                 s->path,
2867                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2868                 s->mode,
2869                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2870         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2871                 x, one ? one : "",
2872                 s->size, s->xfrm_flags);
2873 }
2874
2875 void diff_debug_filepair(const struct diff_filepair *p, int i)
2876 {
2877         diff_debug_filespec(p->one, i, "one");
2878         diff_debug_filespec(p->two, i, "two");
2879         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2880                 p->score, p->status ? p->status : '?',
2881                 p->one->rename_used, p->broken_pair);
2882 }
2883
2884 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2885 {
2886         int i;
2887         if (msg)
2888                 fprintf(stderr, "%s\n", msg);
2889         fprintf(stderr, "q->nr = %d\n", q->nr);
2890         for (i = 0; i < q->nr; i++) {
2891                 struct diff_filepair *p = q->queue[i];
2892                 diff_debug_filepair(p, i);
2893         }
2894 }
2895 #endif
2896
2897 static void diff_resolve_rename_copy(void)
2898 {
2899         int i;
2900         struct diff_filepair *p;
2901         struct diff_queue_struct *q = &diff_queued_diff;
2902
2903         diff_debug_queue("resolve-rename-copy", q);
2904
2905         for (i = 0; i < q->nr; i++) {
2906                 p = q->queue[i];
2907                 p->status = 0; /* undecided */
2908                 if (DIFF_PAIR_UNMERGED(p))
2909                         p->status = DIFF_STATUS_UNMERGED;
2910                 else if (!DIFF_FILE_VALID(p->one))
2911                         p->status = DIFF_STATUS_ADDED;
2912                 else if (!DIFF_FILE_VALID(p->two))
2913                         p->status = DIFF_STATUS_DELETED;
2914                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2915                         p->status = DIFF_STATUS_TYPE_CHANGED;
2916
2917                 /* from this point on, we are dealing with a pair
2918                  * whose both sides are valid and of the same type, i.e.
2919                  * either in-place edit or rename/copy edit.
2920                  */
2921                 else if (DIFF_PAIR_RENAME(p)) {
2922                         /*
2923                          * A rename might have re-connected a broken
2924                          * pair up, causing the pathnames to be the
2925                          * same again. If so, that's not a rename at
2926                          * all, just a modification..
2927                          *
2928                          * Otherwise, see if this source was used for
2929                          * multiple renames, in which case we decrement
2930                          * the count, and call it a copy.
2931                          */
2932                         if (!strcmp(p->one->path, p->two->path))
2933                                 p->status = DIFF_STATUS_MODIFIED;
2934                         else if (--p->one->rename_used > 0)
2935                                 p->status = DIFF_STATUS_COPIED;
2936                         else
2937                                 p->status = DIFF_STATUS_RENAMED;
2938                 }
2939                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2940                          p->one->mode != p->two->mode ||
2941                          is_null_sha1(p->one->sha1))
2942                         p->status = DIFF_STATUS_MODIFIED;
2943                 else {
2944                         /* This is a "no-change" entry and should not
2945                          * happen anymore, but prepare for broken callers.
2946                          */
2947                         error("feeding unmodified %s to diffcore",
2948                               p->one->path);
2949                         p->status = DIFF_STATUS_UNKNOWN;
2950                 }
2951         }
2952         diff_debug_queue("resolve-rename-copy done", q);
2953 }
2954
2955 static int check_pair_status(struct diff_filepair *p)
2956 {
2957         switch (p->status) {
2958         case DIFF_STATUS_UNKNOWN:
2959                 return 0;
2960         case 0:
2961                 die("internal error in diff-resolve-rename-copy");
2962         default:
2963                 return 1;
2964         }
2965 }
2966
2967 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2968 {
2969         int fmt = opt->output_format;
2970
2971         if (fmt & DIFF_FORMAT_CHECKDIFF)
2972                 diff_flush_checkdiff(p, opt);
2973         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2974                 diff_flush_raw(p, opt);
2975         else if (fmt & DIFF_FORMAT_NAME) {
2976                 const char *name_a, *name_b;
2977                 name_a = p->two->path;
2978                 name_b = NULL;
2979                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2980                 write_name_quoted(name_a, opt->file, opt->line_termination);
2981         }
2982 }
2983
2984 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2985 {
2986         if (fs->mode)
2987                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2988         else
2989                 fprintf(file, " %s ", newdelete);
2990         write_name_quoted(fs->path, file, '\n');
2991 }
2992
2993
2994 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
2995 {
2996         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2997                 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
2998                         show_name ? ' ' : '\n');
2999                 if (show_name) {
3000                         write_name_quoted(p->two->path, file, '\n');
3001                 }
3002         }
3003 }
3004
3005 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3006 {
3007         char *names = pprint_rename(p->one->path, p->two->path);
3008
3009         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3010         free(names);
3011         show_mode_change(file, p, 0);
3012 }
3013
3014 static void diff_summary(FILE *file, struct diff_filepair *p)
3015 {
3016         switch(p->status) {
3017         case DIFF_STATUS_DELETED:
3018                 show_file_mode_name(file, "delete", p->one);
3019                 break;
3020         case DIFF_STATUS_ADDED:
3021                 show_file_mode_name(file, "create", p->two);
3022                 break;
3023         case DIFF_STATUS_COPIED:
3024                 show_rename_copy(file, "copy", p);
3025                 break;
3026         case DIFF_STATUS_RENAMED:
3027                 show_rename_copy(file, "rename", p);
3028                 break;
3029         default:
3030                 if (p->score) {
3031                         fputs(" rewrite ", file);
3032                         write_name_quoted(p->two->path, file, ' ');
3033                         fprintf(file, "(%d%%)\n", similarity_index(p));
3034                 }
3035                 show_mode_change(file, p, !p->score);
3036                 break;
3037         }
3038 }
3039
3040 struct patch_id_t {
3041         struct xdiff_emit_state xm;
3042         SHA_CTX *ctx;
3043         int patchlen;
3044 };
3045
3046 static int remove_space(char *line, int len)
3047 {
3048         int i;
3049         char *dst = line;
3050         unsigned char c;
3051
3052         for (i = 0; i < len; i++)
3053                 if (!isspace((c = line[i])))
3054                         *dst++ = c;
3055
3056         return dst - line;
3057 }
3058
3059 static void patch_id_consume(void *priv, char *line, unsigned long len)
3060 {
3061         struct patch_id_t *data = priv;
3062         int new_len;
3063
3064         /* Ignore line numbers when computing the SHA1 of the patch */
3065         if (!prefixcmp(line, "@@ -"))
3066                 return;
3067
3068         new_len = remove_space(line, len);
3069
3070         SHA1_Update(data->ctx, line, new_len);
3071         data->patchlen += new_len;
3072 }
3073
3074 /* returns 0 upon success, and writes result into sha1 */
3075 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3076 {
3077         struct diff_queue_struct *q = &diff_queued_diff;
3078         int i;
3079         SHA_CTX ctx;
3080         struct patch_id_t data;
3081         char buffer[PATH_MAX * 4 + 20];
3082
3083         SHA1_Init(&ctx);
3084         memset(&data, 0, sizeof(struct patch_id_t));
3085         data.ctx = &ctx;
3086         data.xm.consume = patch_id_consume;
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                 ecb.outf = xdiff_outf;
3152                 ecb.priv = &data;
3153                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
3154         }
3155
3156         SHA1_Final(sha1, &ctx);
3157         return 0;
3158 }
3159
3160 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3161 {
3162         struct diff_queue_struct *q = &diff_queued_diff;
3163         int i;
3164         int result = diff_get_patch_id(options, sha1);
3165
3166         for (i = 0; i < q->nr; i++)
3167                 diff_free_filepair(q->queue[i]);
3168
3169         free(q->queue);
3170         q->queue = NULL;
3171         q->nr = q->alloc = 0;
3172
3173         return result;
3174 }
3175
3176 static int is_summary_empty(const struct diff_queue_struct *q)
3177 {
3178         int i;
3179
3180         for (i = 0; i < q->nr; i++) {
3181                 const struct diff_filepair *p = q->queue[i];
3182
3183                 switch (p->status) {
3184                 case DIFF_STATUS_DELETED:
3185                 case DIFF_STATUS_ADDED:
3186                 case DIFF_STATUS_COPIED:
3187                 case DIFF_STATUS_RENAMED:
3188                         return 0;
3189                 default:
3190                         if (p->score)
3191                                 return 0;
3192                         if (p->one->mode && p->two->mode &&
3193                             p->one->mode != p->two->mode)
3194                                 return 0;
3195                         break;
3196                 }
3197         }
3198         return 1;
3199 }
3200
3201 void diff_flush(struct diff_options *options)
3202 {
3203         struct diff_queue_struct *q = &diff_queued_diff;
3204         int i, output_format = options->output_format;
3205         int separator = 0;
3206
3207         /*
3208          * Order: raw, stat, summary, patch
3209          * or:    name/name-status/checkdiff (other bits clear)
3210          */
3211         if (!q->nr)
3212                 goto free_queue;
3213
3214         if (output_format & (DIFF_FORMAT_RAW |
3215                              DIFF_FORMAT_NAME |
3216                              DIFF_FORMAT_NAME_STATUS |
3217                              DIFF_FORMAT_CHECKDIFF)) {
3218                 for (i = 0; i < q->nr; i++) {
3219                         struct diff_filepair *p = q->queue[i];
3220                         if (check_pair_status(p))
3221                                 flush_one_pair(p, options);
3222                 }
3223                 separator++;
3224         }
3225
3226         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3227                 struct diffstat_t diffstat;
3228
3229                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3230                 diffstat.xm.consume = diffstat_consume;
3231                 for (i = 0; i < q->nr; i++) {
3232                         struct diff_filepair *p = q->queue[i];
3233                         if (check_pair_status(p))
3234                                 diff_flush_stat(p, options, &diffstat);
3235                 }
3236                 if (output_format & DIFF_FORMAT_NUMSTAT)
3237                         show_numstat(&diffstat, options);
3238                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3239                         show_stats(&diffstat, options);
3240                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3241                         show_shortstats(&diffstat, options);
3242                 free_diffstat_info(&diffstat);
3243                 separator++;
3244         }
3245         if (output_format & DIFF_FORMAT_DIRSTAT)
3246                 show_dirstat(options);
3247
3248         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3249                 for (i = 0; i < q->nr; i++)
3250                         diff_summary(options->file, q->queue[i]);
3251                 separator++;
3252         }
3253
3254         if (output_format & DIFF_FORMAT_PATCH) {
3255                 if (separator) {
3256                         putc(options->line_termination, options->file);
3257                         if (options->stat_sep) {
3258                                 /* attach patch instead of inline */
3259                                 fputs(options->stat_sep, options->file);
3260                         }
3261                 }
3262
3263                 for (i = 0; i < q->nr; i++) {
3264                         struct diff_filepair *p = q->queue[i];
3265                         if (check_pair_status(p))
3266                                 diff_flush_patch(p, options);
3267                 }
3268         }
3269
3270         if (output_format & DIFF_FORMAT_CALLBACK)
3271                 options->format_callback(q, options, options->format_callback_data);
3272
3273         for (i = 0; i < q->nr; i++)
3274                 diff_free_filepair(q->queue[i]);
3275 free_queue:
3276         free(q->queue);
3277         q->queue = NULL;
3278         q->nr = q->alloc = 0;
3279         if (options->close_file)
3280                 fclose(options->file);
3281 }
3282
3283 static void diffcore_apply_filter(const char *filter)
3284 {
3285         int i;
3286         struct diff_queue_struct *q = &diff_queued_diff;
3287         struct diff_queue_struct outq;
3288         outq.queue = NULL;
3289         outq.nr = outq.alloc = 0;
3290
3291         if (!filter)
3292                 return;
3293
3294         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3295                 int found;
3296                 for (i = found = 0; !found && i < q->nr; i++) {
3297                         struct diff_filepair *p = q->queue[i];
3298                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3299                              ((p->score &&
3300                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3301                               (!p->score &&
3302                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3303                             ((p->status != DIFF_STATUS_MODIFIED) &&
3304                              strchr(filter, p->status)))
3305                                 found++;
3306                 }
3307                 if (found)
3308                         return;
3309
3310                 /* otherwise we will clear the whole queue
3311                  * by copying the empty outq at the end of this
3312                  * function, but first clear the current entries
3313                  * in the queue.
3314                  */
3315                 for (i = 0; i < q->nr; i++)
3316                         diff_free_filepair(q->queue[i]);
3317         }
3318         else {
3319                 /* Only the matching ones */
3320                 for (i = 0; i < q->nr; i++) {
3321                         struct diff_filepair *p = q->queue[i];
3322
3323                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3324                              ((p->score &&
3325                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3326                               (!p->score &&
3327                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3328                             ((p->status != DIFF_STATUS_MODIFIED) &&
3329                              strchr(filter, p->status)))
3330                                 diff_q(&outq, p);
3331                         else
3332                                 diff_free_filepair(p);
3333                 }
3334         }
3335         free(q->queue);
3336         *q = outq;
3337 }
3338
3339 /* Check whether two filespecs with the same mode and size are identical */
3340 static int diff_filespec_is_identical(struct diff_filespec *one,
3341                                       struct diff_filespec *two)
3342 {
3343         if (S_ISGITLINK(one->mode))
3344                 return 0;
3345         if (diff_populate_filespec(one, 0))
3346                 return 0;
3347         if (diff_populate_filespec(two, 0))
3348                 return 0;
3349         return !memcmp(one->data, two->data, one->size);
3350 }
3351
3352 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3353 {
3354         int i;
3355         struct diff_queue_struct *q = &diff_queued_diff;
3356         struct diff_queue_struct outq;
3357         outq.queue = NULL;
3358         outq.nr = outq.alloc = 0;
3359
3360         for (i = 0; i < q->nr; i++) {
3361                 struct diff_filepair *p = q->queue[i];
3362
3363                 /*
3364                  * 1. Entries that come from stat info dirtyness
3365                  *    always have both sides (iow, not create/delete),
3366                  *    one side of the object name is unknown, with
3367                  *    the same mode and size.  Keep the ones that
3368                  *    do not match these criteria.  They have real
3369                  *    differences.
3370                  *
3371                  * 2. At this point, the file is known to be modified,
3372                  *    with the same mode and size, and the object
3373                  *    name of one side is unknown.  Need to inspect
3374                  *    the identical contents.
3375                  */
3376                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3377                     !DIFF_FILE_VALID(p->two) ||
3378                     (p->one->sha1_valid && p->two->sha1_valid) ||
3379                     (p->one->mode != p->two->mode) ||
3380                     diff_populate_filespec(p->one, 1) ||
3381                     diff_populate_filespec(p->two, 1) ||
3382                     (p->one->size != p->two->size) ||
3383                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3384                         diff_q(&outq, p);
3385                 else {
3386                         /*
3387                          * The caller can subtract 1 from skip_stat_unmatch
3388                          * to determine how many paths were dirty only
3389                          * due to stat info mismatch.
3390                          */
3391                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3392                                 diffopt->skip_stat_unmatch++;
3393                         diff_free_filepair(p);
3394                 }
3395         }
3396         free(q->queue);
3397         *q = outq;
3398 }
3399
3400 void diffcore_std(struct diff_options *options)
3401 {
3402         if (DIFF_OPT_TST(options, QUIET))
3403                 return;
3404
3405         if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3406                 diffcore_skip_stat_unmatch(options);
3407         if (options->break_opt != -1)
3408                 diffcore_break(options->break_opt);
3409         if (options->detect_rename)
3410                 diffcore_rename(options);
3411         if (options->break_opt != -1)
3412                 diffcore_merge_broken();
3413         if (options->pickaxe)
3414                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3415         if (options->orderfile)
3416                 diffcore_order(options->orderfile);
3417         diff_resolve_rename_copy();
3418         diffcore_apply_filter(options->filter);
3419
3420         if (diff_queued_diff.nr)
3421                 DIFF_OPT_SET(options, HAS_CHANGES);
3422         else
3423                 DIFF_OPT_CLR(options, HAS_CHANGES);
3424 }
3425
3426 int diff_result_code(struct diff_options *opt, int status)
3427 {
3428         int result = 0;
3429         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3430             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3431                 return status;
3432         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3433             DIFF_OPT_TST(opt, HAS_CHANGES))
3434                 result |= 01;
3435         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3436             DIFF_OPT_TST(opt, CHECK_FAILED))
3437                 result |= 02;
3438         return result;
3439 }
3440
3441 void diff_addremove(struct diff_options *options,
3442                     int addremove, unsigned mode,
3443                     const unsigned char *sha1,
3444                     const char *concatpath)
3445 {
3446         struct diff_filespec *one, *two;
3447
3448         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3449                 return;
3450
3451         /* This may look odd, but it is a preparation for
3452          * feeding "there are unchanged files which should
3453          * not produce diffs, but when you are doing copy
3454          * detection you would need them, so here they are"
3455          * entries to the diff-core.  They will be prefixed
3456          * with something like '=' or '*' (I haven't decided
3457          * which but should not make any difference).
3458          * Feeding the same new and old to diff_change()
3459          * also has the same effect.
3460          * Before the final output happens, they are pruned after
3461          * merged into rename/copy pairs as appropriate.
3462          */
3463         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3464                 addremove = (addremove == '+' ? '-' :
3465                              addremove == '-' ? '+' : addremove);
3466
3467         if (options->prefix &&
3468             strncmp(concatpath, options->prefix, options->prefix_length))
3469                 return;
3470
3471         one = alloc_filespec(concatpath);
3472         two = alloc_filespec(concatpath);
3473
3474         if (addremove != '+')
3475                 fill_filespec(one, sha1, mode);
3476         if (addremove != '-')
3477                 fill_filespec(two, sha1, mode);
3478
3479         diff_queue(&diff_queued_diff, one, two);
3480         DIFF_OPT_SET(options, HAS_CHANGES);
3481 }
3482
3483 void diff_change(struct diff_options *options,
3484                  unsigned old_mode, unsigned new_mode,
3485                  const unsigned char *old_sha1,
3486                  const unsigned char *new_sha1,
3487                  const char *concatpath)
3488 {
3489         struct diff_filespec *one, *two;
3490
3491         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3492                         && S_ISGITLINK(new_mode))
3493                 return;
3494
3495         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3496                 unsigned tmp;
3497                 const unsigned char *tmp_c;
3498                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3499                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3500         }
3501
3502         if (options->prefix &&
3503             strncmp(concatpath, options->prefix, options->prefix_length))
3504                 return;
3505
3506         one = alloc_filespec(concatpath);
3507         two = alloc_filespec(concatpath);
3508         fill_filespec(one, old_sha1, old_mode);
3509         fill_filespec(two, new_sha1, new_mode);
3510
3511         diff_queue(&diff_queued_diff, one, two);
3512         DIFF_OPT_SET(options, HAS_CHANGES);
3513 }
3514
3515 void diff_unmerge(struct diff_options *options,
3516                   const char *path,
3517                   unsigned mode, const unsigned char *sha1)
3518 {
3519         struct diff_filespec *one, *two;
3520
3521         if (options->prefix &&
3522             strncmp(path, options->prefix, options->prefix_length))
3523                 return;
3524
3525         one = alloc_filespec(path);
3526         two = alloc_filespec(path);
3527         fill_filespec(one, sha1, mode);
3528         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3529 }