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
  14 #ifdef NO_FAST_WORKING_DIRECTORY
  15 #define FAST_WORKING_DIRECTORY 0
  16 #else
  17 #define FAST_WORKING_DIRECTORY 1
  18 #endif
  19
  20 static int diff_detect_rename_default;
  21 static int diff_rename_limit_default = 100;
  22 static int diff_use_color_default;
  23 static const char *external_diff_cmd_cfg;
  24 int diff_auto_refresh_index = 1;
  25
  26 static char diff_colors[][COLOR_MAXLEN] = {
  27         "\033[m",       /* reset */
  28         "",             /* PLAIN (normal) */
  29         "\033[1m",      /* METAINFO (bold) */
  30         "\033[36m",     /* FRAGINFO (cyan) */
  31         "\033[31m",     /* OLD (red) */
  32         "\033[32m",     /* NEW (green) */
  33         "\033[33m",     /* COMMIT (yellow) */
  34         "\033[41m",     /* WHITESPACE (red background) */
  35 };
  36
  37 static int parse_diff_color_slot(const char *var, int ofs)
  38 {
  39         if (!strcasecmp(var+ofs, "plain"))
  40                 return DIFF_PLAIN;
  41         if (!strcasecmp(var+ofs, "meta"))
  42                 return DIFF_METAINFO;
  43         if (!strcasecmp(var+ofs, "frag"))
  44                 return DIFF_FRAGINFO;
  45         if (!strcasecmp(var+ofs, "old"))
  46                 return DIFF_FILE_OLD;
  47         if (!strcasecmp(var+ofs, "new"))
  48                 return DIFF_FILE_NEW;
  49         if (!strcasecmp(var+ofs, "commit"))
  50                 return DIFF_COMMIT;
  51         if (!strcasecmp(var+ofs, "whitespace"))
  52                 return DIFF_WHITESPACE;
  53         die("bad config variable '%s'", var);
  54 }
  55
  56 static struct ll_diff_driver {
  57         const char *name;
  58         struct ll_diff_driver *next;
  59         char *cmd;
  60 } *user_diff, **user_diff_tail;
  61
  62 static void read_config_if_needed(void)
  63 {
  64         if (!user_diff_tail) {
  65                 user_diff_tail = &user_diff;
  66                 git_config(git_diff_ui_config);
  67         }
  68 }
  69
  70 /*
  71  * Currently there is only "diff.<drivername>.command" variable;
  72  * because there are "diff.color.<slot>" variables, we are parsing
  73  * this in a bit convoluted way to allow low level diff driver
  74  * called "color".
  75  */
  76 static int parse_lldiff_command(const char *var, const char *ep, const char *value)
  77 {
  78         const char *name;
  79         int namelen;
  80         struct ll_diff_driver *drv;
  81
  82         name = var + 5;
  83         namelen = ep - name;
  84         for (drv = user_diff; drv; drv = drv->next)
  85                 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
  86                         break;
  87         if (!drv) {
  88                 drv = xcalloc(1, sizeof(struct ll_diff_driver));
  89                 drv->name = xmemdupz(name, namelen);
  90                 if (!user_diff_tail)
  91                         user_diff_tail = &user_diff;
  92                 *user_diff_tail = drv;
  93                 user_diff_tail = &(drv->next);
  94         }
  95
  96         if (!value)
  97                 return error("%s: lacks value", var);
  98         drv->cmd = strdup(value);
  99         return 0;
 100 }
 101
 102 /*
 103  * 'diff.<what>.funcname' attribute can be specified in the configuration
 104  * to define a customized regexp to find the beginning of a function to
 105  * be used for hunk header lines of "diff -p" style output.
 106  */
 107 static struct funcname_pattern {
 108         char *name;
 109         char *pattern;
 110         struct funcname_pattern *next;
 111 } *funcname_pattern_list;
 112
 113 static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
 114 {
 115         const char *name;
 116         int namelen;
 117         struct funcname_pattern *pp;
 118
 119         name = var + 5; /* "diff." */
 120         namelen = ep - name;
 121
 122         for (pp = funcname_pattern_list; pp; pp = pp->next)
 123                 if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
 124                         break;
 125         if (!pp) {
 126                 pp = xcalloc(1, sizeof(*pp));
 127                 pp->name = xmemdupz(name, namelen);
 128                 pp->next = funcname_pattern_list;
 129                 funcname_pattern_list = pp;
 130         }
 131         if (pp->pattern)
 132                 free(pp->pattern);
 133         pp->pattern = xstrdup(value);
 134         return 0;
 135 }
 136
 137 /*
 138  * These are to give UI layer defaults.
 139  * The core-level commands such as git-diff-files should
 140  * never be affected by the setting of diff.renames
 141  * the user happens to have in the configuration file.
 142  */
 143 int git_diff_ui_config(const char *var, const char *value)
 144 {
 145         if (!strcmp(var, "diff.renamelimit")) {
 146                 diff_rename_limit_default = git_config_int(var, value);
 147                 return 0;
 148         }
 149         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 150                 diff_use_color_default = git_config_colorbool(var, value, -1);
 151                 return 0;
 152         }
 153         if (!strcmp(var, "diff.renames")) {
 154                 if (!value)
 155                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 156                 else if (!strcasecmp(value, "copies") ||
 157                          !strcasecmp(value, "copy"))
 158                         diff_detect_rename_default = DIFF_DETECT_COPY;
 159                 else if (git_config_bool(var,value))
 160                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 161                 return 0;
 162         }
 163         if (!strcmp(var, "diff.autorefreshindex")) {
 164                 diff_auto_refresh_index = git_config_bool(var, value);
 165                 return 0;
 166         }
 167         if (!strcmp(var, "diff.external")) {
 168                 external_diff_cmd_cfg = xstrdup(value);
 169                 return 0;
 170         }
 171         if (!prefixcmp(var, "diff.")) {
 172                 const char *ep = strrchr(var, '.');
 173
 174                 if (ep != var + 4) {
 175                         if (!strcmp(ep, ".command"))
 176                                 return parse_lldiff_command(var, ep, value);
 177                         if (!strcmp(ep, ".funcname"))
 178                                 return parse_funcname_pattern(var, ep, value);
 179                 }
 180         }
 181
 182         return git_diff_basic_config(var, value);
 183 }
 184
 185 int git_diff_basic_config(const char *var, const char *value)
 186 {
 187         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 188                 int slot = parse_diff_color_slot(var, 11);
 189                 color_parse(value, var, diff_colors[slot]);
 190                 return 0;
 191         }
 192
 193         return git_default_config(var, value);
 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(int count)
 260 {
 261         switch (count) {
 262         case 0:
 263                 printf("0,0");
 264                 break;
 265         case 1:
 266                 printf("1");
 267                 break;
 268         default:
 269                 printf("1,%d", count);
 270                 break;
 271         }
 272 }
 273
 274 static void copy_file(int prefix, const char *data, int size,
 275                 const char *set, const char *reset)
 276 {
 277         int ch, nl_just_seen = 1;
 278         while (0 < size--) {
 279                 ch = *data++;
 280                 if (nl_just_seen) {
 281                         fputs(set, stdout);
 282                         putchar(prefix);
 283                 }
 284                 if (ch == '\n') {
 285                         nl_just_seen = 1;
 286                         fputs(reset, stdout);
 287                 } else
 288                         nl_just_seen = 0;
 289                 putchar(ch);
 290         }
 291         if (!nl_just_seen)
 292                 printf("%s\n\\ No newline at end of file\n", reset);
 293 }
 294
 295 static void emit_rewrite_diff(const char *name_a,
 296                               const char *name_b,
 297                               struct diff_filespec *one,
 298                               struct diff_filespec *two,
 299                               struct diff_options *o)
 300 {
 301         int lc_a, lc_b;
 302         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
 303         const char *name_a_tab, *name_b_tab;
 304         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 305         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 306         const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
 307         const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
 308         const char *reset = diff_get_color(color_diff, DIFF_RESET);
 309         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 310
 311         name_a += (*name_a == '/');
 312         name_b += (*name_b == '/');
 313         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 314         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 315
 316         strbuf_reset(&a_name);
 317         strbuf_reset(&b_name);
 318         quote_two_c_style(&a_name, o->a_prefix, name_a, 0);
 319         quote_two_c_style(&b_name, o->b_prefix, name_b, 0);
 320
 321         diff_populate_filespec(one, 0);
 322         diff_populate_filespec(two, 0);
 323         lc_a = count_lines(one->data, one->size);
 324         lc_b = count_lines(two->data, two->size);
 325         printf("%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
 326                metainfo, a_name.buf, name_a_tab, reset,
 327                metainfo, b_name.buf, name_b_tab, reset, fraginfo);
 328         print_line_count(lc_a);
 329         printf(" +");
 330         print_line_count(lc_b);
 331         printf(" @@%s\n", reset);
 332         if (lc_a)
 333                 copy_file('-', one->data, one->size, old, reset);
 334         if (lc_b)
 335                 copy_file('+', two->data, two->size, new, reset);
 336 }
 337
 338 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 339 {
 340         if (!DIFF_FILE_VALID(one)) {
 341                 mf->ptr = (char *)""; /* does not matter */
 342                 mf->size = 0;
 343                 return 0;
 344         }
 345         else if (diff_populate_filespec(one, 0))
 346                 return -1;
 347         mf->ptr = one->data;
 348         mf->size = one->size;
 349         return 0;
 350 }
 351
 352 struct diff_words_buffer {
 353         mmfile_t text;
 354         long alloc;
 355         long current; /* output pointer */
 356         int suppressed_newline;
 357 };
 358
 359 static void diff_words_append(char *line, unsigned long len,
 360                 struct diff_words_buffer *buffer)
 361 {
 362         if (buffer->text.size + len > buffer->alloc) {
 363                 buffer->alloc = (buffer->text.size + len) * 3 / 2;
 364                 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
 365         }
 366         line++;
 367         len--;
 368         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 369         buffer->text.size += len;
 370 }
 371
 372 struct diff_words_data {
 373         struct xdiff_emit_state xm;
 374         struct diff_words_buffer minus, plus;
 375 };
 376
 377 static void print_word(struct diff_words_buffer *buffer, int len, int color,
 378                 int suppress_newline)
 379 {
 380         const char *ptr;
 381         int eol = 0;
 382
 383         if (len == 0)
 384                 return;
 385
 386         ptr  = buffer->text.ptr + buffer->current;
 387         buffer->current += len;
 388
 389         if (ptr[len - 1] == '\n') {
 390                 eol = 1;
 391                 len--;
 392         }
 393
 394         fputs(diff_get_color(1, color), stdout);
 395         fwrite(ptr, len, 1, stdout);
 396         fputs(diff_get_color(1, DIFF_RESET), stdout);
 397
 398         if (eol) {
 399                 if (suppress_newline)
 400                         buffer->suppressed_newline = 1;
 401                 else
 402                         putchar('\n');
 403         }
 404 }
 405
 406 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 407 {
 408         struct diff_words_data *diff_words = priv;
 409
 410         if (diff_words->minus.suppressed_newline) {
 411                 if (line[0] != '+')
 412                         putchar('\n');
 413                 diff_words->minus.suppressed_newline = 0;
 414         }
 415
 416         len--;
 417         switch (line[0]) {
 418                 case '-':
 419                         print_word(&diff_words->minus, len, DIFF_FILE_OLD, 1);
 420                         break;
 421                 case '+':
 422                         print_word(&diff_words->plus, len, DIFF_FILE_NEW, 0);
 423                         break;
 424                 case ' ':
 425                         print_word(&diff_words->plus, len, DIFF_PLAIN, 0);
 426                         diff_words->minus.current += len;
 427                         break;
 428         }
 429 }
 430
 431 /* this executes the word diff on the accumulated buffers */
 432 static void diff_words_show(struct diff_words_data *diff_words)
 433 {
 434         xpparam_t xpp;
 435         xdemitconf_t xecfg;
 436         xdemitcb_t ecb;
 437         mmfile_t minus, plus;
 438         int i;
 439
 440         memset(&xecfg, 0, sizeof(xecfg));
 441         minus.size = diff_words->minus.text.size;
 442         minus.ptr = xmalloc(minus.size);
 443         memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
 444         for (i = 0; i < minus.size; i++)
 445                 if (isspace(minus.ptr[i]))
 446                         minus.ptr[i] = '\n';
 447         diff_words->minus.current = 0;
 448
 449         plus.size = diff_words->plus.text.size;
 450         plus.ptr = xmalloc(plus.size);
 451         memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
 452         for (i = 0; i < plus.size; i++)
 453                 if (isspace(plus.ptr[i]))
 454                         plus.ptr[i] = '\n';
 455         diff_words->plus.current = 0;
 456
 457         xpp.flags = XDF_NEED_MINIMAL;
 458         xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
 459         ecb.outf = xdiff_outf;
 460         ecb.priv = diff_words;
 461         diff_words->xm.consume = fn_out_diff_words_aux;
 462         xdi_diff(&minus, &plus, &xpp, &xecfg, &ecb);
 463
 464         free(minus.ptr);
 465         free(plus.ptr);
 466         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 467
 468         if (diff_words->minus.suppressed_newline) {
 469                 putchar('\n');
 470                 diff_words->minus.suppressed_newline = 0;
 471         }
 472 }
 473
 474 struct emit_callback {
 475         struct xdiff_emit_state xm;
 476         int nparents, color_diff;
 477         unsigned ws_rule;
 478         const char **label_path;
 479         struct diff_words_data *diff_words;
 480         int *found_changesp;
 481 };
 482
 483 static void free_diff_words_data(struct emit_callback *ecbdata)
 484 {
 485         if (ecbdata->diff_words) {
 486                 /* flush buffers */
 487                 if (ecbdata->diff_words->minus.text.size ||
 488                                 ecbdata->diff_words->plus.text.size)
 489                         diff_words_show(ecbdata->diff_words);
 490
 491                 if (ecbdata->diff_words->minus.text.ptr)
 492                         free (ecbdata->diff_words->minus.text.ptr);
 493                 if (ecbdata->diff_words->plus.text.ptr)
 494                         free (ecbdata->diff_words->plus.text.ptr);
 495                 free(ecbdata->diff_words);
 496                 ecbdata->diff_words = NULL;
 497         }
 498 }
 499
 500 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 501 {
 502         if (diff_use_color)
 503                 return diff_colors[ix];
 504         return "";
 505 }
 506
 507 static void emit_line(const char *set, const char *reset, const char *line, int len)
 508 {
 509         fputs(set, stdout);
 510         fwrite(line, len, 1, stdout);
 511         fputs(reset, stdout);
 512 }
 513
 514 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
 515 {
 516         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 517         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 518
 519         if (!*ws)
 520                 emit_line(set, reset, line, len);
 521         else {
 522                 /* Emit just the prefix, then the rest. */
 523                 emit_line(set, reset, line, ecbdata->nparents);
 524                 (void)check_and_emit_line(line + ecbdata->nparents,
 525                     len - ecbdata->nparents, ecbdata->ws_rule,
 526                     stdout, set, reset, ws);
 527         }
 528 }
 529
 530 static void fn_out_consume(void *priv, char *line, unsigned long len)
 531 {
 532         int i;
 533         int color;
 534         struct emit_callback *ecbdata = priv;
 535         const char *set = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 536         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 537
 538         *(ecbdata->found_changesp) = 1;
 539
 540         if (ecbdata->label_path[0]) {
 541                 const char *name_a_tab, *name_b_tab;
 542
 543                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 544                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 545
 546                 printf("%s--- %s%s%s\n",
 547                        set, ecbdata->label_path[0], reset, name_a_tab);
 548                 printf("%s+++ %s%s%s\n",
 549                        set, ecbdata->label_path[1], reset, name_b_tab);
 550                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 551         }
 552
 553         /* This is not really necessary for now because
 554          * this codepath only deals with two-way diffs.
 555          */
 556         for (i = 0; i < len && line[i] == '@'; i++)
 557                 ;
 558         if (2 <= i && i < len && line[i] == ' ') {
 559                 ecbdata->nparents = i - 1;
 560                 emit_line(diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
 561                           reset, line, len);
 562                 return;
 563         }
 564
 565         if (len < ecbdata->nparents) {
 566                 set = reset;
 567                 emit_line(reset, reset, line, len);
 568                 return;
 569         }
 570
 571         color = DIFF_PLAIN;
 572         if (ecbdata->diff_words && ecbdata->nparents != 1)
 573                 /* fall back to normal diff */
 574                 free_diff_words_data(ecbdata);
 575         if (ecbdata->diff_words) {
 576                 if (line[0] == '-') {
 577                         diff_words_append(line, len,
 578                                           &ecbdata->diff_words->minus);
 579                         return;
 580                 } else if (line[0] == '+') {
 581                         diff_words_append(line, len,
 582                                           &ecbdata->diff_words->plus);
 583                         return;
 584                 }
 585                 if (ecbdata->diff_words->minus.text.size ||
 586                     ecbdata->diff_words->plus.text.size)
 587                         diff_words_show(ecbdata->diff_words);
 588                 line++;
 589                 len--;
 590                 emit_line(set, reset, line, len);
 591                 return;
 592         }
 593         for (i = 0; i < ecbdata->nparents && len; i++) {
 594                 if (line[i] == '-')
 595                         color = DIFF_FILE_OLD;
 596                 else if (line[i] == '+')
 597                         color = DIFF_FILE_NEW;
 598         }
 599
 600         if (color != DIFF_FILE_NEW) {
 601                 emit_line(diff_get_color(ecbdata->color_diff, color),
 602                           reset, line, len);
 603                 return;
 604         }
 605         emit_add_line(reset, ecbdata, line, len);
 606 }
 607
 608 static char *pprint_rename(const char *a, const char *b)
 609 {
 610         const char *old = a;
 611         const char *new = b;
 612         struct strbuf name;
 613         int pfx_length, sfx_length;
 614         int len_a = strlen(a);
 615         int len_b = strlen(b);
 616         int a_midlen, b_midlen;
 617         int qlen_a = quote_c_style(a, NULL, NULL, 0);
 618         int qlen_b = quote_c_style(b, NULL, NULL, 0);
 619
 620         strbuf_init(&name, 0);
 621         if (qlen_a || qlen_b) {
 622                 quote_c_style(a, &name, NULL, 0);
 623                 strbuf_addstr(&name, " => ");
 624                 quote_c_style(b, &name, NULL, 0);
 625                 return strbuf_detach(&name, NULL);
 626         }
 627
 628         /* Find common prefix */
 629         pfx_length = 0;
 630         while (*old && *new && *old == *new) {
 631                 if (*old == '/')
 632                         pfx_length = old - a + 1;
 633                 old++;
 634                 new++;
 635         }
 636
 637         /* Find common suffix */
 638         old = a + len_a;
 639         new = b + len_b;
 640         sfx_length = 0;
 641         while (a <= old && b <= new && *old == *new) {
 642                 if (*old == '/')
 643                         sfx_length = len_a - (old - a);
 644                 old--;
 645                 new--;
 646         }
 647
 648         /*
 649          * pfx{mid-a => mid-b}sfx
 650          * {pfx-a => pfx-b}sfx
 651          * pfx{sfx-a => sfx-b}
 652          * name-a => name-b
 653          */
 654         a_midlen = len_a - pfx_length - sfx_length;
 655         b_midlen = len_b - pfx_length - sfx_length;
 656         if (a_midlen < 0)
 657                 a_midlen = 0;
 658         if (b_midlen < 0)
 659                 b_midlen = 0;
 660
 661         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 662         if (pfx_length + sfx_length) {
 663                 strbuf_add(&name, a, pfx_length);
 664                 strbuf_addch(&name, '{');
 665         }
 666         strbuf_add(&name, a + pfx_length, a_midlen);
 667         strbuf_addstr(&name, " => ");
 668         strbuf_add(&name, b + pfx_length, b_midlen);
 669         if (pfx_length + sfx_length) {
 670                 strbuf_addch(&name, '}');
 671                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 672         }
 673         return strbuf_detach(&name, NULL);
 674 }
 675
 676 struct diffstat_t {
 677         struct xdiff_emit_state xm;
 678
 679         int nr;
 680         int alloc;
 681         struct diffstat_file {
 682                 char *from_name;
 683                 char *name;
 684                 char *print_name;
 685                 unsigned is_unmerged:1;
 686                 unsigned is_binary:1;
 687                 unsigned is_renamed:1;
 688                 unsigned int added, deleted;
 689         } **files;
 690 };
 691
 692 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 693                                           const char *name_a,
 694                                           const char *name_b)
 695 {
 696         struct diffstat_file *x;
 697         x = xcalloc(sizeof (*x), 1);
 698         if (diffstat->nr == diffstat->alloc) {
 699                 diffstat->alloc = alloc_nr(diffstat->alloc);
 700                 diffstat->files = xrealloc(diffstat->files,
 701                                 diffstat->alloc * sizeof(x));
 702         }
 703         diffstat->files[diffstat->nr++] = x;
 704         if (name_b) {
 705                 x->from_name = xstrdup(name_a);
 706                 x->name = xstrdup(name_b);
 707                 x->is_renamed = 1;
 708         }
 709         else {
 710                 x->from_name = NULL;
 711                 x->name = xstrdup(name_a);
 712         }
 713         return x;
 714 }
 715
 716 static void diffstat_consume(void *priv, char *line, unsigned long len)
 717 {
 718         struct diffstat_t *diffstat = priv;
 719         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 720
 721         if (line[0] == '+')
 722                 x->added++;
 723         else if (line[0] == '-')
 724                 x->deleted++;
 725 }
 726
 727 const char mime_boundary_leader[] = "------------";
 728
 729 static int scale_linear(int it, int width, int max_change)
 730 {
 731         /*
 732          * make sure that at least one '-' is printed if there were deletions,
 733          * and likewise for '+'.
 734          */
 735         if (max_change < 2)
 736                 return it;
 737         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
 738 }
 739
 740 static void show_name(const char *prefix, const char *name, int len,
 741                       const char *reset, const char *set)
 742 {
 743         printf(" %s%s%-*s%s |", set, prefix, len, name, reset);
 744 }
 745
 746 static void show_graph(char ch, int cnt, const char *set, const char *reset)
 747 {
 748         if (cnt <= 0)
 749                 return;
 750         printf("%s", set);
 751         while (cnt--)
 752                 putchar(ch);
 753         printf("%s", reset);
 754 }
 755
 756 static void fill_print_name(struct diffstat_file *file)
 757 {
 758         char *pname;
 759
 760         if (file->print_name)
 761                 return;
 762
 763         if (!file->is_renamed) {
 764                 struct strbuf buf;
 765                 strbuf_init(&buf, 0);
 766                 if (quote_c_style(file->name, &buf, NULL, 0)) {
 767                         pname = strbuf_detach(&buf, NULL);
 768                 } else {
 769                         pname = file->name;
 770                         strbuf_release(&buf);
 771                 }
 772         } else {
 773                 pname = pprint_rename(file->from_name, file->name);
 774         }
 775         file->print_name = pname;
 776 }
 777
 778 static void show_stats(struct diffstat_t* data, struct diff_options *options)
 779 {
 780         int i, len, add, del, total, adds = 0, dels = 0;
 781         int max_change = 0, max_len = 0;
 782         int total_files = data->nr;
 783         int width, name_width;
 784         const char *reset, *set, *add_c, *del_c;
 785
 786         if (data->nr == 0)
 787                 return;
 788
 789         width = options->stat_width ? options->stat_width : 80;
 790         name_width = options->stat_name_width ? options->stat_name_width : 50;
 791
 792         /* Sanity: give at least 5 columns to the graph,
 793          * but leave at least 10 columns for the name.
 794          */
 795         if (width < name_width + 15) {
 796                 if (name_width <= 25)
 797                         width = name_width + 15;
 798                 else
 799                         name_width = width - 15;
 800         }
 801
 802         /* Find the longest filename and max number of changes */
 803         reset = diff_get_color_opt(options, DIFF_RESET);
 804         set   = diff_get_color_opt(options, DIFF_PLAIN);
 805         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
 806         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
 807
 808         for (i = 0; i < data->nr; i++) {
 809                 struct diffstat_file *file = data->files[i];
 810                 int change = file->added + file->deleted;
 811                 fill_print_name(file);
 812                 len = strlen(file->print_name);
 813                 if (max_len < len)
 814                         max_len = len;
 815
 816                 if (file->is_binary || file->is_unmerged)
 817                         continue;
 818                 if (max_change < change)
 819                         max_change = change;
 820         }
 821
 822         /* Compute the width of the graph part;
 823          * 10 is for one blank at the beginning of the line plus
 824          * " | count " between the name and the graph.
 825          *
 826          * From here on, name_width is the width of the name area,
 827          * and width is the width of the graph area.
 828          */
 829         name_width = (name_width < max_len) ? name_width : max_len;
 830         if (width < (name_width + 10) + max_change)
 831                 width = width - (name_width + 10);
 832         else
 833                 width = max_change;
 834
 835         for (i = 0; i < data->nr; i++) {
 836                 const char *prefix = "";
 837                 char *name = data->files[i]->print_name;
 838                 int added = data->files[i]->added;
 839                 int deleted = data->files[i]->deleted;
 840                 int name_len;
 841
 842                 /*
 843                  * "scale" the filename
 844                  */
 845                 len = name_width;
 846                 name_len = strlen(name);
 847                 if (name_width < name_len) {
 848                         char *slash;
 849                         prefix = "...";
 850                         len -= 3;
 851                         name += name_len - len;
 852                         slash = strchr(name, '/');
 853                         if (slash)
 854                                 name = slash;
 855                 }
 856
 857                 if (data->files[i]->is_binary) {
 858                         show_name(prefix, name, len, reset, set);
 859                         printf("  Bin ");
 860                         printf("%s%d%s", del_c, deleted, reset);
 861                         printf(" -> ");
 862                         printf("%s%d%s", add_c, added, reset);
 863                         printf(" bytes");
 864                         printf("\n");
 865                         continue;
 866                 }
 867                 else if (data->files[i]->is_unmerged) {
 868                         show_name(prefix, name, len, reset, set);
 869                         printf("  Unmerged\n");
 870                         continue;
 871                 }
 872                 else if (!data->files[i]->is_renamed &&
 873                          (added + deleted == 0)) {
 874                         total_files--;
 875                         continue;
 876                 }
 877
 878                 /*
 879                  * scale the add/delete
 880                  */
 881                 add = added;
 882                 del = deleted;
 883                 total = add + del;
 884                 adds += add;
 885                 dels += del;
 886
 887                 if (width <= max_change) {
 888                         add = scale_linear(add, width, max_change);
 889                         del = scale_linear(del, width, max_change);
 890                         total = add + del;
 891                 }
 892                 show_name(prefix, name, len, reset, set);
 893                 printf("%5d ", added + deleted);
 894                 show_graph('+', add, add_c, reset);
 895                 show_graph('-', del, del_c, reset);
 896                 putchar('\n');
 897         }
 898         printf("%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
 899                set, total_files, adds, dels, reset);
 900 }
 901
 902 static void show_shortstats(struct diffstat_t* data)
 903 {
 904         int i, adds = 0, dels = 0, total_files = data->nr;
 905
 906         if (data->nr == 0)
 907                 return;
 908
 909         for (i = 0; i < data->nr; i++) {
 910                 if (!data->files[i]->is_binary &&
 911                     !data->files[i]->is_unmerged) {
 912                         int added = data->files[i]->added;
 913                         int deleted= data->files[i]->deleted;
 914                         if (!data->files[i]->is_renamed &&
 915                             (added + deleted == 0)) {
 916                                 total_files--;
 917                         } else {
 918                                 adds += added;
 919                                 dels += deleted;
 920                         }
 921                 }
 922         }
 923         printf(" %d files changed, %d insertions(+), %d deletions(-)\n",
 924                total_files, adds, dels);
 925 }
 926
 927 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
 928 {
 929         int i;
 930
 931         if (data->nr == 0)
 932                 return;
 933
 934         for (i = 0; i < data->nr; i++) {
 935                 struct diffstat_file *file = data->files[i];
 936
 937                 if (file->is_binary)
 938                         printf("-\t-\t");
 939                 else
 940                         printf("%d\t%d\t", file->added, file->deleted);
 941                 if (options->line_termination) {
 942                         fill_print_name(file);
 943                         if (!file->is_renamed)
 944                                 write_name_quoted(file->name, stdout,
 945                                                   options->line_termination);
 946                         else {
 947                                 fputs(file->print_name, stdout);
 948                                 putchar(options->line_termination);
 949                         }
 950                 } else {
 951                         if (file->is_renamed) {
 952                                 putchar('\0');
 953                                 write_name_quoted(file->from_name, stdout, '\0');
 954                         }
 955                         write_name_quoted(file->name, stdout, '\0');
 956                 }
 957         }
 958 }
 959
 960 static void free_diffstat_info(struct diffstat_t *diffstat)
 961 {
 962         int i;
 963         for (i = 0; i < diffstat->nr; i++) {
 964                 struct diffstat_file *f = diffstat->files[i];
 965                 if (f->name != f->print_name)
 966                         free(f->print_name);
 967                 free(f->name);
 968                 free(f->from_name);
 969                 free(f);
 970         }
 971         free(diffstat->files);
 972 }
 973
 974 struct checkdiff_t {
 975         struct xdiff_emit_state xm;
 976         const char *filename;
 977         int lineno, color_diff;
 978         unsigned ws_rule;
 979         unsigned status;
 980 };
 981
 982 static void checkdiff_consume(void *priv, char *line, unsigned long len)
 983 {
 984         struct checkdiff_t *data = priv;
 985         const char *ws = diff_get_color(data->color_diff, DIFF_WHITESPACE);
 986         const char *reset = diff_get_color(data->color_diff, DIFF_RESET);
 987         const char *set = diff_get_color(data->color_diff, DIFF_FILE_NEW);
 988         char *err;
 989
 990         if (line[0] == '+') {
 991                 data->status = check_and_emit_line(line + 1, len - 1,
 992                     data->ws_rule, NULL, NULL, NULL, NULL);
 993                 if (!data->status)
 994                         return;
 995                 err = whitespace_error_string(data->status);
 996                 printf("%s:%d: %s.\n", data->filename, data->lineno, err);
 997                 free(err);
 998                 emit_line(set, reset, line, 1);
 999                 (void)check_and_emit_line(line + 1, len - 1, data->ws_rule,
1000                     stdout, set, reset, ws);
1001                 data->lineno++;
1002         } else if (line[0] == ' ')
1003                 data->lineno++;
1004         else if (line[0] == '@') {
1005                 char *plus = strchr(line, '+');
1006                 if (plus)
1007                         data->lineno = strtol(plus, NULL, 10);
1008                 else
1009                         die("invalid diff");
1010         }
1011 }
1012
1013 static unsigned char *deflate_it(char *data,
1014                                  unsigned long size,
1015                                  unsigned long *result_size)
1016 {
1017         int bound;
1018         unsigned char *deflated;
1019         z_stream stream;
1020
1021         memset(&stream, 0, sizeof(stream));
1022         deflateInit(&stream, zlib_compression_level);
1023         bound = deflateBound(&stream, size);
1024         deflated = xmalloc(bound);
1025         stream.next_out = deflated;
1026         stream.avail_out = bound;
1027
1028         stream.next_in = (unsigned char *)data;
1029         stream.avail_in = size;
1030         while (deflate(&stream, Z_FINISH) == Z_OK)
1031                 ; /* nothing */
1032         deflateEnd(&stream);
1033         *result_size = stream.total_out;
1034         return deflated;
1035 }
1036
1037 static void emit_binary_diff_body(mmfile_t *one, mmfile_t *two)
1038 {
1039         void *cp;
1040         void *delta;
1041         void *deflated;
1042         void *data;
1043         unsigned long orig_size;
1044         unsigned long delta_size;
1045         unsigned long deflate_size;
1046         unsigned long data_size;
1047
1048         /* We could do deflated delta, or we could do just deflated two,
1049          * whichever is smaller.
1050          */
1051         delta = NULL;
1052         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1053         if (one->size && two->size) {
1054                 delta = diff_delta(one->ptr, one->size,
1055                                    two->ptr, two->size,
1056                                    &delta_size, deflate_size);
1057                 if (delta) {
1058                         void *to_free = delta;
1059                         orig_size = delta_size;
1060                         delta = deflate_it(delta, delta_size, &delta_size);
1061                         free(to_free);
1062                 }
1063         }
1064
1065         if (delta && delta_size < deflate_size) {
1066                 printf("delta %lu\n", orig_size);
1067                 free(deflated);
1068                 data = delta;
1069                 data_size = delta_size;
1070         }
1071         else {
1072                 printf("literal %lu\n", two->size);
1073                 free(delta);
1074                 data = deflated;
1075                 data_size = deflate_size;
1076         }
1077
1078         /* emit data encoded in base85 */
1079         cp = data;
1080         while (data_size) {
1081                 int bytes = (52 < data_size) ? 52 : data_size;
1082                 char line[70];
1083                 data_size -= bytes;
1084                 if (bytes <= 26)
1085                         line[0] = bytes + 'A' - 1;
1086                 else
1087                         line[0] = bytes - 26 + 'a' - 1;
1088                 encode_85(line + 1, cp, bytes);
1089                 cp = (char *) cp + bytes;
1090                 puts(line);
1091         }
1092         printf("\n");
1093         free(data);
1094 }
1095
1096 static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
1097 {
1098         printf("GIT binary patch\n");
1099         emit_binary_diff_body(one, two);
1100         emit_binary_diff_body(two, one);
1101 }
1102
1103 static void setup_diff_attr_check(struct git_attr_check *check)
1104 {
1105         static struct git_attr *attr_diff;
1106
1107         if (!attr_diff) {
1108                 attr_diff = git_attr("diff", 4);
1109         }
1110         check[0].attr = attr_diff;
1111 }
1112
1113 static void diff_filespec_check_attr(struct diff_filespec *one)
1114 {
1115         struct git_attr_check attr_diff_check;
1116         int check_from_data = 0;
1117
1118         if (one->checked_attr)
1119                 return;
1120
1121         setup_diff_attr_check(&attr_diff_check);
1122         one->is_binary = 0;
1123         one->funcname_pattern_ident = NULL;
1124
1125         if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1126                 const char *value;
1127
1128                 /* binaryness */
1129                 value = attr_diff_check.value;
1130                 if (ATTR_TRUE(value))
1131                         ;
1132                 else if (ATTR_FALSE(value))
1133                         one->is_binary = 1;
1134                 else
1135                         check_from_data = 1;
1136
1137                 /* funcname pattern ident */
1138                 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1139                         ;
1140                 else
1141                         one->funcname_pattern_ident = value;
1142         }
1143
1144         if (check_from_data) {
1145                 if (!one->data && DIFF_FILE_VALID(one))
1146                         diff_populate_filespec(one, 0);
1147
1148                 if (one->data)
1149                         one->is_binary = buffer_is_binary(one->data, one->size);
1150         }
1151 }
1152
1153 int diff_filespec_is_binary(struct diff_filespec *one)
1154 {
1155         diff_filespec_check_attr(one);
1156         return one->is_binary;
1157 }
1158
1159 static const char *funcname_pattern(const char *ident)
1160 {
1161         struct funcname_pattern *pp;
1162
1163         read_config_if_needed();
1164         for (pp = funcname_pattern_list; pp; pp = pp->next)
1165                 if (!strcmp(ident, pp->name))
1166                         return pp->pattern;
1167         return NULL;
1168 }
1169
1170 static struct builtin_funcname_pattern {
1171         const char *name;
1172         const char *pattern;
1173 } builtin_funcname_pattern[] = {
1174         { "java", "!^[  ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1175                         "new\\|return\\|switch\\|throw\\|while\\)\n"
1176                         "^[     ]*\\(\\([       ]*"
1177                         "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1178                         "[      ]*([^;]*$\\)" },
1179         { "tex", "^\\(\\\\\\(sub\\)*section{.*\\)$" },
1180 };
1181
1182 static const char *diff_funcname_pattern(struct diff_filespec *one)
1183 {
1184         const char *ident, *pattern;
1185         int i;
1186
1187         diff_filespec_check_attr(one);
1188         ident = one->funcname_pattern_ident;
1189
1190         if (!ident)
1191                 /*
1192                  * If the config file has "funcname.default" defined, that
1193                  * regexp is used; otherwise NULL is returned and xemit uses
1194                  * the built-in default.
1195                  */
1196                 return funcname_pattern("default");
1197
1198         /* Look up custom "funcname.$ident" regexp from config. */
1199         pattern = funcname_pattern(ident);
1200         if (pattern)
1201                 return pattern;
1202
1203         /*
1204          * And define built-in fallback patterns here.  Note that
1205          * these can be overriden by the user's config settings.
1206          */
1207         for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1208                 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1209                         return builtin_funcname_pattern[i].pattern;
1210
1211         return NULL;
1212 }
1213
1214 static void builtin_diff(const char *name_a,
1215                          const char *name_b,
1216                          struct diff_filespec *one,
1217                          struct diff_filespec *two,
1218                          const char *xfrm_msg,
1219                          struct diff_options *o,
1220                          int complete_rewrite)
1221 {
1222         mmfile_t mf1, mf2;
1223         const char *lbl[2];
1224         char *a_one, *b_two;
1225         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1226         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1227
1228         a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1229         b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1230         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1231         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1232         printf("%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1233         if (lbl[0][0] == '/') {
1234                 /* /dev/null */
1235                 printf("%snew file mode %06o%s\n", set, two->mode, reset);
1236                 if (xfrm_msg && xfrm_msg[0])
1237                         printf("%s%s%s\n", set, xfrm_msg, reset);
1238         }
1239         else if (lbl[1][0] == '/') {
1240                 printf("%sdeleted file mode %06o%s\n", set, one->mode, reset);
1241                 if (xfrm_msg && xfrm_msg[0])
1242                         printf("%s%s%s\n", set, xfrm_msg, reset);
1243         }
1244         else {
1245                 if (one->mode != two->mode) {
1246                         printf("%sold mode %06o%s\n", set, one->mode, reset);
1247                         printf("%snew mode %06o%s\n", set, two->mode, reset);
1248                 }
1249                 if (xfrm_msg && xfrm_msg[0])
1250                         printf("%s%s%s\n", set, xfrm_msg, reset);
1251                 /*
1252                  * we do not run diff between different kind
1253                  * of objects.
1254                  */
1255                 if ((one->mode ^ two->mode) & S_IFMT)
1256                         goto free_ab_and_return;
1257                 if (complete_rewrite) {
1258                         emit_rewrite_diff(name_a, name_b, one, two, o);
1259                         o->found_changes = 1;
1260                         goto free_ab_and_return;
1261                 }
1262         }
1263
1264         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1265                 die("unable to read files to diff");
1266
1267         if (!DIFF_OPT_TST(o, TEXT) &&
1268             (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1269                 /* Quite common confusing case */
1270                 if (mf1.size == mf2.size &&
1271                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1272                         goto free_ab_and_return;
1273                 if (DIFF_OPT_TST(o, BINARY))
1274                         emit_binary_diff(&mf1, &mf2);
1275                 else
1276                         printf("Binary files %s and %s differ\n",
1277                                lbl[0], lbl[1]);
1278                 o->found_changes = 1;
1279         }
1280         else {
1281                 /* Crazy xdl interfaces.. */
1282                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1283                 xpparam_t xpp;
1284                 xdemitconf_t xecfg;
1285                 xdemitcb_t ecb;
1286                 struct emit_callback ecbdata;
1287                 const char *funcname_pattern;
1288
1289                 funcname_pattern = diff_funcname_pattern(one);
1290                 if (!funcname_pattern)
1291                         funcname_pattern = diff_funcname_pattern(two);
1292
1293                 memset(&xecfg, 0, sizeof(xecfg));
1294                 memset(&ecbdata, 0, sizeof(ecbdata));
1295                 ecbdata.label_path = lbl;
1296                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1297                 ecbdata.found_changesp = &o->found_changes;
1298                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1299                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1300                 xecfg.ctxlen = o->context;
1301                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1302                 if (funcname_pattern)
1303                         xdiff_set_find_func(&xecfg, funcname_pattern);
1304                 if (!diffopts)
1305                         ;
1306                 else if (!prefixcmp(diffopts, "--unified="))
1307                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1308                 else if (!prefixcmp(diffopts, "-u"))
1309                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1310                 ecb.outf = xdiff_outf;
1311                 ecb.priv = &ecbdata;
1312                 ecbdata.xm.consume = fn_out_consume;
1313                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1314                         ecbdata.diff_words =
1315                                 xcalloc(1, sizeof(struct diff_words_data));
1316                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1317                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1318                         free_diff_words_data(&ecbdata);
1319         }
1320
1321  free_ab_and_return:
1322         diff_free_filespec_data(one);
1323         diff_free_filespec_data(two);
1324         free(a_one);
1325         free(b_two);
1326         return;
1327 }
1328
1329 static void builtin_diffstat(const char *name_a, const char *name_b,
1330                              struct diff_filespec *one,
1331                              struct diff_filespec *two,
1332                              struct diffstat_t *diffstat,
1333                              struct diff_options *o,
1334                              int complete_rewrite)
1335 {
1336         mmfile_t mf1, mf2;
1337         struct diffstat_file *data;
1338
1339         data = diffstat_add(diffstat, name_a, name_b);
1340
1341         if (!one || !two) {
1342                 data->is_unmerged = 1;
1343                 return;
1344         }
1345         if (complete_rewrite) {
1346                 diff_populate_filespec(one, 0);
1347                 diff_populate_filespec(two, 0);
1348                 data->deleted = count_lines(one->data, one->size);
1349                 data->added = count_lines(two->data, two->size);
1350                 goto free_and_return;
1351         }
1352         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1353                 die("unable to read files to diff");
1354
1355         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1356                 data->is_binary = 1;
1357                 data->added = mf2.size;
1358                 data->deleted = mf1.size;
1359         } else {
1360                 /* Crazy xdl interfaces.. */
1361                 xpparam_t xpp;
1362                 xdemitconf_t xecfg;
1363                 xdemitcb_t ecb;
1364
1365                 memset(&xecfg, 0, sizeof(xecfg));
1366                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1367                 ecb.outf = xdiff_outf;
1368                 ecb.priv = diffstat;
1369                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1370         }
1371
1372  free_and_return:
1373         diff_free_filespec_data(one);
1374         diff_free_filespec_data(two);
1375 }
1376
1377 static void builtin_checkdiff(const char *name_a, const char *name_b,
1378                              struct diff_filespec *one,
1379                              struct diff_filespec *two, struct diff_options *o)
1380 {
1381         mmfile_t mf1, mf2;
1382         struct checkdiff_t data;
1383
1384         if (!two)
1385                 return;
1386
1387         memset(&data, 0, sizeof(data));
1388         data.xm.consume = checkdiff_consume;
1389         data.filename = name_b ? name_b : name_a;
1390         data.lineno = 0;
1391         data.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1392         data.ws_rule = whitespace_rule(data.filename);
1393
1394         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1395                 die("unable to read files to diff");
1396
1397         if (diff_filespec_is_binary(two))
1398                 goto free_and_return;
1399         else {
1400                 /* Crazy xdl interfaces.. */
1401                 xpparam_t xpp;
1402                 xdemitconf_t xecfg;
1403                 xdemitcb_t ecb;
1404
1405                 memset(&xecfg, 0, sizeof(xecfg));
1406                 xpp.flags = XDF_NEED_MINIMAL;
1407                 ecb.outf = xdiff_outf;
1408                 ecb.priv = &data;
1409                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1410         }
1411  free_and_return:
1412         diff_free_filespec_data(one);
1413         diff_free_filespec_data(two);
1414         if (data.status)
1415                 DIFF_OPT_SET(o, CHECK_FAILED);
1416 }
1417
1418 struct diff_filespec *alloc_filespec(const char *path)
1419 {
1420         int namelen = strlen(path);
1421         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1422
1423         memset(spec, 0, sizeof(*spec));
1424         spec->path = (char *)(spec + 1);
1425         memcpy(spec->path, path, namelen+1);
1426         spec->count = 1;
1427         return spec;
1428 }
1429
1430 void free_filespec(struct diff_filespec *spec)
1431 {
1432         if (!--spec->count) {
1433                 diff_free_filespec_data(spec);
1434                 free(spec);
1435         }
1436 }
1437
1438 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1439                    unsigned short mode)
1440 {
1441         if (mode) {
1442                 spec->mode = canon_mode(mode);
1443                 hashcpy(spec->sha1, sha1);
1444                 spec->sha1_valid = !is_null_sha1(sha1);
1445         }
1446 }
1447
1448 /*
1449  * Given a name and sha1 pair, if the index tells us the file in
1450  * the work tree has that object contents, return true, so that
1451  * prepare_temp_file() does not have to inflate and extract.
1452  */
1453 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1454 {
1455         struct cache_entry *ce;
1456         struct stat st;
1457         int pos, len;
1458
1459         /* We do not read the cache ourselves here, because the
1460          * benchmark with my previous version that always reads cache
1461          * shows that it makes things worse for diff-tree comparing
1462          * two linux-2.6 kernel trees in an already checked out work
1463          * tree.  This is because most diff-tree comparisons deal with
1464          * only a small number of files, while reading the cache is
1465          * expensive for a large project, and its cost outweighs the
1466          * savings we get by not inflating the object to a temporary
1467          * file.  Practically, this code only helps when we are used
1468          * by diff-cache --cached, which does read the cache before
1469          * calling us.
1470          */
1471         if (!active_cache)
1472                 return 0;
1473
1474         /* We want to avoid the working directory if our caller
1475          * doesn't need the data in a normal file, this system
1476          * is rather slow with its stat/open/mmap/close syscalls,
1477          * and the object is contained in a pack file.  The pack
1478          * is probably already open and will be faster to obtain
1479          * the data through than the working directory.  Loose
1480          * objects however would tend to be slower as they need
1481          * to be individually opened and inflated.
1482          */
1483         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1484                 return 0;
1485
1486         len = strlen(name);
1487         pos = cache_name_pos(name, len);
1488         if (pos < 0)
1489                 return 0;
1490         ce = active_cache[pos];
1491         if ((lstat(name, &st) < 0) ||
1492             !S_ISREG(st.st_mode) || /* careful! */
1493             ce_match_stat(ce, &st, 0) ||
1494             hashcmp(sha1, ce->sha1))
1495                 return 0;
1496         /* we return 1 only when we can stat, it is a regular file,
1497          * stat information matches, and sha1 recorded in the cache
1498          * matches.  I.e. we know the file in the work tree really is
1499          * the same as the <name, sha1> pair.
1500          */
1501         return 1;
1502 }
1503
1504 static int populate_from_stdin(struct diff_filespec *s)
1505 {
1506         struct strbuf buf;
1507         size_t size = 0;
1508
1509         strbuf_init(&buf, 0);
1510         if (strbuf_read(&buf, 0, 0) < 0)
1511                 return error("error while reading from stdin %s",
1512                                      strerror(errno));
1513
1514         s->should_munmap = 0;
1515         s->data = strbuf_detach(&buf, &size);
1516         s->size = size;
1517         s->should_free = 1;
1518         return 0;
1519 }
1520
1521 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1522 {
1523         int len;
1524         char *data = xmalloc(100);
1525         len = snprintf(data, 100,
1526                 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1527         s->data = data;
1528         s->size = len;
1529         s->should_free = 1;
1530         if (size_only) {
1531                 s->data = NULL;
1532                 free(data);
1533         }
1534         return 0;
1535 }
1536
1537 /*
1538  * While doing rename detection and pickaxe operation, we may need to
1539  * grab the data for the blob (or file) for our own in-core comparison.
1540  * diff_filespec has data and size fields for this purpose.
1541  */
1542 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1543 {
1544         int err = 0;
1545         if (!DIFF_FILE_VALID(s))
1546                 die("internal error: asking to populate invalid file.");
1547         if (S_ISDIR(s->mode))
1548                 return -1;
1549
1550         if (s->data)
1551                 return 0;
1552
1553         if (size_only && 0 < s->size)
1554                 return 0;
1555
1556         if (S_ISGITLINK(s->mode))
1557                 return diff_populate_gitlink(s, size_only);
1558
1559         if (!s->sha1_valid ||
1560             reuse_worktree_file(s->path, s->sha1, 0)) {
1561                 struct strbuf buf;
1562                 struct stat st;
1563                 int fd;
1564
1565                 if (!strcmp(s->path, "-"))
1566                         return populate_from_stdin(s);
1567
1568                 if (lstat(s->path, &st) < 0) {
1569                         if (errno == ENOENT) {
1570                         err_empty:
1571                                 err = -1;
1572                         empty:
1573                                 s->data = (char *)"";
1574                                 s->size = 0;
1575                                 return err;
1576                         }
1577                 }
1578                 s->size = xsize_t(st.st_size);
1579                 if (!s->size)
1580                         goto empty;
1581                 if (size_only)
1582                         return 0;
1583                 if (S_ISLNK(st.st_mode)) {
1584                         int ret;
1585                         s->data = xmalloc(s->size);
1586                         s->should_free = 1;
1587                         ret = readlink(s->path, s->data, s->size);
1588                         if (ret < 0) {
1589                                 free(s->data);
1590                                 goto err_empty;
1591                         }
1592                         return 0;
1593                 }
1594                 fd = open(s->path, O_RDONLY);
1595                 if (fd < 0)
1596                         goto err_empty;
1597                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1598                 close(fd);
1599                 s->should_munmap = 1;
1600
1601                 /*
1602                  * Convert from working tree format to canonical git format
1603                  */
1604                 strbuf_init(&buf, 0);
1605                 if (convert_to_git(s->path, s->data, s->size, &buf)) {
1606                         size_t size = 0;
1607                         munmap(s->data, s->size);
1608                         s->should_munmap = 0;
1609                         s->data = strbuf_detach(&buf, &size);
1610                         s->size = size;
1611                         s->should_free = 1;
1612                 }
1613         }
1614         else {
1615                 enum object_type type;
1616                 if (size_only)
1617                         type = sha1_object_info(s->sha1, &s->size);
1618                 else {
1619                         s->data = read_sha1_file(s->sha1, &type, &s->size);
1620                         s->should_free = 1;
1621                 }
1622         }
1623         return 0;
1624 }
1625
1626 void diff_free_filespec_blob(struct diff_filespec *s)
1627 {
1628         if (s->should_free)
1629                 free(s->data);
1630         else if (s->should_munmap)
1631                 munmap(s->data, s->size);
1632
1633         if (s->should_free || s->should_munmap) {
1634                 s->should_free = s->should_munmap = 0;
1635                 s->data = NULL;
1636         }
1637 }
1638
1639 void diff_free_filespec_data(struct diff_filespec *s)
1640 {
1641         diff_free_filespec_blob(s);
1642         free(s->cnt_data);
1643         s->cnt_data = NULL;
1644 }
1645
1646 static void prep_temp_blob(struct diff_tempfile *temp,
1647                            void *blob,
1648                            unsigned long size,
1649                            const unsigned char *sha1,
1650                            int mode)
1651 {
1652         int fd;
1653
1654         fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1655         if (fd < 0)
1656                 die("unable to create temp-file: %s", strerror(errno));
1657         if (write_in_full(fd, blob, size) != size)
1658                 die("unable to write temp-file");
1659         close(fd);
1660         temp->name = temp->tmp_path;
1661         strcpy(temp->hex, sha1_to_hex(sha1));
1662         temp->hex[40] = 0;
1663         sprintf(temp->mode, "%06o", mode);
1664 }
1665
1666 static void prepare_temp_file(const char *name,
1667                               struct diff_tempfile *temp,
1668                               struct diff_filespec *one)
1669 {
1670         if (!DIFF_FILE_VALID(one)) {
1671         not_a_valid_file:
1672                 /* A '-' entry produces this for file-2, and
1673                  * a '+' entry produces this for file-1.
1674                  */
1675                 temp->name = "/dev/null";
1676                 strcpy(temp->hex, ".");
1677                 strcpy(temp->mode, ".");
1678                 return;
1679         }
1680
1681         if (!one->sha1_valid ||
1682             reuse_worktree_file(name, one->sha1, 1)) {
1683                 struct stat st;
1684                 if (lstat(name, &st) < 0) {
1685                         if (errno == ENOENT)
1686                                 goto not_a_valid_file;
1687                         die("stat(%s): %s", name, strerror(errno));
1688                 }
1689                 if (S_ISLNK(st.st_mode)) {
1690                         int ret;
1691                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1692                         size_t sz = xsize_t(st.st_size);
1693                         if (sizeof(buf) <= st.st_size)
1694                                 die("symlink too long: %s", name);
1695                         ret = readlink(name, buf, sz);
1696                         if (ret < 0)
1697                                 die("readlink(%s)", name);
1698                         prep_temp_blob(temp, buf, sz,
1699                                        (one->sha1_valid ?
1700                                         one->sha1 : null_sha1),
1701                                        (one->sha1_valid ?
1702                                         one->mode : S_IFLNK));
1703                 }
1704                 else {
1705                         /* we can borrow from the file in the work tree */
1706                         temp->name = name;
1707                         if (!one->sha1_valid)
1708                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
1709                         else
1710                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
1711                         /* Even though we may sometimes borrow the
1712                          * contents from the work tree, we always want
1713                          * one->mode.  mode is trustworthy even when
1714                          * !(one->sha1_valid), as long as
1715                          * DIFF_FILE_VALID(one).
1716                          */
1717                         sprintf(temp->mode, "%06o", one->mode);
1718                 }
1719                 return;
1720         }
1721         else {
1722                 if (diff_populate_filespec(one, 0))
1723                         die("cannot read data blob for %s", one->path);
1724                 prep_temp_blob(temp, one->data, one->size,
1725                                one->sha1, one->mode);
1726         }
1727 }
1728
1729 static void remove_tempfile(void)
1730 {
1731         int i;
1732
1733         for (i = 0; i < 2; i++)
1734                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1735                         unlink(diff_temp[i].name);
1736                         diff_temp[i].name = NULL;
1737                 }
1738 }
1739
1740 static void remove_tempfile_on_signal(int signo)
1741 {
1742         remove_tempfile();
1743         signal(SIGINT, SIG_DFL);
1744         raise(signo);
1745 }
1746
1747 /* An external diff command takes:
1748  *
1749  * diff-cmd name infile1 infile1-sha1 infile1-mode \
1750  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
1751  *
1752  */
1753 static void run_external_diff(const char *pgm,
1754                               const char *name,
1755                               const char *other,
1756                               struct diff_filespec *one,
1757                               struct diff_filespec *two,
1758                               const char *xfrm_msg,
1759                               int complete_rewrite)
1760 {
1761         const char *spawn_arg[10];
1762         struct diff_tempfile *temp = diff_temp;
1763         int retval;
1764         static int atexit_asked = 0;
1765         const char *othername;
1766         const char **arg = &spawn_arg[0];
1767
1768         othername = (other? other : name);
1769         if (one && two) {
1770                 prepare_temp_file(name, &temp[0], one);
1771                 prepare_temp_file(othername, &temp[1], two);
1772                 if (! atexit_asked &&
1773                     (temp[0].name == temp[0].tmp_path ||
1774                      temp[1].name == temp[1].tmp_path)) {
1775                         atexit_asked = 1;
1776                         atexit(remove_tempfile);
1777                 }
1778                 signal(SIGINT, remove_tempfile_on_signal);
1779         }
1780
1781         if (one && two) {
1782                 *arg++ = pgm;
1783                 *arg++ = name;
1784                 *arg++ = temp[0].name;
1785                 *arg++ = temp[0].hex;
1786                 *arg++ = temp[0].mode;
1787                 *arg++ = temp[1].name;
1788                 *arg++ = temp[1].hex;
1789                 *arg++ = temp[1].mode;
1790                 if (other) {
1791                         *arg++ = other;
1792                         *arg++ = xfrm_msg;
1793                 }
1794         } else {
1795                 *arg++ = pgm;
1796                 *arg++ = name;
1797         }
1798         *arg = NULL;
1799         fflush(NULL);
1800         retval = run_command_v_opt(spawn_arg, 0);
1801         remove_tempfile();
1802         if (retval) {
1803                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1804                 exit(1);
1805         }
1806 }
1807
1808 static const char *external_diff_attr(const char *name)
1809 {
1810         struct git_attr_check attr_diff_check;
1811
1812         setup_diff_attr_check(&attr_diff_check);
1813         if (!git_checkattr(name, 1, &attr_diff_check)) {
1814                 const char *value = attr_diff_check.value;
1815                 if (!ATTR_TRUE(value) &&
1816                     !ATTR_FALSE(value) &&
1817                     !ATTR_UNSET(value)) {
1818                         struct ll_diff_driver *drv;
1819
1820                         read_config_if_needed();
1821                         for (drv = user_diff; drv; drv = drv->next)
1822                                 if (!strcmp(drv->name, value))
1823                                         return drv->cmd;
1824                 }
1825         }
1826         return NULL;
1827 }
1828
1829 static void run_diff_cmd(const char *pgm,
1830                          const char *name,
1831                          const char *other,
1832                          struct diff_filespec *one,
1833                          struct diff_filespec *two,
1834                          const char *xfrm_msg,
1835                          struct diff_options *o,
1836                          int complete_rewrite)
1837 {
1838         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
1839                 pgm = NULL;
1840         else {
1841                 const char *cmd = external_diff_attr(name);
1842                 if (cmd)
1843                         pgm = cmd;
1844         }
1845
1846         if (pgm) {
1847                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
1848                                   complete_rewrite);
1849                 return;
1850         }
1851         if (one && two)
1852                 builtin_diff(name, other ? other : name,
1853                              one, two, xfrm_msg, o, complete_rewrite);
1854         else
1855                 printf("* Unmerged path %s\n", name);
1856 }
1857
1858 static void diff_fill_sha1_info(struct diff_filespec *one)
1859 {
1860         if (DIFF_FILE_VALID(one)) {
1861                 if (!one->sha1_valid) {
1862                         struct stat st;
1863                         if (!strcmp(one->path, "-")) {
1864                                 hashcpy(one->sha1, null_sha1);
1865                                 return;
1866                         }
1867                         if (lstat(one->path, &st) < 0)
1868                                 die("stat %s", one->path);
1869                         if (index_path(one->sha1, one->path, &st, 0))
1870                                 die("cannot hash %s\n", one->path);
1871                 }
1872         }
1873         else
1874                 hashclr(one->sha1);
1875 }
1876
1877 static int similarity_index(struct diff_filepair *p)
1878 {
1879         return p->score * 100 / MAX_SCORE;
1880 }
1881
1882 static void run_diff(struct diff_filepair *p, struct diff_options *o)
1883 {
1884         const char *pgm = external_diff();
1885         struct strbuf msg;
1886         char *xfrm_msg;
1887         struct diff_filespec *one = p->one;
1888         struct diff_filespec *two = p->two;
1889         const char *name;
1890         const char *other;
1891         int complete_rewrite = 0;
1892
1893
1894         if (DIFF_PAIR_UNMERGED(p)) {
1895                 run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, o, 0);
1896                 return;
1897         }
1898
1899         name  = p->one->path;
1900         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1901         diff_fill_sha1_info(one);
1902         diff_fill_sha1_info(two);
1903
1904         strbuf_init(&msg, PATH_MAX * 2 + 300);
1905         switch (p->status) {
1906         case DIFF_STATUS_COPIED:
1907                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
1908                 strbuf_addstr(&msg, "\ncopy from ");
1909                 quote_c_style(name, &msg, NULL, 0);
1910                 strbuf_addstr(&msg, "\ncopy to ");
1911                 quote_c_style(other, &msg, NULL, 0);
1912                 strbuf_addch(&msg, '\n');
1913                 break;
1914         case DIFF_STATUS_RENAMED:
1915                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
1916                 strbuf_addstr(&msg, "\nrename from ");
1917                 quote_c_style(name, &msg, NULL, 0);
1918                 strbuf_addstr(&msg, "\nrename to ");
1919                 quote_c_style(other, &msg, NULL, 0);
1920                 strbuf_addch(&msg, '\n');
1921                 break;
1922         case DIFF_STATUS_MODIFIED:
1923                 if (p->score) {
1924                         strbuf_addf(&msg, "dissimilarity index %d%%\n",
1925                                         similarity_index(p));
1926                         complete_rewrite = 1;
1927                         break;
1928                 }
1929                 /* fallthru */
1930         default:
1931                 /* nothing */
1932                 ;
1933         }
1934
1935         if (hashcmp(one->sha1, two->sha1)) {
1936                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
1937
1938                 if (DIFF_OPT_TST(o, BINARY)) {
1939                         mmfile_t mf;
1940                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
1941                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
1942                                 abbrev = 40;
1943                 }
1944                 strbuf_addf(&msg, "index %.*s..%.*s",
1945                                 abbrev, sha1_to_hex(one->sha1),
1946                                 abbrev, sha1_to_hex(two->sha1));
1947                 if (one->mode == two->mode)
1948                         strbuf_addf(&msg, " %06o", one->mode);
1949                 strbuf_addch(&msg, '\n');
1950         }
1951
1952         if (msg.len)
1953                 strbuf_setlen(&msg, msg.len - 1);
1954         xfrm_msg = msg.len ? msg.buf : NULL;
1955
1956         if (!pgm &&
1957             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
1958             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
1959                 /* a filepair that changes between file and symlink
1960                  * needs to be split into deletion and creation.
1961                  */
1962                 struct diff_filespec *null = alloc_filespec(two->path);
1963                 run_diff_cmd(NULL, name, other, one, null, xfrm_msg, o, 0);
1964                 free(null);
1965                 null = alloc_filespec(one->path);
1966                 run_diff_cmd(NULL, name, other, null, two, xfrm_msg, o, 0);
1967                 free(null);
1968         }
1969         else
1970                 run_diff_cmd(pgm, name, other, one, two, xfrm_msg, o,
1971                              complete_rewrite);
1972
1973         strbuf_release(&msg);
1974 }
1975
1976 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
1977                          struct diffstat_t *diffstat)
1978 {
1979         const char *name;
1980         const char *other;
1981         int complete_rewrite = 0;
1982
1983         if (DIFF_PAIR_UNMERGED(p)) {
1984                 /* unmerged */
1985                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
1986                 return;
1987         }
1988
1989         name = p->one->path;
1990         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1991
1992         diff_fill_sha1_info(p->one);
1993         diff_fill_sha1_info(p->two);
1994
1995         if (p->status == DIFF_STATUS_MODIFIED && p->score)
1996                 complete_rewrite = 1;
1997         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
1998 }
1999
2000 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2001 {
2002         const char *name;
2003         const char *other;
2004
2005         if (DIFF_PAIR_UNMERGED(p)) {
2006                 /* unmerged */
2007                 return;
2008         }
2009
2010         name = p->one->path;
2011         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2012
2013         diff_fill_sha1_info(p->one);
2014         diff_fill_sha1_info(p->two);
2015
2016         builtin_checkdiff(name, other, p->one, p->two, o);
2017 }
2018
2019 void diff_setup(struct diff_options *options)
2020 {
2021         memset(options, 0, sizeof(*options));
2022         options->line_termination = '\n';
2023         options->break_opt = -1;
2024         options->rename_limit = -1;
2025         options->context = 3;
2026         options->msg_sep = "";
2027
2028         options->change = diff_change;
2029         options->add_remove = diff_addremove;
2030         if (diff_use_color_default)
2031                 DIFF_OPT_SET(options, COLOR_DIFF);
2032         else
2033                 DIFF_OPT_CLR(options, COLOR_DIFF);
2034         options->detect_rename = diff_detect_rename_default;
2035
2036         options->a_prefix = "a/";
2037         options->b_prefix = "b/";
2038 }
2039
2040 int diff_setup_done(struct diff_options *options)
2041 {
2042         int count = 0;
2043
2044         if (options->output_format & DIFF_FORMAT_NAME)
2045                 count++;
2046         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2047                 count++;
2048         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2049                 count++;
2050         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2051                 count++;
2052         if (count > 1)
2053                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2054
2055         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2056                 options->detect_rename = DIFF_DETECT_COPY;
2057
2058         if (options->output_format & (DIFF_FORMAT_NAME |
2059                                       DIFF_FORMAT_NAME_STATUS |
2060                                       DIFF_FORMAT_CHECKDIFF |
2061                                       DIFF_FORMAT_NO_OUTPUT))
2062                 options->output_format &= ~(DIFF_FORMAT_RAW |
2063                                             DIFF_FORMAT_NUMSTAT |
2064                                             DIFF_FORMAT_DIFFSTAT |
2065                                             DIFF_FORMAT_SHORTSTAT |
2066                                             DIFF_FORMAT_SUMMARY |
2067                                             DIFF_FORMAT_PATCH);
2068
2069         /*
2070          * These cases always need recursive; we do not drop caller-supplied
2071          * recursive bits for other formats here.
2072          */
2073         if (options->output_format & (DIFF_FORMAT_PATCH |
2074                                       DIFF_FORMAT_NUMSTAT |
2075                                       DIFF_FORMAT_DIFFSTAT |
2076                                       DIFF_FORMAT_SHORTSTAT |
2077                                       DIFF_FORMAT_SUMMARY |
2078                                       DIFF_FORMAT_CHECKDIFF))
2079                 DIFF_OPT_SET(options, RECURSIVE);
2080         /*
2081          * Also pickaxe would not work very well if you do not say recursive
2082          */
2083         if (options->pickaxe)
2084                 DIFF_OPT_SET(options, RECURSIVE);
2085
2086         if (options->detect_rename && options->rename_limit < 0)
2087                 options->rename_limit = diff_rename_limit_default;
2088         if (options->setup & DIFF_SETUP_USE_CACHE) {
2089                 if (!active_cache)
2090                         /* read-cache does not die even when it fails
2091                          * so it is safe for us to do this here.  Also
2092                          * it does not smudge active_cache or active_nr
2093                          * when it fails, so we do not have to worry about
2094                          * cleaning it up ourselves either.
2095                          */
2096                         read_cache();
2097         }
2098         if (options->abbrev <= 0 || 40 < options->abbrev)
2099                 options->abbrev = 40; /* full */
2100
2101         /*
2102          * It does not make sense to show the first hit we happened
2103          * to have found.  It does not make sense not to return with
2104          * exit code in such a case either.
2105          */
2106         if (DIFF_OPT_TST(options, QUIET)) {
2107                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2108                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2109         }
2110
2111         /*
2112          * If we postprocess in diffcore, we cannot simply return
2113          * upon the first hit.  We need to run diff as usual.
2114          */
2115         if (options->pickaxe || options->filter)
2116                 DIFF_OPT_CLR(options, QUIET);
2117
2118         return 0;
2119 }
2120
2121 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2122 {
2123         char c, *eq;
2124         int len;
2125
2126         if (*arg != '-')
2127                 return 0;
2128         c = *++arg;
2129         if (!c)
2130                 return 0;
2131         if (c == arg_short) {
2132                 c = *++arg;
2133                 if (!c)
2134                         return 1;
2135                 if (val && isdigit(c)) {
2136                         char *end;
2137                         int n = strtoul(arg, &end, 10);
2138                         if (*end)
2139                                 return 0;
2140                         *val = n;
2141                         return 1;
2142                 }
2143                 return 0;
2144         }
2145         if (c != '-')
2146                 return 0;
2147         arg++;
2148         eq = strchr(arg, '=');
2149         if (eq)
2150                 len = eq - arg;
2151         else
2152                 len = strlen(arg);
2153         if (!len || strncmp(arg, arg_long, len))
2154                 return 0;
2155         if (eq) {
2156                 int n;
2157                 char *end;
2158                 if (!isdigit(*++eq))
2159                         return 0;
2160                 n = strtoul(eq, &end, 10);
2161                 if (*end)
2162                         return 0;
2163                 *val = n;
2164         }
2165         return 1;
2166 }
2167
2168 static int diff_scoreopt_parse(const char *opt);
2169
2170 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2171 {
2172         const char *arg = av[0];
2173
2174         /* Output format options */
2175         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2176                 options->output_format |= DIFF_FORMAT_PATCH;
2177         else if (opt_arg(arg, 'U', "unified", &options->context))
2178                 options->output_format |= DIFF_FORMAT_PATCH;
2179         else if (!strcmp(arg, "--raw"))
2180                 options->output_format |= DIFF_FORMAT_RAW;
2181         else if (!strcmp(arg, "--patch-with-raw"))
2182                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2183         else if (!strcmp(arg, "--numstat"))
2184                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2185         else if (!strcmp(arg, "--shortstat"))
2186                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2187         else if (!strcmp(arg, "--check"))
2188                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2189         else if (!strcmp(arg, "--summary"))
2190                 options->output_format |= DIFF_FORMAT_SUMMARY;
2191         else if (!strcmp(arg, "--patch-with-stat"))
2192                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2193         else if (!strcmp(arg, "--name-only"))
2194                 options->output_format |= DIFF_FORMAT_NAME;
2195         else if (!strcmp(arg, "--name-status"))
2196                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2197         else if (!strcmp(arg, "-s"))
2198                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2199         else if (!prefixcmp(arg, "--stat")) {
2200                 char *end;
2201                 int width = options->stat_width;
2202                 int name_width = options->stat_name_width;
2203                 arg += 6;
2204                 end = (char *)arg;
2205
2206                 switch (*arg) {
2207                 case '-':
2208                         if (!prefixcmp(arg, "-width="))
2209                                 width = strtoul(arg + 7, &end, 10);
2210                         else if (!prefixcmp(arg, "-name-width="))
2211                                 name_width = strtoul(arg + 12, &end, 10);
2212                         break;
2213                 case '=':
2214                         width = strtoul(arg+1, &end, 10);
2215                         if (*end == ',')
2216                                 name_width = strtoul(end+1, &end, 10);
2217                 }
2218
2219                 /* Important! This checks all the error cases! */
2220                 if (*end)
2221                         return 0;
2222                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2223                 options->stat_name_width = name_width;
2224                 options->stat_width = width;
2225         }
2226
2227         /* renames options */
2228         else if (!prefixcmp(arg, "-B")) {
2229                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2230                         return -1;
2231         }
2232         else if (!prefixcmp(arg, "-M")) {
2233                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2234                         return -1;
2235                 options->detect_rename = DIFF_DETECT_RENAME;
2236         }
2237         else if (!prefixcmp(arg, "-C")) {
2238                 if (options->detect_rename == DIFF_DETECT_COPY)
2239                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2240                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2241                         return -1;
2242                 options->detect_rename = DIFF_DETECT_COPY;
2243         }
2244         else if (!strcmp(arg, "--no-renames"))
2245                 options->detect_rename = 0;
2246
2247         /* xdiff options */
2248         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2249                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2250         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2251                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2252         else if (!strcmp(arg, "--ignore-space-at-eol"))
2253                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2254
2255         /* flags options */
2256         else if (!strcmp(arg, "--binary")) {
2257                 options->output_format |= DIFF_FORMAT_PATCH;
2258                 DIFF_OPT_SET(options, BINARY);
2259         }
2260         else if (!strcmp(arg, "--full-index"))
2261                 DIFF_OPT_SET(options, FULL_INDEX);
2262         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2263                 DIFF_OPT_SET(options, TEXT);
2264         else if (!strcmp(arg, "-R"))
2265                 DIFF_OPT_SET(options, REVERSE_DIFF);
2266         else if (!strcmp(arg, "--find-copies-harder"))
2267                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2268         else if (!strcmp(arg, "--follow"))
2269                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2270         else if (!strcmp(arg, "--color"))
2271                 DIFF_OPT_SET(options, COLOR_DIFF);
2272         else if (!strcmp(arg, "--no-color"))
2273                 DIFF_OPT_CLR(options, COLOR_DIFF);
2274         else if (!strcmp(arg, "--color-words"))
2275                 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2276         else if (!strcmp(arg, "--exit-code"))
2277                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2278         else if (!strcmp(arg, "--quiet"))
2279                 DIFF_OPT_SET(options, QUIET);
2280         else if (!strcmp(arg, "--ext-diff"))
2281                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2282         else if (!strcmp(arg, "--no-ext-diff"))
2283                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2284
2285         /* misc options */
2286         else if (!strcmp(arg, "-z"))
2287                 options->line_termination = 0;
2288         else if (!prefixcmp(arg, "-l"))
2289                 options->rename_limit = strtoul(arg+2, NULL, 10);
2290         else if (!prefixcmp(arg, "-S"))
2291                 options->pickaxe = arg + 2;
2292         else if (!strcmp(arg, "--pickaxe-all"))
2293                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2294         else if (!strcmp(arg, "--pickaxe-regex"))
2295                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2296         else if (!prefixcmp(arg, "-O"))
2297                 options->orderfile = arg + 2;
2298         else if (!prefixcmp(arg, "--diff-filter="))
2299                 options->filter = arg + 14;
2300         else if (!strcmp(arg, "--abbrev"))
2301                 options->abbrev = DEFAULT_ABBREV;
2302         else if (!prefixcmp(arg, "--abbrev=")) {
2303                 options->abbrev = strtoul(arg + 9, NULL, 10);
2304                 if (options->abbrev < MINIMUM_ABBREV)
2305                         options->abbrev = MINIMUM_ABBREV;
2306                 else if (40 < options->abbrev)
2307                         options->abbrev = 40;
2308         }
2309         else if (!prefixcmp(arg, "--src-prefix="))
2310                 options->a_prefix = arg + 13;
2311         else if (!prefixcmp(arg, "--dst-prefix="))
2312                 options->b_prefix = arg + 13;
2313         else if (!strcmp(arg, "--no-prefix"))
2314                 options->a_prefix = options->b_prefix = "";
2315         else
2316                 return 0;
2317         return 1;
2318 }
2319
2320 static int parse_num(const char **cp_p)
2321 {
2322         unsigned long num, scale;
2323         int ch, dot;
2324         const char *cp = *cp_p;
2325
2326         num = 0;
2327         scale = 1;
2328         dot = 0;
2329         for(;;) {
2330                 ch = *cp;
2331                 if ( !dot && ch == '.' ) {
2332                         scale = 1;
2333                         dot = 1;
2334                 } else if ( ch == '%' ) {
2335                         scale = dot ? scale*100 : 100;
2336                         cp++;   /* % is always at the end */
2337                         break;
2338                 } else if ( ch >= '0' && ch <= '9' ) {
2339                         if ( scale < 100000 ) {
2340                                 scale *= 10;
2341                                 num = (num*10) + (ch-'0');
2342                         }
2343                 } else {
2344                         break;
2345                 }
2346                 cp++;
2347         }
2348         *cp_p = cp;
2349
2350         /* user says num divided by scale and we say internally that
2351          * is MAX_SCORE * num / scale.
2352          */
2353         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2354 }
2355
2356 static int diff_scoreopt_parse(const char *opt)
2357 {
2358         int opt1, opt2, cmd;
2359
2360         if (*opt++ != '-')
2361                 return -1;
2362         cmd = *opt++;
2363         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2364                 return -1; /* that is not a -M, -C nor -B option */
2365
2366         opt1 = parse_num(&opt);
2367         if (cmd != 'B')
2368                 opt2 = 0;
2369         else {
2370                 if (*opt == 0)
2371                         opt2 = 0;
2372                 else if (*opt != '/')
2373                         return -1; /* we expect -B80/99 or -B80 */
2374                 else {
2375                         opt++;
2376                         opt2 = parse_num(&opt);
2377                 }
2378         }
2379         if (*opt != 0)
2380                 return -1;
2381         return opt1 | (opt2 << 16);
2382 }
2383
2384 struct diff_queue_struct diff_queued_diff;
2385
2386 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2387 {
2388         if (queue->alloc <= queue->nr) {
2389                 queue->alloc = alloc_nr(queue->alloc);
2390                 queue->queue = xrealloc(queue->queue,
2391                                         sizeof(dp) * queue->alloc);
2392         }
2393         queue->queue[queue->nr++] = dp;
2394 }
2395
2396 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2397                                  struct diff_filespec *one,
2398                                  struct diff_filespec *two)
2399 {
2400         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2401         dp->one = one;
2402         dp->two = two;
2403         if (queue)
2404                 diff_q(queue, dp);
2405         return dp;
2406 }
2407
2408 void diff_free_filepair(struct diff_filepair *p)
2409 {
2410         free_filespec(p->one);
2411         free_filespec(p->two);
2412         free(p);
2413 }
2414
2415 /* This is different from find_unique_abbrev() in that
2416  * it stuffs the result with dots for alignment.
2417  */
2418 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2419 {
2420         int abblen;
2421         const char *abbrev;
2422         if (len == 40)
2423                 return sha1_to_hex(sha1);
2424
2425         abbrev = find_unique_abbrev(sha1, len);
2426         if (!abbrev)
2427                 return sha1_to_hex(sha1);
2428         abblen = strlen(abbrev);
2429         if (abblen < 37) {
2430                 static char hex[41];
2431                 if (len < abblen && abblen <= len + 2)
2432                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2433                 else
2434                         sprintf(hex, "%s...", abbrev);
2435                 return hex;
2436         }
2437         return sha1_to_hex(sha1);
2438 }
2439
2440 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2441 {
2442         int line_termination = opt->line_termination;
2443         int inter_name_termination = line_termination ? '\t' : '\0';
2444
2445         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2446                 printf(":%06o %06o %s ", p->one->mode, p->two->mode,
2447                        diff_unique_abbrev(p->one->sha1, opt->abbrev));
2448                 printf("%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2449         }
2450         if (p->score) {
2451                 printf("%c%03d%c", p->status, similarity_index(p),
2452                            inter_name_termination);
2453         } else {
2454                 printf("%c%c", p->status, inter_name_termination);
2455         }
2456
2457         if (p->status == DIFF_STATUS_COPIED || p->status == DIFF_STATUS_RENAMED) {
2458                 write_name_quoted(p->one->path, stdout, inter_name_termination);
2459                 write_name_quoted(p->two->path, stdout, line_termination);
2460         } else {
2461                 const char *path = p->one->mode ? p->one->path : p->two->path;
2462                 write_name_quoted(path, stdout, line_termination);
2463         }
2464 }
2465
2466 int diff_unmodified_pair(struct diff_filepair *p)
2467 {
2468         /* This function is written stricter than necessary to support
2469          * the currently implemented transformers, but the idea is to
2470          * let transformers to produce diff_filepairs any way they want,
2471          * and filter and clean them up here before producing the output.
2472          */
2473         struct diff_filespec *one = p->one, *two = p->two;
2474
2475         if (DIFF_PAIR_UNMERGED(p))
2476                 return 0; /* unmerged is interesting */
2477
2478         /* deletion, addition, mode or type change
2479          * and rename are all interesting.
2480          */
2481         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2482             DIFF_PAIR_MODE_CHANGED(p) ||
2483             strcmp(one->path, two->path))
2484                 return 0;
2485
2486         /* both are valid and point at the same path.  that is, we are
2487          * dealing with a change.
2488          */
2489         if (one->sha1_valid && two->sha1_valid &&
2490             !hashcmp(one->sha1, two->sha1))
2491                 return 1; /* no change */
2492         if (!one->sha1_valid && !two->sha1_valid)
2493                 return 1; /* both look at the same file on the filesystem. */
2494         return 0;
2495 }
2496
2497 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2498 {
2499         if (diff_unmodified_pair(p))
2500                 return;
2501
2502         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2503             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2504                 return; /* no tree diffs in patch format */
2505
2506         run_diff(p, o);
2507 }
2508
2509 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2510                             struct diffstat_t *diffstat)
2511 {
2512         if (diff_unmodified_pair(p))
2513                 return;
2514
2515         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2516             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2517                 return; /* no tree diffs in patch format */
2518
2519         run_diffstat(p, o, diffstat);
2520 }
2521
2522 static void diff_flush_checkdiff(struct diff_filepair *p,
2523                 struct diff_options *o)
2524 {
2525         if (diff_unmodified_pair(p))
2526                 return;
2527
2528         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2529             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2530                 return; /* no tree diffs in patch format */
2531
2532         run_checkdiff(p, o);
2533 }
2534
2535 int diff_queue_is_empty(void)
2536 {
2537         struct diff_queue_struct *q = &diff_queued_diff;
2538         int i;
2539         for (i = 0; i < q->nr; i++)
2540                 if (!diff_unmodified_pair(q->queue[i]))
2541                         return 0;
2542         return 1;
2543 }
2544
2545 #if DIFF_DEBUG
2546 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2547 {
2548         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2549                 x, one ? one : "",
2550                 s->path,
2551                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2552                 s->mode,
2553                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2554         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2555                 x, one ? one : "",
2556                 s->size, s->xfrm_flags);
2557 }
2558
2559 void diff_debug_filepair(const struct diff_filepair *p, int i)
2560 {
2561         diff_debug_filespec(p->one, i, "one");
2562         diff_debug_filespec(p->two, i, "two");
2563         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2564                 p->score, p->status ? p->status : '?',
2565                 p->one->rename_used, p->broken_pair);
2566 }
2567
2568 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2569 {
2570         int i;
2571         if (msg)
2572                 fprintf(stderr, "%s\n", msg);
2573         fprintf(stderr, "q->nr = %d\n", q->nr);
2574         for (i = 0; i < q->nr; i++) {
2575                 struct diff_filepair *p = q->queue[i];
2576                 diff_debug_filepair(p, i);
2577         }
2578 }
2579 #endif
2580
2581 static void diff_resolve_rename_copy(void)
2582 {
2583         int i;
2584         struct diff_filepair *p;
2585         struct diff_queue_struct *q = &diff_queued_diff;
2586
2587         diff_debug_queue("resolve-rename-copy", q);
2588
2589         for (i = 0; i < q->nr; i++) {
2590                 p = q->queue[i];
2591                 p->status = 0; /* undecided */
2592                 if (DIFF_PAIR_UNMERGED(p))
2593                         p->status = DIFF_STATUS_UNMERGED;
2594                 else if (!DIFF_FILE_VALID(p->one))
2595                         p->status = DIFF_STATUS_ADDED;
2596                 else if (!DIFF_FILE_VALID(p->two))
2597                         p->status = DIFF_STATUS_DELETED;
2598                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2599                         p->status = DIFF_STATUS_TYPE_CHANGED;
2600
2601                 /* from this point on, we are dealing with a pair
2602                  * whose both sides are valid and of the same type, i.e.
2603                  * either in-place edit or rename/copy edit.
2604                  */
2605                 else if (DIFF_PAIR_RENAME(p)) {
2606                         /*
2607                          * A rename might have re-connected a broken
2608                          * pair up, causing the pathnames to be the
2609                          * same again. If so, that's not a rename at
2610                          * all, just a modification..
2611                          *
2612                          * Otherwise, see if this source was used for
2613                          * multiple renames, in which case we decrement
2614                          * the count, and call it a copy.
2615                          */
2616                         if (!strcmp(p->one->path, p->two->path))
2617                                 p->status = DIFF_STATUS_MODIFIED;
2618                         else if (--p->one->rename_used > 0)
2619                                 p->status = DIFF_STATUS_COPIED;
2620                         else
2621                                 p->status = DIFF_STATUS_RENAMED;
2622                 }
2623                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2624                          p->one->mode != p->two->mode ||
2625                          is_null_sha1(p->one->sha1))
2626                         p->status = DIFF_STATUS_MODIFIED;
2627                 else {
2628                         /* This is a "no-change" entry and should not
2629                          * happen anymore, but prepare for broken callers.
2630                          */
2631                         error("feeding unmodified %s to diffcore",
2632                               p->one->path);
2633                         p->status = DIFF_STATUS_UNKNOWN;
2634                 }
2635         }
2636         diff_debug_queue("resolve-rename-copy done", q);
2637 }
2638
2639 static int check_pair_status(struct diff_filepair *p)
2640 {
2641         switch (p->status) {
2642         case DIFF_STATUS_UNKNOWN:
2643                 return 0;
2644         case 0:
2645                 die("internal error in diff-resolve-rename-copy");
2646         default:
2647                 return 1;
2648         }
2649 }
2650
2651 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2652 {
2653         int fmt = opt->output_format;
2654
2655         if (fmt & DIFF_FORMAT_CHECKDIFF)
2656                 diff_flush_checkdiff(p, opt);
2657         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2658                 diff_flush_raw(p, opt);
2659         else if (fmt & DIFF_FORMAT_NAME)
2660                 write_name_quoted(p->two->path, stdout, opt->line_termination);
2661 }
2662
2663 static void show_file_mode_name(const char *newdelete, struct diff_filespec *fs)
2664 {
2665         if (fs->mode)
2666                 printf(" %s mode %06o ", newdelete, fs->mode);
2667         else
2668                 printf(" %s ", newdelete);
2669         write_name_quoted(fs->path, stdout, '\n');
2670 }
2671
2672
2673 static void show_mode_change(struct diff_filepair *p, int show_name)
2674 {
2675         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2676                 printf(" mode change %06o => %06o%c", p->one->mode, p->two->mode,
2677                         show_name ? ' ' : '\n');
2678                 if (show_name) {
2679                         write_name_quoted(p->two->path, stdout, '\n');
2680                 }
2681         }
2682 }
2683
2684 static void show_rename_copy(const char *renamecopy, struct diff_filepair *p)
2685 {
2686         char *names = pprint_rename(p->one->path, p->two->path);
2687
2688         printf(" %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2689         free(names);
2690         show_mode_change(p, 0);
2691 }
2692
2693 static void diff_summary(struct diff_filepair *p)
2694 {
2695         switch(p->status) {
2696         case DIFF_STATUS_DELETED:
2697                 show_file_mode_name("delete", p->one);
2698                 break;
2699         case DIFF_STATUS_ADDED:
2700                 show_file_mode_name("create", p->two);
2701                 break;
2702         case DIFF_STATUS_COPIED:
2703                 show_rename_copy("copy", p);
2704                 break;
2705         case DIFF_STATUS_RENAMED:
2706                 show_rename_copy("rename", p);
2707                 break;
2708         default:
2709                 if (p->score) {
2710                         fputs(" rewrite ", stdout);
2711                         write_name_quoted(p->two->path, stdout, ' ');
2712                         printf("(%d%%)\n", similarity_index(p));
2713                 }
2714                 show_mode_change(p, !p->score);
2715                 break;
2716         }
2717 }
2718
2719 struct patch_id_t {
2720         struct xdiff_emit_state xm;
2721         SHA_CTX *ctx;
2722         int patchlen;
2723 };
2724
2725 static int remove_space(char *line, int len)
2726 {
2727         int i;
2728         char *dst = line;
2729         unsigned char c;
2730
2731         for (i = 0; i < len; i++)
2732                 if (!isspace((c = line[i])))
2733                         *dst++ = c;
2734
2735         return dst - line;
2736 }
2737
2738 static void patch_id_consume(void *priv, char *line, unsigned long len)
2739 {
2740         struct patch_id_t *data = priv;
2741         int new_len;
2742
2743         /* Ignore line numbers when computing the SHA1 of the patch */
2744         if (!prefixcmp(line, "@@ -"))
2745                 return;
2746
2747         new_len = remove_space(line, len);
2748
2749         SHA1_Update(data->ctx, line, new_len);
2750         data->patchlen += new_len;
2751 }
2752
2753 /* returns 0 upon success, and writes result into sha1 */
2754 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2755 {
2756         struct diff_queue_struct *q = &diff_queued_diff;
2757         int i;
2758         SHA_CTX ctx;
2759         struct patch_id_t data;
2760         char buffer[PATH_MAX * 4 + 20];
2761
2762         SHA1_Init(&ctx);
2763         memset(&data, 0, sizeof(struct patch_id_t));
2764         data.ctx = &ctx;
2765         data.xm.consume = patch_id_consume;
2766
2767         for (i = 0; i < q->nr; i++) {
2768                 xpparam_t xpp;
2769                 xdemitconf_t xecfg;
2770                 xdemitcb_t ecb;
2771                 mmfile_t mf1, mf2;
2772                 struct diff_filepair *p = q->queue[i];
2773                 int len1, len2;
2774
2775                 memset(&xecfg, 0, sizeof(xecfg));
2776                 if (p->status == 0)
2777                         return error("internal diff status error");
2778                 if (p->status == DIFF_STATUS_UNKNOWN)
2779                         continue;
2780                 if (diff_unmodified_pair(p))
2781                         continue;
2782                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2783                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2784                         continue;
2785                 if (DIFF_PAIR_UNMERGED(p))
2786                         continue;
2787
2788                 diff_fill_sha1_info(p->one);
2789                 diff_fill_sha1_info(p->two);
2790                 if (fill_mmfile(&mf1, p->one) < 0 ||
2791                                 fill_mmfile(&mf2, p->two) < 0)
2792                         return error("unable to read files to diff");
2793
2794                 len1 = remove_space(p->one->path, strlen(p->one->path));
2795                 len2 = remove_space(p->two->path, strlen(p->two->path));
2796                 if (p->one->mode == 0)
2797                         len1 = snprintf(buffer, sizeof(buffer),
2798                                         "diff--gita/%.*sb/%.*s"
2799                                         "newfilemode%06o"
2800                                         "---/dev/null"
2801                                         "+++b/%.*s",
2802                                         len1, p->one->path,
2803                                         len2, p->two->path,
2804                                         p->two->mode,
2805                                         len2, p->two->path);
2806                 else if (p->two->mode == 0)
2807                         len1 = snprintf(buffer, sizeof(buffer),
2808                                         "diff--gita/%.*sb/%.*s"
2809                                         "deletedfilemode%06o"
2810                                         "---a/%.*s"
2811                                         "+++/dev/null",
2812                                         len1, p->one->path,
2813                                         len2, p->two->path,
2814                                         p->one->mode,
2815                                         len1, p->one->path);
2816                 else
2817                         len1 = snprintf(buffer, sizeof(buffer),
2818                                         "diff--gita/%.*sb/%.*s"
2819                                         "---a/%.*s"
2820                                         "+++b/%.*s",
2821                                         len1, p->one->path,
2822                                         len2, p->two->path,
2823                                         len1, p->one->path,
2824                                         len2, p->two->path);
2825                 SHA1_Update(&ctx, buffer, len1);
2826
2827                 xpp.flags = XDF_NEED_MINIMAL;
2828                 xecfg.ctxlen = 3;
2829                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2830                 ecb.outf = xdiff_outf;
2831                 ecb.priv = &data;
2832                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
2833         }
2834
2835         SHA1_Final(sha1, &ctx);
2836         return 0;
2837 }
2838
2839 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
2840 {
2841         struct diff_queue_struct *q = &diff_queued_diff;
2842         int i;
2843         int result = diff_get_patch_id(options, sha1);
2844
2845         for (i = 0; i < q->nr; i++)
2846                 diff_free_filepair(q->queue[i]);
2847
2848         free(q->queue);
2849         q->queue = NULL;
2850         q->nr = q->alloc = 0;
2851
2852         return result;
2853 }
2854
2855 static int is_summary_empty(const struct diff_queue_struct *q)
2856 {
2857         int i;
2858
2859         for (i = 0; i < q->nr; i++) {
2860                 const struct diff_filepair *p = q->queue[i];
2861
2862                 switch (p->status) {
2863                 case DIFF_STATUS_DELETED:
2864                 case DIFF_STATUS_ADDED:
2865                 case DIFF_STATUS_COPIED:
2866                 case DIFF_STATUS_RENAMED:
2867                         return 0;
2868                 default:
2869                         if (p->score)
2870                                 return 0;
2871                         if (p->one->mode && p->two->mode &&
2872                             p->one->mode != p->two->mode)
2873                                 return 0;
2874                         break;
2875                 }
2876         }
2877         return 1;
2878 }
2879
2880 void diff_flush(struct diff_options *options)
2881 {
2882         struct diff_queue_struct *q = &diff_queued_diff;
2883         int i, output_format = options->output_format;
2884         int separator = 0;
2885
2886         /*
2887          * Order: raw, stat, summary, patch
2888          * or:    name/name-status/checkdiff (other bits clear)
2889          */
2890         if (!q->nr)
2891                 goto free_queue;
2892
2893         if (output_format & (DIFF_FORMAT_RAW |
2894                              DIFF_FORMAT_NAME |
2895                              DIFF_FORMAT_NAME_STATUS |
2896                              DIFF_FORMAT_CHECKDIFF)) {
2897                 for (i = 0; i < q->nr; i++) {
2898                         struct diff_filepair *p = q->queue[i];
2899                         if (check_pair_status(p))
2900                                 flush_one_pair(p, options);
2901                 }
2902                 separator++;
2903         }
2904
2905         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
2906                 struct diffstat_t diffstat;
2907
2908                 memset(&diffstat, 0, sizeof(struct diffstat_t));
2909                 diffstat.xm.consume = diffstat_consume;
2910                 for (i = 0; i < q->nr; i++) {
2911                         struct diff_filepair *p = q->queue[i];
2912                         if (check_pair_status(p))
2913                                 diff_flush_stat(p, options, &diffstat);
2914                 }
2915                 if (output_format & DIFF_FORMAT_NUMSTAT)
2916                         show_numstat(&diffstat, options);
2917                 if (output_format & DIFF_FORMAT_DIFFSTAT)
2918                         show_stats(&diffstat, options);
2919                 if (output_format & DIFF_FORMAT_SHORTSTAT)
2920                         show_shortstats(&diffstat);
2921                 free_diffstat_info(&diffstat);
2922                 separator++;
2923         }
2924
2925         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
2926                 for (i = 0; i < q->nr; i++)
2927                         diff_summary(q->queue[i]);
2928                 separator++;
2929         }
2930
2931         if (output_format & DIFF_FORMAT_PATCH) {
2932                 if (separator) {
2933                         if (options->stat_sep) {
2934                                 /* attach patch instead of inline */
2935                                 fputs(options->stat_sep, stdout);
2936                         } else {
2937                                 putchar(options->line_termination);
2938                         }
2939                 }
2940
2941                 for (i = 0; i < q->nr; i++) {
2942                         struct diff_filepair *p = q->queue[i];
2943                         if (check_pair_status(p))
2944                                 diff_flush_patch(p, options);
2945                 }
2946         }
2947
2948         if (output_format & DIFF_FORMAT_CALLBACK)
2949                 options->format_callback(q, options, options->format_callback_data);
2950
2951         for (i = 0; i < q->nr; i++)
2952                 diff_free_filepair(q->queue[i]);
2953 free_queue:
2954         free(q->queue);
2955         q->queue = NULL;
2956         q->nr = q->alloc = 0;
2957 }
2958
2959 static void diffcore_apply_filter(const char *filter)
2960 {
2961         int i;
2962         struct diff_queue_struct *q = &diff_queued_diff;
2963         struct diff_queue_struct outq;
2964         outq.queue = NULL;
2965         outq.nr = outq.alloc = 0;
2966
2967         if (!filter)
2968                 return;
2969
2970         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
2971                 int found;
2972                 for (i = found = 0; !found && i < q->nr; i++) {
2973                         struct diff_filepair *p = q->queue[i];
2974                         if (((p->status == DIFF_STATUS_MODIFIED) &&
2975                              ((p->score &&
2976                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
2977                               (!p->score &&
2978                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
2979                             ((p->status != DIFF_STATUS_MODIFIED) &&
2980                              strchr(filter, p->status)))
2981                                 found++;
2982                 }
2983                 if (found)
2984                         return;
2985
2986                 /* otherwise we will clear the whole queue
2987                  * by copying the empty outq at the end of this
2988                  * function, but first clear the current entries
2989                  * in the queue.
2990                  */
2991                 for (i = 0; i < q->nr; i++)
2992                         diff_free_filepair(q->queue[i]);
2993         }
2994         else {
2995                 /* Only the matching ones */
2996                 for (i = 0; i < q->nr; i++) {
2997                         struct diff_filepair *p = q->queue[i];
2998
2999                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3000                              ((p->score &&
3001                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3002                               (!p->score &&
3003                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3004                             ((p->status != DIFF_STATUS_MODIFIED) &&
3005                              strchr(filter, p->status)))
3006                                 diff_q(&outq, p);
3007                         else
3008                                 diff_free_filepair(p);
3009                 }
3010         }
3011         free(q->queue);
3012         *q = outq;
3013 }
3014
3015 /* Check whether two filespecs with the same mode and size are identical */
3016 static int diff_filespec_is_identical(struct diff_filespec *one,
3017                                       struct diff_filespec *two)
3018 {
3019         if (S_ISGITLINK(one->mode)) {
3020                 diff_fill_sha1_info(one);
3021                 diff_fill_sha1_info(two);
3022                 return !hashcmp(one->sha1, two->sha1);
3023         }
3024         if (diff_populate_filespec(one, 0))
3025                 return 0;
3026         if (diff_populate_filespec(two, 0))
3027                 return 0;
3028         return !memcmp(one->data, two->data, one->size);
3029 }
3030
3031 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3032 {
3033         int i;
3034         struct diff_queue_struct *q = &diff_queued_diff;
3035         struct diff_queue_struct outq;
3036         outq.queue = NULL;
3037         outq.nr = outq.alloc = 0;
3038
3039         for (i = 0; i < q->nr; i++) {
3040                 struct diff_filepair *p = q->queue[i];
3041
3042                 /*
3043                  * 1. Entries that come from stat info dirtyness
3044                  *    always have both sides (iow, not create/delete),
3045                  *    one side of the object name is unknown, with
3046                  *    the same mode and size.  Keep the ones that
3047                  *    do not match these criteria.  They have real
3048                  *    differences.
3049                  *
3050                  * 2. At this point, the file is known to be modified,
3051                  *    with the same mode and size, and the object
3052                  *    name of one side is unknown.  Need to inspect
3053                  *    the identical contents.
3054                  */
3055                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3056                     !DIFF_FILE_VALID(p->two) ||
3057                     (p->one->sha1_valid && p->two->sha1_valid) ||
3058                     (p->one->mode != p->two->mode) ||
3059                     diff_populate_filespec(p->one, 1) ||
3060                     diff_populate_filespec(p->two, 1) ||
3061                     (p->one->size != p->two->size) ||
3062                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3063                         diff_q(&outq, p);
3064                 else {
3065                         /*
3066                          * The caller can subtract 1 from skip_stat_unmatch
3067                          * to determine how many paths were dirty only
3068                          * due to stat info mismatch.
3069                          */
3070                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3071                                 diffopt->skip_stat_unmatch++;
3072                         diff_free_filepair(p);
3073                 }
3074         }
3075         free(q->queue);
3076         *q = outq;
3077 }
3078
3079 void diffcore_std(struct diff_options *options)
3080 {
3081         if (DIFF_OPT_TST(options, QUIET))
3082                 return;
3083
3084         if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3085                 diffcore_skip_stat_unmatch(options);
3086         if (options->break_opt != -1)
3087                 diffcore_break(options->break_opt);
3088         if (options->detect_rename)
3089                 diffcore_rename(options);
3090         if (options->break_opt != -1)
3091                 diffcore_merge_broken();
3092         if (options->pickaxe)
3093                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3094         if (options->orderfile)
3095                 diffcore_order(options->orderfile);
3096         diff_resolve_rename_copy();
3097         diffcore_apply_filter(options->filter);
3098
3099         if (diff_queued_diff.nr)
3100                 DIFF_OPT_SET(options, HAS_CHANGES);
3101         else
3102                 DIFF_OPT_CLR(options, HAS_CHANGES);
3103 }
3104
3105 int diff_result_code(struct diff_options *opt, int status)
3106 {
3107         int result = 0;
3108         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3109             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3110                 return status;
3111         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3112             DIFF_OPT_TST(opt, HAS_CHANGES))
3113                 result |= 01;
3114         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3115             DIFF_OPT_TST(opt, CHECK_FAILED))
3116                 result |= 02;
3117         return result;
3118 }
3119
3120 void diff_addremove(struct diff_options *options,
3121                     int addremove, unsigned mode,
3122                     const unsigned char *sha1,
3123                     const char *base, const char *path)
3124 {
3125         char concatpath[PATH_MAX];
3126         struct diff_filespec *one, *two;
3127
3128         /* This may look odd, but it is a preparation for
3129          * feeding "there are unchanged files which should
3130          * not produce diffs, but when you are doing copy
3131          * detection you would need them, so here they are"
3132          * entries to the diff-core.  They will be prefixed
3133          * with something like '=' or '*' (I haven't decided
3134          * which but should not make any difference).
3135          * Feeding the same new and old to diff_change()
3136          * also has the same effect.
3137          * Before the final output happens, they are pruned after
3138          * merged into rename/copy pairs as appropriate.
3139          */
3140         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3141                 addremove = (addremove == '+' ? '-' :
3142                              addremove == '-' ? '+' : addremove);
3143
3144         if (!path) path = "";
3145         sprintf(concatpath, "%s%s", base, path);
3146         one = alloc_filespec(concatpath);
3147         two = alloc_filespec(concatpath);
3148
3149         if (addremove != '+')
3150                 fill_filespec(one, sha1, mode);
3151         if (addremove != '-')
3152                 fill_filespec(two, sha1, mode);
3153
3154         diff_queue(&diff_queued_diff, one, two);
3155         DIFF_OPT_SET(options, HAS_CHANGES);
3156 }
3157
3158 void diff_change(struct diff_options *options,
3159                  unsigned old_mode, unsigned new_mode,
3160                  const unsigned char *old_sha1,
3161                  const unsigned char *new_sha1,
3162                  const char *base, const char *path)
3163 {
3164         char concatpath[PATH_MAX];
3165         struct diff_filespec *one, *two;
3166
3167         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3168                 unsigned tmp;
3169                 const unsigned char *tmp_c;
3170                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3171                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3172         }
3173         if (!path) path = "";
3174         sprintf(concatpath, "%s%s", base, path);
3175         one = alloc_filespec(concatpath);
3176         two = alloc_filespec(concatpath);
3177         fill_filespec(one, old_sha1, old_mode);
3178         fill_filespec(two, new_sha1, new_mode);
3179
3180         diff_queue(&diff_queued_diff, one, two);
3181         DIFF_OPT_SET(options, HAS_CHANGES);
3182 }
3183
3184 void diff_unmerge(struct diff_options *options,
3185                   const char *path,
3186                   unsigned mode, const unsigned char *sha1)
3187 {
3188         struct diff_filespec *one, *two;
3189         one = alloc_filespec(path);
3190         two = alloc_filespec(path);
3191         fill_filespec(one, sha1, mode);
3192         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3193 }