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

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "quote.h"
   6 #include "diff.h"
   7 #include "diffcore.h"
   8 #include "delta.h"
   9 #include "xdiff-interface.h"
  10 #include "color.h"
  11 #include "attr.h"
  12 #include "run-command.h"
  13 #include "utf8.h"
  14
  15 #ifdef NO_FAST_WORKING_DIRECTORY
  16 #define FAST_WORKING_DIRECTORY 0
  17 #else
  18 #define FAST_WORKING_DIRECTORY 1
  19 #endif
  20
  21 static int diff_detect_rename_default;
  22 static int diff_rename_limit_default = 200;
  23 static int diff_suppress_blank_empty;
  24 int diff_use_color_default = -1;
  25 static const char *external_diff_cmd_cfg;
  26 int diff_auto_refresh_index = 1;
  27 static int diff_mnemonic_prefix;
  28
  29 static char diff_colors[][COLOR_MAXLEN] = {
  30         "\033[m",       /* reset */
  31         "",             /* PLAIN (normal) */
  32         "\033[1m",      /* METAINFO (bold) */
  33         "\033[36m",     /* FRAGINFO (cyan) */
  34         "\033[31m",     /* OLD (red) */
  35         "\033[32m",     /* NEW (green) */
  36         "\033[33m",     /* COMMIT (yellow) */
  37         "\033[41m",     /* WHITESPACE (red background) */
  38 };
  39
  40 static int parse_diff_color_slot(const char *var, int ofs)
  41 {
  42         if (!strcasecmp(var+ofs, "plain"))
  43                 return DIFF_PLAIN;
  44         if (!strcasecmp(var+ofs, "meta"))
  45                 return DIFF_METAINFO;
  46         if (!strcasecmp(var+ofs, "frag"))
  47                 return DIFF_FRAGINFO;
  48         if (!strcasecmp(var+ofs, "old"))
  49                 return DIFF_FILE_OLD;
  50         if (!strcasecmp(var+ofs, "new"))
  51                 return DIFF_FILE_NEW;
  52         if (!strcasecmp(var+ofs, "commit"))
  53                 return DIFF_COMMIT;
  54         if (!strcasecmp(var+ofs, "whitespace"))
  55                 return DIFF_WHITESPACE;
  56         die("bad config variable '%s'", var);
  57 }
  58
  59 static struct ll_diff_driver {
  60         const char *name;
  61         struct ll_diff_driver *next;
  62         const char *cmd;
  63 } *user_diff, **user_diff_tail;
  64
  65 /*
  66  * Currently there is only "diff.<drivername>.command" variable;
  67  * because there are "diff.color.<slot>" variables, we are parsing
  68  * this in a bit convoluted way to allow low level diff driver
  69  * called "color".
  70  */
  71 static int parse_lldiff_command(const char *var, const char *ep, const char *value)
  72 {
  73         const char *name;
  74         int namelen;
  75         struct ll_diff_driver *drv;
  76
  77         name = var + 5;
  78         namelen = ep - name;
  79         for (drv = user_diff; drv; drv = drv->next)
  80                 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
  81                         break;
  82         if (!drv) {
  83                 drv = xcalloc(1, sizeof(struct ll_diff_driver));
  84                 drv->name = xmemdupz(name, namelen);
  85                 if (!user_diff_tail)
  86                         user_diff_tail = &user_diff;
  87                 *user_diff_tail = drv;
  88                 user_diff_tail = &(drv->next);
  89         }
  90
  91         return git_config_string(&(drv->cmd), var, value);
  92 }
  93
  94 /*
  95  * 'diff.<what>.funcname' attribute can be specified in the configuration
  96  * to define a customized regexp to find the beginning of a function to
  97  * be used for hunk header lines of "diff -p" style output.
  98  */
  99 static struct funcname_pattern {
 100         char *name;
 101         char *pattern;
 102         struct funcname_pattern *next;
 103 } *funcname_pattern_list;
 104
 105 static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
 106 {
 107         const char *name;
 108         int namelen;
 109         struct funcname_pattern *pp;
 110
 111         name = var + 5; /* "diff." */
 112         namelen = ep - name;
 113
 114         for (pp = funcname_pattern_list; pp; pp = pp->next)
 115                 if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
 116                         break;
 117         if (!pp) {
 118                 pp = xcalloc(1, sizeof(*pp));
 119                 pp->name = xmemdupz(name, namelen);
 120                 pp->next = funcname_pattern_list;
 121                 funcname_pattern_list = pp;
 122         }
 123         free(pp->pattern);
 124         pp->pattern = xstrdup(value);
 125         return 0;
 126 }
 127
 128 /*
 129  * These are to give UI layer defaults.
 130  * The core-level commands such as git-diff-files should
 131  * never be affected by the setting of diff.renames
 132  * the user happens to have in the configuration file.
 133  */
 134 int git_diff_ui_config(const char *var, const char *value, void *cb)
 135 {
 136         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 137                 diff_use_color_default = git_config_colorbool(var, value, -1);
 138                 return 0;
 139         }
 140         if (!strcmp(var, "diff.renames")) {
 141                 if (!value)
 142                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 143                 else if (!strcasecmp(value, "copies") ||
 144                          !strcasecmp(value, "copy"))
 145                         diff_detect_rename_default = DIFF_DETECT_COPY;
 146                 else if (git_config_bool(var,value))
 147                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 148                 return 0;
 149         }
 150         if (!strcmp(var, "diff.autorefreshindex")) {
 151                 diff_auto_refresh_index = git_config_bool(var, value);
 152                 return 0;
 153         }
 154         if (!strcmp(var, "diff.mnemonicprefix")) {
 155                 diff_mnemonic_prefix = git_config_bool(var, value);
 156                 return 0;
 157         }
 158         if (!strcmp(var, "diff.external"))
 159                 return git_config_string(&external_diff_cmd_cfg, var, value);
 160         if (!prefixcmp(var, "diff.")) {
 161                 const char *ep = strrchr(var, '.');
 162
 163                 if (ep != var + 4 && !strcmp(ep, ".command"))
 164                         return parse_lldiff_command(var, ep, value);
 165         }
 166
 167         return git_diff_basic_config(var, value, cb);
 168 }
 169
 170 int git_diff_basic_config(const char *var, const char *value, void *cb)
 171 {
 172         if (!strcmp(var, "diff.renamelimit")) {
 173                 diff_rename_limit_default = git_config_int(var, value);
 174                 return 0;
 175         }
 176
 177         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 178                 int slot = parse_diff_color_slot(var, 11);
 179                 if (!value)
 180                         return config_error_nonbool(var);
 181                 color_parse(value, var, diff_colors[slot]);
 182                 return 0;
 183         }
 184
 185         /* like GNU diff's --suppress-blank-empty option  */
 186         if (!strcmp(var, "diff.suppress-blank-empty")) {
 187                 diff_suppress_blank_empty = git_config_bool(var, value);
 188                 return 0;
 189         }
 190
 191         if (!prefixcmp(var, "diff.")) {
 192                 const char *ep = strrchr(var, '.');
 193                 if (ep != var + 4) {
 194                         if (!strcmp(ep, ".funcname")) {
 195                                 if (!value)
 196                                         return config_error_nonbool(var);
 197                                 return parse_funcname_pattern(var, ep, value);
 198                         }
 199                 }
 200         }
 201
 202         return git_color_default_config(var, value, cb);
 203 }
 204
 205 static char *quote_two(const char *one, const char *two)
 206 {
 207         int need_one = quote_c_style(one, NULL, NULL, 1);
 208         int need_two = quote_c_style(two, NULL, NULL, 1);
 209         struct strbuf res;
 210
 211         strbuf_init(&res, 0);
 212         if (need_one + need_two) {
 213                 strbuf_addch(&res, '"');
 214                 quote_c_style(one, &res, NULL, 1);
 215                 quote_c_style(two, &res, NULL, 1);
 216                 strbuf_addch(&res, '"');
 217         } else {
 218                 strbuf_addstr(&res, one);
 219                 strbuf_addstr(&res, two);
 220         }
 221         return strbuf_detach(&res, NULL);
 222 }
 223
 224 static const char *external_diff(void)
 225 {
 226         static const char *external_diff_cmd = NULL;
 227         static int done_preparing = 0;
 228
 229         if (done_preparing)
 230                 return external_diff_cmd;
 231         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 232         if (!external_diff_cmd)
 233                 external_diff_cmd = external_diff_cmd_cfg;
 234         done_preparing = 1;
 235         return external_diff_cmd;
 236 }
 237
 238 static struct diff_tempfile {
 239         const char *name; /* filename external diff should read from */
 240         char hex[41];
 241         char mode[10];
 242         char tmp_path[PATH_MAX];
 243 } diff_temp[2];
 244
 245 static int count_lines(const char *data, int size)
 246 {
 247         int count, ch, completely_empty = 1, nl_just_seen = 0;
 248         count = 0;
 249         while (0 < size--) {
 250                 ch = *data++;
 251                 if (ch == '\n') {
 252                         count++;
 253                         nl_just_seen = 1;
 254                         completely_empty = 0;
 255                 }
 256                 else {
 257                         nl_just_seen = 0;
 258                         completely_empty = 0;
 259                 }
 260         }
 261         if (completely_empty)
 262                 return 0;
 263         if (!nl_just_seen)
 264                 count++; /* no trailing newline */
 265         return count;
 266 }
 267
 268 static void print_line_count(FILE *file, int count)
 269 {
 270         switch (count) {
 271         case 0:
 272                 fprintf(file, "0,0");
 273                 break;
 274         case 1:
 275                 fprintf(file, "1");
 276                 break;
 277         default:
 278                 fprintf(file, "1,%d", count);
 279                 break;
 280         }
 281 }
 282
 283 static void copy_file_with_prefix(FILE *file,
 284                                   int prefix, const char *data, int size,
 285                                   const char *set, const char *reset)
 286 {
 287         int ch, nl_just_seen = 1;
 288         while (0 < size--) {
 289                 ch = *data++;
 290                 if (nl_just_seen) {
 291                         fputs(set, file);
 292                         putc(prefix, file);
 293                 }
 294                 if (ch == '\n') {
 295                         nl_just_seen = 1;
 296                         fputs(reset, file);
 297                 } else
 298                         nl_just_seen = 0;
 299                 putc(ch, file);
 300         }
 301         if (!nl_just_seen)
 302                 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
 303 }
 304
 305 static void emit_rewrite_diff(const char *name_a,
 306                               const char *name_b,
 307                               struct diff_filespec *one,
 308                               struct diff_filespec *two,
 309                               struct diff_options *o)
 310 {
 311         int lc_a, lc_b;
 312         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
 313         const char *name_a_tab, *name_b_tab;
 314         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 315         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 316         const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
 317         const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
 318         const char *reset = diff_get_color(color_diff, DIFF_RESET);
 319         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 320         const char *a_prefix, *b_prefix;
 321
 322         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 323                 a_prefix = o->b_prefix;
 324                 b_prefix = o->a_prefix;
 325         } else {
 326                 a_prefix = o->a_prefix;
 327                 b_prefix = o->b_prefix;
 328         }
 329
 330         name_a += (*name_a == '/');
 331         name_b += (*name_b == '/');
 332         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 333         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 334
 335         strbuf_reset(&a_name);
 336         strbuf_reset(&b_name);
 337         quote_two_c_style(&a_name, a_prefix, name_a, 0);
 338         quote_two_c_style(&b_name, b_prefix, name_b, 0);
 339
 340         diff_populate_filespec(one, 0);
 341         diff_populate_filespec(two, 0);
 342         lc_a = count_lines(one->data, one->size);
 343         lc_b = count_lines(two->data, two->size);
 344         fprintf(o->file,
 345                 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
 346                 metainfo, a_name.buf, name_a_tab, reset,
 347                 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
 348         print_line_count(o->file, lc_a);
 349         fprintf(o->file, " +");
 350         print_line_count(o->file, lc_b);
 351         fprintf(o->file, " @@%s\n", reset);
 352         if (lc_a)
 353                 copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
 354         if (lc_b)
 355                 copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
 356 }
 357
 358 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 359 {
 360         if (!DIFF_FILE_VALID(one)) {
 361                 mf->ptr = (char *)""; /* does not matter */
 362                 mf->size = 0;
 363                 return 0;
 364         }
 365         else if (diff_populate_filespec(one, 0))
 366                 return -1;
 367         mf->ptr = one->data;
 368         mf->size = one->size;
 369         return 0;
 370 }
 371
 372 struct diff_words_buffer {
 373         mmfile_t text;
 374         long alloc;
 375         long current; /* output pointer */
 376         int suppressed_newline;
 377 };
 378
 379 static void diff_words_append(char *line, unsigned long len,
 380                 struct diff_words_buffer *buffer)
 381 {
 382         if (buffer->text.size + len > buffer->alloc) {
 383                 buffer->alloc = (buffer->text.size + len) * 3 / 2;
 384                 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
 385         }
 386         line++;
 387         len--;
 388         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 389         buffer->text.size += len;
 390 }
 391
 392 struct diff_words_data {
 393         struct diff_words_buffer minus, plus;
 394         FILE *file;
 395 };
 396
 397 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
 398                 int suppress_newline)
 399 {
 400         const char *ptr;
 401         int eol = 0;
 402
 403         if (len == 0)
 404                 return;
 405
 406         ptr  = buffer->text.ptr + buffer->current;
 407         buffer->current += len;
 408
 409         if (ptr[len - 1] == '\n') {
 410                 eol = 1;
 411                 len--;
 412         }
 413
 414         fputs(diff_get_color(1, color), file);
 415         fwrite(ptr, len, 1, file);
 416         fputs(diff_get_color(1, DIFF_RESET), file);
 417
 418         if (eol) {
 419                 if (suppress_newline)
 420                         buffer->suppressed_newline = 1;
 421                 else
 422                         putc('\n', file);
 423         }
 424 }
 425
 426 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 427 {
 428         struct diff_words_data *diff_words = priv;
 429
 430         if (diff_words->minus.suppressed_newline) {
 431                 if (line[0] != '+')
 432                         putc('\n', diff_words->file);
 433                 diff_words->minus.suppressed_newline = 0;
 434         }
 435
 436         len--;
 437         switch (line[0]) {
 438                 case '-':
 439                         print_word(diff_words->file,
 440                                    &diff_words->minus, len, DIFF_FILE_OLD, 1);
 441                         break;
 442                 case '+':
 443                         print_word(diff_words->file,
 444                                    &diff_words->plus, len, DIFF_FILE_NEW, 0);
 445                         break;
 446                 case ' ':
 447                         print_word(diff_words->file,
 448                                    &diff_words->plus, len, DIFF_PLAIN, 0);
 449                         diff_words->minus.current += len;
 450                         break;
 451         }
 452 }
 453
 454 /* this executes the word diff on the accumulated buffers */
 455 static void diff_words_show(struct diff_words_data *diff_words)
 456 {
 457         xpparam_t xpp;
 458         xdemitconf_t xecfg;
 459         xdemitcb_t ecb;
 460         mmfile_t minus, plus;
 461         int i;
 462
 463         memset(&xecfg, 0, sizeof(xecfg));
 464         minus.size = diff_words->minus.text.size;
 465         minus.ptr = xmalloc(minus.size);
 466         memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
 467         for (i = 0; i < minus.size; i++)
 468                 if (isspace(minus.ptr[i]))
 469                         minus.ptr[i] = '\n';
 470         diff_words->minus.current = 0;
 471
 472         plus.size = diff_words->plus.text.size;
 473         plus.ptr = xmalloc(plus.size);
 474         memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
 475         for (i = 0; i < plus.size; i++)
 476                 if (isspace(plus.ptr[i]))
 477                         plus.ptr[i] = '\n';
 478         diff_words->plus.current = 0;
 479
 480         xpp.flags = XDF_NEED_MINIMAL;
 481         xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
 482         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
 483                       &xpp, &xecfg, &ecb);
 484         free(minus.ptr);
 485         free(plus.ptr);
 486         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 487
 488         if (diff_words->minus.suppressed_newline) {
 489                 putc('\n', diff_words->file);
 490                 diff_words->minus.suppressed_newline = 0;
 491         }
 492 }
 493
 494 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 495
 496 struct emit_callback {
 497         int nparents, color_diff;
 498         unsigned ws_rule;
 499         sane_truncate_fn truncate;
 500         const char **label_path;
 501         struct diff_words_data *diff_words;
 502         int *found_changesp;
 503         FILE *file;
 504 };
 505
 506 static void free_diff_words_data(struct emit_callback *ecbdata)
 507 {
 508         if (ecbdata->diff_words) {
 509                 /* flush buffers */
 510                 if (ecbdata->diff_words->minus.text.size ||
 511                                 ecbdata->diff_words->plus.text.size)
 512                         diff_words_show(ecbdata->diff_words);
 513
 514                 free (ecbdata->diff_words->minus.text.ptr);
 515                 free (ecbdata->diff_words->plus.text.ptr);
 516                 free(ecbdata->diff_words);
 517                 ecbdata->diff_words = NULL;
 518         }
 519 }
 520
 521 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 522 {
 523         if (diff_use_color)
 524                 return diff_colors[ix];
 525         return "";
 526 }
 527
 528 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
 529 {
 530         int has_trailing_newline, has_trailing_carriage_return;
 531
 532         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 533         if (has_trailing_newline)
 534                 len--;
 535         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 536         if (has_trailing_carriage_return)
 537                 len--;
 538
 539         fputs(set, file);
 540         fwrite(line, len, 1, file);
 541         fputs(reset, file);
 542         if (has_trailing_carriage_return)
 543                 fputc('\r', file);
 544         if (has_trailing_newline)
 545                 fputc('\n', file);
 546 }
 547
 548 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
 549 {
 550         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 551         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 552
 553         if (!*ws)
 554                 emit_line(ecbdata->file, set, reset, line, len);
 555         else {
 556                 /* Emit just the prefix, then the rest. */
 557                 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
 558                 ws_check_emit(line + ecbdata->nparents,
 559                               len - ecbdata->nparents, ecbdata->ws_rule,
 560                               ecbdata->file, set, reset, ws);
 561         }
 562 }
 563
 564 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
 565 {
 566         const char *cp;
 567         unsigned long allot;
 568         size_t l = len;
 569
 570         if (ecb->truncate)
 571                 return ecb->truncate(line, len);
 572         cp = line;
 573         allot = l;
 574         while (0 < l) {
 575                 (void) utf8_width(&cp, &l);
 576                 if (!cp)
 577                         break; /* truncated in the middle? */
 578         }
 579         return allot - l;
 580 }
 581
 582 static void fn_out_consume(void *priv, char *line, unsigned long len)
 583 {
 584         int i;
 585         int color;
 586         struct emit_callback *ecbdata = priv;
 587         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 588         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 589         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 590
 591         *(ecbdata->found_changesp) = 1;
 592
 593         if (ecbdata->label_path[0]) {
 594                 const char *name_a_tab, *name_b_tab;
 595
 596                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 597                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 598
 599                 fprintf(ecbdata->file, "%s--- %s%s%s\n",
 600                         meta, ecbdata->label_path[0], reset, name_a_tab);
 601                 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
 602                         meta, ecbdata->label_path[1], reset, name_b_tab);
 603                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 604         }
 605
 606         if (diff_suppress_blank_empty
 607             && len == 2 && line[0] == ' ' && line[1] == '\n') {
 608                 line[0] = '\n';
 609                 len = 1;
 610         }
 611
 612         /* This is not really necessary for now because
 613          * this codepath only deals with two-way diffs.
 614          */
 615         for (i = 0; i < len && line[i] == '@'; i++)
 616                 ;
 617         if (2 <= i && i < len && line[i] == ' ') {
 618                 ecbdata->nparents = i - 1;
 619                 len = sane_truncate_line(ecbdata, line, len);
 620                 emit_line(ecbdata->file,
 621                           diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
 622                           reset, line, len);
 623                 if (line[len-1] != '\n')
 624                         putc('\n', ecbdata->file);
 625                 return;
 626         }
 627
 628         if (len < ecbdata->nparents) {
 629                 emit_line(ecbdata->file, reset, reset, line, len);
 630                 return;
 631         }
 632
 633         color = DIFF_PLAIN;
 634         if (ecbdata->diff_words && ecbdata->nparents != 1)
 635                 /* fall back to normal diff */
 636                 free_diff_words_data(ecbdata);
 637         if (ecbdata->diff_words) {
 638                 if (line[0] == '-') {
 639                         diff_words_append(line, len,
 640                                           &ecbdata->diff_words->minus);
 641                         return;
 642                 } else if (line[0] == '+') {
 643                         diff_words_append(line, len,
 644                                           &ecbdata->diff_words->plus);
 645                         return;
 646                 }
 647                 if (ecbdata->diff_words->minus.text.size ||
 648                     ecbdata->diff_words->plus.text.size)
 649                         diff_words_show(ecbdata->diff_words);
 650                 line++;
 651                 len--;
 652                 emit_line(ecbdata->file, plain, reset, line, len);
 653                 return;
 654         }
 655         for (i = 0; i < ecbdata->nparents && len; i++) {
 656                 if (line[i] == '-')
 657                         color = DIFF_FILE_OLD;
 658                 else if (line[i] == '+')
 659                         color = DIFF_FILE_NEW;
 660         }
 661
 662         if (color != DIFF_FILE_NEW) {
 663                 emit_line(ecbdata->file,
 664                           diff_get_color(ecbdata->color_diff, color),
 665                           reset, line, len);
 666                 return;
 667         }
 668         emit_add_line(reset, ecbdata, line, len);
 669 }
 670
 671 static char *pprint_rename(const char *a, const char *b)
 672 {
 673         const char *old = a;
 674         const char *new = b;
 675         struct strbuf name;
 676         int pfx_length, sfx_length;
 677         int len_a = strlen(a);
 678         int len_b = strlen(b);
 679         int a_midlen, b_midlen;
 680         int qlen_a = quote_c_style(a, NULL, NULL, 0);
 681         int qlen_b = quote_c_style(b, NULL, NULL, 0);
 682
 683         strbuf_init(&name, 0);
 684         if (qlen_a || qlen_b) {
 685                 quote_c_style(a, &name, NULL, 0);
 686                 strbuf_addstr(&name, " => ");
 687                 quote_c_style(b, &name, NULL, 0);
 688                 return strbuf_detach(&name, NULL);
 689         }
 690
 691         /* Find common prefix */
 692         pfx_length = 0;
 693         while (*old && *new && *old == *new) {
 694                 if (*old == '/')
 695                         pfx_length = old - a + 1;
 696                 old++;
 697                 new++;
 698         }
 699
 700         /* Find common suffix */
 701         old = a + len_a;
 702         new = b + len_b;
 703         sfx_length = 0;
 704         while (a <= old && b <= new && *old == *new) {
 705                 if (*old == '/')
 706                         sfx_length = len_a - (old - a);
 707                 old--;
 708                 new--;
 709         }
 710
 711         /*
 712          * pfx{mid-a => mid-b}sfx
 713          * {pfx-a => pfx-b}sfx
 714          * pfx{sfx-a => sfx-b}
 715          * name-a => name-b
 716          */
 717         a_midlen = len_a - pfx_length - sfx_length;
 718         b_midlen = len_b - pfx_length - sfx_length;
 719         if (a_midlen < 0)
 720                 a_midlen = 0;
 721         if (b_midlen < 0)
 722                 b_midlen = 0;
 723
 724         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 725         if (pfx_length + sfx_length) {
 726                 strbuf_add(&name, a, pfx_length);
 727                 strbuf_addch(&name, '{');
 728         }
 729         strbuf_add(&name, a + pfx_length, a_midlen);
 730         strbuf_addstr(&name, " => ");
 731         strbuf_add(&name, b + pfx_length, b_midlen);
 732         if (pfx_length + sfx_length) {
 733                 strbuf_addch(&name, '}');
 734                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 735         }
 736         return strbuf_detach(&name, NULL);
 737 }
 738
 739 struct diffstat_t {
 740         int nr;
 741         int alloc;
 742         struct diffstat_file {
 743                 char *from_name;
 744                 char *name;
 745                 char *print_name;
 746                 unsigned is_unmerged:1;
 747                 unsigned is_binary:1;
 748                 unsigned is_renamed:1;
 749                 unsigned int added, deleted;
 750         } **files;
 751 };
 752
 753 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 754                                           const char *name_a,
 755                                           const char *name_b)
 756 {
 757         struct diffstat_file *x;
 758         x = xcalloc(sizeof (*x), 1);
 759         if (diffstat->nr == diffstat->alloc) {
 760                 diffstat->alloc = alloc_nr(diffstat->alloc);
 761                 diffstat->files = xrealloc(diffstat->files,
 762                                 diffstat->alloc * sizeof(x));
 763         }
 764         diffstat->files[diffstat->nr++] = x;
 765         if (name_b) {
 766                 x->from_name = xstrdup(name_a);
 767                 x->name = xstrdup(name_b);
 768                 x->is_renamed = 1;
 769         }
 770         else {
 771                 x->from_name = NULL;
 772                 x->name = xstrdup(name_a);
 773         }
 774         return x;
 775 }
 776
 777 static void diffstat_consume(void *priv, char *line, unsigned long len)
 778 {
 779         struct diffstat_t *diffstat = priv;
 780         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 781
 782         if (line[0] == '+')
 783                 x->added++;
 784         else if (line[0] == '-')
 785                 x->deleted++;
 786 }
 787
 788 const char mime_boundary_leader[] = "------------";
 789
 790 static int scale_linear(int it, int width, int max_change)
 791 {
 792         /*
 793          * make sure that at least one '-' is printed if there were deletions,
 794          * and likewise for '+'.
 795          */
 796         if (max_change < 2)
 797                 return it;
 798         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
 799 }
 800
 801 static void show_name(FILE *file,
 802                       const char *prefix, const char *name, int len,
 803                       const char *reset, const char *set)
 804 {
 805         fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
 806 }
 807
 808 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
 809 {
 810         if (cnt <= 0)
 811                 return;
 812         fprintf(file, "%s", set);
 813         while (cnt--)
 814                 putc(ch, file);
 815         fprintf(file, "%s", reset);
 816 }
 817
 818 static void fill_print_name(struct diffstat_file *file)
 819 {
 820         char *pname;
 821
 822         if (file->print_name)
 823                 return;
 824
 825         if (!file->is_renamed) {
 826                 struct strbuf buf;
 827                 strbuf_init(&buf, 0);
 828                 if (quote_c_style(file->name, &buf, NULL, 0)) {
 829                         pname = strbuf_detach(&buf, NULL);
 830                 } else {
 831                         pname = file->name;
 832                         strbuf_release(&buf);
 833                 }
 834         } else {
 835                 pname = pprint_rename(file->from_name, file->name);
 836         }
 837         file->print_name = pname;
 838 }
 839
 840 static void show_stats(struct diffstat_t* data, struct diff_options *options)
 841 {
 842         int i, len, add, del, total, adds = 0, dels = 0;
 843         int max_change = 0, max_len = 0;
 844         int total_files = data->nr;
 845         int width, name_width;
 846         const char *reset, *set, *add_c, *del_c;
 847
 848         if (data->nr == 0)
 849                 return;
 850
 851         width = options->stat_width ? options->stat_width : 80;
 852         name_width = options->stat_name_width ? options->stat_name_width : 50;
 853
 854         /* Sanity: give at least 5 columns to the graph,
 855          * but leave at least 10 columns for the name.
 856          */
 857         if (width < 25)
 858                 width = 25;
 859         if (name_width < 10)
 860                 name_width = 10;
 861         else if (width < name_width + 15)
 862                 name_width = width - 15;
 863
 864         /* Find the longest filename and max number of changes */
 865         reset = diff_get_color_opt(options, DIFF_RESET);
 866         set   = diff_get_color_opt(options, DIFF_PLAIN);
 867         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
 868         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
 869
 870         for (i = 0; i < data->nr; i++) {
 871                 struct diffstat_file *file = data->files[i];
 872                 int change = file->added + file->deleted;
 873                 fill_print_name(file);
 874                 len = strlen(file->print_name);
 875                 if (max_len < len)
 876                         max_len = len;
 877
 878                 if (file->is_binary || file->is_unmerged)
 879                         continue;
 880                 if (max_change < change)
 881                         max_change = change;
 882         }
 883
 884         /* Compute the width of the graph part;
 885          * 10 is for one blank at the beginning of the line plus
 886          * " | count " between the name and the graph.
 887          *
 888          * From here on, name_width is the width of the name area,
 889          * and width is the width of the graph area.
 890          */
 891         name_width = (name_width < max_len) ? name_width : max_len;
 892         if (width < (name_width + 10) + max_change)
 893                 width = width - (name_width + 10);
 894         else
 895                 width = max_change;
 896
 897         for (i = 0; i < data->nr; i++) {
 898                 const char *prefix = "";
 899                 char *name = data->files[i]->print_name;
 900                 int added = data->files[i]->added;
 901                 int deleted = data->files[i]->deleted;
 902                 int name_len;
 903
 904                 /*
 905                  * "scale" the filename
 906                  */
 907                 len = name_width;
 908                 name_len = strlen(name);
 909                 if (name_width < name_len) {
 910                         char *slash;
 911                         prefix = "...";
 912                         len -= 3;
 913                         name += name_len - len;
 914                         slash = strchr(name, '/');
 915                         if (slash)
 916                                 name = slash;
 917                 }
 918
 919                 if (data->files[i]->is_binary) {
 920                         show_name(options->file, prefix, name, len, reset, set);
 921                         fprintf(options->file, "  Bin ");
 922                         fprintf(options->file, "%s%d%s", del_c, deleted, reset);
 923                         fprintf(options->file, " -> ");
 924                         fprintf(options->file, "%s%d%s", add_c, added, reset);
 925                         fprintf(options->file, " bytes");
 926                         fprintf(options->file, "\n");
 927                         continue;
 928                 }
 929                 else if (data->files[i]->is_unmerged) {
 930                         show_name(options->file, prefix, name, len, reset, set);
 931                         fprintf(options->file, "  Unmerged\n");
 932                         continue;
 933                 }
 934                 else if (!data->files[i]->is_renamed &&
 935                          (added + deleted == 0)) {
 936                         total_files--;
 937                         continue;
 938                 }
 939
 940                 /*
 941                  * scale the add/delete
 942                  */
 943                 add = added;
 944                 del = deleted;
 945                 total = add + del;
 946                 adds += add;
 947                 dels += del;
 948
 949                 if (width <= max_change) {
 950                         add = scale_linear(add, width, max_change);
 951                         del = scale_linear(del, width, max_change);
 952                         total = add + del;
 953                 }
 954                 show_name(options->file, prefix, name, len, reset, set);
 955                 fprintf(options->file, "%5d%s", added + deleted,
 956                                 added + deleted ? " " : "");
 957                 show_graph(options->file, '+', add, add_c, reset);
 958                 show_graph(options->file, '-', del, del_c, reset);
 959                 fprintf(options->file, "\n");
 960         }
 961         fprintf(options->file,
 962                "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
 963                set, total_files, adds, dels, reset);
 964 }
 965
 966 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
 967 {
 968         int i, adds = 0, dels = 0, total_files = data->nr;
 969
 970         if (data->nr == 0)
 971                 return;
 972
 973         for (i = 0; i < data->nr; i++) {
 974                 if (!data->files[i]->is_binary &&
 975                     !data->files[i]->is_unmerged) {
 976                         int added = data->files[i]->added;
 977                         int deleted= data->files[i]->deleted;
 978                         if (!data->files[i]->is_renamed &&
 979                             (added + deleted == 0)) {
 980                                 total_files--;
 981                         } else {
 982                                 adds += added;
 983                                 dels += deleted;
 984                         }
 985                 }
 986         }
 987         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
 988                total_files, adds, dels);
 989 }
 990
 991 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
 992 {
 993         int i;
 994
 995         if (data->nr == 0)
 996                 return;
 997
 998         for (i = 0; i < data->nr; i++) {
 999                 struct diffstat_file *file = data->files[i];
1000
1001                 if (file->is_binary)
1002                         fprintf(options->file, "-\t-\t");
1003                 else
1004                         fprintf(options->file,
1005                                 "%d\t%d\t", file->added, file->deleted);
1006                 if (options->line_termination) {
1007                         fill_print_name(file);
1008                         if (!file->is_renamed)
1009                                 write_name_quoted(file->name, options->file,
1010                                                   options->line_termination);
1011                         else {
1012                                 fputs(file->print_name, options->file);
1013                                 putc(options->line_termination, options->file);
1014                         }
1015                 } else {
1016                         if (file->is_renamed) {
1017                                 putc('\0', options->file);
1018                                 write_name_quoted(file->from_name, options->file, '\0');
1019                         }
1020                         write_name_quoted(file->name, options->file, '\0');
1021                 }
1022         }
1023 }
1024
1025 struct dirstat_file {
1026         const char *name;
1027         unsigned long changed;
1028 };
1029
1030 struct dirstat_dir {
1031         struct dirstat_file *files;
1032         int alloc, nr, percent, cumulative;
1033 };
1034
1035 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1036 {
1037         unsigned long this_dir = 0;
1038         unsigned int sources = 0;
1039
1040         while (dir->nr) {
1041                 struct dirstat_file *f = dir->files;
1042                 int namelen = strlen(f->name);
1043                 unsigned long this;
1044                 char *slash;
1045
1046                 if (namelen < baselen)
1047                         break;
1048                 if (memcmp(f->name, base, baselen))
1049                         break;
1050                 slash = strchr(f->name + baselen, '/');
1051                 if (slash) {
1052                         int newbaselen = slash + 1 - f->name;
1053                         this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1054                         sources++;
1055                 } else {
1056                         this = f->changed;
1057                         dir->files++;
1058                         dir->nr--;
1059                         sources += 2;
1060                 }
1061                 this_dir += this;
1062         }
1063
1064         /*
1065          * We don't report dirstat's for
1066          *  - the top level
1067          *  - or cases where everything came from a single directory
1068          *    under this directory (sources == 1).
1069          */
1070         if (baselen && sources != 1) {
1071                 int permille = this_dir * 1000 / changed;
1072                 if (permille) {
1073                         int percent = permille / 10;
1074                         if (percent >= dir->percent) {
1075                                 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1076                                 if (!dir->cumulative)
1077                                         return 0;
1078                         }
1079                 }
1080         }
1081         return this_dir;
1082 }
1083
1084 static int dirstat_compare(const void *_a, const void *_b)
1085 {
1086         const struct dirstat_file *a = _a;
1087         const struct dirstat_file *b = _b;
1088         return strcmp(a->name, b->name);
1089 }
1090
1091 static void show_dirstat(struct diff_options *options)
1092 {
1093         int i;
1094         unsigned long changed;
1095         struct dirstat_dir dir;
1096         struct diff_queue_struct *q = &diff_queued_diff;
1097
1098         dir.files = NULL;
1099         dir.alloc = 0;
1100         dir.nr = 0;
1101         dir.percent = options->dirstat_percent;
1102         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1103
1104         changed = 0;
1105         for (i = 0; i < q->nr; i++) {
1106                 struct diff_filepair *p = q->queue[i];
1107                 const char *name;
1108                 unsigned long copied, added, damage;
1109
1110                 name = p->one->path ? p->one->path : p->two->path;
1111
1112                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1113                         diff_populate_filespec(p->one, 0);
1114                         diff_populate_filespec(p->two, 0);
1115                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1116                                                &copied, &added);
1117                         diff_free_filespec_data(p->one);
1118                         diff_free_filespec_data(p->two);
1119                 } else if (DIFF_FILE_VALID(p->one)) {
1120                         diff_populate_filespec(p->one, 1);
1121                         copied = added = 0;
1122                         diff_free_filespec_data(p->one);
1123                 } else if (DIFF_FILE_VALID(p->two)) {
1124                         diff_populate_filespec(p->two, 1);
1125                         copied = 0;
1126                         added = p->two->size;
1127                         diff_free_filespec_data(p->two);
1128                 } else
1129                         continue;
1130
1131                 /*
1132                  * Original minus copied is the removed material,
1133                  * added is the new material.  They are both damages
1134                  * made to the preimage.
1135                  */
1136                 damage = (p->one->size - copied) + added;
1137
1138                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1139                 dir.files[dir.nr].name = name;
1140                 dir.files[dir.nr].changed = damage;
1141                 changed += damage;
1142                 dir.nr++;
1143         }
1144
1145         /* This can happen even with many files, if everything was renames */
1146         if (!changed)
1147                 return;
1148
1149         /* Show all directories with more than x% of the changes */
1150         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1151         gather_dirstat(options->file, &dir, changed, "", 0);
1152 }
1153
1154 static void free_diffstat_info(struct diffstat_t *diffstat)
1155 {
1156         int i;
1157         for (i = 0; i < diffstat->nr; i++) {
1158                 struct diffstat_file *f = diffstat->files[i];
1159                 if (f->name != f->print_name)
1160                         free(f->print_name);
1161                 free(f->name);
1162                 free(f->from_name);
1163                 free(f);
1164         }
1165         free(diffstat->files);
1166 }
1167
1168 struct checkdiff_t {
1169         const char *filename;
1170         int lineno;
1171         struct diff_options *o;
1172         unsigned ws_rule;
1173         unsigned status;
1174         int trailing_blanks_start;
1175 };
1176
1177 static int is_conflict_marker(const char *line, unsigned long len)
1178 {
1179         char firstchar;
1180         int cnt;
1181
1182         if (len < 8)
1183                 return 0;
1184         firstchar = line[0];
1185         switch (firstchar) {
1186         case '=': case '>': case '<':
1187                 break;
1188         default:
1189                 return 0;
1190         }
1191         for (cnt = 1; cnt < 7; cnt++)
1192                 if (line[cnt] != firstchar)
1193                         return 0;
1194         /* line[0] thru line[6] are same as firstchar */
1195         if (firstchar == '=') {
1196                 /* divider between ours and theirs? */
1197                 if (len != 8 || line[7] != '\n')
1198                         return 0;
1199         } else if (len < 8 || !isspace(line[7])) {
1200                 /* not divider before ours nor after theirs */
1201                 return 0;
1202         }
1203         return 1;
1204 }
1205
1206 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1207 {
1208         struct checkdiff_t *data = priv;
1209         int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1210         const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1211         const char *reset = diff_get_color(color_diff, DIFF_RESET);
1212         const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1213         char *err;
1214
1215         if (line[0] == '+') {
1216                 unsigned bad;
1217                 data->lineno++;
1218                 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1219                         data->trailing_blanks_start = 0;
1220                 else if (!data->trailing_blanks_start)
1221                         data->trailing_blanks_start = data->lineno;
1222                 if (is_conflict_marker(line + 1, len - 1)) {
1223                         data->status |= 1;
1224                         fprintf(data->o->file,
1225                                 "%s:%d: leftover conflict marker\n",
1226                                 data->filename, data->lineno);
1227                 }
1228                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1229                 if (!bad)
1230                         return;
1231                 data->status |= bad;
1232                 err = whitespace_error_string(bad);
1233                 fprintf(data->o->file, "%s:%d: %s.\n",
1234                         data->filename, data->lineno, err);
1235                 free(err);
1236                 emit_line(data->o->file, set, reset, line, 1);
1237                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1238                               data->o->file, set, reset, ws);
1239         } else if (line[0] == ' ') {
1240                 data->lineno++;
1241                 data->trailing_blanks_start = 0;
1242         } else if (line[0] == '@') {
1243                 char *plus = strchr(line, '+');
1244                 if (plus)
1245                         data->lineno = strtol(plus, NULL, 10) - 1;
1246                 else
1247                         die("invalid diff");
1248                 data->trailing_blanks_start = 0;
1249         }
1250 }
1251
1252 static unsigned char *deflate_it(char *data,
1253                                  unsigned long size,
1254                                  unsigned long *result_size)
1255 {
1256         int bound;
1257         unsigned char *deflated;
1258         z_stream stream;
1259
1260         memset(&stream, 0, sizeof(stream));
1261         deflateInit(&stream, zlib_compression_level);
1262         bound = deflateBound(&stream, size);
1263         deflated = xmalloc(bound);
1264         stream.next_out = deflated;
1265         stream.avail_out = bound;
1266
1267         stream.next_in = (unsigned char *)data;
1268         stream.avail_in = size;
1269         while (deflate(&stream, Z_FINISH) == Z_OK)
1270                 ; /* nothing */
1271         deflateEnd(&stream);
1272         *result_size = stream.total_out;
1273         return deflated;
1274 }
1275
1276 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1277 {
1278         void *cp;
1279         void *delta;
1280         void *deflated;
1281         void *data;
1282         unsigned long orig_size;
1283         unsigned long delta_size;
1284         unsigned long deflate_size;
1285         unsigned long data_size;
1286
1287         /* We could do deflated delta, or we could do just deflated two,
1288          * whichever is smaller.
1289          */
1290         delta = NULL;
1291         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1292         if (one->size && two->size) {
1293                 delta = diff_delta(one->ptr, one->size,
1294                                    two->ptr, two->size,
1295                                    &delta_size, deflate_size);
1296                 if (delta) {
1297                         void *to_free = delta;
1298                         orig_size = delta_size;
1299                         delta = deflate_it(delta, delta_size, &delta_size);
1300                         free(to_free);
1301                 }
1302         }
1303
1304         if (delta && delta_size < deflate_size) {
1305                 fprintf(file, "delta %lu\n", orig_size);
1306                 free(deflated);
1307                 data = delta;
1308                 data_size = delta_size;
1309         }
1310         else {
1311                 fprintf(file, "literal %lu\n", two->size);
1312                 free(delta);
1313                 data = deflated;
1314                 data_size = deflate_size;
1315         }
1316
1317         /* emit data encoded in base85 */
1318         cp = data;
1319         while (data_size) {
1320                 int bytes = (52 < data_size) ? 52 : data_size;
1321                 char line[70];
1322                 data_size -= bytes;
1323                 if (bytes <= 26)
1324                         line[0] = bytes + 'A' - 1;
1325                 else
1326                         line[0] = bytes - 26 + 'a' - 1;
1327                 encode_85(line + 1, cp, bytes);
1328                 cp = (char *) cp + bytes;
1329                 fputs(line, file);
1330                 fputc('\n', file);
1331         }
1332         fprintf(file, "\n");
1333         free(data);
1334 }
1335
1336 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1337 {
1338         fprintf(file, "GIT binary patch\n");
1339         emit_binary_diff_body(file, one, two);
1340         emit_binary_diff_body(file, two, one);
1341 }
1342
1343 static void setup_diff_attr_check(struct git_attr_check *check)
1344 {
1345         static struct git_attr *attr_diff;
1346
1347         if (!attr_diff) {
1348                 attr_diff = git_attr("diff", 4);
1349         }
1350         check[0].attr = attr_diff;
1351 }
1352
1353 static void diff_filespec_check_attr(struct diff_filespec *one)
1354 {
1355         struct git_attr_check attr_diff_check;
1356         int check_from_data = 0;
1357
1358         if (one->checked_attr)
1359                 return;
1360
1361         setup_diff_attr_check(&attr_diff_check);
1362         one->is_binary = 0;
1363         one->funcname_pattern_ident = NULL;
1364
1365         if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1366                 const char *value;
1367
1368                 /* binaryness */
1369                 value = attr_diff_check.value;
1370                 if (ATTR_TRUE(value))
1371                         ;
1372                 else if (ATTR_FALSE(value))
1373                         one->is_binary = 1;
1374                 else
1375                         check_from_data = 1;
1376
1377                 /* funcname pattern ident */
1378                 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1379                         ;
1380                 else
1381                         one->funcname_pattern_ident = value;
1382         }
1383
1384         if (check_from_data) {
1385                 if (!one->data && DIFF_FILE_VALID(one))
1386                         diff_populate_filespec(one, 0);
1387
1388                 if (one->data)
1389                         one->is_binary = buffer_is_binary(one->data, one->size);
1390         }
1391 }
1392
1393 int diff_filespec_is_binary(struct diff_filespec *one)
1394 {
1395         diff_filespec_check_attr(one);
1396         return one->is_binary;
1397 }
1398
1399 static const char *funcname_pattern(const char *ident)
1400 {
1401         struct funcname_pattern *pp;
1402
1403         for (pp = funcname_pattern_list; pp; pp = pp->next)
1404                 if (!strcmp(ident, pp->name))
1405                         return pp->pattern;
1406         return NULL;
1407 }
1408
1409 static struct builtin_funcname_pattern {
1410         const char *name;
1411         const char *pattern;
1412 } builtin_funcname_pattern[] = {
1413         { "bibtex", "\\(@[a-zA-Z]\\{1,\\}[ \t]*{\\{0,1\\}[ \t]*[^ \t\"@',\\#}{~%]*\\).*$" },
1414         { "html", "^\\s*\\(<[Hh][1-6]\\s.*>.*\\)$" },
1415         { "java", "!^[  ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1416                         "new\\|return\\|switch\\|throw\\|while\\)\n"
1417                         "^[     ]*\\(\\([       ]*"
1418                         "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1419                         "[      ]*([^;]*\\)$" },
1420         { "pascal", "^\\(\\(procedure\\|function\\|constructor\\|"
1421                         "destructor\\|interface\\|implementation\\|"
1422                         "initialization\\|finalization\\)[ \t]*.*\\)$"
1423                         "\\|"
1424                         "^\\(.*=[ \t]*\\(class\\|record\\).*\\)$"
1425                         },
1426         { "php", "^[\t ]*\\(\\(function\\|class\\).*\\)" },
1427         { "python", "^\\s*\\(\\(class\\|def\\)\\s.*\\)$" },
1428         { "ruby", "^\\s*\\(\\(class\\|module\\|def\\)\\s.*\\)$" },
1429         { "tex", "^\\(\\\\\\(\\(sub\\)*section\\|chapter\\|part\\)\\*\\{0,1\\}{.*\\)$" },
1430 };
1431
1432 static const char *diff_funcname_pattern(struct diff_filespec *one)
1433 {
1434         const char *ident, *pattern;
1435         int i;
1436
1437         diff_filespec_check_attr(one);
1438         ident = one->funcname_pattern_ident;
1439
1440         if (!ident)
1441                 /*
1442                  * If the config file has "funcname.default" defined, that
1443                  * regexp is used; otherwise NULL is returned and xemit uses
1444                  * the built-in default.
1445                  */
1446                 return funcname_pattern("default");
1447
1448         /* Look up custom "funcname.$ident" regexp from config. */
1449         pattern = funcname_pattern(ident);
1450         if (pattern)
1451                 return pattern;
1452
1453         /*
1454          * And define built-in fallback patterns here.  Note that
1455          * these can be overridden by the user's config settings.
1456          */
1457         for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1458                 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1459                         return builtin_funcname_pattern[i].pattern;
1460
1461         return NULL;
1462 }
1463
1464 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1465 {
1466         if (!options->a_prefix)
1467                 options->a_prefix = a;
1468         if (!options->b_prefix)
1469                 options->b_prefix = b;
1470 }
1471
1472 static void builtin_diff(const char *name_a,
1473                          const char *name_b,
1474                          struct diff_filespec *one,
1475                          struct diff_filespec *two,
1476                          const char *xfrm_msg,
1477                          struct diff_options *o,
1478                          int complete_rewrite)
1479 {
1480         mmfile_t mf1, mf2;
1481         const char *lbl[2];
1482         char *a_one, *b_two;
1483         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1484         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1485         const char *a_prefix, *b_prefix;
1486
1487         diff_set_mnemonic_prefix(o, "a/", "b/");
1488         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1489                 a_prefix = o->b_prefix;
1490                 b_prefix = o->a_prefix;
1491         } else {
1492                 a_prefix = o->a_prefix;
1493                 b_prefix = o->b_prefix;
1494         }
1495
1496         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1497         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1498         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1499         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1500         fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1501         if (lbl[0][0] == '/') {
1502                 /* /dev/null */
1503                 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1504                 if (xfrm_msg && xfrm_msg[0])
1505                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1506         }
1507         else if (lbl[1][0] == '/') {
1508                 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1509                 if (xfrm_msg && xfrm_msg[0])
1510                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1511         }
1512         else {
1513                 if (one->mode != two->mode) {
1514                         fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1515                         fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1516                 }
1517                 if (xfrm_msg && xfrm_msg[0])
1518                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1519                 /*
1520                  * we do not run diff between different kind
1521                  * of objects.
1522                  */
1523                 if ((one->mode ^ two->mode) & S_IFMT)
1524                         goto free_ab_and_return;
1525                 if (complete_rewrite) {
1526                         emit_rewrite_diff(name_a, name_b, one, two, o);
1527                         o->found_changes = 1;
1528                         goto free_ab_and_return;
1529                 }
1530         }
1531
1532         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1533                 die("unable to read files to diff");
1534
1535         if (!DIFF_OPT_TST(o, TEXT) &&
1536             (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1537                 /* Quite common confusing case */
1538                 if (mf1.size == mf2.size &&
1539                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1540                         goto free_ab_and_return;
1541                 if (DIFF_OPT_TST(o, BINARY))
1542                         emit_binary_diff(o->file, &mf1, &mf2);
1543                 else
1544                         fprintf(o->file, "Binary files %s and %s differ\n",
1545                                 lbl[0], lbl[1]);
1546                 o->found_changes = 1;
1547         }
1548         else {
1549                 /* Crazy xdl interfaces.. */
1550                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1551                 xpparam_t xpp;
1552                 xdemitconf_t xecfg;
1553                 xdemitcb_t ecb;
1554                 struct emit_callback ecbdata;
1555                 const char *funcname_pattern;
1556
1557                 funcname_pattern = diff_funcname_pattern(one);
1558                 if (!funcname_pattern)
1559                         funcname_pattern = diff_funcname_pattern(two);
1560
1561                 memset(&xecfg, 0, sizeof(xecfg));
1562                 memset(&ecbdata, 0, sizeof(ecbdata));
1563                 ecbdata.label_path = lbl;
1564                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1565                 ecbdata.found_changesp = &o->found_changes;
1566                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1567                 ecbdata.file = o->file;
1568                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1569                 xecfg.ctxlen = o->context;
1570                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1571                 if (funcname_pattern)
1572                         xdiff_set_find_func(&xecfg, funcname_pattern);
1573                 if (!diffopts)
1574                         ;
1575                 else if (!prefixcmp(diffopts, "--unified="))
1576                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1577                 else if (!prefixcmp(diffopts, "-u"))
1578                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1579                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1580                         ecbdata.diff_words =
1581                                 xcalloc(1, sizeof(struct diff_words_data));
1582                         ecbdata.diff_words->file = o->file;
1583                 }
1584                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1585                               &xpp, &xecfg, &ecb);
1586                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1587                         free_diff_words_data(&ecbdata);
1588         }
1589
1590  free_ab_and_return:
1591         diff_free_filespec_data(one);
1592         diff_free_filespec_data(two);
1593         free(a_one);
1594         free(b_two);
1595         return;
1596 }
1597
1598 static void builtin_diffstat(const char *name_a, const char *name_b,
1599                              struct diff_filespec *one,
1600                              struct diff_filespec *two,
1601                              struct diffstat_t *diffstat,
1602                              struct diff_options *o,
1603                              int complete_rewrite)
1604 {
1605         mmfile_t mf1, mf2;
1606         struct diffstat_file *data;
1607
1608         data = diffstat_add(diffstat, name_a, name_b);
1609
1610         if (!one || !two) {
1611                 data->is_unmerged = 1;
1612                 return;
1613         }
1614         if (complete_rewrite) {
1615                 diff_populate_filespec(one, 0);
1616                 diff_populate_filespec(two, 0);
1617                 data->deleted = count_lines(one->data, one->size);
1618                 data->added = count_lines(two->data, two->size);
1619                 goto free_and_return;
1620         }
1621         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1622                 die("unable to read files to diff");
1623
1624         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1625                 data->is_binary = 1;
1626                 data->added = mf2.size;
1627                 data->deleted = mf1.size;
1628         } else {
1629                 /* Crazy xdl interfaces.. */
1630                 xpparam_t xpp;
1631                 xdemitconf_t xecfg;
1632                 xdemitcb_t ecb;
1633
1634                 memset(&xecfg, 0, sizeof(xecfg));
1635                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1636                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1637                               &xpp, &xecfg, &ecb);
1638         }
1639
1640  free_and_return:
1641         diff_free_filespec_data(one);
1642         diff_free_filespec_data(two);
1643 }
1644
1645 static void builtin_checkdiff(const char *name_a, const char *name_b,
1646                               const char *attr_path,
1647                               struct diff_filespec *one,
1648                               struct diff_filespec *two,
1649                               struct diff_options *o)
1650 {
1651         mmfile_t mf1, mf2;
1652         struct checkdiff_t data;
1653
1654         if (!two)
1655                 return;
1656
1657         memset(&data, 0, sizeof(data));
1658         data.filename = name_b ? name_b : name_a;
1659         data.lineno = 0;
1660         data.o = o;
1661         data.ws_rule = whitespace_rule(attr_path);
1662
1663         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1664                 die("unable to read files to diff");
1665
1666         /*
1667          * All the other codepaths check both sides, but not checking
1668          * the "old" side here is deliberate.  We are checking the newly
1669          * introduced changes, and as long as the "new" side is text, we
1670          * can and should check what it introduces.
1671          */
1672         if (diff_filespec_is_binary(two))
1673                 goto free_and_return;
1674         else {
1675                 /* Crazy xdl interfaces.. */
1676                 xpparam_t xpp;
1677                 xdemitconf_t xecfg;
1678                 xdemitcb_t ecb;
1679
1680                 memset(&xecfg, 0, sizeof(xecfg));
1681                 xecfg.ctxlen = 1; /* at least one context line */
1682                 xpp.flags = XDF_NEED_MINIMAL;
1683                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1684                               &xpp, &xecfg, &ecb);
1685
1686                 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1687                     data.trailing_blanks_start) {
1688                         fprintf(o->file, "%s:%d: ends with blank lines.\n",
1689                                 data.filename, data.trailing_blanks_start);
1690                         data.status = 1; /* report errors */
1691                 }
1692         }
1693  free_and_return:
1694         diff_free_filespec_data(one);
1695         diff_free_filespec_data(two);
1696         if (data.status)
1697                 DIFF_OPT_SET(o, CHECK_FAILED);
1698 }
1699
1700 struct diff_filespec *alloc_filespec(const char *path)
1701 {
1702         int namelen = strlen(path);
1703         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1704
1705         memset(spec, 0, sizeof(*spec));
1706         spec->path = (char *)(spec + 1);
1707         memcpy(spec->path, path, namelen+1);
1708         spec->count = 1;
1709         return spec;
1710 }
1711
1712 void free_filespec(struct diff_filespec *spec)
1713 {
1714         if (!--spec->count) {
1715                 diff_free_filespec_data(spec);
1716                 free(spec);
1717         }
1718 }
1719
1720 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1721                    unsigned short mode)
1722 {
1723         if (mode) {
1724                 spec->mode = canon_mode(mode);
1725                 hashcpy(spec->sha1, sha1);
1726                 spec->sha1_valid = !is_null_sha1(sha1);
1727         }
1728 }
1729
1730 /*
1731  * Given a name and sha1 pair, if the index tells us the file in
1732  * the work tree has that object contents, return true, so that
1733  * prepare_temp_file() does not have to inflate and extract.
1734  */
1735 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1736 {
1737         struct cache_entry *ce;
1738         struct stat st;
1739         int pos, len;
1740
1741         /* We do not read the cache ourselves here, because the
1742          * benchmark with my previous version that always reads cache
1743          * shows that it makes things worse for diff-tree comparing
1744          * two linux-2.6 kernel trees in an already checked out work
1745          * tree.  This is because most diff-tree comparisons deal with
1746          * only a small number of files, while reading the cache is
1747          * expensive for a large project, and its cost outweighs the
1748          * savings we get by not inflating the object to a temporary
1749          * file.  Practically, this code only helps when we are used
1750          * by diff-cache --cached, which does read the cache before
1751          * calling us.
1752          */
1753         if (!active_cache)
1754                 return 0;
1755
1756         /* We want to avoid the working directory if our caller
1757          * doesn't need the data in a normal file, this system
1758          * is rather slow with its stat/open/mmap/close syscalls,
1759          * and the object is contained in a pack file.  The pack
1760          * is probably already open and will be faster to obtain
1761          * the data through than the working directory.  Loose
1762          * objects however would tend to be slower as they need
1763          * to be individually opened and inflated.
1764          */
1765         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1766                 return 0;
1767
1768         len = strlen(name);
1769         pos = cache_name_pos(name, len);
1770         if (pos < 0)
1771                 return 0;
1772         ce = active_cache[pos];
1773
1774         /*
1775          * This is not the sha1 we are looking for, or
1776          * unreusable because it is not a regular file.
1777          */
1778         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1779                 return 0;
1780
1781         /*
1782          * If ce matches the file in the work tree, we can reuse it.
1783          */
1784         if (ce_uptodate(ce) ||
1785             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1786                 return 1;
1787
1788         return 0;
1789 }
1790
1791 static int populate_from_stdin(struct diff_filespec *s)
1792 {
1793         struct strbuf buf;
1794         size_t size = 0;
1795
1796         strbuf_init(&buf, 0);
1797         if (strbuf_read(&buf, 0, 0) < 0)
1798                 return error("error while reading from stdin %s",
1799                                      strerror(errno));
1800
1801         s->should_munmap = 0;
1802         s->data = strbuf_detach(&buf, &size);
1803         s->size = size;
1804         s->should_free = 1;
1805         return 0;
1806 }
1807
1808 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1809 {
1810         int len;
1811         char *data = xmalloc(100);
1812         len = snprintf(data, 100,
1813                 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1814         s->data = data;
1815         s->size = len;
1816         s->should_free = 1;
1817         if (size_only) {
1818                 s->data = NULL;
1819                 free(data);
1820         }
1821         return 0;
1822 }
1823
1824 /*
1825  * While doing rename detection and pickaxe operation, we may need to
1826  * grab the data for the blob (or file) for our own in-core comparison.
1827  * diff_filespec has data and size fields for this purpose.
1828  */
1829 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1830 {
1831         int err = 0;
1832         if (!DIFF_FILE_VALID(s))
1833                 die("internal error: asking to populate invalid file.");
1834         if (S_ISDIR(s->mode))
1835                 return -1;
1836
1837         if (s->data)
1838                 return 0;
1839
1840         if (size_only && 0 < s->size)
1841                 return 0;
1842
1843         if (S_ISGITLINK(s->mode))
1844                 return diff_populate_gitlink(s, size_only);
1845
1846         if (!s->sha1_valid ||
1847             reuse_worktree_file(s->path, s->sha1, 0)) {
1848                 struct strbuf buf;
1849                 struct stat st;
1850                 int fd;
1851
1852                 if (!strcmp(s->path, "-"))
1853                         return populate_from_stdin(s);
1854
1855                 if (lstat(s->path, &st) < 0) {
1856                         if (errno == ENOENT) {
1857                         err_empty:
1858                                 err = -1;
1859                         empty:
1860                                 s->data = (char *)"";
1861                                 s->size = 0;
1862                                 return err;
1863                         }
1864                 }
1865                 s->size = xsize_t(st.st_size);
1866                 if (!s->size)
1867                         goto empty;
1868                 if (size_only)
1869                         return 0;
1870                 if (S_ISLNK(st.st_mode)) {
1871                         int ret;
1872                         s->data = xmalloc(s->size);
1873                         s->should_free = 1;
1874                         ret = readlink(s->path, s->data, s->size);
1875                         if (ret < 0) {
1876                                 free(s->data);
1877                                 goto err_empty;
1878                         }
1879                         return 0;
1880                 }
1881                 fd = open(s->path, O_RDONLY);
1882                 if (fd < 0)
1883                         goto err_empty;
1884                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1885                 close(fd);
1886                 s->should_munmap = 1;
1887
1888                 /*
1889                  * Convert from working tree format to canonical git format
1890                  */
1891                 strbuf_init(&buf, 0);
1892                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1893                         size_t size = 0;
1894                         munmap(s->data, s->size);
1895                         s->should_munmap = 0;
1896                         s->data = strbuf_detach(&buf, &size);
1897                         s->size = size;
1898                         s->should_free = 1;
1899                 }
1900         }
1901         else {
1902                 enum object_type type;
1903                 if (size_only)
1904                         type = sha1_object_info(s->sha1, &s->size);
1905                 else {
1906                         s->data = read_sha1_file(s->sha1, &type, &s->size);
1907                         s->should_free = 1;
1908                 }
1909         }
1910         return 0;
1911 }
1912
1913 void diff_free_filespec_blob(struct diff_filespec *s)
1914 {
1915         if (s->should_free)
1916                 free(s->data);
1917         else if (s->should_munmap)
1918                 munmap(s->data, s->size);
1919
1920         if (s->should_free || s->should_munmap) {
1921                 s->should_free = s->should_munmap = 0;
1922                 s->data = NULL;
1923         }
1924 }
1925
1926 void diff_free_filespec_data(struct diff_filespec *s)
1927 {
1928         diff_free_filespec_blob(s);
1929         free(s->cnt_data);
1930         s->cnt_data = NULL;
1931 }
1932
1933 static void prep_temp_blob(struct diff_tempfile *temp,
1934                            void *blob,
1935                            unsigned long size,
1936                            const unsigned char *sha1,
1937                            int mode)
1938 {
1939         int fd;
1940
1941         fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1942         if (fd < 0)
1943                 die("unable to create temp-file: %s", strerror(errno));
1944         if (write_in_full(fd, blob, size) != size)
1945                 die("unable to write temp-file");
1946         close(fd);
1947         temp->name = temp->tmp_path;
1948         strcpy(temp->hex, sha1_to_hex(sha1));
1949         temp->hex[40] = 0;
1950         sprintf(temp->mode, "%06o", mode);
1951 }
1952
1953 static void prepare_temp_file(const char *name,
1954                               struct diff_tempfile *temp,
1955                               struct diff_filespec *one)
1956 {
1957         if (!DIFF_FILE_VALID(one)) {
1958         not_a_valid_file:
1959                 /* A '-' entry produces this for file-2, and
1960                  * a '+' entry produces this for file-1.
1961                  */
1962                 temp->name = "/dev/null";
1963                 strcpy(temp->hex, ".");
1964                 strcpy(temp->mode, ".");
1965                 return;
1966         }
1967
1968         if (!one->sha1_valid ||
1969             reuse_worktree_file(name, one->sha1, 1)) {
1970                 struct stat st;
1971                 if (lstat(name, &st) < 0) {
1972                         if (errno == ENOENT)
1973                                 goto not_a_valid_file;
1974                         die("stat(%s): %s", name, strerror(errno));
1975                 }
1976                 if (S_ISLNK(st.st_mode)) {
1977                         int ret;
1978                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1979                         size_t sz = xsize_t(st.st_size);
1980                         if (sizeof(buf) <= st.st_size)
1981                                 die("symlink too long: %s", name);
1982                         ret = readlink(name, buf, sz);
1983                         if (ret < 0)
1984                                 die("readlink(%s)", name);
1985                         prep_temp_blob(temp, buf, sz,
1986                                        (one->sha1_valid ?
1987                                         one->sha1 : null_sha1),
1988                                        (one->sha1_valid ?
1989                                         one->mode : S_IFLNK));
1990                 }
1991                 else {
1992                         /* we can borrow from the file in the work tree */
1993                         temp->name = name;
1994                         if (!one->sha1_valid)
1995                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
1996                         else
1997                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
1998                         /* Even though we may sometimes borrow the
1999                          * contents from the work tree, we always want
2000                          * one->mode.  mode is trustworthy even when
2001                          * !(one->sha1_valid), as long as
2002                          * DIFF_FILE_VALID(one).
2003                          */
2004                         sprintf(temp->mode, "%06o", one->mode);
2005                 }
2006                 return;
2007         }
2008         else {
2009                 if (diff_populate_filespec(one, 0))
2010                         die("cannot read data blob for %s", one->path);
2011                 prep_temp_blob(temp, one->data, one->size,
2012                                one->sha1, one->mode);
2013         }
2014 }
2015
2016 static void remove_tempfile(void)
2017 {
2018         int i;
2019
2020         for (i = 0; i < 2; i++)
2021                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
2022                         unlink(diff_temp[i].name);
2023                         diff_temp[i].name = NULL;
2024                 }
2025 }
2026
2027 static void remove_tempfile_on_signal(int signo)
2028 {
2029         remove_tempfile();
2030         signal(SIGINT, SIG_DFL);
2031         raise(signo);
2032 }
2033
2034 /* An external diff command takes:
2035  *
2036  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2037  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2038  *
2039  */
2040 static void run_external_diff(const char *pgm,
2041                               const char *name,
2042                               const char *other,
2043                               struct diff_filespec *one,
2044                               struct diff_filespec *two,
2045                               const char *xfrm_msg,
2046                               int complete_rewrite)
2047 {
2048         const char *spawn_arg[10];
2049         struct diff_tempfile *temp = diff_temp;
2050         int retval;
2051         static int atexit_asked = 0;
2052         const char *othername;
2053         const char **arg = &spawn_arg[0];
2054
2055         othername = (other? other : name);
2056         if (one && two) {
2057                 prepare_temp_file(name, &temp[0], one);
2058                 prepare_temp_file(othername, &temp[1], two);
2059                 if (! atexit_asked &&
2060                     (temp[0].name == temp[0].tmp_path ||
2061                      temp[1].name == temp[1].tmp_path)) {
2062                         atexit_asked = 1;
2063                         atexit(remove_tempfile);
2064                 }
2065                 signal(SIGINT, remove_tempfile_on_signal);
2066         }
2067
2068         if (one && two) {
2069                 *arg++ = pgm;
2070                 *arg++ = name;
2071                 *arg++ = temp[0].name;
2072                 *arg++ = temp[0].hex;
2073                 *arg++ = temp[0].mode;
2074                 *arg++ = temp[1].name;
2075                 *arg++ = temp[1].hex;
2076                 *arg++ = temp[1].mode;
2077                 if (other) {
2078                         *arg++ = other;
2079                         *arg++ = xfrm_msg;
2080                 }
2081         } else {
2082                 *arg++ = pgm;
2083                 *arg++ = name;
2084         }
2085         *arg = NULL;
2086         fflush(NULL);
2087         retval = run_command_v_opt(spawn_arg, 0);
2088         remove_tempfile();
2089         if (retval) {
2090                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2091                 exit(1);
2092         }
2093 }
2094
2095 static const char *external_diff_attr(const char *name)
2096 {
2097         struct git_attr_check attr_diff_check;
2098
2099         if (!name)
2100                 return NULL;
2101
2102         setup_diff_attr_check(&attr_diff_check);
2103         if (!git_checkattr(name, 1, &attr_diff_check)) {
2104                 const char *value = attr_diff_check.value;
2105                 if (!ATTR_TRUE(value) &&
2106                     !ATTR_FALSE(value) &&
2107                     !ATTR_UNSET(value)) {
2108                         struct ll_diff_driver *drv;
2109
2110                         for (drv = user_diff; drv; drv = drv->next)
2111                                 if (!strcmp(drv->name, value))
2112                                         return drv->cmd;
2113                 }
2114         }
2115         return NULL;
2116 }
2117
2118 static void run_diff_cmd(const char *pgm,
2119                          const char *name,
2120                          const char *other,
2121                          const char *attr_path,
2122                          struct diff_filespec *one,
2123                          struct diff_filespec *two,
2124                          const char *xfrm_msg,
2125                          struct diff_options *o,
2126                          int complete_rewrite)
2127 {
2128         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2129                 pgm = NULL;
2130         else {
2131                 const char *cmd = external_diff_attr(attr_path);
2132                 if (cmd)
2133                         pgm = cmd;
2134         }
2135
2136         if (pgm) {
2137                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2138                                   complete_rewrite);
2139                 return;
2140         }
2141         if (one && two)
2142                 builtin_diff(name, other ? other : name,
2143                              one, two, xfrm_msg, o, complete_rewrite);
2144         else
2145                 fprintf(o->file, "* Unmerged path %s\n", name);
2146 }
2147
2148 static void diff_fill_sha1_info(struct diff_filespec *one)
2149 {
2150         if (DIFF_FILE_VALID(one)) {
2151                 if (!one->sha1_valid) {
2152                         struct stat st;
2153                         if (!strcmp(one->path, "-")) {
2154                                 hashcpy(one->sha1, null_sha1);
2155                                 return;
2156                         }
2157                         if (lstat(one->path, &st) < 0)
2158                                 die("stat %s", one->path);
2159                         if (index_path(one->sha1, one->path, &st, 0))
2160                                 die("cannot hash %s\n", one->path);
2161                 }
2162         }
2163         else
2164                 hashclr(one->sha1);
2165 }
2166
2167 static int similarity_index(struct diff_filepair *p)
2168 {
2169         return p->score * 100 / MAX_SCORE;
2170 }
2171
2172 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2173 {
2174         /* Strip the prefix but do not molest /dev/null and absolute paths */
2175         if (*namep && **namep != '/')
2176                 *namep += prefix_length;
2177         if (*otherp && **otherp != '/')
2178                 *otherp += prefix_length;
2179 }
2180
2181 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2182 {
2183         const char *pgm = external_diff();
2184         struct strbuf msg;
2185         char *xfrm_msg;
2186         struct diff_filespec *one = p->one;
2187         struct diff_filespec *two = p->two;
2188         const char *name;
2189         const char *other;
2190         const char *attr_path;
2191         int complete_rewrite = 0;
2192
2193         name  = p->one->path;
2194         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2195         attr_path = name;
2196         if (o->prefix_length)
2197                 strip_prefix(o->prefix_length, &name, &other);
2198
2199         if (DIFF_PAIR_UNMERGED(p)) {
2200                 run_diff_cmd(pgm, name, NULL, attr_path,
2201                              NULL, NULL, NULL, o, 0);
2202                 return;
2203         }
2204
2205         diff_fill_sha1_info(one);
2206         diff_fill_sha1_info(two);
2207
2208         strbuf_init(&msg, PATH_MAX * 2 + 300);
2209         switch (p->status) {
2210         case DIFF_STATUS_COPIED:
2211                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2212                 strbuf_addstr(&msg, "\ncopy from ");
2213                 quote_c_style(name, &msg, NULL, 0);
2214                 strbuf_addstr(&msg, "\ncopy to ");
2215                 quote_c_style(other, &msg, NULL, 0);
2216                 strbuf_addch(&msg, '\n');
2217                 break;
2218         case DIFF_STATUS_RENAMED:
2219                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2220                 strbuf_addstr(&msg, "\nrename from ");
2221                 quote_c_style(name, &msg, NULL, 0);
2222                 strbuf_addstr(&msg, "\nrename to ");
2223                 quote_c_style(other, &msg, NULL, 0);
2224                 strbuf_addch(&msg, '\n');
2225                 break;
2226         case DIFF_STATUS_MODIFIED:
2227                 if (p->score) {
2228                         strbuf_addf(&msg, "dissimilarity index %d%%\n",
2229                                         similarity_index(p));
2230                         complete_rewrite = 1;
2231                         break;
2232                 }
2233                 /* fallthru */
2234         default:
2235                 /* nothing */
2236                 ;
2237         }
2238
2239         if (hashcmp(one->sha1, two->sha1)) {
2240                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2241
2242                 if (DIFF_OPT_TST(o, BINARY)) {
2243                         mmfile_t mf;
2244                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2245                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2246                                 abbrev = 40;
2247                 }
2248                 strbuf_addf(&msg, "index %.*s..%.*s",
2249                                 abbrev, sha1_to_hex(one->sha1),
2250                                 abbrev, sha1_to_hex(two->sha1));
2251                 if (one->mode == two->mode)
2252                         strbuf_addf(&msg, " %06o", one->mode);
2253                 strbuf_addch(&msg, '\n');
2254         }
2255
2256         if (msg.len)
2257                 strbuf_setlen(&msg, msg.len - 1);
2258         xfrm_msg = msg.len ? msg.buf : NULL;
2259
2260         if (!pgm &&
2261             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2262             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2263                 /* a filepair that changes between file and symlink
2264                  * needs to be split into deletion and creation.
2265                  */
2266                 struct diff_filespec *null = alloc_filespec(two->path);
2267                 run_diff_cmd(NULL, name, other, attr_path,
2268                              one, null, xfrm_msg, o, 0);
2269                 free(null);
2270                 null = alloc_filespec(one->path);
2271                 run_diff_cmd(NULL, name, other, attr_path,
2272                              null, two, xfrm_msg, o, 0);
2273                 free(null);
2274         }
2275         else
2276                 run_diff_cmd(pgm, name, other, attr_path,
2277                              one, two, xfrm_msg, o, complete_rewrite);
2278
2279         strbuf_release(&msg);
2280 }
2281
2282 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2283                          struct diffstat_t *diffstat)
2284 {
2285         const char *name;
2286         const char *other;
2287         int complete_rewrite = 0;
2288
2289         if (DIFF_PAIR_UNMERGED(p)) {
2290                 /* unmerged */
2291                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2292                 return;
2293         }
2294
2295         name = p->one->path;
2296         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2297
2298         if (o->prefix_length)
2299                 strip_prefix(o->prefix_length, &name, &other);
2300
2301         diff_fill_sha1_info(p->one);
2302         diff_fill_sha1_info(p->two);
2303
2304         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2305                 complete_rewrite = 1;
2306         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2307 }
2308
2309 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2310 {
2311         const char *name;
2312         const char *other;
2313         const char *attr_path;
2314
2315         if (DIFF_PAIR_UNMERGED(p)) {
2316                 /* unmerged */
2317                 return;
2318         }
2319
2320         name = p->one->path;
2321         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2322         attr_path = other ? other : name;
2323
2324         if (o->prefix_length)
2325                 strip_prefix(o->prefix_length, &name, &other);
2326
2327         diff_fill_sha1_info(p->one);
2328         diff_fill_sha1_info(p->two);
2329
2330         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2331 }
2332
2333 void diff_setup(struct diff_options *options)
2334 {
2335         memset(options, 0, sizeof(*options));
2336
2337         options->file = stdout;
2338
2339         options->line_termination = '\n';
2340         options->break_opt = -1;
2341         options->rename_limit = -1;
2342         options->dirstat_percent = 3;
2343         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
2344         options->context = 3;
2345
2346         options->change = diff_change;
2347         options->add_remove = diff_addremove;
2348         if (diff_use_color_default > 0)
2349                 DIFF_OPT_SET(options, COLOR_DIFF);
2350         else
2351                 DIFF_OPT_CLR(options, COLOR_DIFF);
2352         options->detect_rename = diff_detect_rename_default;
2353
2354         if (!diff_mnemonic_prefix) {
2355                 options->a_prefix = "a/";
2356                 options->b_prefix = "b/";
2357         }
2358 }
2359
2360 int diff_setup_done(struct diff_options *options)
2361 {
2362         int count = 0;
2363
2364         if (options->output_format & DIFF_FORMAT_NAME)
2365                 count++;
2366         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2367                 count++;
2368         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2369                 count++;
2370         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2371                 count++;
2372         if (count > 1)
2373                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2374
2375         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2376                 options->detect_rename = DIFF_DETECT_COPY;
2377
2378         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2379                 options->prefix = NULL;
2380         if (options->prefix)
2381                 options->prefix_length = strlen(options->prefix);
2382         else
2383                 options->prefix_length = 0;
2384
2385         if (options->output_format & (DIFF_FORMAT_NAME |
2386                                       DIFF_FORMAT_NAME_STATUS |
2387                                       DIFF_FORMAT_CHECKDIFF |
2388                                       DIFF_FORMAT_NO_OUTPUT))
2389                 options->output_format &= ~(DIFF_FORMAT_RAW |
2390                                             DIFF_FORMAT_NUMSTAT |
2391                                             DIFF_FORMAT_DIFFSTAT |
2392                                             DIFF_FORMAT_SHORTSTAT |
2393                                             DIFF_FORMAT_DIRSTAT |
2394                                             DIFF_FORMAT_SUMMARY |
2395                                             DIFF_FORMAT_PATCH);
2396
2397         /*
2398          * These cases always need recursive; we do not drop caller-supplied
2399          * recursive bits for other formats here.
2400          */
2401         if (options->output_format & (DIFF_FORMAT_PATCH |
2402                                       DIFF_FORMAT_NUMSTAT |
2403                                       DIFF_FORMAT_DIFFSTAT |
2404                                       DIFF_FORMAT_SHORTSTAT |
2405                                       DIFF_FORMAT_DIRSTAT |
2406                                       DIFF_FORMAT_SUMMARY |
2407                                       DIFF_FORMAT_CHECKDIFF))
2408                 DIFF_OPT_SET(options, RECURSIVE);
2409         /*
2410          * Also pickaxe would not work very well if you do not say recursive
2411          */
2412         if (options->pickaxe)
2413                 DIFF_OPT_SET(options, RECURSIVE);
2414
2415         if (options->detect_rename && options->rename_limit < 0)
2416                 options->rename_limit = diff_rename_limit_default;
2417         if (options->setup & DIFF_SETUP_USE_CACHE) {
2418                 if (!active_cache)
2419                         /* read-cache does not die even when it fails
2420                          * so it is safe for us to do this here.  Also
2421                          * it does not smudge active_cache or active_nr
2422                          * when it fails, so we do not have to worry about
2423                          * cleaning it up ourselves either.
2424                          */
2425                         read_cache();
2426         }
2427         if (options->abbrev <= 0 || 40 < options->abbrev)
2428                 options->abbrev = 40; /* full */
2429
2430         /*
2431          * It does not make sense to show the first hit we happened
2432          * to have found.  It does not make sense not to return with
2433          * exit code in such a case either.
2434          */
2435         if (DIFF_OPT_TST(options, QUIET)) {
2436                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2437                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2438         }
2439
2440         return 0;
2441 }
2442
2443 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2444 {
2445         char c, *eq;
2446         int len;
2447
2448         if (*arg != '-')
2449                 return 0;
2450         c = *++arg;
2451         if (!c)
2452                 return 0;
2453         if (c == arg_short) {
2454                 c = *++arg;
2455                 if (!c)
2456                         return 1;
2457                 if (val && isdigit(c)) {
2458                         char *end;
2459                         int n = strtoul(arg, &end, 10);
2460                         if (*end)
2461                                 return 0;
2462                         *val = n;
2463                         return 1;
2464                 }
2465                 return 0;
2466         }
2467         if (c != '-')
2468                 return 0;
2469         arg++;
2470         eq = strchr(arg, '=');
2471         if (eq)
2472                 len = eq - arg;
2473         else
2474                 len = strlen(arg);
2475         if (!len || strncmp(arg, arg_long, len))
2476                 return 0;
2477         if (eq) {
2478                 int n;
2479                 char *end;
2480                 if (!isdigit(*++eq))
2481                         return 0;
2482                 n = strtoul(eq, &end, 10);
2483                 if (*end)
2484                         return 0;
2485                 *val = n;
2486         }
2487         return 1;
2488 }
2489
2490 static int diff_scoreopt_parse(const char *opt);
2491
2492 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2493 {
2494         const char *arg = av[0];
2495
2496         /* Output format options */
2497         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2498                 options->output_format |= DIFF_FORMAT_PATCH;
2499         else if (opt_arg(arg, 'U', "unified", &options->context))
2500                 options->output_format |= DIFF_FORMAT_PATCH;
2501         else if (!strcmp(arg, "--raw"))
2502                 options->output_format |= DIFF_FORMAT_RAW;
2503         else if (!strcmp(arg, "--patch-with-raw"))
2504                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2505         else if (!strcmp(arg, "--numstat"))
2506                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2507         else if (!strcmp(arg, "--shortstat"))
2508                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2509         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2510                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2511         else if (!strcmp(arg, "--cumulative")) {
2512                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2513                 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2514         }
2515         else if (!strcmp(arg, "--check"))
2516                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2517         else if (!strcmp(arg, "--summary"))
2518                 options->output_format |= DIFF_FORMAT_SUMMARY;
2519         else if (!strcmp(arg, "--patch-with-stat"))
2520                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2521         else if (!strcmp(arg, "--name-only"))
2522                 options->output_format |= DIFF_FORMAT_NAME;
2523         else if (!strcmp(arg, "--name-status"))
2524                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2525         else if (!strcmp(arg, "-s"))
2526                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2527         else if (!prefixcmp(arg, "--stat")) {
2528                 char *end;
2529                 int width = options->stat_width;
2530                 int name_width = options->stat_name_width;
2531                 arg += 6;
2532                 end = (char *)arg;
2533
2534                 switch (*arg) {
2535                 case '-':
2536                         if (!prefixcmp(arg, "-width="))
2537                                 width = strtoul(arg + 7, &end, 10);
2538                         else if (!prefixcmp(arg, "-name-width="))
2539                                 name_width = strtoul(arg + 12, &end, 10);
2540                         break;
2541                 case '=':
2542                         width = strtoul(arg+1, &end, 10);
2543                         if (*end == ',')
2544                                 name_width = strtoul(end+1, &end, 10);
2545                 }
2546
2547                 /* Important! This checks all the error cases! */
2548                 if (*end)
2549                         return 0;
2550                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2551                 options->stat_name_width = name_width;
2552                 options->stat_width = width;
2553         }
2554
2555         /* renames options */
2556         else if (!prefixcmp(arg, "-B")) {
2557                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2558                         return -1;
2559         }
2560         else if (!prefixcmp(arg, "-M")) {
2561                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2562                         return -1;
2563                 options->detect_rename = DIFF_DETECT_RENAME;
2564         }
2565         else if (!prefixcmp(arg, "-C")) {
2566                 if (options->detect_rename == DIFF_DETECT_COPY)
2567                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2568                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2569                         return -1;
2570                 options->detect_rename = DIFF_DETECT_COPY;
2571         }
2572         else if (!strcmp(arg, "--no-renames"))
2573                 options->detect_rename = 0;
2574         else if (!strcmp(arg, "--relative"))
2575                 DIFF_OPT_SET(options, RELATIVE_NAME);
2576         else if (!prefixcmp(arg, "--relative=")) {
2577                 DIFF_OPT_SET(options, RELATIVE_NAME);
2578                 options->prefix = arg + 11;
2579         }
2580
2581         /* xdiff options */
2582         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2583                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2584         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2585                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2586         else if (!strcmp(arg, "--ignore-space-at-eol"))
2587                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2588
2589         /* flags options */
2590         else if (!strcmp(arg, "--binary")) {
2591                 options->output_format |= DIFF_FORMAT_PATCH;
2592                 DIFF_OPT_SET(options, BINARY);
2593         }
2594         else if (!strcmp(arg, "--full-index"))
2595                 DIFF_OPT_SET(options, FULL_INDEX);
2596         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2597                 DIFF_OPT_SET(options, TEXT);
2598         else if (!strcmp(arg, "-R"))
2599                 DIFF_OPT_SET(options, REVERSE_DIFF);
2600         else if (!strcmp(arg, "--find-copies-harder"))
2601                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2602         else if (!strcmp(arg, "--follow"))
2603                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2604         else if (!strcmp(arg, "--color"))
2605                 DIFF_OPT_SET(options, COLOR_DIFF);
2606         else if (!strcmp(arg, "--no-color"))
2607                 DIFF_OPT_CLR(options, COLOR_DIFF);
2608         else if (!strcmp(arg, "--color-words"))
2609                 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2610         else if (!strcmp(arg, "--exit-code"))
2611                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2612         else if (!strcmp(arg, "--quiet"))
2613                 DIFF_OPT_SET(options, QUIET);
2614         else if (!strcmp(arg, "--ext-diff"))
2615                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2616         else if (!strcmp(arg, "--no-ext-diff"))
2617                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2618         else if (!strcmp(arg, "--ignore-submodules"))
2619                 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2620
2621         /* misc options */
2622         else if (!strcmp(arg, "-z"))
2623                 options->line_termination = 0;
2624         else if (!prefixcmp(arg, "-l"))
2625                 options->rename_limit = strtoul(arg+2, NULL, 10);
2626         else if (!prefixcmp(arg, "-S"))
2627                 options->pickaxe = arg + 2;
2628         else if (!strcmp(arg, "--pickaxe-all"))
2629                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2630         else if (!strcmp(arg, "--pickaxe-regex"))
2631                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2632         else if (!prefixcmp(arg, "-O"))
2633                 options->orderfile = arg + 2;
2634         else if (!prefixcmp(arg, "--diff-filter="))
2635                 options->filter = arg + 14;
2636         else if (!strcmp(arg, "--abbrev"))
2637                 options->abbrev = DEFAULT_ABBREV;
2638         else if (!prefixcmp(arg, "--abbrev=")) {
2639                 options->abbrev = strtoul(arg + 9, NULL, 10);
2640                 if (options->abbrev < MINIMUM_ABBREV)
2641                         options->abbrev = MINIMUM_ABBREV;
2642                 else if (40 < options->abbrev)
2643                         options->abbrev = 40;
2644         }
2645         else if (!prefixcmp(arg, "--src-prefix="))
2646                 options->a_prefix = arg + 13;
2647         else if (!prefixcmp(arg, "--dst-prefix="))
2648                 options->b_prefix = arg + 13;
2649         else if (!strcmp(arg, "--no-prefix"))
2650                 options->a_prefix = options->b_prefix = "";
2651         else if (!prefixcmp(arg, "--output=")) {
2652                 options->file = fopen(arg + strlen("--output="), "w");
2653                 options->close_file = 1;
2654         } else
2655                 return 0;
2656         return 1;
2657 }
2658
2659 static int parse_num(const char **cp_p)
2660 {
2661         unsigned long num, scale;
2662         int ch, dot;
2663         const char *cp = *cp_p;
2664
2665         num = 0;
2666         scale = 1;
2667         dot = 0;
2668         for(;;) {
2669                 ch = *cp;
2670                 if ( !dot && ch == '.' ) {
2671                         scale = 1;
2672                         dot = 1;
2673                 } else if ( ch == '%' ) {
2674                         scale = dot ? scale*100 : 100;
2675                         cp++;   /* % is always at the end */
2676                         break;
2677                 } else if ( ch >= '0' && ch <= '9' ) {
2678                         if ( scale < 100000 ) {
2679                                 scale *= 10;
2680                                 num = (num*10) + (ch-'0');
2681                         }
2682                 } else {
2683                         break;
2684                 }
2685                 cp++;
2686         }
2687         *cp_p = cp;
2688
2689         /* user says num divided by scale and we say internally that
2690          * is MAX_SCORE * num / scale.
2691          */
2692         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2693 }
2694
2695 static int diff_scoreopt_parse(const char *opt)
2696 {
2697         int opt1, opt2, cmd;
2698
2699         if (*opt++ != '-')
2700                 return -1;
2701         cmd = *opt++;
2702         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2703                 return -1; /* that is not a -M, -C nor -B option */
2704
2705         opt1 = parse_num(&opt);
2706         if (cmd != 'B')
2707                 opt2 = 0;
2708         else {
2709                 if (*opt == 0)
2710                         opt2 = 0;
2711                 else if (*opt != '/')
2712                         return -1; /* we expect -B80/99 or -B80 */
2713                 else {
2714                         opt++;
2715                         opt2 = parse_num(&opt);
2716                 }
2717         }
2718         if (*opt != 0)
2719                 return -1;
2720         return opt1 | (opt2 << 16);
2721 }
2722
2723 struct diff_queue_struct diff_queued_diff;
2724
2725 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2726 {
2727         if (queue->alloc <= queue->nr) {
2728                 queue->alloc = alloc_nr(queue->alloc);
2729                 queue->queue = xrealloc(queue->queue,
2730                                         sizeof(dp) * queue->alloc);
2731         }
2732         queue->queue[queue->nr++] = dp;
2733 }
2734
2735 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2736                                  struct diff_filespec *one,
2737                                  struct diff_filespec *two)
2738 {
2739         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2740         dp->one = one;
2741         dp->two = two;
2742         if (queue)
2743                 diff_q(queue, dp);
2744         return dp;
2745 }
2746
2747 void diff_free_filepair(struct diff_filepair *p)
2748 {
2749         free_filespec(p->one);
2750         free_filespec(p->two);
2751         free(p);
2752 }
2753
2754 /* This is different from find_unique_abbrev() in that
2755  * it stuffs the result with dots for alignment.
2756  */
2757 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2758 {
2759         int abblen;
2760         const char *abbrev;
2761         if (len == 40)
2762                 return sha1_to_hex(sha1);
2763
2764         abbrev = find_unique_abbrev(sha1, len);
2765         abblen = strlen(abbrev);
2766         if (abblen < 37) {
2767                 static char hex[41];
2768                 if (len < abblen && abblen <= len + 2)
2769                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2770                 else
2771                         sprintf(hex, "%s...", abbrev);
2772                 return hex;
2773         }
2774         return sha1_to_hex(sha1);
2775 }
2776
2777 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2778 {
2779         int line_termination = opt->line_termination;
2780         int inter_name_termination = line_termination ? '\t' : '\0';
2781
2782         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2783                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2784                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
2785                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2786         }
2787         if (p->score) {
2788                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2789                         inter_name_termination);
2790         } else {
2791                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2792         }
2793
2794         if (p->status == DIFF_STATUS_COPIED ||
2795             p->status == DIFF_STATUS_RENAMED) {
2796                 const char *name_a, *name_b;
2797                 name_a = p->one->path;
2798                 name_b = p->two->path;
2799                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2800                 write_name_quoted(name_a, opt->file, inter_name_termination);
2801                 write_name_quoted(name_b, opt->file, line_termination);
2802         } else {
2803                 const char *name_a, *name_b;
2804                 name_a = p->one->mode ? p->one->path : p->two->path;
2805                 name_b = NULL;
2806                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2807                 write_name_quoted(name_a, opt->file, line_termination);
2808         }
2809 }
2810
2811 int diff_unmodified_pair(struct diff_filepair *p)
2812 {
2813         /* This function is written stricter than necessary to support
2814          * the currently implemented transformers, but the idea is to
2815          * let transformers to produce diff_filepairs any way they want,
2816          * and filter and clean them up here before producing the output.
2817          */
2818         struct diff_filespec *one = p->one, *two = p->two;
2819
2820         if (DIFF_PAIR_UNMERGED(p))
2821                 return 0; /* unmerged is interesting */
2822
2823         /* deletion, addition, mode or type change
2824          * and rename are all interesting.
2825          */
2826         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2827             DIFF_PAIR_MODE_CHANGED(p) ||
2828             strcmp(one->path, two->path))
2829                 return 0;
2830
2831         /* both are valid and point at the same path.  that is, we are
2832          * dealing with a change.
2833          */
2834         if (one->sha1_valid && two->sha1_valid &&
2835             !hashcmp(one->sha1, two->sha1))
2836                 return 1; /* no change */
2837         if (!one->sha1_valid && !two->sha1_valid)
2838                 return 1; /* both look at the same file on the filesystem. */
2839         return 0;
2840 }
2841
2842 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2843 {
2844         if (diff_unmodified_pair(p))
2845                 return;
2846
2847         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2848             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2849                 return; /* no tree diffs in patch format */
2850
2851         run_diff(p, o);
2852 }
2853
2854 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2855                             struct diffstat_t *diffstat)
2856 {
2857         if (diff_unmodified_pair(p))
2858                 return;
2859
2860         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2861             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2862                 return; /* no tree diffs in patch format */
2863
2864         run_diffstat(p, o, diffstat);
2865 }
2866
2867 static void diff_flush_checkdiff(struct diff_filepair *p,
2868                 struct diff_options *o)
2869 {
2870         if (diff_unmodified_pair(p))
2871                 return;
2872
2873         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2874             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2875                 return; /* no tree diffs in patch format */
2876
2877         run_checkdiff(p, o);
2878 }
2879
2880 int diff_queue_is_empty(void)
2881 {
2882         struct diff_queue_struct *q = &diff_queued_diff;
2883         int i;
2884         for (i = 0; i < q->nr; i++)
2885                 if (!diff_unmodified_pair(q->queue[i]))
2886                         return 0;
2887         return 1;
2888 }
2889
2890 #if DIFF_DEBUG
2891 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2892 {
2893         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2894                 x, one ? one : "",
2895                 s->path,
2896                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2897                 s->mode,
2898                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2899         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2900                 x, one ? one : "",
2901                 s->size, s->xfrm_flags);
2902 }
2903
2904 void diff_debug_filepair(const struct diff_filepair *p, int i)
2905 {
2906         diff_debug_filespec(p->one, i, "one");
2907         diff_debug_filespec(p->two, i, "two");
2908         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2909                 p->score, p->status ? p->status : '?',
2910                 p->one->rename_used, p->broken_pair);
2911 }
2912
2913 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2914 {
2915         int i;
2916         if (msg)
2917                 fprintf(stderr, "%s\n", msg);
2918         fprintf(stderr, "q->nr = %d\n", q->nr);
2919         for (i = 0; i < q->nr; i++) {
2920                 struct diff_filepair *p = q->queue[i];
2921                 diff_debug_filepair(p, i);
2922         }
2923 }
2924 #endif
2925
2926 static void diff_resolve_rename_copy(void)
2927 {
2928         int i;
2929         struct diff_filepair *p;
2930         struct diff_queue_struct *q = &diff_queued_diff;
2931
2932         diff_debug_queue("resolve-rename-copy", q);
2933
2934         for (i = 0; i < q->nr; i++) {
2935                 p = q->queue[i];
2936                 p->status = 0; /* undecided */
2937                 if (DIFF_PAIR_UNMERGED(p))
2938                         p->status = DIFF_STATUS_UNMERGED;
2939                 else if (!DIFF_FILE_VALID(p->one))
2940                         p->status = DIFF_STATUS_ADDED;
2941                 else if (!DIFF_FILE_VALID(p->two))
2942                         p->status = DIFF_STATUS_DELETED;
2943                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2944                         p->status = DIFF_STATUS_TYPE_CHANGED;
2945
2946                 /* from this point on, we are dealing with a pair
2947                  * whose both sides are valid and of the same type, i.e.
2948                  * either in-place edit or rename/copy edit.
2949                  */
2950                 else if (DIFF_PAIR_RENAME(p)) {
2951                         /*
2952                          * A rename might have re-connected a broken
2953                          * pair up, causing the pathnames to be the
2954                          * same again. If so, that's not a rename at
2955                          * all, just a modification..
2956                          *
2957                          * Otherwise, see if this source was used for
2958                          * multiple renames, in which case we decrement
2959                          * the count, and call it a copy.
2960                          */
2961                         if (!strcmp(p->one->path, p->two->path))
2962                                 p->status = DIFF_STATUS_MODIFIED;
2963                         else if (--p->one->rename_used > 0)
2964                                 p->status = DIFF_STATUS_COPIED;
2965                         else
2966                                 p->status = DIFF_STATUS_RENAMED;
2967                 }
2968                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2969                          p->one->mode != p->two->mode ||
2970                          is_null_sha1(p->one->sha1))
2971                         p->status = DIFF_STATUS_MODIFIED;
2972                 else {
2973                         /* This is a "no-change" entry and should not
2974                          * happen anymore, but prepare for broken callers.
2975                          */
2976                         error("feeding unmodified %s to diffcore",
2977                               p->one->path);
2978                         p->status = DIFF_STATUS_UNKNOWN;
2979                 }
2980         }
2981         diff_debug_queue("resolve-rename-copy done", q);
2982 }
2983
2984 static int check_pair_status(struct diff_filepair *p)
2985 {
2986         switch (p->status) {
2987         case DIFF_STATUS_UNKNOWN:
2988                 return 0;
2989         case 0:
2990                 die("internal error in diff-resolve-rename-copy");
2991         default:
2992                 return 1;
2993         }
2994 }
2995
2996 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2997 {
2998         int fmt = opt->output_format;
2999
3000         if (fmt & DIFF_FORMAT_CHECKDIFF)
3001                 diff_flush_checkdiff(p, opt);
3002         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3003                 diff_flush_raw(p, opt);
3004         else if (fmt & DIFF_FORMAT_NAME) {
3005                 const char *name_a, *name_b;
3006                 name_a = p->two->path;
3007                 name_b = NULL;
3008                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3009                 write_name_quoted(name_a, opt->file, opt->line_termination);
3010         }
3011 }
3012
3013 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3014 {
3015         if (fs->mode)
3016                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3017         else
3018                 fprintf(file, " %s ", newdelete);
3019         write_name_quoted(fs->path, file, '\n');
3020 }
3021
3022
3023 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
3024 {
3025         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3026                 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
3027                         show_name ? ' ' : '\n');
3028                 if (show_name) {
3029                         write_name_quoted(p->two->path, file, '\n');
3030                 }
3031         }
3032 }
3033
3034 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3035 {
3036         char *names = pprint_rename(p->one->path, p->two->path);
3037
3038         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3039         free(names);
3040         show_mode_change(file, p, 0);
3041 }
3042
3043 static void diff_summary(FILE *file, struct diff_filepair *p)
3044 {
3045         switch(p->status) {
3046         case DIFF_STATUS_DELETED:
3047                 show_file_mode_name(file, "delete", p->one);
3048                 break;
3049         case DIFF_STATUS_ADDED:
3050                 show_file_mode_name(file, "create", p->two);
3051                 break;
3052         case DIFF_STATUS_COPIED:
3053                 show_rename_copy(file, "copy", p);
3054                 break;
3055         case DIFF_STATUS_RENAMED:
3056                 show_rename_copy(file, "rename", p);
3057                 break;
3058         default:
3059                 if (p->score) {
3060                         fputs(" rewrite ", file);
3061                         write_name_quoted(p->two->path, file, ' ');
3062                         fprintf(file, "(%d%%)\n", similarity_index(p));
3063                 }
3064                 show_mode_change(file, p, !p->score);
3065                 break;
3066         }
3067 }
3068
3069 struct patch_id_t {
3070         SHA_CTX *ctx;
3071         int patchlen;
3072 };
3073
3074 static int remove_space(char *line, int len)
3075 {
3076         int i;
3077         char *dst = line;
3078         unsigned char c;
3079
3080         for (i = 0; i < len; i++)
3081                 if (!isspace((c = line[i])))
3082                         *dst++ = c;
3083
3084         return dst - line;
3085 }
3086
3087 static void patch_id_consume(void *priv, char *line, unsigned long len)
3088 {
3089         struct patch_id_t *data = priv;
3090         int new_len;
3091
3092         /* Ignore line numbers when computing the SHA1 of the patch */
3093         if (!prefixcmp(line, "@@ -"))
3094                 return;
3095
3096         new_len = remove_space(line, len);
3097
3098         SHA1_Update(data->ctx, line, new_len);
3099         data->patchlen += new_len;
3100 }
3101
3102 /* returns 0 upon success, and writes result into sha1 */
3103 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3104 {
3105         struct diff_queue_struct *q = &diff_queued_diff;
3106         int i;
3107         SHA_CTX ctx;
3108         struct patch_id_t data;
3109         char buffer[PATH_MAX * 4 + 20];
3110
3111         SHA1_Init(&ctx);
3112         memset(&data, 0, sizeof(struct patch_id_t));
3113         data.ctx = &ctx;
3114
3115         for (i = 0; i < q->nr; i++) {
3116                 xpparam_t xpp;
3117                 xdemitconf_t xecfg;
3118                 xdemitcb_t ecb;
3119                 mmfile_t mf1, mf2;
3120                 struct diff_filepair *p = q->queue[i];
3121                 int len1, len2;
3122
3123                 memset(&xecfg, 0, sizeof(xecfg));
3124                 if (p->status == 0)
3125                         return error("internal diff status error");
3126                 if (p->status == DIFF_STATUS_UNKNOWN)
3127                         continue;
3128                 if (diff_unmodified_pair(p))
3129                         continue;
3130                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3131                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3132                         continue;
3133                 if (DIFF_PAIR_UNMERGED(p))
3134                         continue;
3135
3136                 diff_fill_sha1_info(p->one);
3137                 diff_fill_sha1_info(p->two);
3138                 if (fill_mmfile(&mf1, p->one) < 0 ||
3139                                 fill_mmfile(&mf2, p->two) < 0)
3140                         return error("unable to read files to diff");
3141
3142                 len1 = remove_space(p->one->path, strlen(p->one->path));
3143                 len2 = remove_space(p->two->path, strlen(p->two->path));
3144                 if (p->one->mode == 0)
3145                         len1 = snprintf(buffer, sizeof(buffer),
3146                                         "diff--gita/%.*sb/%.*s"
3147                                         "newfilemode%06o"
3148                                         "---/dev/null"
3149                                         "+++b/%.*s",
3150                                         len1, p->one->path,
3151                                         len2, p->two->path,
3152                                         p->two->mode,
3153                                         len2, p->two->path);
3154                 else if (p->two->mode == 0)
3155                         len1 = snprintf(buffer, sizeof(buffer),
3156                                         "diff--gita/%.*sb/%.*s"
3157                                         "deletedfilemode%06o"
3158                                         "---a/%.*s"
3159                                         "+++/dev/null",
3160                                         len1, p->one->path,
3161                                         len2, p->two->path,
3162                                         p->one->mode,
3163                                         len1, p->one->path);
3164                 else
3165                         len1 = snprintf(buffer, sizeof(buffer),
3166                                         "diff--gita/%.*sb/%.*s"
3167                                         "---a/%.*s"
3168                                         "+++b/%.*s",
3169                                         len1, p->one->path,
3170                                         len2, p->two->path,
3171                                         len1, p->one->path,
3172                                         len2, p->two->path);
3173                 SHA1_Update(&ctx, buffer, len1);
3174
3175                 xpp.flags = XDF_NEED_MINIMAL;
3176                 xecfg.ctxlen = 3;
3177                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3178                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3179                               &xpp, &xecfg, &ecb);
3180         }
3181
3182         SHA1_Final(sha1, &ctx);
3183         return 0;
3184 }
3185
3186 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3187 {
3188         struct diff_queue_struct *q = &diff_queued_diff;
3189         int i;
3190         int result = diff_get_patch_id(options, sha1);
3191
3192         for (i = 0; i < q->nr; i++)
3193                 diff_free_filepair(q->queue[i]);
3194
3195         free(q->queue);
3196         q->queue = NULL;
3197         q->nr = q->alloc = 0;
3198
3199         return result;
3200 }
3201
3202 static int is_summary_empty(const struct diff_queue_struct *q)
3203 {
3204         int i;
3205
3206         for (i = 0; i < q->nr; i++) {
3207                 const struct diff_filepair *p = q->queue[i];
3208
3209                 switch (p->status) {
3210                 case DIFF_STATUS_DELETED:
3211                 case DIFF_STATUS_ADDED:
3212                 case DIFF_STATUS_COPIED:
3213                 case DIFF_STATUS_RENAMED:
3214                         return 0;
3215                 default:
3216                         if (p->score)
3217                                 return 0;
3218                         if (p->one->mode && p->two->mode &&
3219                             p->one->mode != p->two->mode)
3220                                 return 0;
3221                         break;
3222                 }
3223         }
3224         return 1;
3225 }
3226
3227 void diff_flush(struct diff_options *options)
3228 {
3229         struct diff_queue_struct *q = &diff_queued_diff;
3230         int i, output_format = options->output_format;
3231         int separator = 0;
3232
3233         /*
3234          * Order: raw, stat, summary, patch
3235          * or:    name/name-status/checkdiff (other bits clear)
3236          */
3237         if (!q->nr)
3238                 goto free_queue;
3239
3240         if (output_format & (DIFF_FORMAT_RAW |
3241                              DIFF_FORMAT_NAME |
3242                              DIFF_FORMAT_NAME_STATUS |
3243                              DIFF_FORMAT_CHECKDIFF)) {
3244                 for (i = 0; i < q->nr; i++) {
3245                         struct diff_filepair *p = q->queue[i];
3246                         if (check_pair_status(p))
3247                                 flush_one_pair(p, options);
3248                 }
3249                 separator++;
3250         }
3251
3252         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3253                 struct diffstat_t diffstat;
3254
3255                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3256                 for (i = 0; i < q->nr; i++) {
3257                         struct diff_filepair *p = q->queue[i];
3258                         if (check_pair_status(p))
3259                                 diff_flush_stat(p, options, &diffstat);
3260                 }
3261                 if (output_format & DIFF_FORMAT_NUMSTAT)
3262                         show_numstat(&diffstat, options);
3263                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3264                         show_stats(&diffstat, options);
3265                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3266                         show_shortstats(&diffstat, options);
3267                 free_diffstat_info(&diffstat);
3268                 separator++;
3269         }
3270         if (output_format & DIFF_FORMAT_DIRSTAT)
3271                 show_dirstat(options);
3272
3273         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3274                 for (i = 0; i < q->nr; i++)
3275                         diff_summary(options->file, q->queue[i]);
3276                 separator++;
3277         }
3278
3279         if (output_format & DIFF_FORMAT_PATCH) {
3280                 if (separator) {
3281                         putc(options->line_termination, options->file);
3282                         if (options->stat_sep) {
3283                                 /* attach patch instead of inline */
3284                                 fputs(options->stat_sep, options->file);
3285                         }
3286                 }
3287
3288                 for (i = 0; i < q->nr; i++) {
3289                         struct diff_filepair *p = q->queue[i];
3290                         if (check_pair_status(p))
3291                                 diff_flush_patch(p, options);
3292                 }
3293         }
3294
3295         if (output_format & DIFF_FORMAT_CALLBACK)
3296                 options->format_callback(q, options, options->format_callback_data);
3297
3298         for (i = 0; i < q->nr; i++)
3299                 diff_free_filepair(q->queue[i]);
3300 free_queue:
3301         free(q->queue);
3302         q->queue = NULL;
3303         q->nr = q->alloc = 0;
3304         if (options->close_file)
3305                 fclose(options->file);
3306 }
3307
3308 static void diffcore_apply_filter(const char *filter)
3309 {
3310         int i;
3311         struct diff_queue_struct *q = &diff_queued_diff;
3312         struct diff_queue_struct outq;
3313         outq.queue = NULL;
3314         outq.nr = outq.alloc = 0;
3315
3316         if (!filter)
3317                 return;
3318
3319         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3320                 int found;
3321                 for (i = found = 0; !found && i < q->nr; i++) {
3322                         struct diff_filepair *p = q->queue[i];
3323                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3324                              ((p->score &&
3325                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3326                               (!p->score &&
3327                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3328                             ((p->status != DIFF_STATUS_MODIFIED) &&
3329                              strchr(filter, p->status)))
3330                                 found++;
3331                 }
3332                 if (found)
3333                         return;
3334
3335                 /* otherwise we will clear the whole queue
3336                  * by copying the empty outq at the end of this
3337                  * function, but first clear the current entries
3338                  * in the queue.
3339                  */
3340                 for (i = 0; i < q->nr; i++)
3341                         diff_free_filepair(q->queue[i]);
3342         }
3343         else {
3344                 /* Only the matching ones */
3345                 for (i = 0; i < q->nr; i++) {
3346                         struct diff_filepair *p = q->queue[i];
3347
3348                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3349                              ((p->score &&
3350                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3351                               (!p->score &&
3352                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3353                             ((p->status != DIFF_STATUS_MODIFIED) &&
3354                              strchr(filter, p->status)))
3355                                 diff_q(&outq, p);
3356                         else
3357                                 diff_free_filepair(p);
3358                 }
3359         }
3360         free(q->queue);
3361         *q = outq;
3362 }
3363
3364 /* Check whether two filespecs with the same mode and size are identical */
3365 static int diff_filespec_is_identical(struct diff_filespec *one,
3366                                       struct diff_filespec *two)
3367 {
3368         if (S_ISGITLINK(one->mode))
3369                 return 0;
3370         if (diff_populate_filespec(one, 0))
3371                 return 0;
3372         if (diff_populate_filespec(two, 0))
3373                 return 0;
3374         return !memcmp(one->data, two->data, one->size);
3375 }
3376
3377 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3378 {
3379         int i;
3380         struct diff_queue_struct *q = &diff_queued_diff;
3381         struct diff_queue_struct outq;
3382         outq.queue = NULL;
3383         outq.nr = outq.alloc = 0;
3384
3385         for (i = 0; i < q->nr; i++) {
3386                 struct diff_filepair *p = q->queue[i];
3387
3388                 /*
3389                  * 1. Entries that come from stat info dirtyness
3390                  *    always have both sides (iow, not create/delete),
3391                  *    one side of the object name is unknown, with
3392                  *    the same mode and size.  Keep the ones that
3393                  *    do not match these criteria.  They have real
3394                  *    differences.
3395                  *
3396                  * 2. At this point, the file is known to be modified,
3397                  *    with the same mode and size, and the object
3398                  *    name of one side is unknown.  Need to inspect
3399                  *    the identical contents.
3400                  */
3401                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3402                     !DIFF_FILE_VALID(p->two) ||
3403                     (p->one->sha1_valid && p->two->sha1_valid) ||
3404                     (p->one->mode != p->two->mode) ||
3405                     diff_populate_filespec(p->one, 1) ||
3406                     diff_populate_filespec(p->two, 1) ||
3407                     (p->one->size != p->two->size) ||
3408                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3409                         diff_q(&outq, p);
3410                 else {
3411                         /*
3412                          * The caller can subtract 1 from skip_stat_unmatch
3413                          * to determine how many paths were dirty only
3414                          * due to stat info mismatch.
3415                          */
3416                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3417                                 diffopt->skip_stat_unmatch++;
3418                         diff_free_filepair(p);
3419                 }
3420         }
3421         free(q->queue);
3422         *q = outq;
3423 }
3424
3425 void diffcore_std(struct diff_options *options)
3426 {
3427         if (options->skip_stat_unmatch)
3428                 diffcore_skip_stat_unmatch(options);
3429         if (options->break_opt != -1)
3430                 diffcore_break(options->break_opt);
3431         if (options->detect_rename)
3432                 diffcore_rename(options);
3433         if (options->break_opt != -1)
3434                 diffcore_merge_broken();
3435         if (options->pickaxe)
3436                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3437         if (options->orderfile)
3438                 diffcore_order(options->orderfile);
3439         diff_resolve_rename_copy();
3440         diffcore_apply_filter(options->filter);
3441
3442         if (diff_queued_diff.nr)
3443                 DIFF_OPT_SET(options, HAS_CHANGES);
3444         else
3445                 DIFF_OPT_CLR(options, HAS_CHANGES);
3446 }
3447
3448 int diff_result_code(struct diff_options *opt, int status)
3449 {
3450         int result = 0;
3451         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3452             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3453                 return status;
3454         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3455             DIFF_OPT_TST(opt, HAS_CHANGES))
3456                 result |= 01;
3457         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3458             DIFF_OPT_TST(opt, CHECK_FAILED))
3459                 result |= 02;
3460         return result;
3461 }
3462
3463 void diff_addremove(struct diff_options *options,
3464                     int addremove, unsigned mode,
3465                     const unsigned char *sha1,
3466                     const char *concatpath)
3467 {
3468         struct diff_filespec *one, *two;
3469
3470         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3471                 return;
3472
3473         /* This may look odd, but it is a preparation for
3474          * feeding "there are unchanged files which should
3475          * not produce diffs, but when you are doing copy
3476          * detection you would need them, so here they are"
3477          * entries to the diff-core.  They will be prefixed
3478          * with something like '=' or '*' (I haven't decided
3479          * which but should not make any difference).
3480          * Feeding the same new and old to diff_change()
3481          * also has the same effect.
3482          * Before the final output happens, they are pruned after
3483          * merged into rename/copy pairs as appropriate.
3484          */
3485         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3486                 addremove = (addremove == '+' ? '-' :
3487                              addremove == '-' ? '+' : addremove);
3488
3489         if (options->prefix &&
3490             strncmp(concatpath, options->prefix, options->prefix_length))
3491                 return;
3492
3493         one = alloc_filespec(concatpath);
3494         two = alloc_filespec(concatpath);
3495
3496         if (addremove != '+')
3497                 fill_filespec(one, sha1, mode);
3498         if (addremove != '-')
3499                 fill_filespec(two, sha1, mode);
3500
3501         diff_queue(&diff_queued_diff, one, two);
3502         DIFF_OPT_SET(options, HAS_CHANGES);
3503 }
3504
3505 void diff_change(struct diff_options *options,
3506                  unsigned old_mode, unsigned new_mode,
3507                  const unsigned char *old_sha1,
3508                  const unsigned char *new_sha1,
3509                  const char *concatpath)
3510 {
3511         struct diff_filespec *one, *two;
3512
3513         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3514                         && S_ISGITLINK(new_mode))
3515                 return;
3516
3517         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3518                 unsigned tmp;
3519                 const unsigned char *tmp_c;
3520                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3521                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3522         }
3523
3524         if (options->prefix &&
3525             strncmp(concatpath, options->prefix, options->prefix_length))
3526                 return;
3527
3528         one = alloc_filespec(concatpath);
3529         two = alloc_filespec(concatpath);
3530         fill_filespec(one, old_sha1, old_mode);
3531         fill_filespec(two, new_sha1, new_mode);
3532
3533         diff_queue(&diff_queued_diff, one, two);
3534         DIFF_OPT_SET(options, HAS_CHANGES);
3535 }
3536
3537 void diff_unmerge(struct diff_options *options,
3538                   const char *path,
3539                   unsigned mode, const unsigned char *sha1)
3540 {
3541         struct diff_filespec *one, *two;
3542
3543         if (options->prefix &&
3544             strncmp(path, options->prefix, options->prefix_length))
3545                 return;
3546
3547         one = alloc_filespec(path);
3548         two = alloc_filespec(path);
3549         fill_filespec(one, sha1, mode);
3550         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3551 }