diff.c

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "quote.h"
   6 #include "diff.h"
   7 #include "diffcore.h"
   8 #include "delta.h"
   9 #include "xdiff-interface.h"
  10 #include "color.h"
  11 #include "attr.h"
  12 #include "run-command.h"
  13 #include "utf8.h"
  14
  15 #ifdef NO_FAST_WORKING_DIRECTORY
  16 #define FAST_WORKING_DIRECTORY 0
  17 #else
  18 #define FAST_WORKING_DIRECTORY 1
  19 #endif
  20
  21 static int diff_detect_rename_default;
  22 static int diff_rename_limit_default = 200;
  23 int diff_use_color_default = -1;
  24 static const char *external_diff_cmd_cfg;
  25 int diff_auto_refresh_index = 1;
  26 static int diff_mnemonic_prefix;
  27
  28 static char diff_colors[][COLOR_MAXLEN] = {
  29         "\033[m",       /* reset */
  30         "",             /* PLAIN (normal) */
  31         "\033[1m",      /* METAINFO (bold) */
  32         "\033[36m",     /* FRAGINFO (cyan) */
  33         "\033[31m",     /* OLD (red) */
  34         "\033[32m",     /* NEW (green) */
  35         "\033[33m",     /* COMMIT (yellow) */
  36         "\033[41m",     /* WHITESPACE (red background) */
  37 };
  38
  39 static int parse_diff_color_slot(const char *var, int ofs)
  40 {
  41         if (!strcasecmp(var+ofs, "plain"))
  42                 return DIFF_PLAIN;
  43         if (!strcasecmp(var+ofs, "meta"))
  44                 return DIFF_METAINFO;
  45         if (!strcasecmp(var+ofs, "frag"))
  46                 return DIFF_FRAGINFO;
  47         if (!strcasecmp(var+ofs, "old"))
  48                 return DIFF_FILE_OLD;
  49         if (!strcasecmp(var+ofs, "new"))
  50                 return DIFF_FILE_NEW;
  51         if (!strcasecmp(var+ofs, "commit"))
  52                 return DIFF_COMMIT;
  53         if (!strcasecmp(var+ofs, "whitespace"))
  54                 return DIFF_WHITESPACE;
  55         die("bad config variable '%s'", var);
  56 }
  57
  58 static struct ll_diff_driver {
  59         const char *name;
  60         struct ll_diff_driver *next;
  61         const char *cmd;
  62 } *user_diff, **user_diff_tail;
  63
  64 /*
  65  * Currently there is only "diff.<drivername>.command" variable;
  66  * because there are "diff.color.<slot>" variables, we are parsing
  67  * this in a bit convoluted way to allow low level diff driver
  68  * called "color".
  69  */
  70 static int parse_lldiff_command(const char *var, const char *ep, const char *value)
  71 {
  72         const char *name;
  73         int namelen;
  74         struct ll_diff_driver *drv;
  75
  76         name = var + 5;
  77         namelen = ep - name;
  78         for (drv = user_diff; drv; drv = drv->next)
  79                 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
  80                         break;
  81         if (!drv) {
  82                 drv = xcalloc(1, sizeof(struct ll_diff_driver));
  83                 drv->name = xmemdupz(name, namelen);
  84                 if (!user_diff_tail)
  85                         user_diff_tail = &user_diff;
  86                 *user_diff_tail = drv;
  87                 user_diff_tail = &(drv->next);
  88         }
  89
  90         return git_config_string(&(drv->cmd), var, value);
  91 }
  92
  93 /*
  94  * 'diff.<what>.funcname' attribute can be specified in the configuration
  95  * to define a customized regexp to find the beginning of a function to
  96  * be used for hunk header lines of "diff -p" style output.
  97  */
  98 static struct funcname_pattern {
  99         char *name;
 100         char *pattern;
 101         struct funcname_pattern *next;
 102 } *funcname_pattern_list;
 103
 104 static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
 105 {
 106         const char *name;
 107         int namelen;
 108         struct funcname_pattern *pp;
 109
 110         name = var + 5; /* "diff." */
 111         namelen = ep - name;
 112
 113         for (pp = funcname_pattern_list; pp; pp = pp->next)
 114                 if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
 115                         break;
 116         if (!pp) {
 117                 pp = xcalloc(1, sizeof(*pp));
 118                 pp->name = xmemdupz(name, namelen);
 119                 pp->next = funcname_pattern_list;
 120                 funcname_pattern_list = pp;
 121         }
 122         free(pp->pattern);
 123         pp->pattern = xstrdup(value);
 124         return 0;
 125 }
 126
 127 /*
 128  * These are to give UI layer defaults.
 129  * The core-level commands such as git-diff-files should
 130  * never be affected by the setting of diff.renames
 131  * the user happens to have in the configuration file.
 132  */
 133 int git_diff_ui_config(const char *var, const char *value, void *cb)
 134 {
 135         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 136                 diff_use_color_default = git_config_colorbool(var, value, -1);
 137                 return 0;
 138         }
 139         if (!strcmp(var, "diff.renames")) {
 140                 if (!value)
 141                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 142                 else if (!strcasecmp(value, "copies") ||
 143                          !strcasecmp(value, "copy"))
 144                         diff_detect_rename_default = DIFF_DETECT_COPY;
 145                 else if (git_config_bool(var,value))
 146                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 147                 return 0;
 148         }
 149         if (!strcmp(var, "diff.autorefreshindex")) {
 150                 diff_auto_refresh_index = git_config_bool(var, value);
 151                 return 0;
 152         }
 153         if (!strcmp(var, "diff.mnemonicprefix")) {
 154                 diff_mnemonic_prefix = git_config_bool(var, value);
 155                 return 0;
 156         }
 157         if (!strcmp(var, "diff.external"))
 158                 return git_config_string(&external_diff_cmd_cfg, var, value);
 159         if (!prefixcmp(var, "diff.")) {
 160                 const char *ep = strrchr(var, '.');
 161
 162                 if (ep != var + 4 && !strcmp(ep, ".command"))
 163                         return parse_lldiff_command(var, ep, value);
 164         }
 165
 166         return git_diff_basic_config(var, value, cb);
 167 }
 168
 169 int git_diff_basic_config(const char *var, const char *value, void *cb)
 170 {
 171         if (!strcmp(var, "diff.renamelimit")) {
 172                 diff_rename_limit_default = git_config_int(var, value);
 173                 return 0;
 174         }
 175
 176         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 177                 int slot = parse_diff_color_slot(var, 11);
 178                 if (!value)
 179                         return config_error_nonbool(var);
 180                 color_parse(value, var, diff_colors[slot]);
 181                 return 0;
 182         }
 183
 184         if (!prefixcmp(var, "diff.")) {
 185                 const char *ep = strrchr(var, '.');
 186                 if (ep != var + 4) {
 187                         if (!strcmp(ep, ".funcname")) {
 188                                 if (!value)
 189                                         return config_error_nonbool(var);
 190                                 return parse_funcname_pattern(var, ep, value);
 191                         }
 192                 }
 193         }
 194
 195         return git_color_default_config(var, value, cb);
 196 }
 197
 198 static char *quote_two(const char *one, const char *two)
 199 {
 200         int need_one = quote_c_style(one, NULL, NULL, 1);
 201         int need_two = quote_c_style(two, NULL, NULL, 1);
 202         struct strbuf res;
 203
 204         strbuf_init(&res, 0);
 205         if (need_one + need_two) {
 206                 strbuf_addch(&res, '"');
 207                 quote_c_style(one, &res, NULL, 1);
 208                 quote_c_style(two, &res, NULL, 1);
 209                 strbuf_addch(&res, '"');
 210         } else {
 211                 strbuf_addstr(&res, one);
 212                 strbuf_addstr(&res, two);
 213         }
 214         return strbuf_detach(&res, NULL);
 215 }
 216
 217 static const char *external_diff(void)
 218 {
 219         static const char *external_diff_cmd = NULL;
 220         static int done_preparing = 0;
 221
 222         if (done_preparing)
 223                 return external_diff_cmd;
 224         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 225         if (!external_diff_cmd)
 226                 external_diff_cmd = external_diff_cmd_cfg;
 227         done_preparing = 1;
 228         return external_diff_cmd;
 229 }
 230
 231 static struct diff_tempfile {
 232         const char *name; /* filename external diff should read from */
 233         char hex[41];
 234         char mode[10];
 235         char tmp_path[PATH_MAX];
 236 } diff_temp[2];
 237
 238 static int count_lines(const char *data, int size)
 239 {
 240         int count, ch, completely_empty = 1, nl_just_seen = 0;
 241         count = 0;
 242         while (0 < size--) {
 243                 ch = *data++;
 244                 if (ch == '\n') {
 245                         count++;
 246                         nl_just_seen = 1;
 247                         completely_empty = 0;
 248                 }
 249                 else {
 250                         nl_just_seen = 0;
 251                         completely_empty = 0;
 252                 }
 253         }
 254         if (completely_empty)
 255                 return 0;
 256         if (!nl_just_seen)
 257                 count++; /* no trailing newline */
 258         return count;
 259 }
 260
 261 static void print_line_count(FILE *file, int count)
 262 {
 263         switch (count) {
 264         case 0:
 265                 fprintf(file, "0,0");
 266                 break;
 267         case 1:
 268                 fprintf(file, "1");
 269                 break;
 270         default:
 271                 fprintf(file, "1,%d", count);
 272                 break;
 273         }
 274 }
 275
 276 static void copy_file_with_prefix(FILE *file,
 277                                   int prefix, const char *data, int size,
 278                                   const char *set, const char *reset)
 279 {
 280         int ch, nl_just_seen = 1;
 281         while (0 < size--) {
 282                 ch = *data++;
 283                 if (nl_just_seen) {
 284                         fputs(set, file);
 285                         putc(prefix, file);
 286                 }
 287                 if (ch == '\n') {
 288                         nl_just_seen = 1;
 289                         fputs(reset, file);
 290                 } else
 291                         nl_just_seen = 0;
 292                 putc(ch, file);
 293         }
 294         if (!nl_just_seen)
 295                 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
 296 }
 297
 298 static void emit_rewrite_diff(const char *name_a,
 299                               const char *name_b,
 300                               struct diff_filespec *one,
 301                               struct diff_filespec *two,
 302                               struct diff_options *o)
 303 {
 304         int lc_a, lc_b;
 305         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
 306         const char *name_a_tab, *name_b_tab;
 307         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 308         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 309         const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
 310         const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
 311         const char *reset = diff_get_color(color_diff, DIFF_RESET);
 312         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 313         const char *a_prefix, *b_prefix;
 314
 315         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 316                 a_prefix = o->b_prefix;
 317                 b_prefix = o->a_prefix;
 318         } else {
 319                 a_prefix = o->a_prefix;
 320                 b_prefix = o->b_prefix;
 321         }
 322
 323         name_a += (*name_a == '/');
 324         name_b += (*name_b == '/');
 325         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 326         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 327
 328         strbuf_reset(&a_name);
 329         strbuf_reset(&b_name);
 330         quote_two_c_style(&a_name, a_prefix, name_a, 0);
 331         quote_two_c_style(&b_name, b_prefix, name_b, 0);
 332
 333         diff_populate_filespec(one, 0);
 334         diff_populate_filespec(two, 0);
 335         lc_a = count_lines(one->data, one->size);
 336         lc_b = count_lines(two->data, two->size);
 337         fprintf(o->file,
 338                 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
 339                 metainfo, a_name.buf, name_a_tab, reset,
 340                 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
 341         print_line_count(o->file, lc_a);
 342         fprintf(o->file, " +");
 343         print_line_count(o->file, lc_b);
 344         fprintf(o->file, " @@%s\n", reset);
 345         if (lc_a)
 346                 copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
 347         if (lc_b)
 348                 copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
 349 }
 350
 351 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 352 {
 353         if (!DIFF_FILE_VALID(one)) {
 354                 mf->ptr = (char *)""; /* does not matter */
 355                 mf->size = 0;
 356                 return 0;
 357         }
 358         else if (diff_populate_filespec(one, 0))
 359                 return -1;
 360         mf->ptr = one->data;
 361         mf->size = one->size;
 362         return 0;
 363 }
 364
 365 struct diff_words_buffer {
 366         mmfile_t text;
 367         long alloc;
 368         long current; /* output pointer */
 369         int suppressed_newline;
 370 };
 371
 372 static void diff_words_append(char *line, unsigned long len,
 373                 struct diff_words_buffer *buffer)
 374 {
 375         if (buffer->text.size + len > buffer->alloc) {
 376                 buffer->alloc = (buffer->text.size + len) * 3 / 2;
 377                 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
 378         }
 379         line++;
 380         len--;
 381         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 382         buffer->text.size += len;
 383 }
 384
 385 struct diff_words_data {
 386         struct xdiff_emit_state xm;
 387         struct diff_words_buffer minus, plus;
 388         FILE *file;
 389 };
 390
 391 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
 392                 int suppress_newline)
 393 {
 394         const char *ptr;
 395         int eol = 0;
 396
 397         if (len == 0)
 398                 return;
 399
 400         ptr  = buffer->text.ptr + buffer->current;
 401         buffer->current += len;
 402
 403         if (ptr[len - 1] == '\n') {
 404                 eol = 1;
 405                 len--;
 406         }
 407
 408         fputs(diff_get_color(1, color), file);
 409         fwrite(ptr, len, 1, file);
 410         fputs(diff_get_color(1, DIFF_RESET), file);
 411
 412         if (eol) {
 413                 if (suppress_newline)
 414                         buffer->suppressed_newline = 1;
 415                 else
 416                         putc('\n', file);
 417         }
 418 }
 419
 420 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 421 {
 422         struct diff_words_data *diff_words = priv;
 423
 424         if (diff_words->minus.suppressed_newline) {
 425                 if (line[0] != '+')
 426                         putc('\n', diff_words->file);
 427                 diff_words->minus.suppressed_newline = 0;
 428         }
 429
 430         len--;
 431         switch (line[0]) {
 432                 case '-':
 433                         print_word(diff_words->file,
 434                                    &diff_words->minus, len, DIFF_FILE_OLD, 1);
 435                         break;
 436                 case '+':
 437                         print_word(diff_words->file,
 438                                    &diff_words->plus, len, DIFF_FILE_NEW, 0);
 439                         break;
 440                 case ' ':
 441                         print_word(diff_words->file,
 442                                    &diff_words->plus, len, DIFF_PLAIN, 0);
 443                         diff_words->minus.current += len;
 444                         break;
 445         }
 446 }
 447
 448 /* this executes the word diff on the accumulated buffers */
 449 static void diff_words_show(struct diff_words_data *diff_words)
 450 {
 451         xpparam_t xpp;
 452         xdemitconf_t xecfg;
 453         xdemitcb_t ecb;
 454         mmfile_t minus, plus;
 455         int i;
 456
 457         memset(&xecfg, 0, sizeof(xecfg));
 458         minus.size = diff_words->minus.text.size;
 459         minus.ptr = xmalloc(minus.size);
 460         memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
 461         for (i = 0; i < minus.size; i++)
 462                 if (isspace(minus.ptr[i]))
 463                         minus.ptr[i] = '\n';
 464         diff_words->minus.current = 0;
 465
 466         plus.size = diff_words->plus.text.size;
 467         plus.ptr = xmalloc(plus.size);
 468         memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
 469         for (i = 0; i < plus.size; i++)
 470                 if (isspace(plus.ptr[i]))
 471                         plus.ptr[i] = '\n';
 472         diff_words->plus.current = 0;
 473
 474         xpp.flags = XDF_NEED_MINIMAL;
 475         xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
 476         ecb.outf = xdiff_outf;
 477         ecb.priv = diff_words;
 478         diff_words->xm.consume = fn_out_diff_words_aux;
 479         xdi_diff(&minus, &plus, &xpp, &xecfg, &ecb);
 480
 481         free(minus.ptr);
 482         free(plus.ptr);
 483         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 484
 485         if (diff_words->minus.suppressed_newline) {
 486                 putc('\n', diff_words->file);
 487                 diff_words->minus.suppressed_newline = 0;
 488         }
 489 }
 490
 491 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 492
 493 struct emit_callback {
 494         struct xdiff_emit_state xm;
 495         int nparents, color_diff;
 496         unsigned ws_rule;
 497         sane_truncate_fn truncate;
 498         const char **label_path;
 499         struct diff_words_data *diff_words;
 500         int *found_changesp;
 501         FILE *file;
 502 };
 503
 504 static void free_diff_words_data(struct emit_callback *ecbdata)
 505 {
 506         if (ecbdata->diff_words) {
 507                 /* flush buffers */
 508                 if (ecbdata->diff_words->minus.text.size ||
 509                                 ecbdata->diff_words->plus.text.size)
 510                         diff_words_show(ecbdata->diff_words);
 511
 512                 free (ecbdata->diff_words->minus.text.ptr);
 513                 free (ecbdata->diff_words->plus.text.ptr);
 514                 free(ecbdata->diff_words);
 515                 ecbdata->diff_words = NULL;
 516         }
 517 }
 518
 519 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 520 {
 521         if (diff_use_color)
 522                 return diff_colors[ix];
 523         return "";
 524 }
 525
 526 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
 527 {
 528         int has_trailing_newline = (len > 0 && line[len-1] == '\n');
 529         if (has_trailing_newline)
 530                 len--;
 531
 532         fputs(set, file);
 533         fwrite(line, len, 1, file);
 534         fputs(reset, file);
 535         if (has_trailing_newline)
 536                 fputc('\n', file);
 537 }
 538
 539 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
 540 {
 541         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 542         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 543
 544         if (!*ws)
 545                 emit_line(ecbdata->file, set, reset, line, len);
 546         else {
 547                 /* Emit just the prefix, then the rest. */
 548                 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
 549                 ws_check_emit(line + ecbdata->nparents,
 550                               len - ecbdata->nparents, ecbdata->ws_rule,
 551                               ecbdata->file, set, reset, ws);
 552         }
 553 }
 554
 555 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
 556 {
 557         const char *cp;
 558         unsigned long allot;
 559         size_t l = len;
 560
 561         if (ecb->truncate)
 562                 return ecb->truncate(line, len);
 563         cp = line;
 564         allot = l;
 565         while (0 < l) {
 566                 (void) utf8_width(&cp, &l);
 567                 if (!cp)
 568                         break; /* truncated in the middle? */
 569         }
 570         return allot - l;
 571 }
 572
 573 static void fn_out_consume(void *priv, char *line, unsigned long len)
 574 {
 575         int i;
 576         int color;
 577         struct emit_callback *ecbdata = priv;
 578         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 579         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 580         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 581
 582         *(ecbdata->found_changesp) = 1;
 583
 584         if (ecbdata->label_path[0]) {
 585                 const char *name_a_tab, *name_b_tab;
 586
 587                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 588                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 589
 590                 fprintf(ecbdata->file, "%s--- %s%s%s\n",
 591                         meta, ecbdata->label_path[0], reset, name_a_tab);
 592                 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
 593                         meta, ecbdata->label_path[1], reset, name_b_tab);
 594                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 595         }
 596
 597         /* This is not really necessary for now because
 598          * this codepath only deals with two-way diffs.
 599          */
 600         for (i = 0; i < len && line[i] == '@'; i++)
 601                 ;
 602         if (2 <= i && i < len && line[i] == ' ') {
 603                 ecbdata->nparents = i - 1;
 604                 len = sane_truncate_line(ecbdata, line, len);
 605                 emit_line(ecbdata->file,
 606                           diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
 607                           reset, line, len);
 608                 if (line[len-1] != '\n')
 609                         putc('\n', ecbdata->file);
 610                 return;
 611         }
 612
 613         if (len < ecbdata->nparents) {
 614                 emit_line(ecbdata->file, reset, reset, line, len);
 615                 return;
 616         }
 617
 618         color = DIFF_PLAIN;
 619         if (ecbdata->diff_words && ecbdata->nparents != 1)
 620                 /* fall back to normal diff */
 621                 free_diff_words_data(ecbdata);
 622         if (ecbdata->diff_words) {
 623                 if (line[0] == '-') {
 624                         diff_words_append(line, len,
 625                                           &ecbdata->diff_words->minus);
 626                         return;
 627                 } else if (line[0] == '+') {
 628                         diff_words_append(line, len,
 629                                           &ecbdata->diff_words->plus);
 630                         return;
 631                 }
 632                 if (ecbdata->diff_words->minus.text.size ||
 633                     ecbdata->diff_words->plus.text.size)
 634                         diff_words_show(ecbdata->diff_words);
 635                 line++;
 636                 len--;
 637                 emit_line(ecbdata->file, plain, reset, line, len);
 638                 return;
 639         }
 640         for (i = 0; i < ecbdata->nparents && len; i++) {
 641                 if (line[i] == '-')
 642                         color = DIFF_FILE_OLD;
 643                 else if (line[i] == '+')
 644                         color = DIFF_FILE_NEW;
 645         }
 646
 647         if (color != DIFF_FILE_NEW) {
 648                 emit_line(ecbdata->file,
 649                           diff_get_color(ecbdata->color_diff, color),
 650                           reset, line, len);
 651                 return;
 652         }
 653         emit_add_line(reset, ecbdata, line, len);
 654 }
 655
 656 static char *pprint_rename(const char *a, const char *b)
 657 {
 658         const char *old = a;
 659         const char *new = b;
 660         struct strbuf name;
 661         int pfx_length, sfx_length;
 662         int len_a = strlen(a);
 663         int len_b = strlen(b);
 664         int a_midlen, b_midlen;
 665         int qlen_a = quote_c_style(a, NULL, NULL, 0);
 666         int qlen_b = quote_c_style(b, NULL, NULL, 0);
 667
 668         strbuf_init(&name, 0);
 669         if (qlen_a || qlen_b) {
 670                 quote_c_style(a, &name, NULL, 0);
 671                 strbuf_addstr(&name, " => ");
 672                 quote_c_style(b, &name, NULL, 0);
 673                 return strbuf_detach(&name, NULL);
 674         }
 675
 676         /* Find common prefix */
 677         pfx_length = 0;
 678         while (*old && *new && *old == *new) {
 679                 if (*old == '/')
 680                         pfx_length = old - a + 1;
 681                 old++;
 682                 new++;
 683         }
 684
 685         /* Find common suffix */
 686         old = a + len_a;
 687         new = b + len_b;
 688         sfx_length = 0;
 689         while (a <= old && b <= new && *old == *new) {
 690                 if (*old == '/')
 691                         sfx_length = len_a - (old - a);
 692                 old--;
 693                 new--;
 694         }
 695
 696         /*
 697          * pfx{mid-a => mid-b}sfx
 698          * {pfx-a => pfx-b}sfx
 699          * pfx{sfx-a => sfx-b}
 700          * name-a => name-b
 701          */
 702         a_midlen = len_a - pfx_length - sfx_length;
 703         b_midlen = len_b - pfx_length - sfx_length;
 704         if (a_midlen < 0)
 705                 a_midlen = 0;
 706         if (b_midlen < 0)
 707                 b_midlen = 0;
 708
 709         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 710         if (pfx_length + sfx_length) {
 711                 strbuf_add(&name, a, pfx_length);
 712                 strbuf_addch(&name, '{');
 713         }
 714         strbuf_add(&name, a + pfx_length, a_midlen);
 715         strbuf_addstr(&name, " => ");
 716         strbuf_add(&name, b + pfx_length, b_midlen);
 717         if (pfx_length + sfx_length) {
 718                 strbuf_addch(&name, '}');
 719                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 720         }
 721         return strbuf_detach(&name, NULL);
 722 }
 723
 724 struct diffstat_t {
 725         struct xdiff_emit_state xm;
 726
 727         int nr;
 728         int alloc;
 729         struct diffstat_file {
 730                 char *from_name;
 731                 char *name;
 732                 char *print_name;
 733                 unsigned is_unmerged:1;
 734                 unsigned is_binary:1;
 735                 unsigned is_renamed:1;
 736                 unsigned int added, deleted;
 737         } **files;
 738 };
 739
 740 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 741                                           const char *name_a,
 742                                           const char *name_b)
 743 {
 744         struct diffstat_file *x;
 745         x = xcalloc(sizeof (*x), 1);
 746         if (diffstat->nr == diffstat->alloc) {
 747                 diffstat->alloc = alloc_nr(diffstat->alloc);
 748                 diffstat->files = xrealloc(diffstat->files,
 749                                 diffstat->alloc * sizeof(x));
 750         }
 751         diffstat->files[diffstat->nr++] = x;
 752         if (name_b) {
 753                 x->from_name = xstrdup(name_a);
 754                 x->name = xstrdup(name_b);
 755                 x->is_renamed = 1;
 756         }
 757         else {
 758                 x->from_name = NULL;
 759                 x->name = xstrdup(name_a);
 760         }
 761         return x;
 762 }
 763
 764 static void diffstat_consume(void *priv, char *line, unsigned long len)
 765 {
 766         struct diffstat_t *diffstat = priv;
 767         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 768
 769         if (line[0] == '+')
 770                 x->added++;
 771         else if (line[0] == '-')
 772                 x->deleted++;
 773 }
 774
 775 const char mime_boundary_leader[] = "------------";
 776
 777 static int scale_linear(int it, int width, int max_change)
 778 {
 779         /*
 780          * make sure that at least one '-' is printed if there were deletions,
 781          * and likewise for '+'.
 782          */
 783         if (max_change < 2)
 784                 return it;
 785         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
 786 }
 787
 788 static void show_name(FILE *file,
 789                       const char *prefix, const char *name, int len,
 790                       const char *reset, const char *set)
 791 {
 792         fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
 793 }
 794
 795 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
 796 {
 797         if (cnt <= 0)
 798                 return;
 799         fprintf(file, "%s", set);
 800         while (cnt--)
 801                 putc(ch, file);
 802         fprintf(file, "%s", reset);
 803 }
 804
 805 static void fill_print_name(struct diffstat_file *file)
 806 {
 807         char *pname;
 808
 809         if (file->print_name)
 810                 return;
 811
 812         if (!file->is_renamed) {
 813                 struct strbuf buf;
 814                 strbuf_init(&buf, 0);
 815                 if (quote_c_style(file->name, &buf, NULL, 0)) {
 816                         pname = strbuf_detach(&buf, NULL);
 817                 } else {
 818                         pname = file->name;
 819                         strbuf_release(&buf);
 820                 }
 821         } else {
 822                 pname = pprint_rename(file->from_name, file->name);
 823         }
 824         file->print_name = pname;
 825 }
 826
 827 static void show_stats(struct diffstat_t* data, struct diff_options *options)
 828 {
 829         int i, len, add, del, total, adds = 0, dels = 0;
 830         int max_change = 0, max_len = 0;
 831         int total_files = data->nr;
 832         int width, name_width;
 833         const char *reset, *set, *add_c, *del_c;
 834
 835         if (data->nr == 0)
 836                 return;
 837
 838         width = options->stat_width ? options->stat_width : 80;
 839         name_width = options->stat_name_width ? options->stat_name_width : 50;
 840
 841         /* Sanity: give at least 5 columns to the graph,
 842          * but leave at least 10 columns for the name.
 843          */
 844         if (width < 25)
 845                 width = 25;
 846         if (name_width < 10)
 847                 name_width = 10;
 848         else if (width < name_width + 15)
 849                 name_width = width - 15;
 850
 851         /* Find the longest filename and max number of changes */
 852         reset = diff_get_color_opt(options, DIFF_RESET);
 853         set   = diff_get_color_opt(options, DIFF_PLAIN);
 854         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
 855         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
 856
 857         for (i = 0; i < data->nr; i++) {
 858                 struct diffstat_file *file = data->files[i];
 859                 int change = file->added + file->deleted;
 860                 fill_print_name(file);
 861                 len = strlen(file->print_name);
 862                 if (max_len < len)
 863                         max_len = len;
 864
 865                 if (file->is_binary || file->is_unmerged)
 866                         continue;
 867                 if (max_change < change)
 868                         max_change = change;
 869         }
 870
 871         /* Compute the width of the graph part;
 872          * 10 is for one blank at the beginning of the line plus
 873          * " | count " between the name and the graph.
 874          *
 875          * From here on, name_width is the width of the name area,
 876          * and width is the width of the graph area.
 877          */
 878         name_width = (name_width < max_len) ? name_width : max_len;
 879         if (width < (name_width + 10) + max_change)
 880                 width = width - (name_width + 10);
 881         else
 882                 width = max_change;
 883
 884         for (i = 0; i < data->nr; i++) {
 885                 const char *prefix = "";
 886                 char *name = data->files[i]->print_name;
 887                 int added = data->files[i]->added;
 888                 int deleted = data->files[i]->deleted;
 889                 int name_len;
 890
 891                 /*
 892                  * "scale" the filename
 893                  */
 894                 len = name_width;
 895                 name_len = strlen(name);
 896                 if (name_width < name_len) {
 897                         char *slash;
 898                         prefix = "...";
 899                         len -= 3;
 900                         name += name_len - len;
 901                         slash = strchr(name, '/');
 902                         if (slash)
 903                                 name = slash;
 904                 }
 905
 906                 if (data->files[i]->is_binary) {
 907                         show_name(options->file, prefix, name, len, reset, set);
 908                         fprintf(options->file, "  Bin ");
 909                         fprintf(options->file, "%s%d%s", del_c, deleted, reset);
 910                         fprintf(options->file, " -> ");
 911                         fprintf(options->file, "%s%d%s", add_c, added, reset);
 912                         fprintf(options->file, " bytes");
 913                         fprintf(options->file, "\n");
 914                         continue;
 915                 }
 916                 else if (data->files[i]->is_unmerged) {
 917                         show_name(options->file, prefix, name, len, reset, set);
 918                         fprintf(options->file, "  Unmerged\n");
 919                         continue;
 920                 }
 921                 else if (!data->files[i]->is_renamed &&
 922                          (added + deleted == 0)) {
 923                         total_files--;
 924                         continue;
 925                 }
 926
 927                 /*
 928                  * scale the add/delete
 929                  */
 930                 add = added;
 931                 del = deleted;
 932                 total = add + del;
 933                 adds += add;
 934                 dels += del;
 935
 936                 if (width <= max_change) {
 937                         add = scale_linear(add, width, max_change);
 938                         del = scale_linear(del, width, max_change);
 939                         total = add + del;
 940                 }
 941                 show_name(options->file, prefix, name, len, reset, set);
 942                 fprintf(options->file, "%5d%s", added + deleted,
 943                                 added + deleted ? " " : "");
 944                 show_graph(options->file, '+', add, add_c, reset);
 945                 show_graph(options->file, '-', del, del_c, reset);
 946                 fprintf(options->file, "\n");
 947         }
 948         fprintf(options->file,
 949                "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
 950                set, total_files, adds, dels, reset);
 951 }
 952
 953 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
 954 {
 955         int i, adds = 0, dels = 0, total_files = data->nr;
 956
 957         if (data->nr == 0)
 958                 return;
 959
 960         for (i = 0; i < data->nr; i++) {
 961                 if (!data->files[i]->is_binary &&
 962                     !data->files[i]->is_unmerged) {
 963                         int added = data->files[i]->added;
 964                         int deleted= data->files[i]->deleted;
 965                         if (!data->files[i]->is_renamed &&
 966                             (added + deleted == 0)) {
 967                                 total_files--;
 968                         } else {
 969                                 adds += added;
 970                                 dels += deleted;
 971                         }
 972                 }
 973         }
 974         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
 975                total_files, adds, dels);
 976 }
 977
 978 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
 979 {
 980         int i;
 981
 982         if (data->nr == 0)
 983                 return;
 984
 985         for (i = 0; i < data->nr; i++) {
 986                 struct diffstat_file *file = data->files[i];
 987
 988                 if (file->is_binary)
 989                         fprintf(options->file, "-\t-\t");
 990                 else
 991                         fprintf(options->file,
 992                                 "%d\t%d\t", file->added, file->deleted);
 993                 if (options->line_termination) {
 994                         fill_print_name(file);
 995                         if (!file->is_renamed)
 996                                 write_name_quoted(file->name, options->file,
 997                                                   options->line_termination);
 998                         else {
 999                                 fputs(file->print_name, options->file);
1000                                 putc(options->line_termination, options->file);
1001                         }
1002                 } else {
1003                         if (file->is_renamed) {
1004                                 putc('\0', options->file);
1005                                 write_name_quoted(file->from_name, options->file, '\0');
1006                         }
1007                         write_name_quoted(file->name, options->file, '\0');
1008                 }
1009         }
1010 }
1011
1012 struct dirstat_file {
1013         const char *name;
1014         unsigned long changed;
1015 };
1016
1017 struct dirstat_dir {
1018         struct dirstat_file *files;
1019         int alloc, nr, percent, cumulative;
1020 };
1021
1022 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1023 {
1024         unsigned long this_dir = 0;
1025         unsigned int sources = 0;
1026
1027         while (dir->nr) {
1028                 struct dirstat_file *f = dir->files;
1029                 int namelen = strlen(f->name);
1030                 unsigned long this;
1031                 char *slash;
1032
1033                 if (namelen < baselen)
1034                         break;
1035                 if (memcmp(f->name, base, baselen))
1036                         break;
1037                 slash = strchr(f->name + baselen, '/');
1038                 if (slash) {
1039                         int newbaselen = slash + 1 - f->name;
1040                         this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1041                         sources++;
1042                 } else {
1043                         this = f->changed;
1044                         dir->files++;
1045                         dir->nr--;
1046                         sources += 2;
1047                 }
1048                 this_dir += this;
1049         }
1050
1051         /*
1052          * We don't report dirstat's for
1053          *  - the top level
1054          *  - or cases where everything came from a single directory
1055          *    under this directory (sources == 1).
1056          */
1057         if (baselen && sources != 1) {
1058                 int permille = this_dir * 1000 / changed;
1059                 if (permille) {
1060                         int percent = permille / 10;
1061                         if (percent >= dir->percent) {
1062                                 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1063                                 if (!dir->cumulative)
1064                                         return 0;
1065                         }
1066                 }
1067         }
1068         return this_dir;
1069 }
1070
1071 static int dirstat_compare(const void *_a, const void *_b)
1072 {
1073         const struct dirstat_file *a = _a;
1074         const struct dirstat_file *b = _b;
1075         return strcmp(a->name, b->name);
1076 }
1077
1078 static void show_dirstat(struct diff_options *options)
1079 {
1080         int i;
1081         unsigned long changed;
1082         struct dirstat_dir dir;
1083         struct diff_queue_struct *q = &diff_queued_diff;
1084
1085         dir.files = NULL;
1086         dir.alloc = 0;
1087         dir.nr = 0;
1088         dir.percent = options->dirstat_percent;
1089         dir.cumulative = options->output_format & DIFF_FORMAT_CUMULATIVE;
1090
1091         changed = 0;
1092         for (i = 0; i < q->nr; i++) {
1093                 struct diff_filepair *p = q->queue[i];
1094                 const char *name;
1095                 unsigned long copied, added, damage;
1096
1097                 name = p->one->path ? p->one->path : p->two->path;
1098
1099                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1100                         diff_populate_filespec(p->one, 0);
1101                         diff_populate_filespec(p->two, 0);
1102                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1103                                                &copied, &added);
1104                         diff_free_filespec_data(p->one);
1105                         diff_free_filespec_data(p->two);
1106                 } else if (DIFF_FILE_VALID(p->one)) {
1107                         diff_populate_filespec(p->one, 1);
1108                         copied = added = 0;
1109                         diff_free_filespec_data(p->one);
1110                 } else if (DIFF_FILE_VALID(p->two)) {
1111                         diff_populate_filespec(p->two, 1);
1112                         copied = 0;
1113                         added = p->two->size;
1114                         diff_free_filespec_data(p->two);
1115                 } else
1116                         continue;
1117
1118                 /*
1119                  * Original minus copied is the removed material,
1120                  * added is the new material.  They are both damages
1121                  * made to the preimage.
1122                  */
1123                 damage = (p->one->size - copied) + added;
1124
1125                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1126                 dir.files[dir.nr].name = name;
1127                 dir.files[dir.nr].changed = damage;
1128                 changed += damage;
1129                 dir.nr++;
1130         }
1131
1132         /* This can happen even with many files, if everything was renames */
1133         if (!changed)
1134                 return;
1135
1136         /* Show all directories with more than x% of the changes */
1137         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1138         gather_dirstat(options->file, &dir, changed, "", 0);
1139 }
1140
1141 static void free_diffstat_info(struct diffstat_t *diffstat)
1142 {
1143         int i;
1144         for (i = 0; i < diffstat->nr; i++) {
1145                 struct diffstat_file *f = diffstat->files[i];
1146                 if (f->name != f->print_name)
1147                         free(f->print_name);
1148                 free(f->name);
1149                 free(f->from_name);
1150                 free(f);
1151         }
1152         free(diffstat->files);
1153 }
1154
1155 struct checkdiff_t {
1156         struct xdiff_emit_state xm;
1157         const char *filename;
1158         int lineno;
1159         struct diff_options *o;
1160         unsigned ws_rule;
1161         unsigned status;
1162         int trailing_blanks_start;
1163 };
1164
1165 static int is_conflict_marker(const char *line, unsigned long len)
1166 {
1167         char firstchar;
1168         int cnt;
1169
1170         if (len < 8)
1171                 return 0;
1172         firstchar = line[0];
1173         switch (firstchar) {
1174         case '=': case '>': case '<':
1175                 break;
1176         default:
1177                 return 0;
1178         }
1179         for (cnt = 1; cnt < 7; cnt++)
1180                 if (line[cnt] != firstchar)
1181                         return 0;
1182         /* line[0] thru line[6] are same as firstchar */
1183         if (firstchar == '=') {
1184                 /* divider between ours and theirs? */
1185                 if (len != 8 || line[7] != '\n')
1186                         return 0;
1187         } else if (len < 8 || !isspace(line[7])) {
1188                 /* not divider before ours nor after theirs */
1189                 return 0;
1190         }
1191         return 1;
1192 }
1193
1194 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1195 {
1196         struct checkdiff_t *data = priv;
1197         int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1198         const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1199         const char *reset = diff_get_color(color_diff, DIFF_RESET);
1200         const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1201         char *err;
1202
1203         if (line[0] == '+') {
1204                 unsigned bad;
1205                 data->lineno++;
1206                 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1207                         data->trailing_blanks_start = 0;
1208                 else if (!data->trailing_blanks_start)
1209                         data->trailing_blanks_start = data->lineno;
1210                 if (is_conflict_marker(line + 1, len - 1)) {
1211                         data->status |= 1;
1212                         fprintf(data->o->file,
1213                                 "%s:%d: leftover conflict marker\n",
1214                                 data->filename, data->lineno);
1215                 }
1216                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1217                 if (!bad)
1218                         return;
1219                 data->status |= bad;
1220                 err = whitespace_error_string(bad);
1221                 fprintf(data->o->file, "%s:%d: %s.\n",
1222                         data->filename, data->lineno, err);
1223                 free(err);
1224                 emit_line(data->o->file, set, reset, line, 1);
1225                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1226                               data->o->file, set, reset, ws);
1227         } else if (line[0] == ' ') {
1228                 data->lineno++;
1229                 data->trailing_blanks_start = 0;
1230         } else if (line[0] == '@') {
1231                 char *plus = strchr(line, '+');
1232                 if (plus)
1233                         data->lineno = strtol(plus, NULL, 10) - 1;
1234                 else
1235                         die("invalid diff");
1236                 data->trailing_blanks_start = 0;
1237         }
1238 }
1239
1240 static unsigned char *deflate_it(char *data,
1241                                  unsigned long size,
1242                                  unsigned long *result_size)
1243 {
1244         int bound;
1245         unsigned char *deflated;
1246         z_stream stream;
1247
1248         memset(&stream, 0, sizeof(stream));
1249         deflateInit(&stream, zlib_compression_level);
1250         bound = deflateBound(&stream, size);
1251         deflated = xmalloc(bound);
1252         stream.next_out = deflated;
1253         stream.avail_out = bound;
1254
1255         stream.next_in = (unsigned char *)data;
1256         stream.avail_in = size;
1257         while (deflate(&stream, Z_FINISH) == Z_OK)
1258                 ; /* nothing */
1259         deflateEnd(&stream);
1260         *result_size = stream.total_out;
1261         return deflated;
1262 }
1263
1264 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1265 {
1266         void *cp;
1267         void *delta;
1268         void *deflated;
1269         void *data;
1270         unsigned long orig_size;
1271         unsigned long delta_size;
1272         unsigned long deflate_size;
1273         unsigned long data_size;
1274
1275         /* We could do deflated delta, or we could do just deflated two,
1276          * whichever is smaller.
1277          */
1278         delta = NULL;
1279         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1280         if (one->size && two->size) {
1281                 delta = diff_delta(one->ptr, one->size,
1282                                    two->ptr, two->size,
1283                                    &delta_size, deflate_size);
1284                 if (delta) {
1285                         void *to_free = delta;
1286                         orig_size = delta_size;
1287                         delta = deflate_it(delta, delta_size, &delta_size);
1288                         free(to_free);
1289                 }
1290         }
1291
1292         if (delta && delta_size < deflate_size) {
1293                 fprintf(file, "delta %lu\n", orig_size);
1294                 free(deflated);
1295                 data = delta;
1296                 data_size = delta_size;
1297         }
1298         else {
1299                 fprintf(file, "literal %lu\n", two->size);
1300                 free(delta);
1301                 data = deflated;
1302                 data_size = deflate_size;
1303         }
1304
1305         /* emit data encoded in base85 */
1306         cp = data;
1307         while (data_size) {
1308                 int bytes = (52 < data_size) ? 52 : data_size;
1309                 char line[70];
1310                 data_size -= bytes;
1311                 if (bytes <= 26)
1312                         line[0] = bytes + 'A' - 1;
1313                 else
1314                         line[0] = bytes - 26 + 'a' - 1;
1315                 encode_85(line + 1, cp, bytes);
1316                 cp = (char *) cp + bytes;
1317                 fputs(line, file);
1318                 fputc('\n', file);
1319         }
1320         fprintf(file, "\n");
1321         free(data);
1322 }
1323
1324 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1325 {
1326         fprintf(file, "GIT binary patch\n");
1327         emit_binary_diff_body(file, one, two);
1328         emit_binary_diff_body(file, two, one);
1329 }
1330
1331 static void setup_diff_attr_check(struct git_attr_check *check)
1332 {
1333         static struct git_attr *attr_diff;
1334
1335         if (!attr_diff) {
1336                 attr_diff = git_attr("diff", 4);
1337         }
1338         check[0].attr = attr_diff;
1339 }
1340
1341 static void diff_filespec_check_attr(struct diff_filespec *one)
1342 {
1343         struct git_attr_check attr_diff_check;
1344         int check_from_data = 0;
1345
1346         if (one->checked_attr)
1347                 return;
1348
1349         setup_diff_attr_check(&attr_diff_check);
1350         one->is_binary = 0;
1351         one->funcname_pattern_ident = NULL;
1352
1353         if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1354                 const char *value;
1355
1356                 /* binaryness */
1357                 value = attr_diff_check.value;
1358                 if (ATTR_TRUE(value))
1359                         ;
1360                 else if (ATTR_FALSE(value))
1361                         one->is_binary = 1;
1362                 else
1363                         check_from_data = 1;
1364
1365                 /* funcname pattern ident */
1366                 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1367                         ;
1368                 else
1369                         one->funcname_pattern_ident = value;
1370         }
1371
1372         if (check_from_data) {
1373                 if (!one->data && DIFF_FILE_VALID(one))
1374                         diff_populate_filespec(one, 0);
1375
1376                 if (one->data)
1377                         one->is_binary = buffer_is_binary(one->data, one->size);
1378         }
1379 }
1380
1381 int diff_filespec_is_binary(struct diff_filespec *one)
1382 {
1383         diff_filespec_check_attr(one);
1384         return one->is_binary;
1385 }
1386
1387 static const char *funcname_pattern(const char *ident)
1388 {
1389         struct funcname_pattern *pp;
1390
1391         for (pp = funcname_pattern_list; pp; pp = pp->next)
1392                 if (!strcmp(ident, pp->name))
1393                         return pp->pattern;
1394         return NULL;
1395 }
1396
1397 static struct builtin_funcname_pattern {
1398         const char *name;
1399         const char *pattern;
1400 } builtin_funcname_pattern[] = {
1401         { "java", "!^[  ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1402                         "new\\|return\\|switch\\|throw\\|while\\)\n"
1403                         "^[     ]*\\(\\([       ]*"
1404                         "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1405                         "[      ]*([^;]*\\)$" },
1406         { "pascal", "^\\(\\(procedure\\|function\\|constructor\\|"
1407                         "destructor\\|interface\\|implementation\\|"
1408                         "initialization\\|finalization\\)[ \t]*.*\\)$"
1409                         "\\|"
1410                         "^\\(.*=[ \t]*\\(class\\|record\\).*\\)$"
1411                         },
1412         { "bibtex", "\\(@[a-zA-Z]\\{1,\\}[ \t]*{\\{0,1\\}[ \t]*[^ \t\"@',\\#}{~%]*\\).*$" },
1413         { "tex", "^\\(\\\\\\(\\(sub\\)*section\\|chapter\\|part\\)\\*\\{0,1\\}{.*\\)$" },
1414         { "ruby", "^\\s*\\(\\(class\\|module\\|def\\)\\s.*\\)$" },
1415 };
1416
1417 static const char *diff_funcname_pattern(struct diff_filespec *one)
1418 {
1419         const char *ident, *pattern;
1420         int i;
1421
1422         diff_filespec_check_attr(one);
1423         ident = one->funcname_pattern_ident;
1424
1425         if (!ident)
1426                 /*
1427                  * If the config file has "funcname.default" defined, that
1428                  * regexp is used; otherwise NULL is returned and xemit uses
1429                  * the built-in default.
1430                  */
1431                 return funcname_pattern("default");
1432
1433         /* Look up custom "funcname.$ident" regexp from config. */
1434         pattern = funcname_pattern(ident);
1435         if (pattern)
1436                 return pattern;
1437
1438         /*
1439          * And define built-in fallback patterns here.  Note that
1440          * these can be overridden by the user's config settings.
1441          */
1442         for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1443                 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1444                         return builtin_funcname_pattern[i].pattern;
1445
1446         return NULL;
1447 }
1448
1449 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1450 {
1451         if (!options->a_prefix)
1452                 options->a_prefix = a;
1453         if (!options->b_prefix)
1454                 options->b_prefix = b;
1455 }
1456
1457 static void builtin_diff(const char *name_a,
1458                          const char *name_b,
1459                          struct diff_filespec *one,
1460                          struct diff_filespec *two,
1461                          const char *xfrm_msg,
1462                          struct diff_options *o,
1463                          int complete_rewrite)
1464 {
1465         mmfile_t mf1, mf2;
1466         const char *lbl[2];
1467         char *a_one, *b_two;
1468         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1469         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1470         const char *a_prefix, *b_prefix;
1471
1472         diff_set_mnemonic_prefix(o, "a/", "b/");
1473         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1474                 a_prefix = o->b_prefix;
1475                 b_prefix = o->a_prefix;
1476         } else {
1477                 a_prefix = o->a_prefix;
1478                 b_prefix = o->b_prefix;
1479         }
1480
1481         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1482         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1483         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1484         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1485         fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1486         if (lbl[0][0] == '/') {
1487                 /* /dev/null */
1488                 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1489                 if (xfrm_msg && xfrm_msg[0])
1490                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1491         }
1492         else if (lbl[1][0] == '/') {
1493                 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1494                 if (xfrm_msg && xfrm_msg[0])
1495                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1496         }
1497         else {
1498                 if (one->mode != two->mode) {
1499                         fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1500                         fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1501                 }
1502                 if (xfrm_msg && xfrm_msg[0])
1503                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1504                 /*
1505                  * we do not run diff between different kind
1506                  * of objects.
1507                  */
1508                 if ((one->mode ^ two->mode) & S_IFMT)
1509                         goto free_ab_and_return;
1510                 if (complete_rewrite) {
1511                         emit_rewrite_diff(name_a, name_b, one, two, o);
1512                         o->found_changes = 1;
1513                         goto free_ab_and_return;
1514                 }
1515         }
1516
1517         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1518                 die("unable to read files to diff");
1519
1520         if (!DIFF_OPT_TST(o, TEXT) &&
1521             (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1522                 /* Quite common confusing case */
1523                 if (mf1.size == mf2.size &&
1524                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1525                         goto free_ab_and_return;
1526                 if (DIFF_OPT_TST(o, BINARY))
1527                         emit_binary_diff(o->file, &mf1, &mf2);
1528                 else
1529                         fprintf(o->file, "Binary files %s and %s differ\n",
1530                                 lbl[0], lbl[1]);
1531                 o->found_changes = 1;
1532         }
1533         else {
1534                 /* Crazy xdl interfaces.. */
1535                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1536                 xpparam_t xpp;
1537                 xdemitconf_t xecfg;
1538                 xdemitcb_t ecb;
1539                 struct emit_callback ecbdata;
1540                 const char *funcname_pattern;
1541
1542                 funcname_pattern = diff_funcname_pattern(one);
1543                 if (!funcname_pattern)
1544                         funcname_pattern = diff_funcname_pattern(two);
1545
1546                 memset(&xecfg, 0, sizeof(xecfg));
1547                 memset(&ecbdata, 0, sizeof(ecbdata));
1548                 ecbdata.label_path = lbl;
1549                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1550                 ecbdata.found_changesp = &o->found_changes;
1551                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1552                 ecbdata.file = o->file;
1553                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1554                 xecfg.ctxlen = o->context;
1555                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1556                 if (funcname_pattern)
1557                         xdiff_set_find_func(&xecfg, funcname_pattern);
1558                 if (!diffopts)
1559                         ;
1560                 else if (!prefixcmp(diffopts, "--unified="))
1561                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1562                 else if (!prefixcmp(diffopts, "-u"))
1563                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1564                 ecb.outf = xdiff_outf;
1565                 ecb.priv = &ecbdata;
1566                 ecbdata.xm.consume = fn_out_consume;
1567                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1568                         ecbdata.diff_words =
1569                                 xcalloc(1, sizeof(struct diff_words_data));
1570                         ecbdata.diff_words->file = o->file;
1571                 }
1572                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1573                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1574                         free_diff_words_data(&ecbdata);
1575         }
1576
1577  free_ab_and_return:
1578         diff_free_filespec_data(one);
1579         diff_free_filespec_data(two);
1580         free(a_one);
1581         free(b_two);
1582         return;
1583 }
1584
1585 static void builtin_diffstat(const char *name_a, const char *name_b,
1586                              struct diff_filespec *one,
1587                              struct diff_filespec *two,
1588                              struct diffstat_t *diffstat,
1589                              struct diff_options *o,
1590                              int complete_rewrite)
1591 {
1592         mmfile_t mf1, mf2;
1593         struct diffstat_file *data;
1594
1595         data = diffstat_add(diffstat, name_a, name_b);
1596
1597         if (!one || !two) {
1598                 data->is_unmerged = 1;
1599                 return;
1600         }
1601         if (complete_rewrite) {
1602                 diff_populate_filespec(one, 0);
1603                 diff_populate_filespec(two, 0);
1604                 data->deleted = count_lines(one->data, one->size);
1605                 data->added = count_lines(two->data, two->size);
1606                 goto free_and_return;
1607         }
1608         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1609                 die("unable to read files to diff");
1610
1611         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1612                 data->is_binary = 1;
1613                 data->added = mf2.size;
1614                 data->deleted = mf1.size;
1615         } else {
1616                 /* Crazy xdl interfaces.. */
1617                 xpparam_t xpp;
1618                 xdemitconf_t xecfg;
1619                 xdemitcb_t ecb;
1620
1621                 memset(&xecfg, 0, sizeof(xecfg));
1622                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1623                 ecb.outf = xdiff_outf;
1624                 ecb.priv = diffstat;
1625                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1626         }
1627
1628  free_and_return:
1629         diff_free_filespec_data(one);
1630         diff_free_filespec_data(two);
1631 }
1632
1633 static void builtin_checkdiff(const char *name_a, const char *name_b,
1634                               const char *attr_path,
1635                               struct diff_filespec *one,
1636                               struct diff_filespec *two,
1637                               struct diff_options *o)
1638 {
1639         mmfile_t mf1, mf2;
1640         struct checkdiff_t data;
1641
1642         if (!two)
1643                 return;
1644
1645         memset(&data, 0, sizeof(data));
1646         data.xm.consume = checkdiff_consume;
1647         data.filename = name_b ? name_b : name_a;
1648         data.lineno = 0;
1649         data.o = o;
1650         data.ws_rule = whitespace_rule(attr_path);
1651
1652         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1653                 die("unable to read files to diff");
1654
1655         /*
1656          * All the other codepaths check both sides, but not checking
1657          * the "old" side here is deliberate.  We are checking the newly
1658          * introduced changes, and as long as the "new" side is text, we
1659          * can and should check what it introduces.
1660          */
1661         if (diff_filespec_is_binary(two))
1662                 goto free_and_return;
1663         else {
1664                 /* Crazy xdl interfaces.. */
1665                 xpparam_t xpp;
1666                 xdemitconf_t xecfg;
1667                 xdemitcb_t ecb;
1668
1669                 memset(&xecfg, 0, sizeof(xecfg));
1670                 xecfg.ctxlen = 1; /* at least one context line */
1671                 xpp.flags = XDF_NEED_MINIMAL;
1672                 ecb.outf = xdiff_outf;
1673                 ecb.priv = &data;
1674                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1675
1676                 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1677                     data.trailing_blanks_start) {
1678                         fprintf(o->file, "%s:%d: ends with blank lines.\n",
1679                                 data.filename, data.trailing_blanks_start);
1680                         data.status = 1; /* report errors */
1681                 }
1682         }
1683  free_and_return:
1684         diff_free_filespec_data(one);
1685         diff_free_filespec_data(two);
1686         if (data.status)
1687                 DIFF_OPT_SET(o, CHECK_FAILED);
1688 }
1689
1690 struct diff_filespec *alloc_filespec(const char *path)
1691 {
1692         int namelen = strlen(path);
1693         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1694
1695         memset(spec, 0, sizeof(*spec));
1696         spec->path = (char *)(spec + 1);
1697         memcpy(spec->path, path, namelen+1);
1698         spec->count = 1;
1699         return spec;
1700 }
1701
1702 void free_filespec(struct diff_filespec *spec)
1703 {
1704         if (!--spec->count) {
1705                 diff_free_filespec_data(spec);
1706                 free(spec);
1707         }
1708 }
1709
1710 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1711                    unsigned short mode)
1712 {
1713         if (mode) {
1714                 spec->mode = canon_mode(mode);
1715                 hashcpy(spec->sha1, sha1);
1716                 spec->sha1_valid = !is_null_sha1(sha1);
1717         }
1718 }
1719
1720 /*
1721  * Given a name and sha1 pair, if the index tells us the file in
1722  * the work tree has that object contents, return true, so that
1723  * prepare_temp_file() does not have to inflate and extract.
1724  */
1725 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1726 {
1727         struct cache_entry *ce;
1728         struct stat st;
1729         int pos, len;
1730
1731         /* We do not read the cache ourselves here, because the
1732          * benchmark with my previous version that always reads cache
1733          * shows that it makes things worse for diff-tree comparing
1734          * two linux-2.6 kernel trees in an already checked out work
1735          * tree.  This is because most diff-tree comparisons deal with
1736          * only a small number of files, while reading the cache is
1737          * expensive for a large project, and its cost outweighs the
1738          * savings we get by not inflating the object to a temporary
1739          * file.  Practically, this code only helps when we are used
1740          * by diff-cache --cached, which does read the cache before
1741          * calling us.
1742          */
1743         if (!active_cache)
1744                 return 0;
1745
1746         /* We want to avoid the working directory if our caller
1747          * doesn't need the data in a normal file, this system
1748          * is rather slow with its stat/open/mmap/close syscalls,
1749          * and the object is contained in a pack file.  The pack
1750          * is probably already open and will be faster to obtain
1751          * the data through than the working directory.  Loose
1752          * objects however would tend to be slower as they need
1753          * to be individually opened and inflated.
1754          */
1755         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1756                 return 0;
1757
1758         len = strlen(name);
1759         pos = cache_name_pos(name, len);
1760         if (pos < 0)
1761                 return 0;
1762         ce = active_cache[pos];
1763
1764         /*
1765          * This is not the sha1 we are looking for, or
1766          * unreusable because it is not a regular file.
1767          */
1768         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1769                 return 0;
1770
1771         /*
1772          * If ce matches the file in the work tree, we can reuse it.
1773          */
1774         if (ce_uptodate(ce) ||
1775             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1776                 return 1;
1777
1778         return 0;
1779 }
1780
1781 static int populate_from_stdin(struct diff_filespec *s)
1782 {
1783         struct strbuf buf;
1784         size_t size = 0;
1785
1786         strbuf_init(&buf, 0);
1787         if (strbuf_read(&buf, 0, 0) < 0)
1788                 return error("error while reading from stdin %s",
1789                                      strerror(errno));
1790
1791         s->should_munmap = 0;
1792         s->data = strbuf_detach(&buf, &size);
1793         s->size = size;
1794         s->should_free = 1;
1795         return 0;
1796 }
1797
1798 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1799 {
1800         int len;
1801         char *data = xmalloc(100);
1802         len = snprintf(data, 100,
1803                 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1804         s->data = data;
1805         s->size = len;
1806         s->should_free = 1;
1807         if (size_only) {
1808                 s->data = NULL;
1809                 free(data);
1810         }
1811         return 0;
1812 }
1813
1814 /*
1815  * While doing rename detection and pickaxe operation, we may need to
1816  * grab the data for the blob (or file) for our own in-core comparison.
1817  * diff_filespec has data and size fields for this purpose.
1818  */
1819 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1820 {
1821         int err = 0;
1822         if (!DIFF_FILE_VALID(s))
1823                 die("internal error: asking to populate invalid file.");
1824         if (S_ISDIR(s->mode))
1825                 return -1;
1826
1827         if (s->data)
1828                 return 0;
1829
1830         if (size_only && 0 < s->size)
1831                 return 0;
1832
1833         if (S_ISGITLINK(s->mode))
1834                 return diff_populate_gitlink(s, size_only);
1835
1836         if (!s->sha1_valid ||
1837             reuse_worktree_file(s->path, s->sha1, 0)) {
1838                 struct strbuf buf;
1839                 struct stat st;
1840                 int fd;
1841
1842                 if (!strcmp(s->path, "-"))
1843                         return populate_from_stdin(s);
1844
1845                 if (lstat(s->path, &st) < 0) {
1846                         if (errno == ENOENT) {
1847                         err_empty:
1848                                 err = -1;
1849                         empty:
1850                                 s->data = (char *)"";
1851                                 s->size = 0;
1852                                 return err;
1853                         }
1854                 }
1855                 s->size = xsize_t(st.st_size);
1856                 if (!s->size)
1857                         goto empty;
1858                 if (size_only)
1859                         return 0;
1860                 if (S_ISLNK(st.st_mode)) {
1861                         int ret;
1862                         s->data = xmalloc(s->size);
1863                         s->should_free = 1;
1864                         ret = readlink(s->path, s->data, s->size);
1865                         if (ret < 0) {
1866                                 free(s->data);
1867                                 goto err_empty;
1868                         }
1869                         return 0;
1870                 }
1871                 fd = open(s->path, O_RDONLY);
1872                 if (fd < 0)
1873                         goto err_empty;
1874                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1875                 close(fd);
1876                 s->should_munmap = 1;
1877
1878                 /*
1879                  * Convert from working tree format to canonical git format
1880                  */
1881                 strbuf_init(&buf, 0);
1882                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1883                         size_t size = 0;
1884                         munmap(s->data, s->size);
1885                         s->should_munmap = 0;
1886                         s->data = strbuf_detach(&buf, &size);
1887                         s->size = size;
1888                         s->should_free = 1;
1889                 }
1890         }
1891         else {
1892                 enum object_type type;
1893                 if (size_only)
1894                         type = sha1_object_info(s->sha1, &s->size);
1895                 else {
1896                         s->data = read_sha1_file(s->sha1, &type, &s->size);
1897                         s->should_free = 1;
1898                 }
1899         }
1900         return 0;
1901 }
1902
1903 void diff_free_filespec_blob(struct diff_filespec *s)
1904 {
1905         if (s->should_free)
1906                 free(s->data);
1907         else if (s->should_munmap)
1908                 munmap(s->data, s->size);
1909
1910         if (s->should_free || s->should_munmap) {
1911                 s->should_free = s->should_munmap = 0;
1912                 s->data = NULL;
1913         }
1914 }
1915
1916 void diff_free_filespec_data(struct diff_filespec *s)
1917 {
1918         diff_free_filespec_blob(s);
1919         free(s->cnt_data);
1920         s->cnt_data = NULL;
1921 }
1922
1923 static void prep_temp_blob(struct diff_tempfile *temp,
1924                            void *blob,
1925                            unsigned long size,
1926                            const unsigned char *sha1,
1927                            int mode)
1928 {
1929         int fd;
1930
1931         fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1932         if (fd < 0)
1933                 die("unable to create temp-file: %s", strerror(errno));
1934         if (write_in_full(fd, blob, size) != size)
1935                 die("unable to write temp-file");
1936         close(fd);
1937         temp->name = temp->tmp_path;
1938         strcpy(temp->hex, sha1_to_hex(sha1));
1939         temp->hex[40] = 0;
1940         sprintf(temp->mode, "%06o", mode);
1941 }
1942
1943 static void prepare_temp_file(const char *name,
1944                               struct diff_tempfile *temp,
1945                               struct diff_filespec *one)
1946 {
1947         if (!DIFF_FILE_VALID(one)) {
1948         not_a_valid_file:
1949                 /* A '-' entry produces this for file-2, and
1950                  * a '+' entry produces this for file-1.
1951                  */
1952                 temp->name = "/dev/null";
1953                 strcpy(temp->hex, ".");
1954                 strcpy(temp->mode, ".");
1955                 return;
1956         }
1957
1958         if (!one->sha1_valid ||
1959             reuse_worktree_file(name, one->sha1, 1)) {
1960                 struct stat st;
1961                 if (lstat(name, &st) < 0) {
1962                         if (errno == ENOENT)
1963                                 goto not_a_valid_file;
1964                         die("stat(%s): %s", name, strerror(errno));
1965                 }
1966                 if (S_ISLNK(st.st_mode)) {
1967                         int ret;
1968                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1969                         size_t sz = xsize_t(st.st_size);
1970                         if (sizeof(buf) <= st.st_size)
1971                                 die("symlink too long: %s", name);
1972                         ret = readlink(name, buf, sz);
1973                         if (ret < 0)
1974                                 die("readlink(%s)", name);
1975                         prep_temp_blob(temp, buf, sz,
1976                                        (one->sha1_valid ?
1977                                         one->sha1 : null_sha1),
1978                                        (one->sha1_valid ?
1979                                         one->mode : S_IFLNK));
1980                 }
1981                 else {
1982                         /* we can borrow from the file in the work tree */
1983                         temp->name = name;
1984                         if (!one->sha1_valid)
1985                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
1986                         else
1987                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
1988                         /* Even though we may sometimes borrow the
1989                          * contents from the work tree, we always want
1990                          * one->mode.  mode is trustworthy even when
1991                          * !(one->sha1_valid), as long as
1992                          * DIFF_FILE_VALID(one).
1993                          */
1994                         sprintf(temp->mode, "%06o", one->mode);
1995                 }
1996                 return;
1997         }
1998         else {
1999                 if (diff_populate_filespec(one, 0))
2000                         die("cannot read data blob for %s", one->path);
2001                 prep_temp_blob(temp, one->data, one->size,
2002                                one->sha1, one->mode);
2003         }
2004 }
2005
2006 static void remove_tempfile(void)
2007 {
2008         int i;
2009
2010         for (i = 0; i < 2; i++)
2011                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
2012                         unlink(diff_temp[i].name);
2013                         diff_temp[i].name = NULL;
2014                 }
2015 }
2016
2017 static void remove_tempfile_on_signal(int signo)
2018 {
2019         remove_tempfile();
2020         signal(SIGINT, SIG_DFL);
2021         raise(signo);
2022 }
2023
2024 /* An external diff command takes:
2025  *
2026  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2027  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2028  *
2029  */
2030 static void run_external_diff(const char *pgm,
2031                               const char *name,
2032                               const char *other,
2033                               struct diff_filespec *one,
2034                               struct diff_filespec *two,
2035                               const char *xfrm_msg,
2036                               int complete_rewrite)
2037 {
2038         const char *spawn_arg[10];
2039         struct diff_tempfile *temp = diff_temp;
2040         int retval;
2041         static int atexit_asked = 0;
2042         const char *othername;
2043         const char **arg = &spawn_arg[0];
2044
2045         othername = (other? other : name);
2046         if (one && two) {
2047                 prepare_temp_file(name, &temp[0], one);
2048                 prepare_temp_file(othername, &temp[1], two);
2049                 if (! atexit_asked &&
2050                     (temp[0].name == temp[0].tmp_path ||
2051                      temp[1].name == temp[1].tmp_path)) {
2052                         atexit_asked = 1;
2053                         atexit(remove_tempfile);
2054                 }
2055                 signal(SIGINT, remove_tempfile_on_signal);
2056         }
2057
2058         if (one && two) {
2059                 *arg++ = pgm;
2060                 *arg++ = name;
2061                 *arg++ = temp[0].name;
2062                 *arg++ = temp[0].hex;
2063                 *arg++ = temp[0].mode;
2064                 *arg++ = temp[1].name;
2065                 *arg++ = temp[1].hex;
2066                 *arg++ = temp[1].mode;
2067                 if (other) {
2068                         *arg++ = other;
2069                         *arg++ = xfrm_msg;
2070                 }
2071         } else {
2072                 *arg++ = pgm;
2073                 *arg++ = name;
2074         }
2075         *arg = NULL;
2076         fflush(NULL);
2077         retval = run_command_v_opt(spawn_arg, 0);
2078         remove_tempfile();
2079         if (retval) {
2080                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2081                 exit(1);
2082         }
2083 }
2084
2085 static const char *external_diff_attr(const char *name)
2086 {
2087         struct git_attr_check attr_diff_check;
2088
2089         if (!name)
2090                 return NULL;
2091
2092         setup_diff_attr_check(&attr_diff_check);
2093         if (!git_checkattr(name, 1, &attr_diff_check)) {
2094                 const char *value = attr_diff_check.value;
2095                 if (!ATTR_TRUE(value) &&
2096                     !ATTR_FALSE(value) &&
2097                     !ATTR_UNSET(value)) {
2098                         struct ll_diff_driver *drv;
2099
2100                         for (drv = user_diff; drv; drv = drv->next)
2101                                 if (!strcmp(drv->name, value))
2102                                         return drv->cmd;
2103                 }
2104         }
2105         return NULL;
2106 }
2107
2108 static void run_diff_cmd(const char *pgm,
2109                          const char *name,
2110                          const char *other,
2111                          const char *attr_path,
2112                          struct diff_filespec *one,
2113                          struct diff_filespec *two,
2114                          const char *xfrm_msg,
2115                          struct diff_options *o,
2116                          int complete_rewrite)
2117 {
2118         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2119                 pgm = NULL;
2120         else {
2121                 const char *cmd = external_diff_attr(attr_path);
2122                 if (cmd)
2123                         pgm = cmd;
2124         }
2125
2126         if (pgm) {
2127                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2128                                   complete_rewrite);
2129                 return;
2130         }
2131         if (one && two)
2132                 builtin_diff(name, other ? other : name,
2133                              one, two, xfrm_msg, o, complete_rewrite);
2134         else
2135                 fprintf(o->file, "* Unmerged path %s\n", name);
2136 }
2137
2138 static void diff_fill_sha1_info(struct diff_filespec *one)
2139 {
2140         if (DIFF_FILE_VALID(one)) {
2141                 if (!one->sha1_valid) {
2142                         struct stat st;
2143                         if (!strcmp(one->path, "-")) {
2144                                 hashcpy(one->sha1, null_sha1);
2145                                 return;
2146                         }
2147                         if (lstat(one->path, &st) < 0)
2148                                 die("stat %s", one->path);
2149                         if (index_path(one->sha1, one->path, &st, 0))
2150                                 die("cannot hash %s\n", one->path);
2151                 }
2152         }
2153         else
2154                 hashclr(one->sha1);
2155 }
2156
2157 static int similarity_index(struct diff_filepair *p)
2158 {
2159         return p->score * 100 / MAX_SCORE;
2160 }
2161
2162 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2163 {
2164         /* Strip the prefix but do not molest /dev/null and absolute paths */
2165         if (*namep && **namep != '/')
2166                 *namep += prefix_length;
2167         if (*otherp && **otherp != '/')
2168                 *otherp += prefix_length;
2169 }
2170
2171 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2172 {
2173         const char *pgm = external_diff();
2174         struct strbuf msg;
2175         char *xfrm_msg;
2176         struct diff_filespec *one = p->one;
2177         struct diff_filespec *two = p->two;
2178         const char *name;
2179         const char *other;
2180         const char *attr_path;
2181         int complete_rewrite = 0;
2182
2183         name  = p->one->path;
2184         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2185         attr_path = name;
2186         if (o->prefix_length)
2187                 strip_prefix(o->prefix_length, &name, &other);
2188
2189         if (DIFF_PAIR_UNMERGED(p)) {
2190                 run_diff_cmd(pgm, name, NULL, attr_path,
2191                              NULL, NULL, NULL, o, 0);
2192                 return;
2193         }
2194
2195         diff_fill_sha1_info(one);
2196         diff_fill_sha1_info(two);
2197
2198         strbuf_init(&msg, PATH_MAX * 2 + 300);
2199         switch (p->status) {
2200         case DIFF_STATUS_COPIED:
2201                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2202                 strbuf_addstr(&msg, "\ncopy from ");
2203                 quote_c_style(name, &msg, NULL, 0);
2204                 strbuf_addstr(&msg, "\ncopy to ");
2205                 quote_c_style(other, &msg, NULL, 0);
2206                 strbuf_addch(&msg, '\n');
2207                 break;
2208         case DIFF_STATUS_RENAMED:
2209                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2210                 strbuf_addstr(&msg, "\nrename from ");
2211                 quote_c_style(name, &msg, NULL, 0);
2212                 strbuf_addstr(&msg, "\nrename to ");
2213                 quote_c_style(other, &msg, NULL, 0);
2214                 strbuf_addch(&msg, '\n');
2215                 break;
2216         case DIFF_STATUS_MODIFIED:
2217                 if (p->score) {
2218                         strbuf_addf(&msg, "dissimilarity index %d%%\n",
2219                                         similarity_index(p));
2220                         complete_rewrite = 1;
2221                         break;
2222                 }
2223                 /* fallthru */
2224         default:
2225                 /* nothing */
2226                 ;
2227         }
2228
2229         if (hashcmp(one->sha1, two->sha1)) {
2230                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2231
2232                 if (DIFF_OPT_TST(o, BINARY)) {
2233                         mmfile_t mf;
2234                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2235                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2236                                 abbrev = 40;
2237                 }
2238                 strbuf_addf(&msg, "index %.*s..%.*s",
2239                                 abbrev, sha1_to_hex(one->sha1),
2240                                 abbrev, sha1_to_hex(two->sha1));
2241                 if (one->mode == two->mode)
2242                         strbuf_addf(&msg, " %06o", one->mode);
2243                 strbuf_addch(&msg, '\n');
2244         }
2245
2246         if (msg.len)
2247                 strbuf_setlen(&msg, msg.len - 1);
2248         xfrm_msg = msg.len ? msg.buf : NULL;
2249
2250         if (!pgm &&
2251             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2252             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2253                 /* a filepair that changes between file and symlink
2254                  * needs to be split into deletion and creation.
2255                  */
2256                 struct diff_filespec *null = alloc_filespec(two->path);
2257                 run_diff_cmd(NULL, name, other, attr_path,
2258                              one, null, xfrm_msg, o, 0);
2259                 free(null);
2260                 null = alloc_filespec(one->path);
2261                 run_diff_cmd(NULL, name, other, attr_path,
2262                              null, two, xfrm_msg, o, 0);
2263                 free(null);
2264         }
2265         else
2266                 run_diff_cmd(pgm, name, other, attr_path,
2267                              one, two, xfrm_msg, o, complete_rewrite);
2268
2269         strbuf_release(&msg);
2270 }
2271
2272 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2273                          struct diffstat_t *diffstat)
2274 {
2275         const char *name;
2276         const char *other;
2277         int complete_rewrite = 0;
2278
2279         if (DIFF_PAIR_UNMERGED(p)) {
2280                 /* unmerged */
2281                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2282                 return;
2283         }
2284
2285         name = p->one->path;
2286         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2287
2288         if (o->prefix_length)
2289                 strip_prefix(o->prefix_length, &name, &other);
2290
2291         diff_fill_sha1_info(p->one);
2292         diff_fill_sha1_info(p->two);
2293
2294         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2295                 complete_rewrite = 1;
2296         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2297 }
2298
2299 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2300 {
2301         const char *name;
2302         const char *other;
2303         const char *attr_path;
2304
2305         if (DIFF_PAIR_UNMERGED(p)) {
2306                 /* unmerged */
2307                 return;
2308         }
2309
2310         name = p->one->path;
2311         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2312         attr_path = other ? other : name;
2313
2314         if (o->prefix_length)
2315                 strip_prefix(o->prefix_length, &name, &other);
2316
2317         diff_fill_sha1_info(p->one);
2318         diff_fill_sha1_info(p->two);
2319
2320         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2321 }
2322
2323 void diff_setup(struct diff_options *options)
2324 {
2325         memset(options, 0, sizeof(*options));
2326
2327         options->file = stdout;
2328
2329         options->line_termination = '\n';
2330         options->break_opt = -1;
2331         options->rename_limit = -1;
2332         options->dirstat_percent = 3;
2333         options->context = 3;
2334
2335         options->change = diff_change;
2336         options->add_remove = diff_addremove;
2337         if (diff_use_color_default > 0)
2338                 DIFF_OPT_SET(options, COLOR_DIFF);
2339         else
2340                 DIFF_OPT_CLR(options, COLOR_DIFF);
2341         options->detect_rename = diff_detect_rename_default;
2342
2343         if (!diff_mnemonic_prefix) {
2344                 options->a_prefix = "a/";
2345                 options->b_prefix = "b/";
2346         }
2347 }
2348
2349 int diff_setup_done(struct diff_options *options)
2350 {
2351         int count = 0;
2352
2353         if (options->output_format & DIFF_FORMAT_NAME)
2354                 count++;
2355         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2356                 count++;
2357         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2358                 count++;
2359         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2360                 count++;
2361         if (count > 1)
2362                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2363
2364         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2365                 options->detect_rename = DIFF_DETECT_COPY;
2366
2367         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2368                 options->prefix = NULL;
2369         if (options->prefix)
2370                 options->prefix_length = strlen(options->prefix);
2371         else
2372                 options->prefix_length = 0;
2373
2374         if (options->output_format & (DIFF_FORMAT_NAME |
2375                                       DIFF_FORMAT_NAME_STATUS |
2376                                       DIFF_FORMAT_CHECKDIFF |
2377                                       DIFF_FORMAT_NO_OUTPUT))
2378                 options->output_format &= ~(DIFF_FORMAT_RAW |
2379                                             DIFF_FORMAT_NUMSTAT |
2380                                             DIFF_FORMAT_DIFFSTAT |
2381                                             DIFF_FORMAT_SHORTSTAT |
2382                                             DIFF_FORMAT_DIRSTAT |
2383                                             DIFF_FORMAT_SUMMARY |
2384                                             DIFF_FORMAT_PATCH);
2385
2386         /*
2387          * These cases always need recursive; we do not drop caller-supplied
2388          * recursive bits for other formats here.
2389          */
2390         if (options->output_format & (DIFF_FORMAT_PATCH |
2391                                       DIFF_FORMAT_NUMSTAT |
2392                                       DIFF_FORMAT_DIFFSTAT |
2393                                       DIFF_FORMAT_SHORTSTAT |
2394                                       DIFF_FORMAT_DIRSTAT |
2395                                       DIFF_FORMAT_SUMMARY |
2396                                       DIFF_FORMAT_CHECKDIFF))
2397                 DIFF_OPT_SET(options, RECURSIVE);
2398         /*
2399          * Also pickaxe would not work very well if you do not say recursive
2400          */
2401         if (options->pickaxe)
2402                 DIFF_OPT_SET(options, RECURSIVE);
2403
2404         if (options->detect_rename && options->rename_limit < 0)
2405                 options->rename_limit = diff_rename_limit_default;
2406         if (options->setup & DIFF_SETUP_USE_CACHE) {
2407                 if (!active_cache)
2408                         /* read-cache does not die even when it fails
2409                          * so it is safe for us to do this here.  Also
2410                          * it does not smudge active_cache or active_nr
2411                          * when it fails, so we do not have to worry about
2412                          * cleaning it up ourselves either.
2413                          */
2414                         read_cache();
2415         }
2416         if (options->abbrev <= 0 || 40 < options->abbrev)
2417                 options->abbrev = 40; /* full */
2418
2419         /*
2420          * It does not make sense to show the first hit we happened
2421          * to have found.  It does not make sense not to return with
2422          * exit code in such a case either.
2423          */
2424         if (DIFF_OPT_TST(options, QUIET)) {
2425                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2426                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2427         }
2428
2429         /*
2430          * If we postprocess in diffcore, we cannot simply return
2431          * upon the first hit.  We need to run diff as usual.
2432          */
2433         if (options->pickaxe || options->filter)
2434                 DIFF_OPT_CLR(options, QUIET);
2435
2436         return 0;
2437 }
2438
2439 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2440 {
2441         char c, *eq;
2442         int len;
2443
2444         if (*arg != '-')
2445                 return 0;
2446         c = *++arg;
2447         if (!c)
2448                 return 0;
2449         if (c == arg_short) {
2450                 c = *++arg;
2451                 if (!c)
2452                         return 1;
2453                 if (val && isdigit(c)) {
2454                         char *end;
2455                         int n = strtoul(arg, &end, 10);
2456                         if (*end)
2457                                 return 0;
2458                         *val = n;
2459                         return 1;
2460                 }
2461                 return 0;
2462         }
2463         if (c != '-')
2464                 return 0;
2465         arg++;
2466         eq = strchr(arg, '=');
2467         if (eq)
2468                 len = eq - arg;
2469         else
2470                 len = strlen(arg);
2471         if (!len || strncmp(arg, arg_long, len))
2472                 return 0;
2473         if (eq) {
2474                 int n;
2475                 char *end;
2476                 if (!isdigit(*++eq))
2477                         return 0;
2478                 n = strtoul(eq, &end, 10);
2479                 if (*end)
2480                         return 0;
2481                 *val = n;
2482         }
2483         return 1;
2484 }
2485
2486 static int diff_scoreopt_parse(const char *opt);
2487
2488 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2489 {
2490         const char *arg = av[0];
2491
2492         /* Output format options */
2493         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2494                 options->output_format |= DIFF_FORMAT_PATCH;
2495         else if (opt_arg(arg, 'U', "unified", &options->context))
2496                 options->output_format |= DIFF_FORMAT_PATCH;
2497         else if (!strcmp(arg, "--raw"))
2498                 options->output_format |= DIFF_FORMAT_RAW;
2499         else if (!strcmp(arg, "--patch-with-raw"))
2500                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2501         else if (!strcmp(arg, "--numstat"))
2502                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2503         else if (!strcmp(arg, "--shortstat"))
2504                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2505         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2506                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2507         else if (!strcmp(arg, "--cumulative"))
2508                 options->output_format |= DIFF_FORMAT_CUMULATIVE;
2509         else if (!strcmp(arg, "--check"))
2510                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2511         else if (!strcmp(arg, "--summary"))
2512                 options->output_format |= DIFF_FORMAT_SUMMARY;
2513         else if (!strcmp(arg, "--patch-with-stat"))
2514                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2515         else if (!strcmp(arg, "--name-only"))
2516                 options->output_format |= DIFF_FORMAT_NAME;
2517         else if (!strcmp(arg, "--name-status"))
2518                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2519         else if (!strcmp(arg, "-s"))
2520                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2521         else if (!prefixcmp(arg, "--stat")) {
2522                 char *end;
2523                 int width = options->stat_width;
2524                 int name_width = options->stat_name_width;
2525                 arg += 6;
2526                 end = (char *)arg;
2527
2528                 switch (*arg) {
2529                 case '-':
2530                         if (!prefixcmp(arg, "-width="))
2531                                 width = strtoul(arg + 7, &end, 10);
2532                         else if (!prefixcmp(arg, "-name-width="))
2533                                 name_width = strtoul(arg + 12, &end, 10);
2534                         break;
2535                 case '=':
2536                         width = strtoul(arg+1, &end, 10);
2537                         if (*end == ',')
2538                                 name_width = strtoul(end+1, &end, 10);
2539                 }
2540
2541                 /* Important! This checks all the error cases! */
2542                 if (*end)
2543                         return 0;
2544                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2545                 options->stat_name_width = name_width;
2546                 options->stat_width = width;
2547         }
2548
2549         /* renames options */
2550         else if (!prefixcmp(arg, "-B")) {
2551                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2552                         return -1;
2553         }
2554         else if (!prefixcmp(arg, "-M")) {
2555                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2556                         return -1;
2557                 options->detect_rename = DIFF_DETECT_RENAME;
2558         }
2559         else if (!prefixcmp(arg, "-C")) {
2560                 if (options->detect_rename == DIFF_DETECT_COPY)
2561                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2562                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2563                         return -1;
2564                 options->detect_rename = DIFF_DETECT_COPY;
2565         }
2566         else if (!strcmp(arg, "--no-renames"))
2567                 options->detect_rename = 0;
2568         else if (!strcmp(arg, "--relative"))
2569                 DIFF_OPT_SET(options, RELATIVE_NAME);
2570         else if (!prefixcmp(arg, "--relative=")) {
2571                 DIFF_OPT_SET(options, RELATIVE_NAME);
2572                 options->prefix = arg + 11;
2573         }
2574
2575         /* xdiff options */
2576         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2577                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2578         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2579                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2580         else if (!strcmp(arg, "--ignore-space-at-eol"))
2581                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2582
2583         /* flags options */
2584         else if (!strcmp(arg, "--binary")) {
2585                 options->output_format |= DIFF_FORMAT_PATCH;
2586                 DIFF_OPT_SET(options, BINARY);
2587         }
2588         else if (!strcmp(arg, "--full-index"))
2589                 DIFF_OPT_SET(options, FULL_INDEX);
2590         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2591                 DIFF_OPT_SET(options, TEXT);
2592         else if (!strcmp(arg, "-R"))
2593                 DIFF_OPT_SET(options, REVERSE_DIFF);
2594         else if (!strcmp(arg, "--find-copies-harder"))
2595                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2596         else if (!strcmp(arg, "--follow"))
2597                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2598         else if (!strcmp(arg, "--color"))
2599                 DIFF_OPT_SET(options, COLOR_DIFF);
2600         else if (!strcmp(arg, "--no-color"))
2601                 DIFF_OPT_CLR(options, COLOR_DIFF);
2602         else if (!strcmp(arg, "--color-words"))
2603                 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2604         else if (!strcmp(arg, "--exit-code"))
2605                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2606         else if (!strcmp(arg, "--quiet"))
2607                 DIFF_OPT_SET(options, QUIET);
2608         else if (!strcmp(arg, "--ext-diff"))
2609                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2610         else if (!strcmp(arg, "--no-ext-diff"))
2611                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2612         else if (!strcmp(arg, "--ignore-submodules"))
2613                 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2614
2615         /* misc options */
2616         else if (!strcmp(arg, "-z"))
2617                 options->line_termination = 0;
2618         else if (!prefixcmp(arg, "-l"))
2619                 options->rename_limit = strtoul(arg+2, NULL, 10);
2620         else if (!prefixcmp(arg, "-S"))
2621                 options->pickaxe = arg + 2;
2622         else if (!strcmp(arg, "--pickaxe-all"))
2623                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2624         else if (!strcmp(arg, "--pickaxe-regex"))
2625                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2626         else if (!prefixcmp(arg, "-O"))
2627                 options->orderfile = arg + 2;
2628         else if (!prefixcmp(arg, "--diff-filter="))
2629                 options->filter = arg + 14;
2630         else if (!strcmp(arg, "--abbrev"))
2631                 options->abbrev = DEFAULT_ABBREV;
2632         else if (!prefixcmp(arg, "--abbrev=")) {
2633                 options->abbrev = strtoul(arg + 9, NULL, 10);
2634                 if (options->abbrev < MINIMUM_ABBREV)
2635                         options->abbrev = MINIMUM_ABBREV;
2636                 else if (40 < options->abbrev)
2637                         options->abbrev = 40;
2638         }
2639         else if (!prefixcmp(arg, "--src-prefix="))
2640                 options->a_prefix = arg + 13;
2641         else if (!prefixcmp(arg, "--dst-prefix="))
2642                 options->b_prefix = arg + 13;
2643         else if (!strcmp(arg, "--no-prefix"))
2644                 options->a_prefix = options->b_prefix = "";
2645         else if (!prefixcmp(arg, "--output=")) {
2646                 options->file = fopen(arg + strlen("--output="), "w");
2647                 options->close_file = 1;
2648         } else
2649                 return 0;
2650         return 1;
2651 }
2652
2653 static int parse_num(const char **cp_p)
2654 {
2655         unsigned long num, scale;
2656         int ch, dot;
2657         const char *cp = *cp_p;
2658
2659         num = 0;
2660         scale = 1;
2661         dot = 0;
2662         for(;;) {
2663                 ch = *cp;
2664                 if ( !dot && ch == '.' ) {
2665                         scale = 1;
2666                         dot = 1;
2667                 } else if ( ch == '%' ) {
2668                         scale = dot ? scale*100 : 100;
2669                         cp++;   /* % is always at the end */
2670                         break;
2671                 } else if ( ch >= '0' && ch <= '9' ) {
2672                         if ( scale < 100000 ) {
2673                                 scale *= 10;
2674                                 num = (num*10) + (ch-'0');
2675                         }
2676                 } else {
2677                         break;
2678                 }
2679                 cp++;
2680         }
2681         *cp_p = cp;
2682
2683         /* user says num divided by scale and we say internally that
2684          * is MAX_SCORE * num / scale.
2685          */
2686         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2687 }
2688
2689 static int diff_scoreopt_parse(const char *opt)
2690 {
2691         int opt1, opt2, cmd;
2692
2693         if (*opt++ != '-')
2694                 return -1;
2695         cmd = *opt++;
2696         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2697                 return -1; /* that is not a -M, -C nor -B option */
2698
2699         opt1 = parse_num(&opt);
2700         if (cmd != 'B')
2701                 opt2 = 0;
2702         else {
2703                 if (*opt == 0)
2704                         opt2 = 0;
2705                 else if (*opt != '/')
2706                         return -1; /* we expect -B80/99 or -B80 */
2707                 else {
2708                         opt++;
2709                         opt2 = parse_num(&opt);
2710                 }
2711         }
2712         if (*opt != 0)
2713                 return -1;
2714         return opt1 | (opt2 << 16);
2715 }
2716
2717 struct diff_queue_struct diff_queued_diff;
2718
2719 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2720 {
2721         if (queue->alloc <= queue->nr) {
2722                 queue->alloc = alloc_nr(queue->alloc);
2723                 queue->queue = xrealloc(queue->queue,
2724                                         sizeof(dp) * queue->alloc);
2725         }
2726         queue->queue[queue->nr++] = dp;
2727 }
2728
2729 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2730                                  struct diff_filespec *one,
2731                                  struct diff_filespec *two)
2732 {
2733         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2734         dp->one = one;
2735         dp->two = two;
2736         if (queue)
2737                 diff_q(queue, dp);
2738         return dp;
2739 }
2740
2741 void diff_free_filepair(struct diff_filepair *p)
2742 {
2743         free_filespec(p->one);
2744         free_filespec(p->two);
2745         free(p);
2746 }
2747
2748 /* This is different from find_unique_abbrev() in that
2749  * it stuffs the result with dots for alignment.
2750  */
2751 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2752 {
2753         int abblen;
2754         const char *abbrev;
2755         if (len == 40)
2756                 return sha1_to_hex(sha1);
2757
2758         abbrev = find_unique_abbrev(sha1, len);
2759         abblen = strlen(abbrev);
2760         if (abblen < 37) {
2761                 static char hex[41];
2762                 if (len < abblen && abblen <= len + 2)
2763                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2764                 else
2765                         sprintf(hex, "%s...", abbrev);
2766                 return hex;
2767         }
2768         return sha1_to_hex(sha1);
2769 }
2770
2771 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2772 {
2773         int line_termination = opt->line_termination;
2774         int inter_name_termination = line_termination ? '\t' : '\0';
2775
2776         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2777                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2778                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
2779                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2780         }
2781         if (p->score) {
2782                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2783                         inter_name_termination);
2784         } else {
2785                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2786         }
2787
2788         if (p->status == DIFF_STATUS_COPIED ||
2789             p->status == DIFF_STATUS_RENAMED) {
2790                 const char *name_a, *name_b;
2791                 name_a = p->one->path;
2792                 name_b = p->two->path;
2793                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2794                 write_name_quoted(name_a, opt->file, inter_name_termination);
2795                 write_name_quoted(name_b, opt->file, line_termination);
2796         } else {
2797                 const char *name_a, *name_b;
2798                 name_a = p->one->mode ? p->one->path : p->two->path;
2799                 name_b = NULL;
2800                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2801                 write_name_quoted(name_a, opt->file, line_termination);
2802         }
2803 }
2804
2805 int diff_unmodified_pair(struct diff_filepair *p)
2806 {
2807         /* This function is written stricter than necessary to support
2808          * the currently implemented transformers, but the idea is to
2809          * let transformers to produce diff_filepairs any way they want,
2810          * and filter and clean them up here before producing the output.
2811          */
2812         struct diff_filespec *one = p->one, *two = p->two;
2813
2814         if (DIFF_PAIR_UNMERGED(p))
2815                 return 0; /* unmerged is interesting */
2816
2817         /* deletion, addition, mode or type change
2818          * and rename are all interesting.
2819          */
2820         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2821             DIFF_PAIR_MODE_CHANGED(p) ||
2822             strcmp(one->path, two->path))
2823                 return 0;
2824
2825         /* both are valid and point at the same path.  that is, we are
2826          * dealing with a change.
2827          */
2828         if (one->sha1_valid && two->sha1_valid &&
2829             !hashcmp(one->sha1, two->sha1))
2830                 return 1; /* no change */
2831         if (!one->sha1_valid && !two->sha1_valid)
2832                 return 1; /* both look at the same file on the filesystem. */
2833         return 0;
2834 }
2835
2836 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2837 {
2838         if (diff_unmodified_pair(p))
2839                 return;
2840
2841         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2842             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2843                 return; /* no tree diffs in patch format */
2844
2845         run_diff(p, o);
2846 }
2847
2848 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2849                             struct diffstat_t *diffstat)
2850 {
2851         if (diff_unmodified_pair(p))
2852                 return;
2853
2854         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2855             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2856                 return; /* no tree diffs in patch format */
2857
2858         run_diffstat(p, o, diffstat);
2859 }
2860
2861 static void diff_flush_checkdiff(struct diff_filepair *p,
2862                 struct diff_options *o)
2863 {
2864         if (diff_unmodified_pair(p))
2865                 return;
2866
2867         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2868             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2869                 return; /* no tree diffs in patch format */
2870
2871         run_checkdiff(p, o);
2872 }
2873
2874 int diff_queue_is_empty(void)
2875 {
2876         struct diff_queue_struct *q = &diff_queued_diff;
2877         int i;
2878         for (i = 0; i < q->nr; i++)
2879                 if (!diff_unmodified_pair(q->queue[i]))
2880                         return 0;
2881         return 1;
2882 }
2883
2884 #if DIFF_DEBUG
2885 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2886 {
2887         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2888                 x, one ? one : "",
2889                 s->path,
2890                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2891                 s->mode,
2892                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2893         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2894                 x, one ? one : "",
2895                 s->size, s->xfrm_flags);
2896 }
2897
2898 void diff_debug_filepair(const struct diff_filepair *p, int i)
2899 {
2900         diff_debug_filespec(p->one, i, "one");
2901         diff_debug_filespec(p->two, i, "two");
2902         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2903                 p->score, p->status ? p->status : '?',
2904                 p->one->rename_used, p->broken_pair);
2905 }
2906
2907 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2908 {
2909         int i;
2910         if (msg)
2911                 fprintf(stderr, "%s\n", msg);
2912         fprintf(stderr, "q->nr = %d\n", q->nr);
2913         for (i = 0; i < q->nr; i++) {
2914                 struct diff_filepair *p = q->queue[i];
2915                 diff_debug_filepair(p, i);
2916         }
2917 }
2918 #endif
2919
2920 static void diff_resolve_rename_copy(void)
2921 {
2922         int i;
2923         struct diff_filepair *p;
2924         struct diff_queue_struct *q = &diff_queued_diff;
2925
2926         diff_debug_queue("resolve-rename-copy", q);
2927
2928         for (i = 0; i < q->nr; i++) {
2929                 p = q->queue[i];
2930                 p->status = 0; /* undecided */
2931                 if (DIFF_PAIR_UNMERGED(p))
2932                         p->status = DIFF_STATUS_UNMERGED;
2933                 else if (!DIFF_FILE_VALID(p->one))
2934                         p->status = DIFF_STATUS_ADDED;
2935                 else if (!DIFF_FILE_VALID(p->two))
2936                         p->status = DIFF_STATUS_DELETED;
2937                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2938                         p->status = DIFF_STATUS_TYPE_CHANGED;
2939
2940                 /* from this point on, we are dealing with a pair
2941                  * whose both sides are valid and of the same type, i.e.
2942                  * either in-place edit or rename/copy edit.
2943                  */
2944                 else if (DIFF_PAIR_RENAME(p)) {
2945                         /*
2946                          * A rename might have re-connected a broken
2947                          * pair up, causing the pathnames to be the
2948                          * same again. If so, that's not a rename at
2949                          * all, just a modification..
2950                          *
2951                          * Otherwise, see if this source was used for
2952                          * multiple renames, in which case we decrement
2953                          * the count, and call it a copy.
2954                          */
2955                         if (!strcmp(p->one->path, p->two->path))
2956                                 p->status = DIFF_STATUS_MODIFIED;
2957                         else if (--p->one->rename_used > 0)
2958                                 p->status = DIFF_STATUS_COPIED;
2959                         else
2960                                 p->status = DIFF_STATUS_RENAMED;
2961                 }
2962                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2963                          p->one->mode != p->two->mode ||
2964                          is_null_sha1(p->one->sha1))
2965                         p->status = DIFF_STATUS_MODIFIED;
2966                 else {
2967                         /* This is a "no-change" entry and should not
2968                          * happen anymore, but prepare for broken callers.
2969                          */
2970                         error("feeding unmodified %s to diffcore",
2971                               p->one->path);
2972                         p->status = DIFF_STATUS_UNKNOWN;
2973                 }
2974         }
2975         diff_debug_queue("resolve-rename-copy done", q);
2976 }
2977
2978 static int check_pair_status(struct diff_filepair *p)
2979 {
2980         switch (p->status) {
2981         case DIFF_STATUS_UNKNOWN:
2982                 return 0;
2983         case 0:
2984                 die("internal error in diff-resolve-rename-copy");
2985         default:
2986                 return 1;
2987         }
2988 }
2989
2990 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2991 {
2992         int fmt = opt->output_format;
2993
2994         if (fmt & DIFF_FORMAT_CHECKDIFF)
2995                 diff_flush_checkdiff(p, opt);
2996         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2997                 diff_flush_raw(p, opt);
2998         else if (fmt & DIFF_FORMAT_NAME) {
2999                 const char *name_a, *name_b;
3000                 name_a = p->two->path;
3001                 name_b = NULL;
3002                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3003                 write_name_quoted(name_a, opt->file, opt->line_termination);
3004         }
3005 }
3006
3007 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3008 {
3009         if (fs->mode)
3010                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3011         else
3012                 fprintf(file, " %s ", newdelete);
3013         write_name_quoted(fs->path, file, '\n');
3014 }
3015
3016
3017 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
3018 {
3019         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3020                 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
3021                         show_name ? ' ' : '\n');
3022                 if (show_name) {
3023                         write_name_quoted(p->two->path, file, '\n');
3024                 }
3025         }
3026 }
3027
3028 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3029 {
3030         char *names = pprint_rename(p->one->path, p->two->path);
3031
3032         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3033         free(names);
3034         show_mode_change(file, p, 0);
3035 }
3036
3037 static void diff_summary(FILE *file, struct diff_filepair *p)
3038 {
3039         switch(p->status) {
3040         case DIFF_STATUS_DELETED:
3041                 show_file_mode_name(file, "delete", p->one);
3042                 break;
3043         case DIFF_STATUS_ADDED:
3044                 show_file_mode_name(file, "create", p->two);
3045                 break;
3046         case DIFF_STATUS_COPIED:
3047                 show_rename_copy(file, "copy", p);
3048                 break;
3049         case DIFF_STATUS_RENAMED:
3050                 show_rename_copy(file, "rename", p);
3051                 break;
3052         default:
3053                 if (p->score) {
3054                         fputs(" rewrite ", file);
3055                         write_name_quoted(p->two->path, file, ' ');
3056                         fprintf(file, "(%d%%)\n", similarity_index(p));
3057                 }
3058                 show_mode_change(file, p, !p->score);
3059                 break;
3060         }
3061 }
3062
3063 struct patch_id_t {
3064         struct xdiff_emit_state xm;
3065         SHA_CTX *ctx;
3066         int patchlen;
3067 };
3068
3069 static int remove_space(char *line, int len)
3070 {
3071         int i;
3072         char *dst = line;
3073         unsigned char c;
3074
3075         for (i = 0; i < len; i++)
3076                 if (!isspace((c = line[i])))
3077                         *dst++ = c;
3078
3079         return dst - line;
3080 }
3081
3082 static void patch_id_consume(void *priv, char *line, unsigned long len)
3083 {
3084         struct patch_id_t *data = priv;
3085         int new_len;
3086
3087         /* Ignore line numbers when computing the SHA1 of the patch */
3088         if (!prefixcmp(line, "@@ -"))
3089                 return;
3090
3091         new_len = remove_space(line, len);
3092
3093         SHA1_Update(data->ctx, line, new_len);
3094         data->patchlen += new_len;
3095 }
3096
3097 /* returns 0 upon success, and writes result into sha1 */
3098 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3099 {
3100         struct diff_queue_struct *q = &diff_queued_diff;
3101         int i;
3102         SHA_CTX ctx;
3103         struct patch_id_t data;
3104         char buffer[PATH_MAX * 4 + 20];
3105
3106         SHA1_Init(&ctx);
3107         memset(&data, 0, sizeof(struct patch_id_t));
3108         data.ctx = &ctx;
3109         data.xm.consume = patch_id_consume;
3110
3111         for (i = 0; i < q->nr; i++) {
3112                 xpparam_t xpp;
3113                 xdemitconf_t xecfg;
3114                 xdemitcb_t ecb;
3115                 mmfile_t mf1, mf2;
3116                 struct diff_filepair *p = q->queue[i];
3117                 int len1, len2;
3118
3119                 memset(&xecfg, 0, sizeof(xecfg));
3120                 if (p->status == 0)
3121                         return error("internal diff status error");
3122                 if (p->status == DIFF_STATUS_UNKNOWN)
3123                         continue;
3124                 if (diff_unmodified_pair(p))
3125                         continue;
3126                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3127                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3128                         continue;
3129                 if (DIFF_PAIR_UNMERGED(p))
3130                         continue;
3131
3132                 diff_fill_sha1_info(p->one);
3133                 diff_fill_sha1_info(p->two);
3134                 if (fill_mmfile(&mf1, p->one) < 0 ||
3135                                 fill_mmfile(&mf2, p->two) < 0)
3136                         return error("unable to read files to diff");
3137
3138                 len1 = remove_space(p->one->path, strlen(p->one->path));
3139                 len2 = remove_space(p->two->path, strlen(p->two->path));
3140                 if (p->one->mode == 0)
3141                         len1 = snprintf(buffer, sizeof(buffer),
3142                                         "diff--gita/%.*sb/%.*s"
3143                                         "newfilemode%06o"
3144                                         "---/dev/null"
3145                                         "+++b/%.*s",
3146                                         len1, p->one->path,
3147                                         len2, p->two->path,
3148                                         p->two->mode,
3149                                         len2, p->two->path);
3150                 else if (p->two->mode == 0)
3151                         len1 = snprintf(buffer, sizeof(buffer),
3152                                         "diff--gita/%.*sb/%.*s"
3153                                         "deletedfilemode%06o"
3154                                         "---a/%.*s"
3155                                         "+++/dev/null",
3156                                         len1, p->one->path,
3157                                         len2, p->two->path,
3158                                         p->one->mode,
3159                                         len1, p->one->path);
3160                 else
3161                         len1 = snprintf(buffer, sizeof(buffer),
3162                                         "diff--gita/%.*sb/%.*s"
3163                                         "---a/%.*s"
3164                                         "+++b/%.*s",
3165                                         len1, p->one->path,
3166                                         len2, p->two->path,
3167                                         len1, p->one->path,
3168                                         len2, p->two->path);
3169                 SHA1_Update(&ctx, buffer, len1);
3170
3171                 xpp.flags = XDF_NEED_MINIMAL;
3172                 xecfg.ctxlen = 3;
3173                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3174                 ecb.outf = xdiff_outf;
3175                 ecb.priv = &data;
3176                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
3177         }
3178
3179         SHA1_Final(sha1, &ctx);
3180         return 0;
3181 }
3182
3183 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3184 {
3185         struct diff_queue_struct *q = &diff_queued_diff;
3186         int i;
3187         int result = diff_get_patch_id(options, sha1);
3188
3189         for (i = 0; i < q->nr; i++)
3190                 diff_free_filepair(q->queue[i]);
3191
3192         free(q->queue);
3193         q->queue = NULL;
3194         q->nr = q->alloc = 0;
3195
3196         return result;
3197 }
3198
3199 static int is_summary_empty(const struct diff_queue_struct *q)
3200 {
3201         int i;
3202
3203         for (i = 0; i < q->nr; i++) {
3204                 const struct diff_filepair *p = q->queue[i];
3205
3206                 switch (p->status) {
3207                 case DIFF_STATUS_DELETED:
3208                 case DIFF_STATUS_ADDED:
3209                 case DIFF_STATUS_COPIED:
3210                 case DIFF_STATUS_RENAMED:
3211                         return 0;
3212                 default:
3213                         if (p->score)
3214                                 return 0;
3215                         if (p->one->mode && p->two->mode &&
3216                             p->one->mode != p->two->mode)
3217                                 return 0;
3218                         break;
3219                 }
3220         }
3221         return 1;
3222 }
3223
3224 void diff_flush(struct diff_options *options)
3225 {
3226         struct diff_queue_struct *q = &diff_queued_diff;
3227         int i, output_format = options->output_format;
3228         int separator = 0;
3229
3230         /*
3231          * Order: raw, stat, summary, patch
3232          * or:    name/name-status/checkdiff (other bits clear)
3233          */
3234         if (!q->nr)
3235                 goto free_queue;
3236
3237         if (output_format & (DIFF_FORMAT_RAW |
3238                              DIFF_FORMAT_NAME |
3239                              DIFF_FORMAT_NAME_STATUS |
3240                              DIFF_FORMAT_CHECKDIFF)) {
3241                 for (i = 0; i < q->nr; i++) {
3242                         struct diff_filepair *p = q->queue[i];
3243                         if (check_pair_status(p))
3244                                 flush_one_pair(p, options);
3245                 }
3246                 separator++;
3247         }
3248
3249         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3250                 struct diffstat_t diffstat;
3251
3252                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3253                 diffstat.xm.consume = diffstat_consume;
3254                 for (i = 0; i < q->nr; i++) {
3255                         struct diff_filepair *p = q->queue[i];
3256                         if (check_pair_status(p))
3257                                 diff_flush_stat(p, options, &diffstat);
3258                 }
3259                 if (output_format & DIFF_FORMAT_NUMSTAT)
3260                         show_numstat(&diffstat, options);
3261                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3262                         show_stats(&diffstat, options);
3263                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3264                         show_shortstats(&diffstat, options);
3265                 free_diffstat_info(&diffstat);
3266                 separator++;
3267         }
3268         if (output_format & DIFF_FORMAT_DIRSTAT)
3269                 show_dirstat(options);
3270
3271         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3272                 for (i = 0; i < q->nr; i++)
3273                         diff_summary(options->file, q->queue[i]);
3274                 separator++;
3275         }
3276
3277         if (output_format & DIFF_FORMAT_PATCH) {
3278                 if (separator) {
3279                         putc(options->line_termination, options->file);
3280                         if (options->stat_sep) {
3281                                 /* attach patch instead of inline */
3282                                 fputs(options->stat_sep, options->file);
3283                         }
3284                 }
3285
3286                 for (i = 0; i < q->nr; i++) {
3287                         struct diff_filepair *p = q->queue[i];
3288                         if (check_pair_status(p))
3289                                 diff_flush_patch(p, options);
3290                 }
3291         }
3292
3293         if (output_format & DIFF_FORMAT_CALLBACK)
3294                 options->format_callback(q, options, options->format_callback_data);
3295
3296         for (i = 0; i < q->nr; i++)
3297                 diff_free_filepair(q->queue[i]);
3298 free_queue:
3299         free(q->queue);
3300         q->queue = NULL;
3301         q->nr = q->alloc = 0;
3302         if (options->close_file)
3303                 fclose(options->file);
3304 }
3305
3306 static void diffcore_apply_filter(const char *filter)
3307 {
3308         int i;
3309         struct diff_queue_struct *q = &diff_queued_diff;
3310         struct diff_queue_struct outq;
3311         outq.queue = NULL;
3312         outq.nr = outq.alloc = 0;
3313
3314         if (!filter)
3315                 return;
3316
3317         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3318                 int found;
3319                 for (i = found = 0; !found && i < q->nr; i++) {
3320                         struct diff_filepair *p = q->queue[i];
3321                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3322                              ((p->score &&
3323                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3324                               (!p->score &&
3325                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3326                             ((p->status != DIFF_STATUS_MODIFIED) &&
3327                              strchr(filter, p->status)))
3328                                 found++;
3329                 }
3330                 if (found)
3331                         return;
3332
3333                 /* otherwise we will clear the whole queue
3334                  * by copying the empty outq at the end of this
3335                  * function, but first clear the current entries
3336                  * in the queue.
3337                  */
3338                 for (i = 0; i < q->nr; i++)
3339                         diff_free_filepair(q->queue[i]);
3340         }
3341         else {
3342                 /* Only the matching ones */
3343                 for (i = 0; i < q->nr; i++) {
3344                         struct diff_filepair *p = q->queue[i];
3345
3346                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3347                              ((p->score &&
3348                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3349                               (!p->score &&
3350                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3351                             ((p->status != DIFF_STATUS_MODIFIED) &&
3352                              strchr(filter, p->status)))
3353                                 diff_q(&outq, p);
3354                         else
3355                                 diff_free_filepair(p);
3356                 }
3357         }
3358         free(q->queue);
3359         *q = outq;
3360 }
3361
3362 /* Check whether two filespecs with the same mode and size are identical */
3363 static int diff_filespec_is_identical(struct diff_filespec *one,
3364                                       struct diff_filespec *two)
3365 {
3366         if (S_ISGITLINK(one->mode))
3367                 return 0;
3368         if (diff_populate_filespec(one, 0))
3369                 return 0;
3370         if (diff_populate_filespec(two, 0))
3371                 return 0;
3372         return !memcmp(one->data, two->data, one->size);
3373 }
3374
3375 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3376 {
3377         int i;
3378         struct diff_queue_struct *q = &diff_queued_diff;
3379         struct diff_queue_struct outq;
3380         outq.queue = NULL;
3381         outq.nr = outq.alloc = 0;
3382
3383         for (i = 0; i < q->nr; i++) {
3384                 struct diff_filepair *p = q->queue[i];
3385
3386                 /*
3387                  * 1. Entries that come from stat info dirtyness
3388                  *    always have both sides (iow, not create/delete),
3389                  *    one side of the object name is unknown, with
3390                  *    the same mode and size.  Keep the ones that
3391                  *    do not match these criteria.  They have real
3392                  *    differences.
3393                  *
3394                  * 2. At this point, the file is known to be modified,
3395                  *    with the same mode and size, and the object
3396                  *    name of one side is unknown.  Need to inspect
3397                  *    the identical contents.
3398                  */
3399                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3400                     !DIFF_FILE_VALID(p->two) ||
3401                     (p->one->sha1_valid && p->two->sha1_valid) ||
3402                     (p->one->mode != p->two->mode) ||
3403                     diff_populate_filespec(p->one, 1) ||
3404                     diff_populate_filespec(p->two, 1) ||
3405                     (p->one->size != p->two->size) ||
3406                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3407                         diff_q(&outq, p);
3408                 else {
3409                         /*
3410                          * The caller can subtract 1 from skip_stat_unmatch
3411                          * to determine how many paths were dirty only
3412                          * due to stat info mismatch.
3413                          */
3414                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3415                                 diffopt->skip_stat_unmatch++;
3416                         diff_free_filepair(p);
3417                 }
3418         }
3419         free(q->queue);
3420         *q = outq;
3421 }
3422
3423 void diffcore_std(struct diff_options *options)
3424 {
3425         if (DIFF_OPT_TST(options, QUIET))
3426                 return;
3427
3428         if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3429                 diffcore_skip_stat_unmatch(options);
3430         if (options->break_opt != -1)
3431                 diffcore_break(options->break_opt);
3432         if (options->detect_rename)
3433                 diffcore_rename(options);
3434         if (options->break_opt != -1)
3435                 diffcore_merge_broken();
3436         if (options->pickaxe)
3437                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3438         if (options->orderfile)
3439                 diffcore_order(options->orderfile);
3440         diff_resolve_rename_copy();
3441         diffcore_apply_filter(options->filter);
3442
3443         if (diff_queued_diff.nr)
3444                 DIFF_OPT_SET(options, HAS_CHANGES);
3445         else
3446                 DIFF_OPT_CLR(options, HAS_CHANGES);
3447 }
3448
3449 int diff_result_code(struct diff_options *opt, int status)
3450 {
3451         int result = 0;
3452         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3453             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3454                 return status;
3455         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3456             DIFF_OPT_TST(opt, HAS_CHANGES))
3457                 result |= 01;
3458         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3459             DIFF_OPT_TST(opt, CHECK_FAILED))
3460                 result |= 02;
3461         return result;
3462 }
3463
3464 void diff_addremove(struct diff_options *options,
3465                     int addremove, unsigned mode,
3466                     const unsigned char *sha1,
3467                     const char *concatpath)
3468 {
3469         struct diff_filespec *one, *two;
3470
3471         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3472                 return;
3473
3474         /* This may look odd, but it is a preparation for
3475          * feeding "there are unchanged files which should
3476          * not produce diffs, but when you are doing copy
3477          * detection you would need them, so here they are"
3478          * entries to the diff-core.  They will be prefixed
3479          * with something like '=' or '*' (I haven't decided
3480          * which but should not make any difference).
3481          * Feeding the same new and old to diff_change()
3482          * also has the same effect.
3483          * Before the final output happens, they are pruned after
3484          * merged into rename/copy pairs as appropriate.
3485          */
3486         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3487                 addremove = (addremove == '+' ? '-' :
3488                              addremove == '-' ? '+' : addremove);
3489
3490         if (options->prefix &&
3491             strncmp(concatpath, options->prefix, options->prefix_length))
3492                 return;
3493
3494         one = alloc_filespec(concatpath);
3495         two = alloc_filespec(concatpath);
3496
3497         if (addremove != '+')
3498                 fill_filespec(one, sha1, mode);
3499         if (addremove != '-')
3500                 fill_filespec(two, sha1, mode);
3501
3502         diff_queue(&diff_queued_diff, one, two);
3503         DIFF_OPT_SET(options, HAS_CHANGES);
3504 }
3505
3506 void diff_change(struct diff_options *options,
3507                  unsigned old_mode, unsigned new_mode,
3508                  const unsigned char *old_sha1,
3509                  const unsigned char *new_sha1,
3510                  const char *concatpath)
3511 {
3512         struct diff_filespec *one, *two;
3513
3514         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3515                         && S_ISGITLINK(new_mode))
3516                 return;
3517
3518         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3519                 unsigned tmp;
3520                 const unsigned char *tmp_c;
3521                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3522                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3523         }
3524
3525         if (options->prefix &&
3526             strncmp(concatpath, options->prefix, options->prefix_length))
3527                 return;
3528
3529         one = alloc_filespec(concatpath);
3530         two = alloc_filespec(concatpath);
3531         fill_filespec(one, old_sha1, old_mode);
3532         fill_filespec(two, new_sha1, new_mode);
3533
3534         diff_queue(&diff_queued_diff, one, two);
3535         DIFF_OPT_SET(options, HAS_CHANGES);
3536 }
3537
3538 void diff_unmerge(struct diff_options *options,
3539                   const char *path,
3540                   unsigned mode, const unsigned char *sha1)
3541 {
3542         struct diff_filespec *one, *two;
3543
3544         if (options->prefix &&
3545             strncmp(path, options->prefix, options->prefix_length))
3546                 return;
3547
3548         one = alloc_filespec(path);
3549         two = alloc_filespec(path);
3550         fill_filespec(one, sha1, mode);
3551         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3552 }