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

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