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