diff.c

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "quote.h"
   6 #include "diff.h"
   7 #include "diffcore.h"
   8 #include "delta.h"
   9 #include "xdiff-interface.h"
  10 #include "color.h"
  11 #include "attr.h"
  12 #include "run-command.h"
  13 #include "utf8.h"
  14
  15 #ifdef NO_FAST_WORKING_DIRECTORY
  16 #define FAST_WORKING_DIRECTORY 0
  17 #else
  18 #define FAST_WORKING_DIRECTORY 1
  19 #endif
  20
  21 static int diff_detect_rename_default;
  22 static int diff_rename_limit_default = 200;
  23 static int diff_suppress_blank_empty;
  24 int diff_use_color_default = -1;
  25 static const char *external_diff_cmd_cfg;
  26 int diff_auto_refresh_index = 1;
  27 static int diff_mnemonic_prefix;
  28
  29 static char diff_colors[][COLOR_MAXLEN] = {
  30         "\033[m",       /* reset */
  31         "",             /* PLAIN (normal) */
  32         "\033[1m",      /* METAINFO (bold) */
  33         "\033[36m",     /* FRAGINFO (cyan) */
  34         "\033[31m",     /* OLD (red) */
  35         "\033[32m",     /* NEW (green) */
  36         "\033[33m",     /* COMMIT (yellow) */
  37         "\033[41m",     /* WHITESPACE (red background) */
  38 };
  39
  40 static int parse_diff_color_slot(const char *var, int ofs)
  41 {
  42         if (!strcasecmp(var+ofs, "plain"))
  43                 return DIFF_PLAIN;
  44         if (!strcasecmp(var+ofs, "meta"))
  45                 return DIFF_METAINFO;
  46         if (!strcasecmp(var+ofs, "frag"))
  47                 return DIFF_FRAGINFO;
  48         if (!strcasecmp(var+ofs, "old"))
  49                 return DIFF_FILE_OLD;
  50         if (!strcasecmp(var+ofs, "new"))
  51                 return DIFF_FILE_NEW;
  52         if (!strcasecmp(var+ofs, "commit"))
  53                 return DIFF_COMMIT;
  54         if (!strcasecmp(var+ofs, "whitespace"))
  55                 return DIFF_WHITESPACE;
  56         die("bad config variable '%s'", var);
  57 }
  58
  59 static struct ll_diff_driver {
  60         const char *name;
  61         struct ll_diff_driver *next;
  62         const char *cmd;
  63 } *user_diff, **user_diff_tail;
  64
  65 /*
  66  * Currently there is only "diff.<drivername>.command" variable;
  67  * because there are "diff.color.<slot>" variables, we are parsing
  68  * this in a bit convoluted way to allow low level diff driver
  69  * called "color".
  70  */
  71 static int parse_lldiff_command(const char *var, const char *ep, const char *value)
  72 {
  73         const char *name;
  74         int namelen;
  75         struct ll_diff_driver *drv;
  76
  77         name = var + 5;
  78         namelen = ep - name;
  79         for (drv = user_diff; drv; drv = drv->next)
  80                 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
  81                         break;
  82         if (!drv) {
  83                 drv = xcalloc(1, sizeof(struct ll_diff_driver));
  84                 drv->name = xmemdupz(name, namelen);
  85                 if (!user_diff_tail)
  86                         user_diff_tail = &user_diff;
  87                 *user_diff_tail = drv;
  88                 user_diff_tail = &(drv->next);
  89         }
  90
  91         return git_config_string(&(drv->cmd), var, value);
  92 }
  93
  94 /*
  95  * 'diff.<what>.funcname' attribute can be specified in the configuration
  96  * to define a customized regexp to find the beginning of a function to
  97  * be used for hunk header lines of "diff -p" style output.
  98  */
  99 struct funcname_pattern_entry {
 100         char *name;
 101         char *pattern;
 102         int cflags;
 103 };
 104 static struct funcname_pattern_list {
 105         struct funcname_pattern_list *next;
 106         struct funcname_pattern_entry e;
 107 } *funcname_pattern_list;
 108
 109 static int parse_funcname_pattern(const char *var, const char *ep, const char *value, int cflags)
 110 {
 111         const char *name;
 112         int namelen;
 113         struct funcname_pattern_list *pp;
 114
 115         name = var + 5; /* "diff." */
 116         namelen = ep - name;
 117
 118         for (pp = funcname_pattern_list; pp; pp = pp->next)
 119                 if (!strncmp(pp->e.name, name, namelen) && !pp->e.name[namelen])
 120                         break;
 121         if (!pp) {
 122                 pp = xcalloc(1, sizeof(*pp));
 123                 pp->e.name = xmemdupz(name, namelen);
 124                 pp->next = funcname_pattern_list;
 125                 funcname_pattern_list = pp;
 126         }
 127         free(pp->e.pattern);
 128         pp->e.pattern = xstrdup(value);
 129         pp->e.cflags = cflags;
 130         return 0;
 131 }
 132
 133 /*
 134  * These are to give UI layer defaults.
 135  * The core-level commands such as git-diff-files should
 136  * never be affected by the setting of diff.renames
 137  * the user happens to have in the configuration file.
 138  */
 139 int git_diff_ui_config(const char *var, const char *value, void *cb)
 140 {
 141         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 142                 diff_use_color_default = git_config_colorbool(var, value, -1);
 143                 return 0;
 144         }
 145         if (!strcmp(var, "diff.renames")) {
 146                 if (!value)
 147                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 148                 else if (!strcasecmp(value, "copies") ||
 149                          !strcasecmp(value, "copy"))
 150                         diff_detect_rename_default = DIFF_DETECT_COPY;
 151                 else if (git_config_bool(var,value))
 152                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 153                 return 0;
 154         }
 155         if (!strcmp(var, "diff.autorefreshindex")) {
 156                 diff_auto_refresh_index = git_config_bool(var, value);
 157                 return 0;
 158         }
 159         if (!strcmp(var, "diff.mnemonicprefix")) {
 160                 diff_mnemonic_prefix = git_config_bool(var, value);
 161                 return 0;
 162         }
 163         if (!strcmp(var, "diff.external"))
 164                 return git_config_string(&external_diff_cmd_cfg, var, value);
 165         if (!prefixcmp(var, "diff.")) {
 166                 const char *ep = strrchr(var, '.');
 167
 168                 if (ep != var + 4 && !strcmp(ep, ".command"))
 169                         return parse_lldiff_command(var, ep, value);
 170         }
 171
 172         return git_diff_basic_config(var, value, cb);
 173 }
 174
 175 int git_diff_basic_config(const char *var, const char *value, void *cb)
 176 {
 177         if (!strcmp(var, "diff.renamelimit")) {
 178                 diff_rename_limit_default = git_config_int(var, value);
 179                 return 0;
 180         }
 181
 182         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 183                 int slot = parse_diff_color_slot(var, 11);
 184                 if (!value)
 185                         return config_error_nonbool(var);
 186                 color_parse(value, var, diff_colors[slot]);
 187                 return 0;
 188         }
 189
 190         /* like GNU diff's --suppress-blank-empty option  */
 191         if (!strcmp(var, "diff.suppress-blank-empty")) {
 192                 diff_suppress_blank_empty = git_config_bool(var, value);
 193                 return 0;
 194         }
 195
 196         if (!prefixcmp(var, "diff.")) {
 197                 const char *ep = strrchr(var, '.');
 198                 if (ep != var + 4) {
 199                         if (!strcmp(ep, ".funcname")) {
 200                                 if (!value)
 201                                         return config_error_nonbool(var);
 202                                 return parse_funcname_pattern(var, ep, value,
 203                                         0);
 204                         } else if (!strcmp(ep, ".xfuncname")) {
 205                                 if (!value)
 206                                         return config_error_nonbool(var);
 207                                 return parse_funcname_pattern(var, ep, value,
 208                                         REG_EXTENDED);
 209                         }
 210                 }
 211         }
 212
 213         return git_color_default_config(var, value, cb);
 214 }
 215
 216 static char *quote_two(const char *one, const char *two)
 217 {
 218         int need_one = quote_c_style(one, NULL, NULL, 1);
 219         int need_two = quote_c_style(two, NULL, NULL, 1);
 220         struct strbuf res = STRBUF_INIT;
 221
 222         if (need_one + need_two) {
 223                 strbuf_addch(&res, '"');
 224                 quote_c_style(one, &res, NULL, 1);
 225                 quote_c_style(two, &res, NULL, 1);
 226                 strbuf_addch(&res, '"');
 227         } else {
 228                 strbuf_addstr(&res, one);
 229                 strbuf_addstr(&res, two);
 230         }
 231         return strbuf_detach(&res, NULL);
 232 }
 233
 234 static const char *external_diff(void)
 235 {
 236         static const char *external_diff_cmd = NULL;
 237         static int done_preparing = 0;
 238
 239         if (done_preparing)
 240                 return external_diff_cmd;
 241         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 242         if (!external_diff_cmd)
 243                 external_diff_cmd = external_diff_cmd_cfg;
 244         done_preparing = 1;
 245         return external_diff_cmd;
 246 }
 247
 248 static struct diff_tempfile {
 249         const char *name; /* filename external diff should read from */
 250         char hex[41];
 251         char mode[10];
 252         char tmp_path[PATH_MAX];
 253 } diff_temp[2];
 254
 255 static int count_lines(const char *data, int size)
 256 {
 257         int count, ch, completely_empty = 1, nl_just_seen = 0;
 258         count = 0;
 259         while (0 < size--) {
 260                 ch = *data++;
 261                 if (ch == '\n') {
 262                         count++;
 263                         nl_just_seen = 1;
 264                         completely_empty = 0;
 265                 }
 266                 else {
 267                         nl_just_seen = 0;
 268                         completely_empty = 0;
 269                 }
 270         }
 271         if (completely_empty)
 272                 return 0;
 273         if (!nl_just_seen)
 274                 count++; /* no trailing newline */
 275         return count;
 276 }
 277
 278 static void print_line_count(FILE *file, int count)
 279 {
 280         switch (count) {
 281         case 0:
 282                 fprintf(file, "0,0");
 283                 break;
 284         case 1:
 285                 fprintf(file, "1");
 286                 break;
 287         default:
 288                 fprintf(file, "1,%d", count);
 289                 break;
 290         }
 291 }
 292
 293 static void copy_file_with_prefix(FILE *file,
 294                                   int prefix, const char *data, int size,
 295                                   const char *set, const char *reset)
 296 {
 297         int ch, nl_just_seen = 1;
 298         while (0 < size--) {
 299                 ch = *data++;
 300                 if (nl_just_seen) {
 301                         fputs(set, file);
 302                         putc(prefix, file);
 303                 }
 304                 if (ch == '\n') {
 305                         nl_just_seen = 1;
 306                         fputs(reset, file);
 307                 } else
 308                         nl_just_seen = 0;
 309                 putc(ch, file);
 310         }
 311         if (!nl_just_seen)
 312                 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
 313 }
 314
 315 static void emit_rewrite_diff(const char *name_a,
 316                               const char *name_b,
 317                               struct diff_filespec *one,
 318                               struct diff_filespec *two,
 319                               struct diff_options *o)
 320 {
 321         int lc_a, lc_b;
 322         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
 323         const char *name_a_tab, *name_b_tab;
 324         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 325         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 326         const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
 327         const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
 328         const char *reset = diff_get_color(color_diff, DIFF_RESET);
 329         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 330         const char *a_prefix, *b_prefix;
 331
 332         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 333                 a_prefix = o->b_prefix;
 334                 b_prefix = o->a_prefix;
 335         } else {
 336                 a_prefix = o->a_prefix;
 337                 b_prefix = o->b_prefix;
 338         }
 339
 340         name_a += (*name_a == '/');
 341         name_b += (*name_b == '/');
 342         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 343         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 344
 345         strbuf_reset(&a_name);
 346         strbuf_reset(&b_name);
 347         quote_two_c_style(&a_name, a_prefix, name_a, 0);
 348         quote_two_c_style(&b_name, b_prefix, name_b, 0);
 349
 350         diff_populate_filespec(one, 0);
 351         diff_populate_filespec(two, 0);
 352         lc_a = count_lines(one->data, one->size);
 353         lc_b = count_lines(two->data, two->size);
 354         fprintf(o->file,
 355                 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
 356                 metainfo, a_name.buf, name_a_tab, reset,
 357                 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
 358         print_line_count(o->file, lc_a);
 359         fprintf(o->file, " +");
 360         print_line_count(o->file, lc_b);
 361         fprintf(o->file, " @@%s\n", reset);
 362         if (lc_a)
 363                 copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
 364         if (lc_b)
 365                 copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
 366 }
 367
 368 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 369 {
 370         if (!DIFF_FILE_VALID(one)) {
 371                 mf->ptr = (char *)""; /* does not matter */
 372                 mf->size = 0;
 373                 return 0;
 374         }
 375         else if (diff_populate_filespec(one, 0))
 376                 return -1;
 377         mf->ptr = one->data;
 378         mf->size = one->size;
 379         return 0;
 380 }
 381
 382 struct diff_words_buffer {
 383         mmfile_t text;
 384         long alloc;
 385         long current; /* output pointer */
 386         int suppressed_newline;
 387 };
 388
 389 static void diff_words_append(char *line, unsigned long len,
 390                 struct diff_words_buffer *buffer)
 391 {
 392         if (buffer->text.size + len > buffer->alloc) {
 393                 buffer->alloc = (buffer->text.size + len) * 3 / 2;
 394                 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
 395         }
 396         line++;
 397         len--;
 398         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 399         buffer->text.size += len;
 400 }
 401
 402 struct diff_words_data {
 403         struct diff_words_buffer minus, plus;
 404         FILE *file;
 405 };
 406
 407 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
 408                 int suppress_newline)
 409 {
 410         const char *ptr;
 411         int eol = 0;
 412
 413         if (len == 0)
 414                 return;
 415
 416         ptr  = buffer->text.ptr + buffer->current;
 417         buffer->current += len;
 418
 419         if (ptr[len - 1] == '\n') {
 420                 eol = 1;
 421                 len--;
 422         }
 423
 424         fputs(diff_get_color(1, color), file);
 425         fwrite(ptr, len, 1, file);
 426         fputs(diff_get_color(1, DIFF_RESET), file);
 427
 428         if (eol) {
 429                 if (suppress_newline)
 430                         buffer->suppressed_newline = 1;
 431                 else
 432                         putc('\n', file);
 433         }
 434 }
 435
 436 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 437 {
 438         struct diff_words_data *diff_words = priv;
 439
 440         if (diff_words->minus.suppressed_newline) {
 441                 if (line[0] != '+')
 442                         putc('\n', diff_words->file);
 443                 diff_words->minus.suppressed_newline = 0;
 444         }
 445
 446         len--;
 447         switch (line[0]) {
 448                 case '-':
 449                         print_word(diff_words->file,
 450                                    &diff_words->minus, len, DIFF_FILE_OLD, 1);
 451                         break;
 452                 case '+':
 453                         print_word(diff_words->file,
 454                                    &diff_words->plus, len, DIFF_FILE_NEW, 0);
 455                         break;
 456                 case ' ':
 457                         print_word(diff_words->file,
 458                                    &diff_words->plus, len, DIFF_PLAIN, 0);
 459                         diff_words->minus.current += len;
 460                         break;
 461         }
 462 }
 463
 464 /* this executes the word diff on the accumulated buffers */
 465 static void diff_words_show(struct diff_words_data *diff_words)
 466 {
 467         xpparam_t xpp;
 468         xdemitconf_t xecfg;
 469         xdemitcb_t ecb;
 470         mmfile_t minus, plus;
 471         int i;
 472
 473         memset(&xpp, 0, sizeof(xpp));
 474         memset(&xecfg, 0, sizeof(xecfg));
 475         minus.size = diff_words->minus.text.size;
 476         minus.ptr = xmalloc(minus.size);
 477         memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
 478         for (i = 0; i < minus.size; i++)
 479                 if (isspace(minus.ptr[i]))
 480                         minus.ptr[i] = '\n';
 481         diff_words->minus.current = 0;
 482
 483         plus.size = diff_words->plus.text.size;
 484         plus.ptr = xmalloc(plus.size);
 485         memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
 486         for (i = 0; i < plus.size; i++)
 487                 if (isspace(plus.ptr[i]))
 488                         plus.ptr[i] = '\n';
 489         diff_words->plus.current = 0;
 490
 491         xpp.flags = XDF_NEED_MINIMAL;
 492         xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
 493         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
 494                       &xpp, &xecfg, &ecb);
 495         free(minus.ptr);
 496         free(plus.ptr);
 497         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 498
 499         if (diff_words->minus.suppressed_newline) {
 500                 putc('\n', diff_words->file);
 501                 diff_words->minus.suppressed_newline = 0;
 502         }
 503 }
 504
 505 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 506
 507 struct emit_callback {
 508         int nparents, color_diff;
 509         unsigned ws_rule;
 510         sane_truncate_fn truncate;
 511         const char **label_path;
 512         struct diff_words_data *diff_words;
 513         int *found_changesp;
 514         FILE *file;
 515 };
 516
 517 static void free_diff_words_data(struct emit_callback *ecbdata)
 518 {
 519         if (ecbdata->diff_words) {
 520                 /* flush buffers */
 521                 if (ecbdata->diff_words->minus.text.size ||
 522                                 ecbdata->diff_words->plus.text.size)
 523                         diff_words_show(ecbdata->diff_words);
 524
 525                 free (ecbdata->diff_words->minus.text.ptr);
 526                 free (ecbdata->diff_words->plus.text.ptr);
 527                 free(ecbdata->diff_words);
 528                 ecbdata->diff_words = NULL;
 529         }
 530 }
 531
 532 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 533 {
 534         if (diff_use_color)
 535                 return diff_colors[ix];
 536         return "";
 537 }
 538
 539 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
 540 {
 541         int has_trailing_newline, has_trailing_carriage_return;
 542
 543         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 544         if (has_trailing_newline)
 545                 len--;
 546         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 547         if (has_trailing_carriage_return)
 548                 len--;
 549
 550         fputs(set, file);
 551         fwrite(line, len, 1, file);
 552         fputs(reset, file);
 553         if (has_trailing_carriage_return)
 554                 fputc('\r', file);
 555         if (has_trailing_newline)
 556                 fputc('\n', file);
 557 }
 558
 559 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
 560 {
 561         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 562         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 563
 564         if (!*ws)
 565                 emit_line(ecbdata->file, set, reset, line, len);
 566         else {
 567                 /* Emit just the prefix, then the rest. */
 568                 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
 569                 ws_check_emit(line + ecbdata->nparents,
 570                               len - ecbdata->nparents, ecbdata->ws_rule,
 571                               ecbdata->file, set, reset, ws);
 572         }
 573 }
 574
 575 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
 576 {
 577         const char *cp;
 578         unsigned long allot;
 579         size_t l = len;
 580
 581         if (ecb->truncate)
 582                 return ecb->truncate(line, len);
 583         cp = line;
 584         allot = l;
 585         while (0 < l) {
 586                 (void) utf8_width(&cp, &l);
 587                 if (!cp)
 588                         break; /* truncated in the middle? */
 589         }
 590         return allot - l;
 591 }
 592
 593 static void fn_out_consume(void *priv, char *line, unsigned long len)
 594 {
 595         int i;
 596         int color;
 597         struct emit_callback *ecbdata = priv;
 598         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 599         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 600         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 601
 602         *(ecbdata->found_changesp) = 1;
 603
 604         if (ecbdata->label_path[0]) {
 605                 const char *name_a_tab, *name_b_tab;
 606
 607                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 608                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 609
 610                 fprintf(ecbdata->file, "%s--- %s%s%s\n",
 611                         meta, ecbdata->label_path[0], reset, name_a_tab);
 612                 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
 613                         meta, ecbdata->label_path[1], reset, name_b_tab);
 614                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 615         }
 616
 617         if (diff_suppress_blank_empty
 618             && len == 2 && line[0] == ' ' && line[1] == '\n') {
 619                 line[0] = '\n';
 620                 len = 1;
 621         }
 622
 623         /* This is not really necessary for now because
 624          * this codepath only deals with two-way diffs.
 625          */
 626         for (i = 0; i < len && line[i] == '@'; i++)
 627                 ;
 628         if (2 <= i && i < len && line[i] == ' ') {
 629                 ecbdata->nparents = i - 1;
 630                 len = sane_truncate_line(ecbdata, line, len);
 631                 emit_line(ecbdata->file,
 632                           diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
 633                           reset, line, len);
 634                 if (line[len-1] != '\n')
 635                         putc('\n', ecbdata->file);
 636                 return;
 637         }
 638
 639         if (len < ecbdata->nparents) {
 640                 emit_line(ecbdata->file, reset, reset, line, len);
 641                 return;
 642         }
 643
 644         color = DIFF_PLAIN;
 645         if (ecbdata->diff_words && ecbdata->nparents != 1)
 646                 /* fall back to normal diff */
 647                 free_diff_words_data(ecbdata);
 648         if (ecbdata->diff_words) {
 649                 if (line[0] == '-') {
 650                         diff_words_append(line, len,
 651                                           &ecbdata->diff_words->minus);
 652                         return;
 653                 } else if (line[0] == '+') {
 654                         diff_words_append(line, len,
 655                                           &ecbdata->diff_words->plus);
 656                         return;
 657                 }
 658                 if (ecbdata->diff_words->minus.text.size ||
 659                     ecbdata->diff_words->plus.text.size)
 660                         diff_words_show(ecbdata->diff_words);
 661                 line++;
 662                 len--;
 663                 emit_line(ecbdata->file, plain, reset, line, len);
 664                 return;
 665         }
 666         for (i = 0; i < ecbdata->nparents && len; i++) {
 667                 if (line[i] == '-')
 668                         color = DIFF_FILE_OLD;
 669                 else if (line[i] == '+')
 670                         color = DIFF_FILE_NEW;
 671         }
 672
 673         if (color != DIFF_FILE_NEW) {
 674                 emit_line(ecbdata->file,
 675                           diff_get_color(ecbdata->color_diff, color),
 676                           reset, line, len);
 677                 return;
 678         }
 679         emit_add_line(reset, ecbdata, line, len);
 680 }
 681
 682 static char *pprint_rename(const char *a, const char *b)
 683 {
 684         const char *old = a;
 685         const char *new = b;
 686         struct strbuf name = STRBUF_INIT;
 687         int pfx_length, sfx_length;
 688         int len_a = strlen(a);
 689         int len_b = strlen(b);
 690         int a_midlen, b_midlen;
 691         int qlen_a = quote_c_style(a, NULL, NULL, 0);
 692         int qlen_b = quote_c_style(b, NULL, NULL, 0);
 693
 694         if (qlen_a || qlen_b) {
 695                 quote_c_style(a, &name, NULL, 0);
 696                 strbuf_addstr(&name, " => ");
 697                 quote_c_style(b, &name, NULL, 0);
 698                 return strbuf_detach(&name, NULL);
 699         }
 700
 701         /* Find common prefix */
 702         pfx_length = 0;
 703         while (*old && *new && *old == *new) {
 704                 if (*old == '/')
 705                         pfx_length = old - a + 1;
 706                 old++;
 707                 new++;
 708         }
 709
 710         /* Find common suffix */
 711         old = a + len_a;
 712         new = b + len_b;
 713         sfx_length = 0;
 714         while (a <= old && b <= new && *old == *new) {
 715                 if (*old == '/')
 716                         sfx_length = len_a - (old - a);
 717                 old--;
 718                 new--;
 719         }
 720
 721         /*
 722          * pfx{mid-a => mid-b}sfx
 723          * {pfx-a => pfx-b}sfx
 724          * pfx{sfx-a => sfx-b}
 725          * name-a => name-b
 726          */
 727         a_midlen = len_a - pfx_length - sfx_length;
 728         b_midlen = len_b - pfx_length - sfx_length;
 729         if (a_midlen < 0)
 730                 a_midlen = 0;
 731         if (b_midlen < 0)
 732                 b_midlen = 0;
 733
 734         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 735         if (pfx_length + sfx_length) {
 736                 strbuf_add(&name, a, pfx_length);
 737                 strbuf_addch(&name, '{');
 738         }
 739         strbuf_add(&name, a + pfx_length, a_midlen);
 740         strbuf_addstr(&name, " => ");
 741         strbuf_add(&name, b + pfx_length, b_midlen);
 742         if (pfx_length + sfx_length) {
 743                 strbuf_addch(&name, '}');
 744                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 745         }
 746         return strbuf_detach(&name, NULL);
 747 }
 748
 749 struct diffstat_t {
 750         int nr;
 751         int alloc;
 752         struct diffstat_file {
 753                 char *from_name;
 754                 char *name;
 755                 char *print_name;
 756                 unsigned is_unmerged:1;
 757                 unsigned is_binary:1;
 758                 unsigned is_renamed:1;
 759                 unsigned int added, deleted;
 760         } **files;
 761 };
 762
 763 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 764                                           const char *name_a,
 765                                           const char *name_b)
 766 {
 767         struct diffstat_file *x;
 768         x = xcalloc(sizeof (*x), 1);
 769         if (diffstat->nr == diffstat->alloc) {
 770                 diffstat->alloc = alloc_nr(diffstat->alloc);
 771                 diffstat->files = xrealloc(diffstat->files,
 772                                 diffstat->alloc * sizeof(x));
 773         }
 774         diffstat->files[diffstat->nr++] = x;
 775         if (name_b) {
 776                 x->from_name = xstrdup(name_a);
 777                 x->name = xstrdup(name_b);
 778                 x->is_renamed = 1;
 779         }
 780         else {
 781                 x->from_name = NULL;
 782                 x->name = xstrdup(name_a);
 783         }
 784         return x;
 785 }
 786
 787 static void diffstat_consume(void *priv, char *line, unsigned long len)
 788 {
 789         struct diffstat_t *diffstat = priv;
 790         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 791
 792         if (line[0] == '+')
 793                 x->added++;
 794         else if (line[0] == '-')
 795                 x->deleted++;
 796 }
 797
 798 const char mime_boundary_leader[] = "------------";
 799
 800 static int scale_linear(int it, int width, int max_change)
 801 {
 802         /*
 803          * make sure that at least one '-' is printed if there were deletions,
 804          * and likewise for '+'.
 805          */
 806         if (max_change < 2)
 807                 return it;
 808         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
 809 }
 810
 811 static void show_name(FILE *file,
 812                       const char *prefix, const char *name, int len,
 813                       const char *reset, const char *set)
 814 {
 815         fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
 816 }
 817
 818 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
 819 {
 820         if (cnt <= 0)
 821                 return;
 822         fprintf(file, "%s", set);
 823         while (cnt--)
 824                 putc(ch, file);
 825         fprintf(file, "%s", reset);
 826 }
 827
 828 static void fill_print_name(struct diffstat_file *file)
 829 {
 830         char *pname;
 831
 832         if (file->print_name)
 833                 return;
 834
 835         if (!file->is_renamed) {
 836                 struct strbuf buf = STRBUF_INIT;
 837                 if (quote_c_style(file->name, &buf, NULL, 0)) {
 838                         pname = strbuf_detach(&buf, NULL);
 839                 } else {
 840                         pname = file->name;
 841                         strbuf_release(&buf);
 842                 }
 843         } else {
 844                 pname = pprint_rename(file->from_name, file->name);
 845         }
 846         file->print_name = pname;
 847 }
 848
 849 static void show_stats(struct diffstat_t* data, struct diff_options *options)
 850 {
 851         int i, len, add, del, total, adds = 0, dels = 0;
 852         int max_change = 0, max_len = 0;
 853         int total_files = data->nr;
 854         int width, name_width;
 855         const char *reset, *set, *add_c, *del_c;
 856
 857         if (data->nr == 0)
 858                 return;
 859
 860         width = options->stat_width ? options->stat_width : 80;
 861         name_width = options->stat_name_width ? options->stat_name_width : 50;
 862
 863         /* Sanity: give at least 5 columns to the graph,
 864          * but leave at least 10 columns for the name.
 865          */
 866         if (width < 25)
 867                 width = 25;
 868         if (name_width < 10)
 869                 name_width = 10;
 870         else if (width < name_width + 15)
 871                 name_width = width - 15;
 872
 873         /* Find the longest filename and max number of changes */
 874         reset = diff_get_color_opt(options, DIFF_RESET);
 875         set   = diff_get_color_opt(options, DIFF_PLAIN);
 876         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
 877         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
 878
 879         for (i = 0; i < data->nr; i++) {
 880                 struct diffstat_file *file = data->files[i];
 881                 int change = file->added + file->deleted;
 882                 fill_print_name(file);
 883                 len = strlen(file->print_name);
 884                 if (max_len < len)
 885                         max_len = len;
 886
 887                 if (file->is_binary || file->is_unmerged)
 888                         continue;
 889                 if (max_change < change)
 890                         max_change = change;
 891         }
 892
 893         /* Compute the width of the graph part;
 894          * 10 is for one blank at the beginning of the line plus
 895          * " | count " between the name and the graph.
 896          *
 897          * From here on, name_width is the width of the name area,
 898          * and width is the width of the graph area.
 899          */
 900         name_width = (name_width < max_len) ? name_width : max_len;
 901         if (width < (name_width + 10) + max_change)
 902                 width = width - (name_width + 10);
 903         else
 904                 width = max_change;
 905
 906         for (i = 0; i < data->nr; i++) {
 907                 const char *prefix = "";
 908                 char *name = data->files[i]->print_name;
 909                 int added = data->files[i]->added;
 910                 int deleted = data->files[i]->deleted;
 911                 int name_len;
 912
 913                 /*
 914                  * "scale" the filename
 915                  */
 916                 len = name_width;
 917                 name_len = strlen(name);
 918                 if (name_width < name_len) {
 919                         char *slash;
 920                         prefix = "...";
 921                         len -= 3;
 922                         name += name_len - len;
 923                         slash = strchr(name, '/');
 924                         if (slash)
 925                                 name = slash;
 926                 }
 927
 928                 if (data->files[i]->is_binary) {
 929                         show_name(options->file, prefix, name, len, reset, set);
 930                         fprintf(options->file, "  Bin ");
 931                         fprintf(options->file, "%s%d%s", del_c, deleted, reset);
 932                         fprintf(options->file, " -> ");
 933                         fprintf(options->file, "%s%d%s", add_c, added, reset);
 934                         fprintf(options->file, " bytes");
 935                         fprintf(options->file, "\n");
 936                         continue;
 937                 }
 938                 else if (data->files[i]->is_unmerged) {
 939                         show_name(options->file, prefix, name, len, reset, set);
 940                         fprintf(options->file, "  Unmerged\n");
 941                         continue;
 942                 }
 943                 else if (!data->files[i]->is_renamed &&
 944                          (added + deleted == 0)) {
 945                         total_files--;
 946                         continue;
 947                 }
 948
 949                 /*
 950                  * scale the add/delete
 951                  */
 952                 add = added;
 953                 del = deleted;
 954                 total = add + del;
 955                 adds += add;
 956                 dels += del;
 957
 958                 if (width <= max_change) {
 959                         add = scale_linear(add, width, max_change);
 960                         del = scale_linear(del, width, max_change);
 961                         total = add + del;
 962                 }
 963                 show_name(options->file, prefix, name, len, reset, set);
 964                 fprintf(options->file, "%5d%s", added + deleted,
 965                                 added + deleted ? " " : "");
 966                 show_graph(options->file, '+', add, add_c, reset);
 967                 show_graph(options->file, '-', del, del_c, reset);
 968                 fprintf(options->file, "\n");
 969         }
 970         fprintf(options->file,
 971                "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
 972                set, total_files, adds, dels, reset);
 973 }
 974
 975 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
 976 {
 977         int i, adds = 0, dels = 0, total_files = data->nr;
 978
 979         if (data->nr == 0)
 980                 return;
 981
 982         for (i = 0; i < data->nr; i++) {
 983                 if (!data->files[i]->is_binary &&
 984                     !data->files[i]->is_unmerged) {
 985                         int added = data->files[i]->added;
 986                         int deleted= data->files[i]->deleted;
 987                         if (!data->files[i]->is_renamed &&
 988                             (added + deleted == 0)) {
 989                                 total_files--;
 990                         } else {
 991                                 adds += added;
 992                                 dels += deleted;
 993                         }
 994                 }
 995         }
 996         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
 997                total_files, adds, dels);
 998 }
 999
1000 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
1001 {
1002         int i;
1003
1004         if (data->nr == 0)
1005                 return;
1006
1007         for (i = 0; i < data->nr; i++) {
1008                 struct diffstat_file *file = data->files[i];
1009
1010                 if (file->is_binary)
1011                         fprintf(options->file, "-\t-\t");
1012                 else
1013                         fprintf(options->file,
1014                                 "%d\t%d\t", file->added, file->deleted);
1015                 if (options->line_termination) {
1016                         fill_print_name(file);
1017                         if (!file->is_renamed)
1018                                 write_name_quoted(file->name, options->file,
1019                                                   options->line_termination);
1020                         else {
1021                                 fputs(file->print_name, options->file);
1022                                 putc(options->line_termination, options->file);
1023                         }
1024                 } else {
1025                         if (file->is_renamed) {
1026                                 putc('\0', options->file);
1027                                 write_name_quoted(file->from_name, options->file, '\0');
1028                         }
1029                         write_name_quoted(file->name, options->file, '\0');
1030                 }
1031         }
1032 }
1033
1034 struct dirstat_file {
1035         const char *name;
1036         unsigned long changed;
1037 };
1038
1039 struct dirstat_dir {
1040         struct dirstat_file *files;
1041         int alloc, nr, percent, cumulative;
1042 };
1043
1044 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1045 {
1046         unsigned long this_dir = 0;
1047         unsigned int sources = 0;
1048
1049         while (dir->nr) {
1050                 struct dirstat_file *f = dir->files;
1051                 int namelen = strlen(f->name);
1052                 unsigned long this;
1053                 char *slash;
1054
1055                 if (namelen < baselen)
1056                         break;
1057                 if (memcmp(f->name, base, baselen))
1058                         break;
1059                 slash = strchr(f->name + baselen, '/');
1060                 if (slash) {
1061                         int newbaselen = slash + 1 - f->name;
1062                         this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1063                         sources++;
1064                 } else {
1065                         this = f->changed;
1066                         dir->files++;
1067                         dir->nr--;
1068                         sources += 2;
1069                 }
1070                 this_dir += this;
1071         }
1072
1073         /*
1074          * We don't report dirstat's for
1075          *  - the top level
1076          *  - or cases where everything came from a single directory
1077          *    under this directory (sources == 1).
1078          */
1079         if (baselen && sources != 1) {
1080                 int permille = this_dir * 1000 / changed;
1081                 if (permille) {
1082                         int percent = permille / 10;
1083                         if (percent >= dir->percent) {
1084                                 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1085                                 if (!dir->cumulative)
1086                                         return 0;
1087                         }
1088                 }
1089         }
1090         return this_dir;
1091 }
1092
1093 static int dirstat_compare(const void *_a, const void *_b)
1094 {
1095         const struct dirstat_file *a = _a;
1096         const struct dirstat_file *b = _b;
1097         return strcmp(a->name, b->name);
1098 }
1099
1100 static void show_dirstat(struct diff_options *options)
1101 {
1102         int i;
1103         unsigned long changed;
1104         struct dirstat_dir dir;
1105         struct diff_queue_struct *q = &diff_queued_diff;
1106
1107         dir.files = NULL;
1108         dir.alloc = 0;
1109         dir.nr = 0;
1110         dir.percent = options->dirstat_percent;
1111         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1112
1113         changed = 0;
1114         for (i = 0; i < q->nr; i++) {
1115                 struct diff_filepair *p = q->queue[i];
1116                 const char *name;
1117                 unsigned long copied, added, damage;
1118
1119                 name = p->one->path ? p->one->path : p->two->path;
1120
1121                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1122                         diff_populate_filespec(p->one, 0);
1123                         diff_populate_filespec(p->two, 0);
1124                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1125                                                &copied, &added);
1126                         diff_free_filespec_data(p->one);
1127                         diff_free_filespec_data(p->two);
1128                 } else if (DIFF_FILE_VALID(p->one)) {
1129                         diff_populate_filespec(p->one, 1);
1130                         copied = added = 0;
1131                         diff_free_filespec_data(p->one);
1132                 } else if (DIFF_FILE_VALID(p->two)) {
1133                         diff_populate_filespec(p->two, 1);
1134                         copied = 0;
1135                         added = p->two->size;
1136                         diff_free_filespec_data(p->two);
1137                 } else
1138                         continue;
1139
1140                 /*
1141                  * Original minus copied is the removed material,
1142                  * added is the new material.  They are both damages
1143                  * made to the preimage. In --dirstat-by-file mode, count
1144                  * damaged files, not damaged lines. This is done by
1145                  * counting only a single damaged line per file.
1146                  */
1147                 damage = (p->one->size - copied) + added;
1148                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1149                         damage = 1;
1150
1151                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1152                 dir.files[dir.nr].name = name;
1153                 dir.files[dir.nr].changed = damage;
1154                 changed += damage;
1155                 dir.nr++;
1156         }
1157
1158         /* This can happen even with many files, if everything was renames */
1159         if (!changed)
1160                 return;
1161
1162         /* Show all directories with more than x% of the changes */
1163         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1164         gather_dirstat(options->file, &dir, changed, "", 0);
1165 }
1166
1167 static void free_diffstat_info(struct diffstat_t *diffstat)
1168 {
1169         int i;
1170         for (i = 0; i < diffstat->nr; i++) {
1171                 struct diffstat_file *f = diffstat->files[i];
1172                 if (f->name != f->print_name)
1173                         free(f->print_name);
1174                 free(f->name);
1175                 free(f->from_name);
1176                 free(f);
1177         }
1178         free(diffstat->files);
1179 }
1180
1181 struct checkdiff_t {
1182         const char *filename;
1183         int lineno;
1184         struct diff_options *o;
1185         unsigned ws_rule;
1186         unsigned status;
1187         int trailing_blanks_start;
1188 };
1189
1190 static int is_conflict_marker(const char *line, unsigned long len)
1191 {
1192         char firstchar;
1193         int cnt;
1194
1195         if (len < 8)
1196                 return 0;
1197         firstchar = line[0];
1198         switch (firstchar) {
1199         case '=': case '>': case '<':
1200                 break;
1201         default:
1202                 return 0;
1203         }
1204         for (cnt = 1; cnt < 7; cnt++)
1205                 if (line[cnt] != firstchar)
1206                         return 0;
1207         /* line[0] thru line[6] are same as firstchar */
1208         if (firstchar == '=') {
1209                 /* divider between ours and theirs? */
1210                 if (len != 8 || line[7] != '\n')
1211                         return 0;
1212         } else if (len < 8 || !isspace(line[7])) {
1213                 /* not divider before ours nor after theirs */
1214                 return 0;
1215         }
1216         return 1;
1217 }
1218
1219 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1220 {
1221         struct checkdiff_t *data = priv;
1222         int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1223         const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1224         const char *reset = diff_get_color(color_diff, DIFF_RESET);
1225         const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1226         char *err;
1227
1228         if (line[0] == '+') {
1229                 unsigned bad;
1230                 data->lineno++;
1231                 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1232                         data->trailing_blanks_start = 0;
1233                 else if (!data->trailing_blanks_start)
1234                         data->trailing_blanks_start = data->lineno;
1235                 if (is_conflict_marker(line + 1, len - 1)) {
1236                         data->status |= 1;
1237                         fprintf(data->o->file,
1238                                 "%s:%d: leftover conflict marker\n",
1239                                 data->filename, data->lineno);
1240                 }
1241                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1242                 if (!bad)
1243                         return;
1244                 data->status |= bad;
1245                 err = whitespace_error_string(bad);
1246                 fprintf(data->o->file, "%s:%d: %s.\n",
1247                         data->filename, data->lineno, err);
1248                 free(err);
1249                 emit_line(data->o->file, set, reset, line, 1);
1250                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1251                               data->o->file, set, reset, ws);
1252         } else if (line[0] == ' ') {
1253                 data->lineno++;
1254                 data->trailing_blanks_start = 0;
1255         } else if (line[0] == '@') {
1256                 char *plus = strchr(line, '+');
1257                 if (plus)
1258                         data->lineno = strtol(plus, NULL, 10) - 1;
1259                 else
1260                         die("invalid diff");
1261                 data->trailing_blanks_start = 0;
1262         }
1263 }
1264
1265 static unsigned char *deflate_it(char *data,
1266                                  unsigned long size,
1267                                  unsigned long *result_size)
1268 {
1269         int bound;
1270         unsigned char *deflated;
1271         z_stream stream;
1272
1273         memset(&stream, 0, sizeof(stream));
1274         deflateInit(&stream, zlib_compression_level);
1275         bound = deflateBound(&stream, size);
1276         deflated = xmalloc(bound);
1277         stream.next_out = deflated;
1278         stream.avail_out = bound;
1279
1280         stream.next_in = (unsigned char *)data;
1281         stream.avail_in = size;
1282         while (deflate(&stream, Z_FINISH) == Z_OK)
1283                 ; /* nothing */
1284         deflateEnd(&stream);
1285         *result_size = stream.total_out;
1286         return deflated;
1287 }
1288
1289 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1290 {
1291         void *cp;
1292         void *delta;
1293         void *deflated;
1294         void *data;
1295         unsigned long orig_size;
1296         unsigned long delta_size;
1297         unsigned long deflate_size;
1298         unsigned long data_size;
1299
1300         /* We could do deflated delta, or we could do just deflated two,
1301          * whichever is smaller.
1302          */
1303         delta = NULL;
1304         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1305         if (one->size && two->size) {
1306                 delta = diff_delta(one->ptr, one->size,
1307                                    two->ptr, two->size,
1308                                    &delta_size, deflate_size);
1309                 if (delta) {
1310                         void *to_free = delta;
1311                         orig_size = delta_size;
1312                         delta = deflate_it(delta, delta_size, &delta_size);
1313                         free(to_free);
1314                 }
1315         }
1316
1317         if (delta && delta_size < deflate_size) {
1318                 fprintf(file, "delta %lu\n", orig_size);
1319                 free(deflated);
1320                 data = delta;
1321                 data_size = delta_size;
1322         }
1323         else {
1324                 fprintf(file, "literal %lu\n", two->size);
1325                 free(delta);
1326                 data = deflated;
1327                 data_size = deflate_size;
1328         }
1329
1330         /* emit data encoded in base85 */
1331         cp = data;
1332         while (data_size) {
1333                 int bytes = (52 < data_size) ? 52 : data_size;
1334                 char line[70];
1335                 data_size -= bytes;
1336                 if (bytes <= 26)
1337                         line[0] = bytes + 'A' - 1;
1338                 else
1339                         line[0] = bytes - 26 + 'a' - 1;
1340                 encode_85(line + 1, cp, bytes);
1341                 cp = (char *) cp + bytes;
1342                 fputs(line, file);
1343                 fputc('\n', file);
1344         }
1345         fprintf(file, "\n");
1346         free(data);
1347 }
1348
1349 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1350 {
1351         fprintf(file, "GIT binary patch\n");
1352         emit_binary_diff_body(file, one, two);
1353         emit_binary_diff_body(file, two, one);
1354 }
1355
1356 static void setup_diff_attr_check(struct git_attr_check *check)
1357 {
1358         static struct git_attr *attr_diff;
1359
1360         if (!attr_diff) {
1361                 attr_diff = git_attr("diff", 4);
1362         }
1363         check[0].attr = attr_diff;
1364 }
1365
1366 static void diff_filespec_check_attr(struct diff_filespec *one)
1367 {
1368         struct git_attr_check attr_diff_check;
1369         int check_from_data = 0;
1370
1371         if (one->checked_attr)
1372                 return;
1373
1374         setup_diff_attr_check(&attr_diff_check);
1375         one->is_binary = 0;
1376         one->funcname_pattern_ident = NULL;
1377
1378         if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1379                 const char *value;
1380
1381                 /* binaryness */
1382                 value = attr_diff_check.value;
1383                 if (ATTR_TRUE(value))
1384                         ;
1385                 else if (ATTR_FALSE(value))
1386                         one->is_binary = 1;
1387                 else
1388                         check_from_data = 1;
1389
1390                 /* funcname pattern ident */
1391                 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1392                         ;
1393                 else
1394                         one->funcname_pattern_ident = value;
1395         }
1396
1397         if (check_from_data) {
1398                 if (!one->data && DIFF_FILE_VALID(one))
1399                         diff_populate_filespec(one, 0);
1400
1401                 if (one->data)
1402                         one->is_binary = buffer_is_binary(one->data, one->size);
1403         }
1404 }
1405
1406 int diff_filespec_is_binary(struct diff_filespec *one)
1407 {
1408         diff_filespec_check_attr(one);
1409         return one->is_binary;
1410 }
1411
1412 static const struct funcname_pattern_entry *funcname_pattern(const char *ident)
1413 {
1414         struct funcname_pattern_list *pp;
1415
1416         for (pp = funcname_pattern_list; pp; pp = pp->next)
1417                 if (!strcmp(ident, pp->e.name))
1418                         return &pp->e;
1419         return NULL;
1420 }
1421
1422 static const struct funcname_pattern_entry builtin_funcname_pattern[] = {
1423         { "bibtex", "(@[a-zA-Z]{1,}[ \t]*\\{{0,1}[ \t]*[^ \t\"@',\\#}{~%]*).*$",
1424           REG_EXTENDED },
1425         { "html", "^[ \t]*(<[Hh][1-6][ \t].*>.*)$", REG_EXTENDED },
1426         { "java",
1427           "!^[ \t]*(catch|do|for|if|instanceof|new|return|switch|throw|while)\n"
1428           "^[ \t]*(([ \t]*[A-Za-z_][A-Za-z_0-9]*){2,}[ \t]*\\([^;]*)$",
1429           REG_EXTENDED },
1430         { "objc",
1431           /* Negate C statements that can look like functions */
1432           "!^[ \t]*(do|for|if|else|return|switch|while)\n"
1433           /* Objective-C methods */
1434           "^[ \t]*([-+][ \t]*\\([ \t]*[A-Za-z_][A-Za-z_0-9* \t]*\\)[ \t]*[A-Za-z_].*)$\n"
1435           /* C functions */
1436           "^[ \t]*(([ \t]*[A-Za-z_][A-Za-z_0-9]*){2,}[ \t]*\\([^;]*)$\n"
1437           /* Objective-C class/protocol definitions */
1438           "^(@(implementation|interface|protocol)[ \t].*)$",
1439           REG_EXTENDED },
1440         { "pascal",
1441           "^((procedure|function|constructor|destructor|interface|"
1442                 "implementation|initialization|finalization)[ \t]*.*)$"
1443           "\n"
1444           "^(.*=[ \t]*(class|record).*)$",
1445           REG_EXTENDED },
1446         { "php", "^[\t ]*((function|class).*)", REG_EXTENDED },
1447         { "python", "^[ \t]*((class|def)[ \t].*)$", REG_EXTENDED },
1448         { "ruby", "^[ \t]*((class|module|def)[ \t].*)$",
1449           REG_EXTENDED },
1450         { "tex",
1451           "^(\\\\((sub)*section|chapter|part)\\*{0,1}\\{.*)$",
1452           REG_EXTENDED },
1453 };
1454
1455 static const struct funcname_pattern_entry *diff_funcname_pattern(struct diff_filespec *one)
1456 {
1457         const char *ident;
1458         const struct funcname_pattern_entry *pe;
1459         int i;
1460
1461         diff_filespec_check_attr(one);
1462         ident = one->funcname_pattern_ident;
1463
1464         if (!ident)
1465                 /*
1466                  * If the config file has "funcname.default" defined, that
1467                  * regexp is used; otherwise NULL is returned and xemit uses
1468                  * the built-in default.
1469                  */
1470                 return funcname_pattern("default");
1471
1472         /* Look up custom "funcname.$ident" regexp from config. */
1473         pe = funcname_pattern(ident);
1474         if (pe)
1475                 return pe;
1476
1477         /*
1478          * And define built-in fallback patterns here.  Note that
1479          * these can be overridden by the user's config settings.
1480          */
1481         for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1482                 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1483                         return &builtin_funcname_pattern[i];
1484
1485         return NULL;
1486 }
1487
1488 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1489 {
1490         if (!options->a_prefix)
1491                 options->a_prefix = a;
1492         if (!options->b_prefix)
1493                 options->b_prefix = b;
1494 }
1495
1496 static void builtin_diff(const char *name_a,
1497                          const char *name_b,
1498                          struct diff_filespec *one,
1499                          struct diff_filespec *two,
1500                          const char *xfrm_msg,
1501                          struct diff_options *o,
1502                          int complete_rewrite)
1503 {
1504         mmfile_t mf1, mf2;
1505         const char *lbl[2];
1506         char *a_one, *b_two;
1507         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1508         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1509         const char *a_prefix, *b_prefix;
1510
1511         diff_set_mnemonic_prefix(o, "a/", "b/");
1512         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1513                 a_prefix = o->b_prefix;
1514                 b_prefix = o->a_prefix;
1515         } else {
1516                 a_prefix = o->a_prefix;
1517                 b_prefix = o->b_prefix;
1518         }
1519
1520         /* Never use a non-valid filename anywhere if at all possible */
1521         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1522         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1523
1524         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1525         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1526         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1527         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1528         fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1529         if (lbl[0][0] == '/') {
1530                 /* /dev/null */
1531                 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1532                 if (xfrm_msg && xfrm_msg[0])
1533                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1534         }
1535         else if (lbl[1][0] == '/') {
1536                 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1537                 if (xfrm_msg && xfrm_msg[0])
1538                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1539         }
1540         else {
1541                 if (one->mode != two->mode) {
1542                         fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1543                         fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1544                 }
1545                 if (xfrm_msg && xfrm_msg[0])
1546                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1547                 /*
1548                  * we do not run diff between different kind
1549                  * of objects.
1550                  */
1551                 if ((one->mode ^ two->mode) & S_IFMT)
1552                         goto free_ab_and_return;
1553                 if (complete_rewrite) {
1554                         emit_rewrite_diff(name_a, name_b, one, two, o);
1555                         o->found_changes = 1;
1556                         goto free_ab_and_return;
1557                 }
1558         }
1559
1560         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1561                 die("unable to read files to diff");
1562
1563         if (!DIFF_OPT_TST(o, TEXT) &&
1564             (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1565                 /* Quite common confusing case */
1566                 if (mf1.size == mf2.size &&
1567                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1568                         goto free_ab_and_return;
1569                 if (DIFF_OPT_TST(o, BINARY))
1570                         emit_binary_diff(o->file, &mf1, &mf2);
1571                 else
1572                         fprintf(o->file, "Binary files %s and %s differ\n",
1573                                 lbl[0], lbl[1]);
1574                 o->found_changes = 1;
1575         }
1576         else {
1577                 /* Crazy xdl interfaces.. */
1578                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1579                 xpparam_t xpp;
1580                 xdemitconf_t xecfg;
1581                 xdemitcb_t ecb;
1582                 struct emit_callback ecbdata;
1583                 const struct funcname_pattern_entry *pe;
1584
1585                 pe = diff_funcname_pattern(one);
1586                 if (!pe)
1587                         pe = diff_funcname_pattern(two);
1588
1589                 memset(&xpp, 0, sizeof(xpp));
1590                 memset(&xecfg, 0, sizeof(xecfg));
1591                 memset(&ecbdata, 0, sizeof(ecbdata));
1592                 ecbdata.label_path = lbl;
1593                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1594                 ecbdata.found_changesp = &o->found_changes;
1595                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1596                 ecbdata.file = o->file;
1597                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1598                 xecfg.ctxlen = o->context;
1599                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1600                 if (pe)
1601                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1602                 if (!diffopts)
1603                         ;
1604                 else if (!prefixcmp(diffopts, "--unified="))
1605                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1606                 else if (!prefixcmp(diffopts, "-u"))
1607                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1608                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1609                         ecbdata.diff_words =
1610                                 xcalloc(1, sizeof(struct diff_words_data));
1611                         ecbdata.diff_words->file = o->file;
1612                 }
1613                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1614                               &xpp, &xecfg, &ecb);
1615                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1616                         free_diff_words_data(&ecbdata);
1617         }
1618
1619  free_ab_and_return:
1620         diff_free_filespec_data(one);
1621         diff_free_filespec_data(two);
1622         free(a_one);
1623         free(b_two);
1624         return;
1625 }
1626
1627 static void builtin_diffstat(const char *name_a, const char *name_b,
1628                              struct diff_filespec *one,
1629                              struct diff_filespec *two,
1630                              struct diffstat_t *diffstat,
1631                              struct diff_options *o,
1632                              int complete_rewrite)
1633 {
1634         mmfile_t mf1, mf2;
1635         struct diffstat_file *data;
1636
1637         data = diffstat_add(diffstat, name_a, name_b);
1638
1639         if (!one || !two) {
1640                 data->is_unmerged = 1;
1641                 return;
1642         }
1643         if (complete_rewrite) {
1644                 diff_populate_filespec(one, 0);
1645                 diff_populate_filespec(two, 0);
1646                 data->deleted = count_lines(one->data, one->size);
1647                 data->added = count_lines(two->data, two->size);
1648                 goto free_and_return;
1649         }
1650         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1651                 die("unable to read files to diff");
1652
1653         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1654                 data->is_binary = 1;
1655                 data->added = mf2.size;
1656                 data->deleted = mf1.size;
1657         } else {
1658                 /* Crazy xdl interfaces.. */
1659                 xpparam_t xpp;
1660                 xdemitconf_t xecfg;
1661                 xdemitcb_t ecb;
1662
1663                 memset(&xpp, 0, sizeof(xpp));
1664                 memset(&xecfg, 0, sizeof(xecfg));
1665                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1666                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1667                               &xpp, &xecfg, &ecb);
1668         }
1669
1670  free_and_return:
1671         diff_free_filespec_data(one);
1672         diff_free_filespec_data(two);
1673 }
1674
1675 static void builtin_checkdiff(const char *name_a, const char *name_b,
1676                               const char *attr_path,
1677                               struct diff_filespec *one,
1678                               struct diff_filespec *two,
1679                               struct diff_options *o)
1680 {
1681         mmfile_t mf1, mf2;
1682         struct checkdiff_t data;
1683
1684         if (!two)
1685                 return;
1686
1687         memset(&data, 0, sizeof(data));
1688         data.filename = name_b ? name_b : name_a;
1689         data.lineno = 0;
1690         data.o = o;
1691         data.ws_rule = whitespace_rule(attr_path);
1692
1693         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1694                 die("unable to read files to diff");
1695
1696         /*
1697          * All the other codepaths check both sides, but not checking
1698          * the "old" side here is deliberate.  We are checking the newly
1699          * introduced changes, and as long as the "new" side is text, we
1700          * can and should check what it introduces.
1701          */
1702         if (diff_filespec_is_binary(two))
1703                 goto free_and_return;
1704         else {
1705                 /* Crazy xdl interfaces.. */
1706                 xpparam_t xpp;
1707                 xdemitconf_t xecfg;
1708                 xdemitcb_t ecb;
1709
1710                 memset(&xpp, 0, sizeof(xpp));
1711                 memset(&xecfg, 0, sizeof(xecfg));
1712                 xecfg.ctxlen = 1; /* at least one context line */
1713                 xpp.flags = XDF_NEED_MINIMAL;
1714                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1715                               &xpp, &xecfg, &ecb);
1716
1717                 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1718                     data.trailing_blanks_start) {
1719                         fprintf(o->file, "%s:%d: ends with blank lines.\n",
1720                                 data.filename, data.trailing_blanks_start);
1721                         data.status = 1; /* report errors */
1722                 }
1723         }
1724  free_and_return:
1725         diff_free_filespec_data(one);
1726         diff_free_filespec_data(two);
1727         if (data.status)
1728                 DIFF_OPT_SET(o, CHECK_FAILED);
1729 }
1730
1731 struct diff_filespec *alloc_filespec(const char *path)
1732 {
1733         int namelen = strlen(path);
1734         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1735
1736         memset(spec, 0, sizeof(*spec));
1737         spec->path = (char *)(spec + 1);
1738         memcpy(spec->path, path, namelen+1);
1739         spec->count = 1;
1740         return spec;
1741 }
1742
1743 void free_filespec(struct diff_filespec *spec)
1744 {
1745         if (!--spec->count) {
1746                 diff_free_filespec_data(spec);
1747                 free(spec);
1748         }
1749 }
1750
1751 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1752                    unsigned short mode)
1753 {
1754         if (mode) {
1755                 spec->mode = canon_mode(mode);
1756                 hashcpy(spec->sha1, sha1);
1757                 spec->sha1_valid = !is_null_sha1(sha1);
1758         }
1759 }
1760
1761 /*
1762  * Given a name and sha1 pair, if the index tells us the file in
1763  * the work tree has that object contents, return true, so that
1764  * prepare_temp_file() does not have to inflate and extract.
1765  */
1766 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1767 {
1768         struct cache_entry *ce;
1769         struct stat st;
1770         int pos, len;
1771
1772         /* We do not read the cache ourselves here, because the
1773          * benchmark with my previous version that always reads cache
1774          * shows that it makes things worse for diff-tree comparing
1775          * two linux-2.6 kernel trees in an already checked out work
1776          * tree.  This is because most diff-tree comparisons deal with
1777          * only a small number of files, while reading the cache is
1778          * expensive for a large project, and its cost outweighs the
1779          * savings we get by not inflating the object to a temporary
1780          * file.  Practically, this code only helps when we are used
1781          * by diff-cache --cached, which does read the cache before
1782          * calling us.
1783          */
1784         if (!active_cache)
1785                 return 0;
1786
1787         /* We want to avoid the working directory if our caller
1788          * doesn't need the data in a normal file, this system
1789          * is rather slow with its stat/open/mmap/close syscalls,
1790          * and the object is contained in a pack file.  The pack
1791          * is probably already open and will be faster to obtain
1792          * the data through than the working directory.  Loose
1793          * objects however would tend to be slower as they need
1794          * to be individually opened and inflated.
1795          */
1796         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1797                 return 0;
1798
1799         len = strlen(name);
1800         pos = cache_name_pos(name, len);
1801         if (pos < 0)
1802                 return 0;
1803         ce = active_cache[pos];
1804
1805         /*
1806          * This is not the sha1 we are looking for, or
1807          * unreusable because it is not a regular file.
1808          */
1809         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1810                 return 0;
1811
1812         /*
1813          * If ce matches the file in the work tree, we can reuse it.
1814          */
1815         if (ce_uptodate(ce) ||
1816             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1817                 return 1;
1818
1819         return 0;
1820 }
1821
1822 static int populate_from_stdin(struct diff_filespec *s)
1823 {
1824         struct strbuf buf = STRBUF_INIT;
1825         size_t size = 0;
1826
1827         if (strbuf_read(&buf, 0, 0) < 0)
1828                 return error("error while reading from stdin %s",
1829                                      strerror(errno));
1830
1831         s->should_munmap = 0;
1832         s->data = strbuf_detach(&buf, &size);
1833         s->size = size;
1834         s->should_free = 1;
1835         return 0;
1836 }
1837
1838 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1839 {
1840         int len;
1841         char *data = xmalloc(100);
1842         len = snprintf(data, 100,
1843                 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1844         s->data = data;
1845         s->size = len;
1846         s->should_free = 1;
1847         if (size_only) {
1848                 s->data = NULL;
1849                 free(data);
1850         }
1851         return 0;
1852 }
1853
1854 /*
1855  * While doing rename detection and pickaxe operation, we may need to
1856  * grab the data for the blob (or file) for our own in-core comparison.
1857  * diff_filespec has data and size fields for this purpose.
1858  */
1859 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1860 {
1861         int err = 0;
1862         if (!DIFF_FILE_VALID(s))
1863                 die("internal error: asking to populate invalid file.");
1864         if (S_ISDIR(s->mode))
1865                 return -1;
1866
1867         if (s->data)
1868                 return 0;
1869
1870         if (size_only && 0 < s->size)
1871                 return 0;
1872
1873         if (S_ISGITLINK(s->mode))
1874                 return diff_populate_gitlink(s, size_only);
1875
1876         if (!s->sha1_valid ||
1877             reuse_worktree_file(s->path, s->sha1, 0)) {
1878                 struct strbuf buf = STRBUF_INIT;
1879                 struct stat st;
1880                 int fd;
1881
1882                 if (!strcmp(s->path, "-"))
1883                         return populate_from_stdin(s);
1884
1885                 if (lstat(s->path, &st) < 0) {
1886                         if (errno == ENOENT) {
1887                         err_empty:
1888                                 err = -1;
1889                         empty:
1890                                 s->data = (char *)"";
1891                                 s->size = 0;
1892                                 return err;
1893                         }
1894                 }
1895                 s->size = xsize_t(st.st_size);
1896                 if (!s->size)
1897                         goto empty;
1898                 if (size_only)
1899                         return 0;
1900                 if (S_ISLNK(st.st_mode)) {
1901                         int ret;
1902                         s->data = xmalloc(s->size);
1903                         s->should_free = 1;
1904                         ret = readlink(s->path, s->data, s->size);
1905                         if (ret < 0) {
1906                                 free(s->data);
1907                                 goto err_empty;
1908                         }
1909                         return 0;
1910                 }
1911                 fd = open(s->path, O_RDONLY);
1912                 if (fd < 0)
1913                         goto err_empty;
1914                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1915                 close(fd);
1916                 s->should_munmap = 1;
1917
1918                 /*
1919                  * Convert from working tree format to canonical git format
1920                  */
1921                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1922                         size_t size = 0;
1923                         munmap(s->data, s->size);
1924                         s->should_munmap = 0;
1925                         s->data = strbuf_detach(&buf, &size);
1926                         s->size = size;
1927                         s->should_free = 1;
1928                 }
1929         }
1930         else {
1931                 enum object_type type;
1932                 if (size_only)
1933                         type = sha1_object_info(s->sha1, &s->size);
1934                 else {
1935                         s->data = read_sha1_file(s->sha1, &type, &s->size);
1936                         s->should_free = 1;
1937                 }
1938         }
1939         return 0;
1940 }
1941
1942 void diff_free_filespec_blob(struct diff_filespec *s)
1943 {
1944         if (s->should_free)
1945                 free(s->data);
1946         else if (s->should_munmap)
1947                 munmap(s->data, s->size);
1948
1949         if (s->should_free || s->should_munmap) {
1950                 s->should_free = s->should_munmap = 0;
1951                 s->data = NULL;
1952         }
1953 }
1954
1955 void diff_free_filespec_data(struct diff_filespec *s)
1956 {
1957         diff_free_filespec_blob(s);
1958         free(s->cnt_data);
1959         s->cnt_data = NULL;
1960 }
1961
1962 static void prep_temp_blob(struct diff_tempfile *temp,
1963                            void *blob,
1964                            unsigned long size,
1965                            const unsigned char *sha1,
1966                            int mode)
1967 {
1968         int fd;
1969
1970         fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1971         if (fd < 0)
1972                 die("unable to create temp-file: %s", strerror(errno));
1973         if (write_in_full(fd, blob, size) != size)
1974                 die("unable to write temp-file");
1975         close(fd);
1976         temp->name = temp->tmp_path;
1977         strcpy(temp->hex, sha1_to_hex(sha1));
1978         temp->hex[40] = 0;
1979         sprintf(temp->mode, "%06o", mode);
1980 }
1981
1982 static void prepare_temp_file(const char *name,
1983                               struct diff_tempfile *temp,
1984                               struct diff_filespec *one)
1985 {
1986         if (!DIFF_FILE_VALID(one)) {
1987         not_a_valid_file:
1988                 /* A '-' entry produces this for file-2, and
1989                  * a '+' entry produces this for file-1.
1990                  */
1991                 temp->name = "/dev/null";
1992                 strcpy(temp->hex, ".");
1993                 strcpy(temp->mode, ".");
1994                 return;
1995         }
1996
1997         if (!one->sha1_valid ||
1998             reuse_worktree_file(name, one->sha1, 1)) {
1999                 struct stat st;
2000                 if (lstat(name, &st) < 0) {
2001                         if (errno == ENOENT)
2002                                 goto not_a_valid_file;
2003                         die("stat(%s): %s", name, strerror(errno));
2004                 }
2005                 if (S_ISLNK(st.st_mode)) {
2006                         int ret;
2007                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
2008                         size_t sz = xsize_t(st.st_size);
2009                         if (sizeof(buf) <= st.st_size)
2010                                 die("symlink too long: %s", name);
2011                         ret = readlink(name, buf, sz);
2012                         if (ret < 0)
2013                                 die("readlink(%s)", name);
2014                         prep_temp_blob(temp, buf, sz,
2015                                        (one->sha1_valid ?
2016                                         one->sha1 : null_sha1),
2017                                        (one->sha1_valid ?
2018                                         one->mode : S_IFLNK));
2019                 }
2020                 else {
2021                         /* we can borrow from the file in the work tree */
2022                         temp->name = name;
2023                         if (!one->sha1_valid)
2024                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2025                         else
2026                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2027                         /* Even though we may sometimes borrow the
2028                          * contents from the work tree, we always want
2029                          * one->mode.  mode is trustworthy even when
2030                          * !(one->sha1_valid), as long as
2031                          * DIFF_FILE_VALID(one).
2032                          */
2033                         sprintf(temp->mode, "%06o", one->mode);
2034                 }
2035                 return;
2036         }
2037         else {
2038                 if (diff_populate_filespec(one, 0))
2039                         die("cannot read data blob for %s", one->path);
2040                 prep_temp_blob(temp, one->data, one->size,
2041                                one->sha1, one->mode);
2042         }
2043 }
2044
2045 static void remove_tempfile(void)
2046 {
2047         int i;
2048
2049         for (i = 0; i < 2; i++)
2050                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
2051                         unlink(diff_temp[i].name);
2052                         diff_temp[i].name = NULL;
2053                 }
2054 }
2055
2056 static void remove_tempfile_on_signal(int signo)
2057 {
2058         remove_tempfile();
2059         signal(SIGINT, SIG_DFL);
2060         raise(signo);
2061 }
2062
2063 /* An external diff command takes:
2064  *
2065  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2066  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2067  *
2068  */
2069 static void run_external_diff(const char *pgm,
2070                               const char *name,
2071                               const char *other,
2072                               struct diff_filespec *one,
2073                               struct diff_filespec *two,
2074                               const char *xfrm_msg,
2075                               int complete_rewrite)
2076 {
2077         const char *spawn_arg[10];
2078         struct diff_tempfile *temp = diff_temp;
2079         int retval;
2080         static int atexit_asked = 0;
2081         const char *othername;
2082         const char **arg = &spawn_arg[0];
2083
2084         othername = (other? other : name);
2085         if (one && two) {
2086                 prepare_temp_file(name, &temp[0], one);
2087                 prepare_temp_file(othername, &temp[1], two);
2088                 if (! atexit_asked &&
2089                     (temp[0].name == temp[0].tmp_path ||
2090                      temp[1].name == temp[1].tmp_path)) {
2091                         atexit_asked = 1;
2092                         atexit(remove_tempfile);
2093                 }
2094                 signal(SIGINT, remove_tempfile_on_signal);
2095         }
2096
2097         if (one && two) {
2098                 *arg++ = pgm;
2099                 *arg++ = name;
2100                 *arg++ = temp[0].name;
2101                 *arg++ = temp[0].hex;
2102                 *arg++ = temp[0].mode;
2103                 *arg++ = temp[1].name;
2104                 *arg++ = temp[1].hex;
2105                 *arg++ = temp[1].mode;
2106                 if (other) {
2107                         *arg++ = other;
2108                         *arg++ = xfrm_msg;
2109                 }
2110         } else {
2111                 *arg++ = pgm;
2112                 *arg++ = name;
2113         }
2114         *arg = NULL;
2115         fflush(NULL);
2116         retval = run_command_v_opt(spawn_arg, 0);
2117         remove_tempfile();
2118         if (retval) {
2119                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2120                 exit(1);
2121         }
2122 }
2123
2124 static const char *external_diff_attr(const char *name)
2125 {
2126         struct git_attr_check attr_diff_check;
2127
2128         if (!name)
2129                 return NULL;
2130
2131         setup_diff_attr_check(&attr_diff_check);
2132         if (!git_checkattr(name, 1, &attr_diff_check)) {
2133                 const char *value = attr_diff_check.value;
2134                 if (!ATTR_TRUE(value) &&
2135                     !ATTR_FALSE(value) &&
2136                     !ATTR_UNSET(value)) {
2137                         struct ll_diff_driver *drv;
2138
2139                         for (drv = user_diff; drv; drv = drv->next)
2140                                 if (!strcmp(drv->name, value))
2141                                         return drv->cmd;
2142                 }
2143         }
2144         return NULL;
2145 }
2146
2147 static void run_diff_cmd(const char *pgm,
2148                          const char *name,
2149                          const char *other,
2150                          const char *attr_path,
2151                          struct diff_filespec *one,
2152                          struct diff_filespec *two,
2153                          const char *xfrm_msg,
2154                          struct diff_options *o,
2155                          int complete_rewrite)
2156 {
2157         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2158                 pgm = NULL;
2159         else {
2160                 const char *cmd = external_diff_attr(attr_path);
2161                 if (cmd)
2162                         pgm = cmd;
2163         }
2164
2165         if (pgm) {
2166                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2167                                   complete_rewrite);
2168                 return;
2169         }
2170         if (one && two)
2171                 builtin_diff(name, other ? other : name,
2172                              one, two, xfrm_msg, o, complete_rewrite);
2173         else
2174                 fprintf(o->file, "* Unmerged path %s\n", name);
2175 }
2176
2177 static void diff_fill_sha1_info(struct diff_filespec *one)
2178 {
2179         if (DIFF_FILE_VALID(one)) {
2180                 if (!one->sha1_valid) {
2181                         struct stat st;
2182                         if (!strcmp(one->path, "-")) {
2183                                 hashcpy(one->sha1, null_sha1);
2184                                 return;
2185                         }
2186                         if (lstat(one->path, &st) < 0)
2187                                 die("stat %s", one->path);
2188                         if (index_path(one->sha1, one->path, &st, 0))
2189                                 die("cannot hash %s\n", one->path);
2190                 }
2191         }
2192         else
2193                 hashclr(one->sha1);
2194 }
2195
2196 static int similarity_index(struct diff_filepair *p)
2197 {
2198         return p->score * 100 / MAX_SCORE;
2199 }
2200
2201 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2202 {
2203         /* Strip the prefix but do not molest /dev/null and absolute paths */
2204         if (*namep && **namep != '/')
2205                 *namep += prefix_length;
2206         if (*otherp && **otherp != '/')
2207                 *otherp += prefix_length;
2208 }
2209
2210 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2211 {
2212         const char *pgm = external_diff();
2213         struct strbuf msg;
2214         char *xfrm_msg;
2215         struct diff_filespec *one = p->one;
2216         struct diff_filespec *two = p->two;
2217         const char *name;
2218         const char *other;
2219         const char *attr_path;
2220         int complete_rewrite = 0;
2221
2222         name  = p->one->path;
2223         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2224         attr_path = name;
2225         if (o->prefix_length)
2226                 strip_prefix(o->prefix_length, &name, &other);
2227
2228         if (DIFF_PAIR_UNMERGED(p)) {
2229                 run_diff_cmd(pgm, name, NULL, attr_path,
2230                              NULL, NULL, NULL, o, 0);
2231                 return;
2232         }
2233
2234         diff_fill_sha1_info(one);
2235         diff_fill_sha1_info(two);
2236
2237         strbuf_init(&msg, PATH_MAX * 2 + 300);
2238         switch (p->status) {
2239         case DIFF_STATUS_COPIED:
2240                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2241                 strbuf_addstr(&msg, "\ncopy from ");
2242                 quote_c_style(name, &msg, NULL, 0);
2243                 strbuf_addstr(&msg, "\ncopy to ");
2244                 quote_c_style(other, &msg, NULL, 0);
2245                 strbuf_addch(&msg, '\n');
2246                 break;
2247         case DIFF_STATUS_RENAMED:
2248                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2249                 strbuf_addstr(&msg, "\nrename from ");
2250                 quote_c_style(name, &msg, NULL, 0);
2251                 strbuf_addstr(&msg, "\nrename to ");
2252                 quote_c_style(other, &msg, NULL, 0);
2253                 strbuf_addch(&msg, '\n');
2254                 break;
2255         case DIFF_STATUS_MODIFIED:
2256                 if (p->score) {
2257                         strbuf_addf(&msg, "dissimilarity index %d%%\n",
2258                                         similarity_index(p));
2259                         complete_rewrite = 1;
2260                         break;
2261                 }
2262                 /* fallthru */
2263         default:
2264                 /* nothing */
2265                 ;
2266         }
2267
2268         if (hashcmp(one->sha1, two->sha1)) {
2269                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2270
2271                 if (DIFF_OPT_TST(o, BINARY)) {
2272                         mmfile_t mf;
2273                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2274                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2275                                 abbrev = 40;
2276                 }
2277                 strbuf_addf(&msg, "index %.*s..%.*s",
2278                                 abbrev, sha1_to_hex(one->sha1),
2279                                 abbrev, sha1_to_hex(two->sha1));
2280                 if (one->mode == two->mode)
2281                         strbuf_addf(&msg, " %06o", one->mode);
2282                 strbuf_addch(&msg, '\n');
2283         }
2284
2285         if (msg.len)
2286                 strbuf_setlen(&msg, msg.len - 1);
2287         xfrm_msg = msg.len ? msg.buf : NULL;
2288
2289         if (!pgm &&
2290             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2291             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2292                 /* a filepair that changes between file and symlink
2293                  * needs to be split into deletion and creation.
2294                  */
2295                 struct diff_filespec *null = alloc_filespec(two->path);
2296                 run_diff_cmd(NULL, name, other, attr_path,
2297                              one, null, xfrm_msg, o, 0);
2298                 free(null);
2299                 null = alloc_filespec(one->path);
2300                 run_diff_cmd(NULL, name, other, attr_path,
2301                              null, two, xfrm_msg, o, 0);
2302                 free(null);
2303         }
2304         else
2305                 run_diff_cmd(pgm, name, other, attr_path,
2306                              one, two, xfrm_msg, o, complete_rewrite);
2307
2308         strbuf_release(&msg);
2309 }
2310
2311 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2312                          struct diffstat_t *diffstat)
2313 {
2314         const char *name;
2315         const char *other;
2316         int complete_rewrite = 0;
2317
2318         if (DIFF_PAIR_UNMERGED(p)) {
2319                 /* unmerged */
2320                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2321                 return;
2322         }
2323
2324         name = p->one->path;
2325         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2326
2327         if (o->prefix_length)
2328                 strip_prefix(o->prefix_length, &name, &other);
2329
2330         diff_fill_sha1_info(p->one);
2331         diff_fill_sha1_info(p->two);
2332
2333         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2334                 complete_rewrite = 1;
2335         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2336 }
2337
2338 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2339 {
2340         const char *name;
2341         const char *other;
2342         const char *attr_path;
2343
2344         if (DIFF_PAIR_UNMERGED(p)) {
2345                 /* unmerged */
2346                 return;
2347         }
2348
2349         name = p->one->path;
2350         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2351         attr_path = other ? other : name;
2352
2353         if (o->prefix_length)
2354                 strip_prefix(o->prefix_length, &name, &other);
2355
2356         diff_fill_sha1_info(p->one);
2357         diff_fill_sha1_info(p->two);
2358
2359         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2360 }
2361
2362 void diff_setup(struct diff_options *options)
2363 {
2364         memset(options, 0, sizeof(*options));
2365
2366         options->file = stdout;
2367
2368         options->line_termination = '\n';
2369         options->break_opt = -1;
2370         options->rename_limit = -1;
2371         options->dirstat_percent = 3;
2372         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
2373         options->context = 3;
2374
2375         options->change = diff_change;
2376         options->add_remove = diff_addremove;
2377         if (diff_use_color_default > 0)
2378                 DIFF_OPT_SET(options, COLOR_DIFF);
2379         else
2380                 DIFF_OPT_CLR(options, COLOR_DIFF);
2381         options->detect_rename = diff_detect_rename_default;
2382
2383         if (!diff_mnemonic_prefix) {
2384                 options->a_prefix = "a/";
2385                 options->b_prefix = "b/";
2386         }
2387 }
2388
2389 int diff_setup_done(struct diff_options *options)
2390 {
2391         int count = 0;
2392
2393         if (options->output_format & DIFF_FORMAT_NAME)
2394                 count++;
2395         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2396                 count++;
2397         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2398                 count++;
2399         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2400                 count++;
2401         if (count > 1)
2402                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2403
2404         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2405                 options->detect_rename = DIFF_DETECT_COPY;
2406
2407         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2408                 options->prefix = NULL;
2409         if (options->prefix)
2410                 options->prefix_length = strlen(options->prefix);
2411         else
2412                 options->prefix_length = 0;
2413
2414         if (options->output_format & (DIFF_FORMAT_NAME |
2415                                       DIFF_FORMAT_NAME_STATUS |
2416                                       DIFF_FORMAT_CHECKDIFF |
2417                                       DIFF_FORMAT_NO_OUTPUT))
2418                 options->output_format &= ~(DIFF_FORMAT_RAW |
2419                                             DIFF_FORMAT_NUMSTAT |
2420                                             DIFF_FORMAT_DIFFSTAT |
2421                                             DIFF_FORMAT_SHORTSTAT |
2422                                             DIFF_FORMAT_DIRSTAT |
2423                                             DIFF_FORMAT_SUMMARY |
2424                                             DIFF_FORMAT_PATCH);
2425
2426         /*
2427          * These cases always need recursive; we do not drop caller-supplied
2428          * recursive bits for other formats here.
2429          */
2430         if (options->output_format & (DIFF_FORMAT_PATCH |
2431                                       DIFF_FORMAT_NUMSTAT |
2432                                       DIFF_FORMAT_DIFFSTAT |
2433                                       DIFF_FORMAT_SHORTSTAT |
2434                                       DIFF_FORMAT_DIRSTAT |
2435                                       DIFF_FORMAT_SUMMARY |
2436                                       DIFF_FORMAT_CHECKDIFF))
2437                 DIFF_OPT_SET(options, RECURSIVE);
2438         /*
2439          * Also pickaxe would not work very well if you do not say recursive
2440          */
2441         if (options->pickaxe)
2442                 DIFF_OPT_SET(options, RECURSIVE);
2443
2444         if (options->detect_rename && options->rename_limit < 0)
2445                 options->rename_limit = diff_rename_limit_default;
2446         if (options->setup & DIFF_SETUP_USE_CACHE) {
2447                 if (!active_cache)
2448                         /* read-cache does not die even when it fails
2449                          * so it is safe for us to do this here.  Also
2450                          * it does not smudge active_cache or active_nr
2451                          * when it fails, so we do not have to worry about
2452                          * cleaning it up ourselves either.
2453                          */
2454                         read_cache();
2455         }
2456         if (options->abbrev <= 0 || 40 < options->abbrev)
2457                 options->abbrev = 40; /* full */
2458
2459         /*
2460          * It does not make sense to show the first hit we happened
2461          * to have found.  It does not make sense not to return with
2462          * exit code in such a case either.
2463          */
2464         if (DIFF_OPT_TST(options, QUIET)) {
2465                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2466                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2467         }
2468
2469         return 0;
2470 }
2471
2472 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2473 {
2474         char c, *eq;
2475         int len;
2476
2477         if (*arg != '-')
2478                 return 0;
2479         c = *++arg;
2480         if (!c)
2481                 return 0;
2482         if (c == arg_short) {
2483                 c = *++arg;
2484                 if (!c)
2485                         return 1;
2486                 if (val && isdigit(c)) {
2487                         char *end;
2488                         int n = strtoul(arg, &end, 10);
2489                         if (*end)
2490                                 return 0;
2491                         *val = n;
2492                         return 1;
2493                 }
2494                 return 0;
2495         }
2496         if (c != '-')
2497                 return 0;
2498         arg++;
2499         eq = strchr(arg, '=');
2500         if (eq)
2501                 len = eq - arg;
2502         else
2503                 len = strlen(arg);
2504         if (!len || strncmp(arg, arg_long, len))
2505                 return 0;
2506         if (eq) {
2507                 int n;
2508                 char *end;
2509                 if (!isdigit(*++eq))
2510                         return 0;
2511                 n = strtoul(eq, &end, 10);
2512                 if (*end)
2513                         return 0;
2514                 *val = n;
2515         }
2516         return 1;
2517 }
2518
2519 static int diff_scoreopt_parse(const char *opt);
2520
2521 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2522 {
2523         const char *arg = av[0];
2524
2525         /* Output format options */
2526         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2527                 options->output_format |= DIFF_FORMAT_PATCH;
2528         else if (opt_arg(arg, 'U', "unified", &options->context))
2529                 options->output_format |= DIFF_FORMAT_PATCH;
2530         else if (!strcmp(arg, "--raw"))
2531                 options->output_format |= DIFF_FORMAT_RAW;
2532         else if (!strcmp(arg, "--patch-with-raw"))
2533                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2534         else if (!strcmp(arg, "--numstat"))
2535                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2536         else if (!strcmp(arg, "--shortstat"))
2537                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2538         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2539                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2540         else if (!strcmp(arg, "--cumulative")) {
2541                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2542                 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2543         } else if (opt_arg(arg, 0, "dirstat-by-file",
2544                            &options->dirstat_percent)) {
2545                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2546                 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
2547         }
2548         else if (!strcmp(arg, "--check"))
2549                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2550         else if (!strcmp(arg, "--summary"))
2551                 options->output_format |= DIFF_FORMAT_SUMMARY;
2552         else if (!strcmp(arg, "--patch-with-stat"))
2553                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2554         else if (!strcmp(arg, "--name-only"))
2555                 options->output_format |= DIFF_FORMAT_NAME;
2556         else if (!strcmp(arg, "--name-status"))
2557                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2558         else if (!strcmp(arg, "-s"))
2559                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2560         else if (!prefixcmp(arg, "--stat")) {
2561                 char *end;
2562                 int width = options->stat_width;
2563                 int name_width = options->stat_name_width;
2564                 arg += 6;
2565                 end = (char *)arg;
2566
2567                 switch (*arg) {
2568                 case '-':
2569                         if (!prefixcmp(arg, "-width="))
2570                                 width = strtoul(arg + 7, &end, 10);
2571                         else if (!prefixcmp(arg, "-name-width="))
2572                                 name_width = strtoul(arg + 12, &end, 10);
2573                         break;
2574                 case '=':
2575                         width = strtoul(arg+1, &end, 10);
2576                         if (*end == ',')
2577                                 name_width = strtoul(end+1, &end, 10);
2578                 }
2579
2580                 /* Important! This checks all the error cases! */
2581                 if (*end)
2582                         return 0;
2583                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2584                 options->stat_name_width = name_width;
2585                 options->stat_width = width;
2586         }
2587
2588         /* renames options */
2589         else if (!prefixcmp(arg, "-B")) {
2590                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2591                         return -1;
2592         }
2593         else if (!prefixcmp(arg, "-M")) {
2594                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2595                         return -1;
2596                 options->detect_rename = DIFF_DETECT_RENAME;
2597         }
2598         else if (!prefixcmp(arg, "-C")) {
2599                 if (options->detect_rename == DIFF_DETECT_COPY)
2600                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2601                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2602                         return -1;
2603                 options->detect_rename = DIFF_DETECT_COPY;
2604         }
2605         else if (!strcmp(arg, "--no-renames"))
2606                 options->detect_rename = 0;
2607         else if (!strcmp(arg, "--relative"))
2608                 DIFF_OPT_SET(options, RELATIVE_NAME);
2609         else if (!prefixcmp(arg, "--relative=")) {
2610                 DIFF_OPT_SET(options, RELATIVE_NAME);
2611                 options->prefix = arg + 11;
2612         }
2613
2614         /* xdiff options */
2615         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2616                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2617         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2618                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2619         else if (!strcmp(arg, "--ignore-space-at-eol"))
2620                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2621
2622         /* flags options */
2623         else if (!strcmp(arg, "--binary")) {
2624                 options->output_format |= DIFF_FORMAT_PATCH;
2625                 DIFF_OPT_SET(options, BINARY);
2626         }
2627         else if (!strcmp(arg, "--full-index"))
2628                 DIFF_OPT_SET(options, FULL_INDEX);
2629         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2630                 DIFF_OPT_SET(options, TEXT);
2631         else if (!strcmp(arg, "-R"))
2632                 DIFF_OPT_SET(options, REVERSE_DIFF);
2633         else if (!strcmp(arg, "--find-copies-harder"))
2634                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2635         else if (!strcmp(arg, "--follow"))
2636                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2637         else if (!strcmp(arg, "--color"))
2638                 DIFF_OPT_SET(options, COLOR_DIFF);
2639         else if (!strcmp(arg, "--no-color"))
2640                 DIFF_OPT_CLR(options, COLOR_DIFF);
2641         else if (!strcmp(arg, "--color-words"))
2642                 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2643         else if (!strcmp(arg, "--exit-code"))
2644                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2645         else if (!strcmp(arg, "--quiet"))
2646                 DIFF_OPT_SET(options, QUIET);
2647         else if (!strcmp(arg, "--ext-diff"))
2648                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2649         else if (!strcmp(arg, "--no-ext-diff"))
2650                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2651         else if (!strcmp(arg, "--ignore-submodules"))
2652                 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2653
2654         /* misc options */
2655         else if (!strcmp(arg, "-z"))
2656                 options->line_termination = 0;
2657         else if (!prefixcmp(arg, "-l"))
2658                 options->rename_limit = strtoul(arg+2, NULL, 10);
2659         else if (!prefixcmp(arg, "-S"))
2660                 options->pickaxe = arg + 2;
2661         else if (!strcmp(arg, "--pickaxe-all"))
2662                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2663         else if (!strcmp(arg, "--pickaxe-regex"))
2664                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2665         else if (!prefixcmp(arg, "-O"))
2666                 options->orderfile = arg + 2;
2667         else if (!prefixcmp(arg, "--diff-filter="))
2668                 options->filter = arg + 14;
2669         else if (!strcmp(arg, "--abbrev"))
2670                 options->abbrev = DEFAULT_ABBREV;
2671         else if (!prefixcmp(arg, "--abbrev=")) {
2672                 options->abbrev = strtoul(arg + 9, NULL, 10);
2673                 if (options->abbrev < MINIMUM_ABBREV)
2674                         options->abbrev = MINIMUM_ABBREV;
2675                 else if (40 < options->abbrev)
2676                         options->abbrev = 40;
2677         }
2678         else if (!prefixcmp(arg, "--src-prefix="))
2679                 options->a_prefix = arg + 13;
2680         else if (!prefixcmp(arg, "--dst-prefix="))
2681                 options->b_prefix = arg + 13;
2682         else if (!strcmp(arg, "--no-prefix"))
2683                 options->a_prefix = options->b_prefix = "";
2684         else if (!prefixcmp(arg, "--output=")) {
2685                 options->file = fopen(arg + strlen("--output="), "w");
2686                 options->close_file = 1;
2687         } else
2688                 return 0;
2689         return 1;
2690 }
2691
2692 static int parse_num(const char **cp_p)
2693 {
2694         unsigned long num, scale;
2695         int ch, dot;
2696         const char *cp = *cp_p;
2697
2698         num = 0;
2699         scale = 1;
2700         dot = 0;
2701         for(;;) {
2702                 ch = *cp;
2703                 if ( !dot && ch == '.' ) {
2704                         scale = 1;
2705                         dot = 1;
2706                 } else if ( ch == '%' ) {
2707                         scale = dot ? scale*100 : 100;
2708                         cp++;   /* % is always at the end */
2709                         break;
2710                 } else if ( ch >= '0' && ch <= '9' ) {
2711                         if ( scale < 100000 ) {
2712                                 scale *= 10;
2713                                 num = (num*10) + (ch-'0');
2714                         }
2715                 } else {
2716                         break;
2717                 }
2718                 cp++;
2719         }
2720         *cp_p = cp;
2721
2722         /* user says num divided by scale and we say internally that
2723          * is MAX_SCORE * num / scale.
2724          */
2725         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2726 }
2727
2728 static int diff_scoreopt_parse(const char *opt)
2729 {
2730         int opt1, opt2, cmd;
2731
2732         if (*opt++ != '-')
2733                 return -1;
2734         cmd = *opt++;
2735         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2736                 return -1; /* that is not a -M, -C nor -B option */
2737
2738         opt1 = parse_num(&opt);
2739         if (cmd != 'B')
2740                 opt2 = 0;
2741         else {
2742                 if (*opt == 0)
2743                         opt2 = 0;
2744                 else if (*opt != '/')
2745                         return -1; /* we expect -B80/99 or -B80 */
2746                 else {
2747                         opt++;
2748                         opt2 = parse_num(&opt);
2749                 }
2750         }
2751         if (*opt != 0)
2752                 return -1;
2753         return opt1 | (opt2 << 16);
2754 }
2755
2756 struct diff_queue_struct diff_queued_diff;
2757
2758 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2759 {
2760         if (queue->alloc <= queue->nr) {
2761                 queue->alloc = alloc_nr(queue->alloc);
2762                 queue->queue = xrealloc(queue->queue,
2763                                         sizeof(dp) * queue->alloc);
2764         }
2765         queue->queue[queue->nr++] = dp;
2766 }
2767
2768 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2769                                  struct diff_filespec *one,
2770                                  struct diff_filespec *two)
2771 {
2772         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2773         dp->one = one;
2774         dp->two = two;
2775         if (queue)
2776                 diff_q(queue, dp);
2777         return dp;
2778 }
2779
2780 void diff_free_filepair(struct diff_filepair *p)
2781 {
2782         free_filespec(p->one);
2783         free_filespec(p->two);
2784         free(p);
2785 }
2786
2787 /* This is different from find_unique_abbrev() in that
2788  * it stuffs the result with dots for alignment.
2789  */
2790 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2791 {
2792         int abblen;
2793         const char *abbrev;
2794         if (len == 40)
2795                 return sha1_to_hex(sha1);
2796
2797         abbrev = find_unique_abbrev(sha1, len);
2798         abblen = strlen(abbrev);
2799         if (abblen < 37) {
2800                 static char hex[41];
2801                 if (len < abblen && abblen <= len + 2)
2802                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2803                 else
2804                         sprintf(hex, "%s...", abbrev);
2805                 return hex;
2806         }
2807         return sha1_to_hex(sha1);
2808 }
2809
2810 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2811 {
2812         int line_termination = opt->line_termination;
2813         int inter_name_termination = line_termination ? '\t' : '\0';
2814
2815         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2816                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2817                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
2818                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2819         }
2820         if (p->score) {
2821                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2822                         inter_name_termination);
2823         } else {
2824                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2825         }
2826
2827         if (p->status == DIFF_STATUS_COPIED ||
2828             p->status == DIFF_STATUS_RENAMED) {
2829                 const char *name_a, *name_b;
2830                 name_a = p->one->path;
2831                 name_b = p->two->path;
2832                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2833                 write_name_quoted(name_a, opt->file, inter_name_termination);
2834                 write_name_quoted(name_b, opt->file, line_termination);
2835         } else {
2836                 const char *name_a, *name_b;
2837                 name_a = p->one->mode ? p->one->path : p->two->path;
2838                 name_b = NULL;
2839                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2840                 write_name_quoted(name_a, opt->file, line_termination);
2841         }
2842 }
2843
2844 int diff_unmodified_pair(struct diff_filepair *p)
2845 {
2846         /* This function is written stricter than necessary to support
2847          * the currently implemented transformers, but the idea is to
2848          * let transformers to produce diff_filepairs any way they want,
2849          * and filter and clean them up here before producing the output.
2850          */
2851         struct diff_filespec *one = p->one, *two = p->two;
2852
2853         if (DIFF_PAIR_UNMERGED(p))
2854                 return 0; /* unmerged is interesting */
2855
2856         /* deletion, addition, mode or type change
2857          * and rename are all interesting.
2858          */
2859         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2860             DIFF_PAIR_MODE_CHANGED(p) ||
2861             strcmp(one->path, two->path))
2862                 return 0;
2863
2864         /* both are valid and point at the same path.  that is, we are
2865          * dealing with a change.
2866          */
2867         if (one->sha1_valid && two->sha1_valid &&
2868             !hashcmp(one->sha1, two->sha1))
2869                 return 1; /* no change */
2870         if (!one->sha1_valid && !two->sha1_valid)
2871                 return 1; /* both look at the same file on the filesystem. */
2872         return 0;
2873 }
2874
2875 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2876 {
2877         if (diff_unmodified_pair(p))
2878                 return;
2879
2880         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2881             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2882                 return; /* no tree diffs in patch format */
2883
2884         run_diff(p, o);
2885 }
2886
2887 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2888                             struct diffstat_t *diffstat)
2889 {
2890         if (diff_unmodified_pair(p))
2891                 return;
2892
2893         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2894             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2895                 return; /* no tree diffs in patch format */
2896
2897         run_diffstat(p, o, diffstat);
2898 }
2899
2900 static void diff_flush_checkdiff(struct diff_filepair *p,
2901                 struct diff_options *o)
2902 {
2903         if (diff_unmodified_pair(p))
2904                 return;
2905
2906         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2907             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2908                 return; /* no tree diffs in patch format */
2909
2910         run_checkdiff(p, o);
2911 }
2912
2913 int diff_queue_is_empty(void)
2914 {
2915         struct diff_queue_struct *q = &diff_queued_diff;
2916         int i;
2917         for (i = 0; i < q->nr; i++)
2918                 if (!diff_unmodified_pair(q->queue[i]))
2919                         return 0;
2920         return 1;
2921 }
2922
2923 #if DIFF_DEBUG
2924 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2925 {
2926         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2927                 x, one ? one : "",
2928                 s->path,
2929                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2930                 s->mode,
2931                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2932         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2933                 x, one ? one : "",
2934                 s->size, s->xfrm_flags);
2935 }
2936
2937 void diff_debug_filepair(const struct diff_filepair *p, int i)
2938 {
2939         diff_debug_filespec(p->one, i, "one");
2940         diff_debug_filespec(p->two, i, "two");
2941         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2942                 p->score, p->status ? p->status : '?',
2943                 p->one->rename_used, p->broken_pair);
2944 }
2945
2946 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2947 {
2948         int i;
2949         if (msg)
2950                 fprintf(stderr, "%s\n", msg);
2951         fprintf(stderr, "q->nr = %d\n", q->nr);
2952         for (i = 0; i < q->nr; i++) {
2953                 struct diff_filepair *p = q->queue[i];
2954                 diff_debug_filepair(p, i);
2955         }
2956 }
2957 #endif
2958
2959 static void diff_resolve_rename_copy(void)
2960 {
2961         int i;
2962         struct diff_filepair *p;
2963         struct diff_queue_struct *q = &diff_queued_diff;
2964
2965         diff_debug_queue("resolve-rename-copy", q);
2966
2967         for (i = 0; i < q->nr; i++) {
2968                 p = q->queue[i];
2969                 p->status = 0; /* undecided */
2970                 if (DIFF_PAIR_UNMERGED(p))
2971                         p->status = DIFF_STATUS_UNMERGED;
2972                 else if (!DIFF_FILE_VALID(p->one))
2973                         p->status = DIFF_STATUS_ADDED;
2974                 else if (!DIFF_FILE_VALID(p->two))
2975                         p->status = DIFF_STATUS_DELETED;
2976                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2977                         p->status = DIFF_STATUS_TYPE_CHANGED;
2978
2979                 /* from this point on, we are dealing with a pair
2980                  * whose both sides are valid and of the same type, i.e.
2981                  * either in-place edit or rename/copy edit.
2982                  */
2983                 else if (DIFF_PAIR_RENAME(p)) {
2984                         /*
2985                          * A rename might have re-connected a broken
2986                          * pair up, causing the pathnames to be the
2987                          * same again. If so, that's not a rename at
2988                          * all, just a modification..
2989                          *
2990                          * Otherwise, see if this source was used for
2991                          * multiple renames, in which case we decrement
2992                          * the count, and call it a copy.
2993                          */
2994                         if (!strcmp(p->one->path, p->two->path))
2995                                 p->status = DIFF_STATUS_MODIFIED;
2996                         else if (--p->one->rename_used > 0)
2997                                 p->status = DIFF_STATUS_COPIED;
2998                         else
2999                                 p->status = DIFF_STATUS_RENAMED;
3000                 }
3001                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
3002                          p->one->mode != p->two->mode ||
3003                          is_null_sha1(p->one->sha1))
3004                         p->status = DIFF_STATUS_MODIFIED;
3005                 else {
3006                         /* This is a "no-change" entry and should not
3007                          * happen anymore, but prepare for broken callers.
3008                          */
3009                         error("feeding unmodified %s to diffcore",
3010                               p->one->path);
3011                         p->status = DIFF_STATUS_UNKNOWN;
3012                 }
3013         }
3014         diff_debug_queue("resolve-rename-copy done", q);
3015 }
3016
3017 static int check_pair_status(struct diff_filepair *p)
3018 {
3019         switch (p->status) {
3020         case DIFF_STATUS_UNKNOWN:
3021                 return 0;
3022         case 0:
3023                 die("internal error in diff-resolve-rename-copy");
3024         default:
3025                 return 1;
3026         }
3027 }
3028
3029 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3030 {
3031         int fmt = opt->output_format;
3032
3033         if (fmt & DIFF_FORMAT_CHECKDIFF)
3034                 diff_flush_checkdiff(p, opt);
3035         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3036                 diff_flush_raw(p, opt);
3037         else if (fmt & DIFF_FORMAT_NAME) {
3038                 const char *name_a, *name_b;
3039                 name_a = p->two->path;
3040                 name_b = NULL;
3041                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3042                 write_name_quoted(name_a, opt->file, opt->line_termination);
3043         }
3044 }
3045
3046 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3047 {
3048         if (fs->mode)
3049                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3050         else
3051                 fprintf(file, " %s ", newdelete);
3052         write_name_quoted(fs->path, file, '\n');
3053 }
3054
3055
3056 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
3057 {
3058         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3059                 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
3060                         show_name ? ' ' : '\n');
3061                 if (show_name) {
3062                         write_name_quoted(p->two->path, file, '\n');
3063                 }
3064         }
3065 }
3066
3067 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3068 {
3069         char *names = pprint_rename(p->one->path, p->two->path);
3070
3071         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3072         free(names);
3073         show_mode_change(file, p, 0);
3074 }
3075
3076 static void diff_summary(FILE *file, struct diff_filepair *p)
3077 {
3078         switch(p->status) {
3079         case DIFF_STATUS_DELETED:
3080                 show_file_mode_name(file, "delete", p->one);
3081                 break;
3082         case DIFF_STATUS_ADDED:
3083                 show_file_mode_name(file, "create", p->two);
3084                 break;
3085         case DIFF_STATUS_COPIED:
3086                 show_rename_copy(file, "copy", p);
3087                 break;
3088         case DIFF_STATUS_RENAMED:
3089                 show_rename_copy(file, "rename", p);
3090                 break;
3091         default:
3092                 if (p->score) {
3093                         fputs(" rewrite ", file);
3094                         write_name_quoted(p->two->path, file, ' ');
3095                         fprintf(file, "(%d%%)\n", similarity_index(p));
3096                 }
3097                 show_mode_change(file, p, !p->score);
3098                 break;
3099         }
3100 }
3101
3102 struct patch_id_t {
3103         git_SHA_CTX *ctx;
3104         int patchlen;
3105 };
3106
3107 static int remove_space(char *line, int len)
3108 {
3109         int i;
3110         char *dst = line;
3111         unsigned char c;
3112
3113         for (i = 0; i < len; i++)
3114                 if (!isspace((c = line[i])))
3115                         *dst++ = c;
3116
3117         return dst - line;
3118 }
3119
3120 static void patch_id_consume(void *priv, char *line, unsigned long len)
3121 {
3122         struct patch_id_t *data = priv;
3123         int new_len;
3124
3125         /* Ignore line numbers when computing the SHA1 of the patch */
3126         if (!prefixcmp(line, "@@ -"))
3127                 return;
3128
3129         new_len = remove_space(line, len);
3130
3131         git_SHA1_Update(data->ctx, line, new_len);
3132         data->patchlen += new_len;
3133 }
3134
3135 /* returns 0 upon success, and writes result into sha1 */
3136 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3137 {
3138         struct diff_queue_struct *q = &diff_queued_diff;
3139         int i;
3140         git_SHA_CTX ctx;
3141         struct patch_id_t data;
3142         char buffer[PATH_MAX * 4 + 20];
3143
3144         git_SHA1_Init(&ctx);
3145         memset(&data, 0, sizeof(struct patch_id_t));
3146         data.ctx = &ctx;
3147
3148         for (i = 0; i < q->nr; i++) {
3149                 xpparam_t xpp;
3150                 xdemitconf_t xecfg;
3151                 xdemitcb_t ecb;
3152                 mmfile_t mf1, mf2;
3153                 struct diff_filepair *p = q->queue[i];
3154                 int len1, len2;
3155
3156                 memset(&xpp, 0, sizeof(xpp));
3157                 memset(&xecfg, 0, sizeof(xecfg));
3158                 if (p->status == 0)
3159                         return error("internal diff status error");
3160                 if (p->status == DIFF_STATUS_UNKNOWN)
3161                         continue;
3162                 if (diff_unmodified_pair(p))
3163                         continue;
3164                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3165                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3166                         continue;
3167                 if (DIFF_PAIR_UNMERGED(p))
3168                         continue;
3169
3170                 diff_fill_sha1_info(p->one);
3171                 diff_fill_sha1_info(p->two);
3172                 if (fill_mmfile(&mf1, p->one) < 0 ||
3173                                 fill_mmfile(&mf2, p->two) < 0)
3174                         return error("unable to read files to diff");
3175
3176                 len1 = remove_space(p->one->path, strlen(p->one->path));
3177                 len2 = remove_space(p->two->path, strlen(p->two->path));
3178                 if (p->one->mode == 0)
3179                         len1 = snprintf(buffer, sizeof(buffer),
3180                                         "diff--gita/%.*sb/%.*s"
3181                                         "newfilemode%06o"
3182                                         "---/dev/null"
3183                                         "+++b/%.*s",
3184                                         len1, p->one->path,
3185                                         len2, p->two->path,
3186                                         p->two->mode,
3187                                         len2, p->two->path);
3188                 else if (p->two->mode == 0)
3189                         len1 = snprintf(buffer, sizeof(buffer),
3190                                         "diff--gita/%.*sb/%.*s"
3191                                         "deletedfilemode%06o"
3192                                         "---a/%.*s"
3193                                         "+++/dev/null",
3194                                         len1, p->one->path,
3195                                         len2, p->two->path,
3196                                         p->one->mode,
3197                                         len1, p->one->path);
3198                 else
3199                         len1 = snprintf(buffer, sizeof(buffer),
3200                                         "diff--gita/%.*sb/%.*s"
3201                                         "---a/%.*s"
3202                                         "+++b/%.*s",
3203                                         len1, p->one->path,
3204                                         len2, p->two->path,
3205                                         len1, p->one->path,
3206                                         len2, p->two->path);
3207                 git_SHA1_Update(&ctx, buffer, len1);
3208
3209                 xpp.flags = XDF_NEED_MINIMAL;
3210                 xecfg.ctxlen = 3;
3211                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3212                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3213                               &xpp, &xecfg, &ecb);
3214         }
3215
3216         git_SHA1_Final(sha1, &ctx);
3217         return 0;
3218 }
3219
3220 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3221 {
3222         struct diff_queue_struct *q = &diff_queued_diff;
3223         int i;
3224         int result = diff_get_patch_id(options, sha1);
3225
3226         for (i = 0; i < q->nr; i++)
3227                 diff_free_filepair(q->queue[i]);
3228
3229         free(q->queue);
3230         q->queue = NULL;
3231         q->nr = q->alloc = 0;
3232
3233         return result;
3234 }
3235
3236 static int is_summary_empty(const struct diff_queue_struct *q)
3237 {
3238         int i;
3239
3240         for (i = 0; i < q->nr; i++) {
3241                 const struct diff_filepair *p = q->queue[i];
3242
3243                 switch (p->status) {
3244                 case DIFF_STATUS_DELETED:
3245                 case DIFF_STATUS_ADDED:
3246                 case DIFF_STATUS_COPIED:
3247                 case DIFF_STATUS_RENAMED:
3248                         return 0;
3249                 default:
3250                         if (p->score)
3251                                 return 0;
3252                         if (p->one->mode && p->two->mode &&
3253                             p->one->mode != p->two->mode)
3254                                 return 0;
3255                         break;
3256                 }
3257         }
3258         return 1;
3259 }
3260
3261 void diff_flush(struct diff_options *options)
3262 {
3263         struct diff_queue_struct *q = &diff_queued_diff;
3264         int i, output_format = options->output_format;
3265         int separator = 0;
3266
3267         /*
3268          * Order: raw, stat, summary, patch
3269          * or:    name/name-status/checkdiff (other bits clear)
3270          */
3271         if (!q->nr)
3272                 goto free_queue;
3273
3274         if (output_format & (DIFF_FORMAT_RAW |
3275                              DIFF_FORMAT_NAME |
3276                              DIFF_FORMAT_NAME_STATUS |
3277                              DIFF_FORMAT_CHECKDIFF)) {
3278                 for (i = 0; i < q->nr; i++) {
3279                         struct diff_filepair *p = q->queue[i];
3280                         if (check_pair_status(p))
3281                                 flush_one_pair(p, options);
3282                 }
3283                 separator++;
3284         }
3285
3286         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3287                 struct diffstat_t diffstat;
3288
3289                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3290                 for (i = 0; i < q->nr; i++) {
3291                         struct diff_filepair *p = q->queue[i];
3292                         if (check_pair_status(p))
3293                                 diff_flush_stat(p, options, &diffstat);
3294                 }
3295                 if (output_format & DIFF_FORMAT_NUMSTAT)
3296                         show_numstat(&diffstat, options);
3297                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3298                         show_stats(&diffstat, options);
3299                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3300                         show_shortstats(&diffstat, options);
3301                 free_diffstat_info(&diffstat);
3302                 separator++;
3303         }
3304         if (output_format & DIFF_FORMAT_DIRSTAT)
3305                 show_dirstat(options);
3306
3307         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3308                 for (i = 0; i < q->nr; i++)
3309                         diff_summary(options->file, q->queue[i]);
3310                 separator++;
3311         }
3312
3313         if (output_format & DIFF_FORMAT_PATCH) {
3314                 if (separator) {
3315                         putc(options->line_termination, options->file);
3316                         if (options->stat_sep) {
3317                                 /* attach patch instead of inline */
3318                                 fputs(options->stat_sep, options->file);
3319                         }
3320                 }
3321
3322                 for (i = 0; i < q->nr; i++) {
3323                         struct diff_filepair *p = q->queue[i];
3324                         if (check_pair_status(p))
3325                                 diff_flush_patch(p, options);
3326                 }
3327         }
3328
3329         if (output_format & DIFF_FORMAT_CALLBACK)
3330                 options->format_callback(q, options, options->format_callback_data);
3331
3332         for (i = 0; i < q->nr; i++)
3333                 diff_free_filepair(q->queue[i]);
3334 free_queue:
3335         free(q->queue);
3336         q->queue = NULL;
3337         q->nr = q->alloc = 0;
3338         if (options->close_file)
3339                 fclose(options->file);
3340 }
3341
3342 static void diffcore_apply_filter(const char *filter)
3343 {
3344         int i;
3345         struct diff_queue_struct *q = &diff_queued_diff;
3346         struct diff_queue_struct outq;
3347         outq.queue = NULL;
3348         outq.nr = outq.alloc = 0;
3349
3350         if (!filter)
3351                 return;
3352
3353         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3354                 int found;
3355                 for (i = found = 0; !found && i < q->nr; i++) {
3356                         struct diff_filepair *p = q->queue[i];
3357                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3358                              ((p->score &&
3359                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3360                               (!p->score &&
3361                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3362                             ((p->status != DIFF_STATUS_MODIFIED) &&
3363                              strchr(filter, p->status)))
3364                                 found++;
3365                 }
3366                 if (found)
3367                         return;
3368
3369                 /* otherwise we will clear the whole queue
3370                  * by copying the empty outq at the end of this
3371                  * function, but first clear the current entries
3372                  * in the queue.
3373                  */
3374                 for (i = 0; i < q->nr; i++)
3375                         diff_free_filepair(q->queue[i]);
3376         }
3377         else {
3378                 /* Only the matching ones */
3379                 for (i = 0; i < q->nr; i++) {
3380                         struct diff_filepair *p = q->queue[i];
3381
3382                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3383                              ((p->score &&
3384                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3385                               (!p->score &&
3386                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3387                             ((p->status != DIFF_STATUS_MODIFIED) &&
3388                              strchr(filter, p->status)))
3389                                 diff_q(&outq, p);
3390                         else
3391                                 diff_free_filepair(p);
3392                 }
3393         }
3394         free(q->queue);
3395         *q = outq;
3396 }
3397
3398 /* Check whether two filespecs with the same mode and size are identical */
3399 static int diff_filespec_is_identical(struct diff_filespec *one,
3400                                       struct diff_filespec *two)
3401 {
3402         if (S_ISGITLINK(one->mode))
3403                 return 0;
3404         if (diff_populate_filespec(one, 0))
3405                 return 0;
3406         if (diff_populate_filespec(two, 0))
3407                 return 0;
3408         return !memcmp(one->data, two->data, one->size);
3409 }
3410
3411 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3412 {
3413         int i;
3414         struct diff_queue_struct *q = &diff_queued_diff;
3415         struct diff_queue_struct outq;
3416         outq.queue = NULL;
3417         outq.nr = outq.alloc = 0;
3418
3419         for (i = 0; i < q->nr; i++) {
3420                 struct diff_filepair *p = q->queue[i];
3421
3422                 /*
3423                  * 1. Entries that come from stat info dirtyness
3424                  *    always have both sides (iow, not create/delete),
3425                  *    one side of the object name is unknown, with
3426                  *    the same mode and size.  Keep the ones that
3427                  *    do not match these criteria.  They have real
3428                  *    differences.
3429                  *
3430                  * 2. At this point, the file is known to be modified,
3431                  *    with the same mode and size, and the object
3432                  *    name of one side is unknown.  Need to inspect
3433                  *    the identical contents.
3434                  */
3435                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3436                     !DIFF_FILE_VALID(p->two) ||
3437                     (p->one->sha1_valid && p->two->sha1_valid) ||
3438                     (p->one->mode != p->two->mode) ||
3439                     diff_populate_filespec(p->one, 1) ||
3440                     diff_populate_filespec(p->two, 1) ||
3441                     (p->one->size != p->two->size) ||
3442                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3443                         diff_q(&outq, p);
3444                 else {
3445                         /*
3446                          * The caller can subtract 1 from skip_stat_unmatch
3447                          * to determine how many paths were dirty only
3448                          * due to stat info mismatch.
3449                          */
3450                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3451                                 diffopt->skip_stat_unmatch++;
3452                         diff_free_filepair(p);
3453                 }
3454         }
3455         free(q->queue);
3456         *q = outq;
3457 }
3458
3459 void diffcore_std(struct diff_options *options)
3460 {
3461         if (options->skip_stat_unmatch)
3462                 diffcore_skip_stat_unmatch(options);
3463         if (options->break_opt != -1)
3464                 diffcore_break(options->break_opt);
3465         if (options->detect_rename)
3466                 diffcore_rename(options);
3467         if (options->break_opt != -1)
3468                 diffcore_merge_broken();
3469         if (options->pickaxe)
3470                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3471         if (options->orderfile)
3472                 diffcore_order(options->orderfile);
3473         diff_resolve_rename_copy();
3474         diffcore_apply_filter(options->filter);
3475
3476         if (diff_queued_diff.nr)
3477                 DIFF_OPT_SET(options, HAS_CHANGES);
3478         else
3479                 DIFF_OPT_CLR(options, HAS_CHANGES);
3480 }
3481
3482 int diff_result_code(struct diff_options *opt, int status)
3483 {
3484         int result = 0;
3485         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3486             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3487                 return status;
3488         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3489             DIFF_OPT_TST(opt, HAS_CHANGES))
3490                 result |= 01;
3491         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3492             DIFF_OPT_TST(opt, CHECK_FAILED))
3493                 result |= 02;
3494         return result;
3495 }
3496
3497 void diff_addremove(struct diff_options *options,
3498                     int addremove, unsigned mode,
3499                     const unsigned char *sha1,
3500                     const char *concatpath)
3501 {
3502         struct diff_filespec *one, *two;
3503
3504         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3505                 return;
3506
3507         /* This may look odd, but it is a preparation for
3508          * feeding "there are unchanged files which should
3509          * not produce diffs, but when you are doing copy
3510          * detection you would need them, so here they are"
3511          * entries to the diff-core.  They will be prefixed
3512          * with something like '=' or '*' (I haven't decided
3513          * which but should not make any difference).
3514          * Feeding the same new and old to diff_change()
3515          * also has the same effect.
3516          * Before the final output happens, they are pruned after
3517          * merged into rename/copy pairs as appropriate.
3518          */
3519         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3520                 addremove = (addremove == '+' ? '-' :
3521                              addremove == '-' ? '+' : addremove);
3522
3523         if (options->prefix &&
3524             strncmp(concatpath, options->prefix, options->prefix_length))
3525                 return;
3526
3527         one = alloc_filespec(concatpath);
3528         two = alloc_filespec(concatpath);
3529
3530         if (addremove != '+')
3531                 fill_filespec(one, sha1, mode);
3532         if (addremove != '-')
3533                 fill_filespec(two, sha1, mode);
3534
3535         diff_queue(&diff_queued_diff, one, two);
3536         DIFF_OPT_SET(options, HAS_CHANGES);
3537 }
3538
3539 void diff_change(struct diff_options *options,
3540                  unsigned old_mode, unsigned new_mode,
3541                  const unsigned char *old_sha1,
3542                  const unsigned char *new_sha1,
3543                  const char *concatpath)
3544 {
3545         struct diff_filespec *one, *two;
3546
3547         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3548                         && S_ISGITLINK(new_mode))
3549                 return;
3550
3551         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3552                 unsigned tmp;
3553                 const unsigned char *tmp_c;
3554                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3555                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3556         }
3557
3558         if (options->prefix &&
3559             strncmp(concatpath, options->prefix, options->prefix_length))
3560                 return;
3561
3562         one = alloc_filespec(concatpath);
3563         two = alloc_filespec(concatpath);
3564         fill_filespec(one, old_sha1, old_mode);
3565         fill_filespec(two, new_sha1, new_mode);
3566
3567         diff_queue(&diff_queued_diff, one, two);
3568         DIFF_OPT_SET(options, HAS_CHANGES);
3569 }
3570
3571 void diff_unmerge(struct diff_options *options,
3572                   const char *path,
3573                   unsigned mode, const unsigned char *sha1)
3574 {
3575         struct diff_filespec *one, *two;
3576
3577         if (options->prefix &&
3578             strncmp(path, options->prefix, options->prefix_length))
3579                 return;
3580
3581         one = alloc_filespec(path);
3582         two = alloc_filespec(path);
3583         fill_filespec(one, sha1, mode);
3584         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3585 }