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

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "quote.h"
   6 #include "diff.h"
   7 #include "diffcore.h"
   8 #include "delta.h"
   9 #include "xdiff-interface.h"
  10 #include "color.h"
  11 #include "attr.h"
  12
  13 #ifdef NO_FAST_WORKING_DIRECTORY
  14 #define FAST_WORKING_DIRECTORY 0
  15 #else
  16 #define FAST_WORKING_DIRECTORY 1
  17 #endif
  18
  19 static int diff_detect_rename_default;
  20 static int diff_rename_limit_default = -1;
  21 static int diff_use_color_default;
  22
  23 static char diff_colors[][COLOR_MAXLEN] = {
  24         "\033[m",       /* reset */
  25         "",             /* PLAIN (normal) */
  26         "\033[1m",      /* METAINFO (bold) */
  27         "\033[36m",     /* FRAGINFO (cyan) */
  28         "\033[31m",     /* OLD (red) */
  29         "\033[32m",     /* NEW (green) */
  30         "\033[33m",     /* COMMIT (yellow) */
  31         "\033[41m",     /* WHITESPACE (red background) */
  32 };
  33
  34 static int parse_diff_color_slot(const char *var, int ofs)
  35 {
  36         if (!strcasecmp(var+ofs, "plain"))
  37                 return DIFF_PLAIN;
  38         if (!strcasecmp(var+ofs, "meta"))
  39                 return DIFF_METAINFO;
  40         if (!strcasecmp(var+ofs, "frag"))
  41                 return DIFF_FRAGINFO;
  42         if (!strcasecmp(var+ofs, "old"))
  43                 return DIFF_FILE_OLD;
  44         if (!strcasecmp(var+ofs, "new"))
  45                 return DIFF_FILE_NEW;
  46         if (!strcasecmp(var+ofs, "commit"))
  47                 return DIFF_COMMIT;
  48         if (!strcasecmp(var+ofs, "whitespace"))
  49                 return DIFF_WHITESPACE;
  50         die("bad config variable '%s'", var);
  51 }
  52
  53 static struct ll_diff_driver {
  54         const char *name;
  55         struct ll_diff_driver *next;
  56         char *cmd;
  57 } *user_diff, **user_diff_tail;
  58
  59 static void read_config_if_needed(void)
  60 {
  61         if (!user_diff_tail) {
  62                 user_diff_tail = &user_diff;
  63                 git_config(git_diff_ui_config);
  64         }
  65 }
  66
  67 /*
  68  * Currently there is only "diff.<drivername>.command" variable;
  69  * because there are "diff.color.<slot>" variables, we are parsing
  70  * this in a bit convoluted way to allow low level diff driver
  71  * called "color".
  72  */
  73 static int parse_lldiff_command(const char *var, const char *ep, const char *value)
  74 {
  75         const char *name;
  76         int namelen;
  77         struct ll_diff_driver *drv;
  78
  79         name = var + 5;
  80         namelen = ep - name;
  81         for (drv = user_diff; drv; drv = drv->next)
  82                 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
  83                         break;
  84         if (!drv) {
  85                 char *namebuf;
  86                 drv = xcalloc(1, sizeof(struct ll_diff_driver));
  87                 namebuf = xmalloc(namelen + 1);
  88                 memcpy(namebuf, name, namelen);
  89                 namebuf[namelen] = 0;
  90                 drv->name = namebuf;
  91                 drv->next = NULL;
  92                 if (!user_diff_tail)
  93                         user_diff_tail = &user_diff;
  94                 *user_diff_tail = drv;
  95                 user_diff_tail = &(drv->next);
  96         }
  97
  98         if (!value)
  99                 return error("%s: lacks value", var);
 100         drv->cmd = strdup(value);
 101         return 0;
 102 }
 103
 104 /*
 105  * 'diff.<what>.funcname' attribute can be specified in the configuration
 106  * to define a customized regexp to find the beginning of a function to
 107  * be used for hunk header lines of "diff -p" style output.
 108  */
 109 static struct funcname_pattern {
 110         char *name;
 111         char *pattern;
 112         struct funcname_pattern *next;
 113 } *funcname_pattern_list;
 114
 115 static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
 116 {
 117         const char *name;
 118         int namelen;
 119         struct funcname_pattern *pp;
 120
 121         name = var + 5; /* "diff." */
 122         namelen = ep - name;
 123
 124         for (pp = funcname_pattern_list; pp; pp = pp->next)
 125                 if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
 126                         break;
 127         if (!pp) {
 128                 char *namebuf;
 129                 pp = xcalloc(1, sizeof(*pp));
 130                 namebuf = xmalloc(namelen + 1);
 131                 memcpy(namebuf, name, namelen);
 132                 namebuf[namelen] = 0;
 133                 pp->name = namebuf;
 134                 pp->next = funcname_pattern_list;
 135                 funcname_pattern_list = pp;
 136         }
 137         if (pp->pattern)
 138                 free(pp->pattern);
 139         pp->pattern = xstrdup(value);
 140         return 0;
 141 }
 142
 143 /*
 144  * These are to give UI layer defaults.
 145  * The core-level commands such as git-diff-files should
 146  * never be affected by the setting of diff.renames
 147  * the user happens to have in the configuration file.
 148  */
 149 int git_diff_ui_config(const char *var, const char *value)
 150 {
 151         if (!strcmp(var, "diff.renamelimit")) {
 152                 diff_rename_limit_default = git_config_int(var, value);
 153                 return 0;
 154         }
 155         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 156                 diff_use_color_default = git_config_colorbool(var, value);
 157                 return 0;
 158         }
 159         if (!strcmp(var, "diff.renames")) {
 160                 if (!value)
 161                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 162                 else if (!strcasecmp(value, "copies") ||
 163                          !strcasecmp(value, "copy"))
 164                         diff_detect_rename_default = DIFF_DETECT_COPY;
 165                 else if (git_config_bool(var,value))
 166                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 167                 return 0;
 168         }
 169         if (!prefixcmp(var, "diff.")) {
 170                 const char *ep = strrchr(var, '.');
 171
 172                 if (ep != var + 4) {
 173                         if (!strcmp(ep, ".command"))
 174                                 return parse_lldiff_command(var, ep, value);
 175                         if (!strcmp(ep, ".funcname"))
 176                                 return parse_funcname_pattern(var, ep, value);
 177                 }
 178         }
 179         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 180                 int slot = parse_diff_color_slot(var, 11);
 181                 color_parse(value, var, diff_colors[slot]);
 182                 return 0;
 183         }
 184
 185         return git_default_config(var, value);
 186 }
 187
 188 static char *quote_one(const char *str)
 189 {
 190         int needlen;
 191         char *xp;
 192
 193         if (!str)
 194                 return NULL;
 195         needlen = quote_c_style(str, NULL, NULL, 0);
 196         if (!needlen)
 197                 return xstrdup(str);
 198         xp = xmalloc(needlen + 1);
 199         quote_c_style(str, xp, NULL, 0);
 200         return xp;
 201 }
 202
 203 static char *quote_two(const char *one, const char *two)
 204 {
 205         int need_one = quote_c_style(one, NULL, NULL, 1);
 206         int need_two = quote_c_style(two, NULL, NULL, 1);
 207         char *xp;
 208
 209         if (need_one + need_two) {
 210                 if (!need_one) need_one = strlen(one);
 211                 if (!need_two) need_one = strlen(two);
 212
 213                 xp = xmalloc(need_one + need_two + 3);
 214                 xp[0] = '"';
 215                 quote_c_style(one, xp + 1, NULL, 1);
 216                 quote_c_style(two, xp + need_one + 1, NULL, 1);
 217                 strcpy(xp + need_one + need_two + 1, "\"");
 218                 return xp;
 219         }
 220         need_one = strlen(one);
 221         need_two = strlen(two);
 222         xp = xmalloc(need_one + need_two + 1);
 223         strcpy(xp, one);
 224         strcpy(xp + need_one, two);
 225         return xp;
 226 }
 227
 228 static const char *external_diff(void)
 229 {
 230         static const char *external_diff_cmd = NULL;
 231         static int done_preparing = 0;
 232
 233         if (done_preparing)
 234                 return external_diff_cmd;
 235         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 236         done_preparing = 1;
 237         return external_diff_cmd;
 238 }
 239
 240 static struct diff_tempfile {
 241         const char *name; /* filename external diff should read from */
 242         char hex[41];
 243         char mode[10];
 244         char tmp_path[PATH_MAX];
 245 } diff_temp[2];
 246
 247 static int count_lines(const char *data, int size)
 248 {
 249         int count, ch, completely_empty = 1, nl_just_seen = 0;
 250         count = 0;
 251         while (0 < size--) {
 252                 ch = *data++;
 253                 if (ch == '\n') {
 254                         count++;
 255                         nl_just_seen = 1;
 256                         completely_empty = 0;
 257                 }
 258                 else {
 259                         nl_just_seen = 0;
 260                         completely_empty = 0;
 261                 }
 262         }
 263         if (completely_empty)
 264                 return 0;
 265         if (!nl_just_seen)
 266                 count++; /* no trailing newline */
 267         return count;
 268 }
 269
 270 static void print_line_count(int count)
 271 {
 272         switch (count) {
 273         case 0:
 274                 printf("0,0");
 275                 break;
 276         case 1:
 277                 printf("1");
 278                 break;
 279         default:
 280                 printf("1,%d", count);
 281                 break;
 282         }
 283 }
 284
 285 static void copy_file(int prefix, const char *data, int size,
 286                 const char *set, const char *reset)
 287 {
 288         int ch, nl_just_seen = 1;
 289         while (0 < size--) {
 290                 ch = *data++;
 291                 if (nl_just_seen) {
 292                         fputs(set, stdout);
 293                         putchar(prefix);
 294                 }
 295                 if (ch == '\n') {
 296                         nl_just_seen = 1;
 297                         fputs(reset, stdout);
 298                 } else
 299                         nl_just_seen = 0;
 300                 putchar(ch);
 301         }
 302         if (!nl_just_seen)
 303                 printf("%s\n\\ No newline at end of file\n", reset);
 304 }
 305
 306 static void emit_rewrite_diff(const char *name_a,
 307                               const char *name_b,
 308                               struct diff_filespec *one,
 309                               struct diff_filespec *two,
 310                               int color_diff)
 311 {
 312         int lc_a, lc_b;
 313         const char *name_a_tab, *name_b_tab;
 314         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 315         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 316         const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
 317         const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
 318         const char *reset = diff_get_color(color_diff, DIFF_RESET);
 319
 320         name_a += (*name_a == '/');
 321         name_b += (*name_b == '/');
 322         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 323         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 324
 325         diff_populate_filespec(one, 0);
 326         diff_populate_filespec(two, 0);
 327         lc_a = count_lines(one->data, one->size);
 328         lc_b = count_lines(two->data, two->size);
 329         printf("%s--- a/%s%s%s\n%s+++ b/%s%s%s\n%s@@ -",
 330                metainfo, name_a, name_a_tab, reset,
 331                metainfo, name_b, name_b_tab, reset, fraginfo);
 332         print_line_count(lc_a);
 333         printf(" +");
 334         print_line_count(lc_b);
 335         printf(" @@%s\n", reset);
 336         if (lc_a)
 337                 copy_file('-', one->data, one->size, old, reset);
 338         if (lc_b)
 339                 copy_file('+', two->data, two->size, new, reset);
 340 }
 341
 342 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 343 {
 344         if (!DIFF_FILE_VALID(one)) {
 345                 mf->ptr = (char *)""; /* does not matter */
 346                 mf->size = 0;
 347                 return 0;
 348         }
 349         else if (diff_populate_filespec(one, 0))
 350                 return -1;
 351         mf->ptr = one->data;
 352         mf->size = one->size;
 353         return 0;
 354 }
 355
 356 struct diff_words_buffer {
 357         mmfile_t text;
 358         long alloc;
 359         long current; /* output pointer */
 360         int suppressed_newline;
 361 };
 362
 363 static void diff_words_append(char *line, unsigned long len,
 364                 struct diff_words_buffer *buffer)
 365 {
 366         if (buffer->text.size + len > buffer->alloc) {
 367                 buffer->alloc = (buffer->text.size + len) * 3 / 2;
 368                 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
 369         }
 370         line++;
 371         len--;
 372         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 373         buffer->text.size += len;
 374 }
 375
 376 struct diff_words_data {
 377         struct xdiff_emit_state xm;
 378         struct diff_words_buffer minus, plus;
 379 };
 380
 381 static void print_word(struct diff_words_buffer *buffer, int len, int color,
 382                 int suppress_newline)
 383 {
 384         const char *ptr;
 385         int eol = 0;
 386
 387         if (len == 0)
 388                 return;
 389
 390         ptr  = buffer->text.ptr + buffer->current;
 391         buffer->current += len;
 392
 393         if (ptr[len - 1] == '\n') {
 394                 eol = 1;
 395                 len--;
 396         }
 397
 398         fputs(diff_get_color(1, color), stdout);
 399         fwrite(ptr, len, 1, stdout);
 400         fputs(diff_get_color(1, DIFF_RESET), stdout);
 401
 402         if (eol) {
 403                 if (suppress_newline)
 404                         buffer->suppressed_newline = 1;
 405                 else
 406                         putchar('\n');
 407         }
 408 }
 409
 410 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 411 {
 412         struct diff_words_data *diff_words = priv;
 413
 414         if (diff_words->minus.suppressed_newline) {
 415                 if (line[0] != '+')
 416                         putchar('\n');
 417                 diff_words->minus.suppressed_newline = 0;
 418         }
 419
 420         len--;
 421         switch (line[0]) {
 422                 case '-':
 423                         print_word(&diff_words->minus, len, DIFF_FILE_OLD, 1);
 424                         break;
 425                 case '+':
 426                         print_word(&diff_words->plus, len, DIFF_FILE_NEW, 0);
 427                         break;
 428                 case ' ':
 429                         print_word(&diff_words->plus, len, DIFF_PLAIN, 0);
 430                         diff_words->minus.current += len;
 431                         break;
 432         }
 433 }
 434
 435 /* this executes the word diff on the accumulated buffers */
 436 static void diff_words_show(struct diff_words_data *diff_words)
 437 {
 438         xpparam_t xpp;
 439         xdemitconf_t xecfg;
 440         xdemitcb_t ecb;
 441         mmfile_t minus, plus;
 442         int i;
 443
 444         memset(&xecfg, 0, sizeof(xecfg));
 445         minus.size = diff_words->minus.text.size;
 446         minus.ptr = xmalloc(minus.size);
 447         memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
 448         for (i = 0; i < minus.size; i++)
 449                 if (isspace(minus.ptr[i]))
 450                         minus.ptr[i] = '\n';
 451         diff_words->minus.current = 0;
 452
 453         plus.size = diff_words->plus.text.size;
 454         plus.ptr = xmalloc(plus.size);
 455         memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
 456         for (i = 0; i < plus.size; i++)
 457                 if (isspace(plus.ptr[i]))
 458                         plus.ptr[i] = '\n';
 459         diff_words->plus.current = 0;
 460
 461         xpp.flags = XDF_NEED_MINIMAL;
 462         xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
 463         ecb.outf = xdiff_outf;
 464         ecb.priv = diff_words;
 465         diff_words->xm.consume = fn_out_diff_words_aux;
 466         xdl_diff(&minus, &plus, &xpp, &xecfg, &ecb);
 467
 468         free(minus.ptr);
 469         free(plus.ptr);
 470         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 471
 472         if (diff_words->minus.suppressed_newline) {
 473                 putchar('\n');
 474                 diff_words->minus.suppressed_newline = 0;
 475         }
 476 }
 477
 478 struct emit_callback {
 479         struct xdiff_emit_state xm;
 480         int nparents, color_diff;
 481         const char **label_path;
 482         struct diff_words_data *diff_words;
 483         int *found_changesp;
 484 };
 485
 486 static void free_diff_words_data(struct emit_callback *ecbdata)
 487 {
 488         if (ecbdata->diff_words) {
 489                 /* flush buffers */
 490                 if (ecbdata->diff_words->minus.text.size ||
 491                                 ecbdata->diff_words->plus.text.size)
 492                         diff_words_show(ecbdata->diff_words);
 493
 494                 if (ecbdata->diff_words->minus.text.ptr)
 495                         free (ecbdata->diff_words->minus.text.ptr);
 496                 if (ecbdata->diff_words->plus.text.ptr)
 497                         free (ecbdata->diff_words->plus.text.ptr);
 498                 free(ecbdata->diff_words);
 499                 ecbdata->diff_words = NULL;
 500         }
 501 }
 502
 503 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 504 {
 505         if (diff_use_color)
 506                 return diff_colors[ix];
 507         return "";
 508 }
 509
 510 static void emit_line(const char *set, const char *reset, const char *line, int len)
 511 {
 512         if (len > 0 && line[len-1] == '\n')
 513                 len--;
 514         fputs(set, stdout);
 515         fwrite(line, len, 1, stdout);
 516         puts(reset);
 517 }
 518
 519 static void emit_line_with_ws(int nparents,
 520                 const char *set, const char *reset, const char *ws,
 521                 const char *line, int len)
 522 {
 523         int col0 = nparents;
 524         int last_tab_in_indent = -1;
 525         int last_space_in_indent = -1;
 526         int i;
 527         int tail = len;
 528         int need_highlight_leading_space = 0;
 529         /* The line is a newly added line.  Does it have funny leading
 530          * whitespaces?  In indent, SP should never precede a TAB.
 531          */
 532         for (i = col0; i < len; i++) {
 533                 if (line[i] == '\t') {
 534                         last_tab_in_indent = i;
 535                         if (0 <= last_space_in_indent)
 536                                 need_highlight_leading_space = 1;
 537                 }
 538                 else if (line[i] == ' ')
 539                         last_space_in_indent = i;
 540                 else
 541                         break;
 542         }
 543         fputs(set, stdout);
 544         fwrite(line, col0, 1, stdout);
 545         fputs(reset, stdout);
 546         if (((i == len) || line[i] == '\n') && i != col0) {
 547                 /* The whole line was indent */
 548                 emit_line(ws, reset, line + col0, len - col0);
 549                 return;
 550         }
 551         i = col0;
 552         if (need_highlight_leading_space) {
 553                 while (i < last_tab_in_indent) {
 554                         if (line[i] == ' ') {
 555                                 fputs(ws, stdout);
 556                                 putchar(' ');
 557                                 fputs(reset, stdout);
 558                         }
 559                         else
 560                                 putchar(line[i]);
 561                         i++;
 562                 }
 563         }
 564         tail = len - 1;
 565         if (line[tail] == '\n' && i < tail)
 566                 tail--;
 567         while (i < tail) {
 568                 if (!isspace(line[tail]))
 569                         break;
 570                 tail--;
 571         }
 572         if ((i < tail && line[tail + 1] != '\n')) {
 573                 /* This has whitespace between tail+1..len */
 574                 fputs(set, stdout);
 575                 fwrite(line + i, tail - i + 1, 1, stdout);
 576                 fputs(reset, stdout);
 577                 emit_line(ws, reset, line + tail + 1, len - tail - 1);
 578         }
 579         else
 580                 emit_line(set, reset, line + i, len - i);
 581 }
 582
 583 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
 584 {
 585         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 586         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 587
 588         if (!*ws)
 589                 emit_line(set, reset, line, len);
 590         else
 591                 emit_line_with_ws(ecbdata->nparents, set, reset, ws,
 592                                 line, len);
 593 }
 594
 595 static void fn_out_consume(void *priv, char *line, unsigned long len)
 596 {
 597         int i;
 598         int color;
 599         struct emit_callback *ecbdata = priv;
 600         const char *set = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 601         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 602
 603         *(ecbdata->found_changesp) = 1;
 604
 605         if (ecbdata->label_path[0]) {
 606                 const char *name_a_tab, *name_b_tab;
 607
 608                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 609                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 610
 611                 printf("%s--- %s%s%s\n",
 612                        set, ecbdata->label_path[0], reset, name_a_tab);
 613                 printf("%s+++ %s%s%s\n",
 614                        set, ecbdata->label_path[1], reset, name_b_tab);
 615                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 616         }
 617
 618         /* This is not really necessary for now because
 619          * this codepath only deals with two-way diffs.
 620          */
 621         for (i = 0; i < len && line[i] == '@'; i++)
 622                 ;
 623         if (2 <= i && i < len && line[i] == ' ') {
 624                 ecbdata->nparents = i - 1;
 625                 emit_line(diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
 626                           reset, line, len);
 627                 return;
 628         }
 629
 630         if (len < ecbdata->nparents) {
 631                 set = reset;
 632                 emit_line(reset, reset, line, len);
 633                 return;
 634         }
 635
 636         color = DIFF_PLAIN;
 637         if (ecbdata->diff_words && ecbdata->nparents != 1)
 638                 /* fall back to normal diff */
 639                 free_diff_words_data(ecbdata);
 640         if (ecbdata->diff_words) {
 641                 if (line[0] == '-') {
 642                         diff_words_append(line, len,
 643                                           &ecbdata->diff_words->minus);
 644                         return;
 645                 } else if (line[0] == '+') {
 646                         diff_words_append(line, len,
 647                                           &ecbdata->diff_words->plus);
 648                         return;
 649                 }
 650                 if (ecbdata->diff_words->minus.text.size ||
 651                     ecbdata->diff_words->plus.text.size)
 652                         diff_words_show(ecbdata->diff_words);
 653                 line++;
 654                 len--;
 655                 emit_line(set, reset, line, len);
 656                 return;
 657         }
 658         for (i = 0; i < ecbdata->nparents && len; i++) {
 659                 if (line[i] == '-')
 660                         color = DIFF_FILE_OLD;
 661                 else if (line[i] == '+')
 662                         color = DIFF_FILE_NEW;
 663         }
 664
 665         if (color != DIFF_FILE_NEW) {
 666                 emit_line(diff_get_color(ecbdata->color_diff, color),
 667                           reset, line, len);
 668                 return;
 669         }
 670         emit_add_line(reset, ecbdata, line, len);
 671 }
 672
 673 static char *pprint_rename(const char *a, const char *b)
 674 {
 675         const char *old = a;
 676         const char *new = b;
 677         char *name = NULL;
 678         int pfx_length, sfx_length;
 679         int len_a = strlen(a);
 680         int len_b = strlen(b);
 681         int qlen_a = quote_c_style(a, NULL, NULL, 0);
 682         int qlen_b = quote_c_style(b, NULL, NULL, 0);
 683
 684         if (qlen_a || qlen_b) {
 685                 if (qlen_a) len_a = qlen_a;
 686                 if (qlen_b) len_b = qlen_b;
 687                 name = xmalloc( len_a + len_b + 5 );
 688                 if (qlen_a)
 689                         quote_c_style(a, name, NULL, 0);
 690                 else
 691                         memcpy(name, a, len_a);
 692                 memcpy(name + len_a, " => ", 4);
 693                 if (qlen_b)
 694                         quote_c_style(b, name + len_a + 4, NULL, 0);
 695                 else
 696                         memcpy(name + len_a + 4, b, len_b + 1);
 697                 return name;
 698         }
 699
 700         /* Find common prefix */
 701         pfx_length = 0;
 702         while (*old && *new && *old == *new) {
 703                 if (*old == '/')
 704                         pfx_length = old - a + 1;
 705                 old++;
 706                 new++;
 707         }
 708
 709         /* Find common suffix */
 710         old = a + len_a;
 711         new = b + len_b;
 712         sfx_length = 0;
 713         while (a <= old && b <= new && *old == *new) {
 714                 if (*old == '/')
 715                         sfx_length = len_a - (old - a);
 716                 old--;
 717                 new--;
 718         }
 719
 720         /*
 721          * pfx{mid-a => mid-b}sfx
 722          * {pfx-a => pfx-b}sfx
 723          * pfx{sfx-a => sfx-b}
 724          * name-a => name-b
 725          */
 726         if (pfx_length + sfx_length) {
 727                 int a_midlen = len_a - pfx_length - sfx_length;
 728                 int b_midlen = len_b - pfx_length - sfx_length;
 729                 if (a_midlen < 0) a_midlen = 0;
 730                 if (b_midlen < 0) b_midlen = 0;
 731
 732                 name = xmalloc(pfx_length + a_midlen + b_midlen + sfx_length + 7);
 733                 sprintf(name, "%.*s{%.*s => %.*s}%s",
 734                         pfx_length, a,
 735                         a_midlen, a + pfx_length,
 736                         b_midlen, b + pfx_length,
 737                         a + len_a - sfx_length);
 738         }
 739         else {
 740                 name = xmalloc(len_a + len_b + 5);
 741                 sprintf(name, "%s => %s", a, b);
 742         }
 743         return name;
 744 }
 745
 746 struct diffstat_t {
 747         struct xdiff_emit_state xm;
 748
 749         int nr;
 750         int alloc;
 751         struct diffstat_file {
 752                 char *name;
 753                 unsigned is_unmerged:1;
 754                 unsigned is_binary:1;
 755                 unsigned is_renamed:1;
 756                 unsigned int added, deleted;
 757         } **files;
 758 };
 759
 760 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 761                                           const char *name_a,
 762                                           const char *name_b)
 763 {
 764         struct diffstat_file *x;
 765         x = xcalloc(sizeof (*x), 1);
 766         if (diffstat->nr == diffstat->alloc) {
 767                 diffstat->alloc = alloc_nr(diffstat->alloc);
 768                 diffstat->files = xrealloc(diffstat->files,
 769                                 diffstat->alloc * sizeof(x));
 770         }
 771         diffstat->files[diffstat->nr++] = x;
 772         if (name_b) {
 773                 x->name = pprint_rename(name_a, name_b);
 774                 x->is_renamed = 1;
 775         }
 776         else
 777                 x->name = xstrdup(name_a);
 778         return x;
 779 }
 780
 781 static void diffstat_consume(void *priv, char *line, unsigned long len)
 782 {
 783         struct diffstat_t *diffstat = priv;
 784         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 785
 786         if (line[0] == '+')
 787                 x->added++;
 788         else if (line[0] == '-')
 789                 x->deleted++;
 790 }
 791
 792 const char mime_boundary_leader[] = "------------";
 793
 794 static int scale_linear(int it, int width, int max_change)
 795 {
 796         /*
 797          * make sure that at least one '-' is printed if there were deletions,
 798          * and likewise for '+'.
 799          */
 800         if (max_change < 2)
 801                 return it;
 802         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
 803 }
 804
 805 static void show_name(const char *prefix, const char *name, int len,
 806                       const char *reset, const char *set)
 807 {
 808         printf(" %s%s%-*s%s |", set, prefix, len, name, reset);
 809 }
 810
 811 static void show_graph(char ch, int cnt, const char *set, const char *reset)
 812 {
 813         if (cnt <= 0)
 814                 return;
 815         printf("%s", set);
 816         while (cnt--)
 817                 putchar(ch);
 818         printf("%s", reset);
 819 }
 820
 821 static void show_stats(struct diffstat_t* data, struct diff_options *options)
 822 {
 823         int i, len, add, del, total, adds = 0, dels = 0;
 824         int max_change = 0, max_len = 0;
 825         int total_files = data->nr;
 826         int width, name_width;
 827         const char *reset, *set, *add_c, *del_c;
 828
 829         if (data->nr == 0)
 830                 return;
 831
 832         width = options->stat_width ? options->stat_width : 80;
 833         name_width = options->stat_name_width ? options->stat_name_width : 50;
 834
 835         /* Sanity: give at least 5 columns to the graph,
 836          * but leave at least 10 columns for the name.
 837          */
 838         if (width < name_width + 15) {
 839                 if (name_width <= 25)
 840                         width = name_width + 15;
 841                 else
 842                         name_width = width - 15;
 843         }
 844
 845         /* Find the longest filename and max number of changes */
 846         reset = diff_get_color(options->color_diff, DIFF_RESET);
 847         set = diff_get_color(options->color_diff, DIFF_PLAIN);
 848         add_c = diff_get_color(options->color_diff, DIFF_FILE_NEW);
 849         del_c = diff_get_color(options->color_diff, DIFF_FILE_OLD);
 850
 851         for (i = 0; i < data->nr; i++) {
 852                 struct diffstat_file *file = data->files[i];
 853                 int change = file->added + file->deleted;
 854
 855                 if (!file->is_renamed) {  /* renames are already quoted by pprint_rename */
 856                         len = quote_c_style(file->name, NULL, NULL, 0);
 857                         if (len) {
 858                                 char *qname = xmalloc(len + 1);
 859                                 quote_c_style(file->name, qname, NULL, 0);
 860                                 free(file->name);
 861                                 file->name = qname;
 862                         }
 863                 }
 864
 865                 len = strlen(file->name);
 866                 if (max_len < len)
 867                         max_len = len;
 868
 869                 if (file->is_binary || file->is_unmerged)
 870                         continue;
 871                 if (max_change < change)
 872                         max_change = change;
 873         }
 874
 875         /* Compute the width of the graph part;
 876          * 10 is for one blank at the beginning of the line plus
 877          * " | count " between the name and the graph.
 878          *
 879          * From here on, name_width is the width of the name area,
 880          * and width is the width of the graph area.
 881          */
 882         name_width = (name_width < max_len) ? name_width : max_len;
 883         if (width < (name_width + 10) + max_change)
 884                 width = width - (name_width + 10);
 885         else
 886                 width = max_change;
 887
 888         for (i = 0; i < data->nr; i++) {
 889                 const char *prefix = "";
 890                 char *name = data->files[i]->name;
 891                 int added = data->files[i]->added;
 892                 int deleted = data->files[i]->deleted;
 893                 int name_len;
 894
 895                 /*
 896                  * "scale" the filename
 897                  */
 898                 len = name_width;
 899                 name_len = strlen(name);
 900                 if (name_width < name_len) {
 901                         char *slash;
 902                         prefix = "...";
 903                         len -= 3;
 904                         name += name_len - len;
 905                         slash = strchr(name, '/');
 906                         if (slash)
 907                                 name = slash;
 908                 }
 909
 910                 if (data->files[i]->is_binary) {
 911                         show_name(prefix, name, len, reset, set);
 912                         printf("  Bin ");
 913                         printf("%s%d%s", del_c, deleted, reset);
 914                         printf(" -> ");
 915                         printf("%s%d%s", add_c, added, reset);
 916                         printf(" bytes");
 917                         printf("\n");
 918                         goto free_diffstat_file;
 919                 }
 920                 else if (data->files[i]->is_unmerged) {
 921                         show_name(prefix, name, len, reset, set);
 922                         printf("  Unmerged\n");
 923                         goto free_diffstat_file;
 924                 }
 925                 else if (!data->files[i]->is_renamed &&
 926                          (added + deleted == 0)) {
 927                         total_files--;
 928                         goto free_diffstat_file;
 929                 }
 930
 931                 /*
 932                  * scale the add/delete
 933                  */
 934                 add = added;
 935                 del = deleted;
 936                 total = add + del;
 937                 adds += add;
 938                 dels += del;
 939
 940                 if (width <= max_change) {
 941                         add = scale_linear(add, width, max_change);
 942                         del = scale_linear(del, width, max_change);
 943                         total = add + del;
 944                 }
 945                 show_name(prefix, name, len, reset, set);
 946                 printf("%5d ", added + deleted);
 947                 show_graph('+', add, add_c, reset);
 948                 show_graph('-', del, del_c, reset);
 949                 putchar('\n');
 950         free_diffstat_file:
 951                 free(data->files[i]->name);
 952                 free(data->files[i]);
 953         }
 954         free(data->files);
 955         printf("%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
 956                set, total_files, adds, dels, reset);
 957 }
 958
 959 static void show_shortstats(struct diffstat_t* data)
 960 {
 961         int i, adds = 0, dels = 0, total_files = data->nr;
 962
 963         if (data->nr == 0)
 964                 return;
 965
 966         for (i = 0; i < data->nr; i++) {
 967                 if (!data->files[i]->is_binary &&
 968                     !data->files[i]->is_unmerged) {
 969                         int added = data->files[i]->added;
 970                         int deleted= data->files[i]->deleted;
 971                         if (!data->files[i]->is_renamed &&
 972                             (added + deleted == 0)) {
 973                                 total_files--;
 974                         } else {
 975                                 adds += added;
 976                                 dels += deleted;
 977                         }
 978                 }
 979                 free(data->files[i]->name);
 980                 free(data->files[i]);
 981         }
 982         free(data->files);
 983
 984         printf(" %d files changed, %d insertions(+), %d deletions(-)\n",
 985                total_files, adds, dels);
 986 }
 987
 988 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
 989 {
 990         int i;
 991
 992         for (i = 0; i < data->nr; i++) {
 993                 struct diffstat_file *file = data->files[i];
 994
 995                 if (file->is_binary)
 996                         printf("-\t-\t");
 997                 else
 998                         printf("%d\t%d\t", file->added, file->deleted);
 999                 if (options->line_termination && !file->is_renamed &&
1000                     quote_c_style(file->name, NULL, NULL, 0))
1001                         quote_c_style(file->name, NULL, stdout, 0);
1002                 else
1003                         fputs(file->name, stdout);
1004                 putchar(options->line_termination);
1005         }
1006 }
1007
1008 struct checkdiff_t {
1009         struct xdiff_emit_state xm;
1010         const char *filename;
1011         int lineno, color_diff;
1012 };
1013
1014 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1015 {
1016         struct checkdiff_t *data = priv;
1017         const char *ws = diff_get_color(data->color_diff, DIFF_WHITESPACE);
1018         const char *reset = diff_get_color(data->color_diff, DIFF_RESET);
1019         const char *set = diff_get_color(data->color_diff, DIFF_FILE_NEW);
1020
1021         if (line[0] == '+') {
1022                 int i, spaces = 0, space_before_tab = 0, white_space_at_end = 0;
1023
1024                 /* check space before tab */
1025                 for (i = 1; i < len && (line[i] == ' ' || line[i] == '\t'); i++)
1026                         if (line[i] == ' ')
1027                                 spaces++;
1028                 if (line[i - 1] == '\t' && spaces)
1029                         space_before_tab = 1;
1030
1031                 /* check white space at line end */
1032                 if (line[len - 1] == '\n')
1033                         len--;
1034                 if (isspace(line[len - 1]))
1035                         white_space_at_end = 1;
1036
1037                 if (space_before_tab || white_space_at_end) {
1038                         printf("%s:%d: %s", data->filename, data->lineno, ws);
1039                         if (space_before_tab) {
1040                                 printf("space before tab");
1041                                 if (white_space_at_end)
1042                                         putchar(',');
1043                         }
1044                         if (white_space_at_end)
1045                                 printf("white space at end");
1046                         printf(":%s ", reset);
1047                         emit_line_with_ws(1, set, reset, ws, line, len);
1048                 }
1049
1050                 data->lineno++;
1051         } else if (line[0] == ' ')
1052                 data->lineno++;
1053         else if (line[0] == '@') {
1054                 char *plus = strchr(line, '+');
1055                 if (plus)
1056                         data->lineno = strtol(plus, NULL, 10);
1057                 else
1058                         die("invalid diff");
1059         }
1060 }
1061
1062 static unsigned char *deflate_it(char *data,
1063                                  unsigned long size,
1064                                  unsigned long *result_size)
1065 {
1066         int bound;
1067         unsigned char *deflated;
1068         z_stream stream;
1069
1070         memset(&stream, 0, sizeof(stream));
1071         deflateInit(&stream, zlib_compression_level);
1072         bound = deflateBound(&stream, size);
1073         deflated = xmalloc(bound);
1074         stream.next_out = deflated;
1075         stream.avail_out = bound;
1076
1077         stream.next_in = (unsigned char *)data;
1078         stream.avail_in = size;
1079         while (deflate(&stream, Z_FINISH) == Z_OK)
1080                 ; /* nothing */
1081         deflateEnd(&stream);
1082         *result_size = stream.total_out;
1083         return deflated;
1084 }
1085
1086 static void emit_binary_diff_body(mmfile_t *one, mmfile_t *two)
1087 {
1088         void *cp;
1089         void *delta;
1090         void *deflated;
1091         void *data;
1092         unsigned long orig_size;
1093         unsigned long delta_size;
1094         unsigned long deflate_size;
1095         unsigned long data_size;
1096
1097         /* We could do deflated delta, or we could do just deflated two,
1098          * whichever is smaller.
1099          */
1100         delta = NULL;
1101         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1102         if (one->size && two->size) {
1103                 delta = diff_delta(one->ptr, one->size,
1104                                    two->ptr, two->size,
1105                                    &delta_size, deflate_size);
1106                 if (delta) {
1107                         void *to_free = delta;
1108                         orig_size = delta_size;
1109                         delta = deflate_it(delta, delta_size, &delta_size);
1110                         free(to_free);
1111                 }
1112         }
1113
1114         if (delta && delta_size < deflate_size) {
1115                 printf("delta %lu\n", orig_size);
1116                 free(deflated);
1117                 data = delta;
1118                 data_size = delta_size;
1119         }
1120         else {
1121                 printf("literal %lu\n", two->size);
1122                 free(delta);
1123                 data = deflated;
1124                 data_size = deflate_size;
1125         }
1126
1127         /* emit data encoded in base85 */
1128         cp = data;
1129         while (data_size) {
1130                 int bytes = (52 < data_size) ? 52 : data_size;
1131                 char line[70];
1132                 data_size -= bytes;
1133                 if (bytes <= 26)
1134                         line[0] = bytes + 'A' - 1;
1135                 else
1136                         line[0] = bytes - 26 + 'a' - 1;
1137                 encode_85(line + 1, cp, bytes);
1138                 cp = (char *) cp + bytes;
1139                 puts(line);
1140         }
1141         printf("\n");
1142         free(data);
1143 }
1144
1145 static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
1146 {
1147         printf("GIT binary patch\n");
1148         emit_binary_diff_body(one, two);
1149         emit_binary_diff_body(two, one);
1150 }
1151
1152 static void setup_diff_attr_check(struct git_attr_check *check)
1153 {
1154         static struct git_attr *attr_diff;
1155
1156         if (!attr_diff) {
1157                 attr_diff = git_attr("diff", 4);
1158         }
1159         check[0].attr = attr_diff;
1160 }
1161
1162 static void diff_filespec_check_attr(struct diff_filespec *one)
1163 {
1164         struct git_attr_check attr_diff_check;
1165         int check_from_data = 0;
1166
1167         if (one->checked_attr)
1168                 return;
1169
1170         setup_diff_attr_check(&attr_diff_check);
1171         one->is_binary = 0;
1172         one->funcname_pattern_ident = NULL;
1173
1174         if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1175                 const char *value;
1176
1177                 /* binaryness */
1178                 value = attr_diff_check.value;
1179                 if (ATTR_TRUE(value))
1180                         ;
1181                 else if (ATTR_FALSE(value))
1182                         one->is_binary = 1;
1183                 else
1184                         check_from_data = 1;
1185
1186                 /* funcname pattern ident */
1187                 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1188                         ;
1189                 else
1190                         one->funcname_pattern_ident = value;
1191         }
1192
1193         if (check_from_data) {
1194                 if (!one->data && DIFF_FILE_VALID(one))
1195                         diff_populate_filespec(one, 0);
1196
1197                 if (one->data)
1198                         one->is_binary = buffer_is_binary(one->data, one->size);
1199         }
1200 }
1201
1202 int diff_filespec_is_binary(struct diff_filespec *one)
1203 {
1204         diff_filespec_check_attr(one);
1205         return one->is_binary;
1206 }
1207
1208 static const char *funcname_pattern(const char *ident)
1209 {
1210         struct funcname_pattern *pp;
1211
1212         read_config_if_needed();
1213         for (pp = funcname_pattern_list; pp; pp = pp->next)
1214                 if (!strcmp(ident, pp->name))
1215                         return pp->pattern;
1216         return NULL;
1217 }
1218
1219 static struct builtin_funcname_pattern {
1220         const char *name;
1221         const char *pattern;
1222 } builtin_funcname_pattern[] = {
1223         { "java", "!^[  ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1224                         "new\\|return\\|switch\\|throw\\|while\\)\n"
1225                         "^[     ]*\\(\\([       ]*"
1226                         "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1227                         "[      ]*([^;]*$\\)" },
1228         { "tex", "^\\(\\\\\\(sub\\)*section{.*\\)$" },
1229 };
1230
1231 static const char *diff_funcname_pattern(struct diff_filespec *one)
1232 {
1233         const char *ident, *pattern;
1234         int i;
1235
1236         diff_filespec_check_attr(one);
1237         ident = one->funcname_pattern_ident;
1238
1239         if (!ident)
1240                 /*
1241                  * If the config file has "funcname.default" defined, that
1242                  * regexp is used; otherwise NULL is returned and xemit uses
1243                  * the built-in default.
1244                  */
1245                 return funcname_pattern("default");
1246
1247         /* Look up custom "funcname.$ident" regexp from config. */
1248         pattern = funcname_pattern(ident);
1249         if (pattern)
1250                 return pattern;
1251
1252         /*
1253          * And define built-in fallback patterns here.  Note that
1254          * these can be overriden by the user's config settings.
1255          */
1256         for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1257                 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1258                         return builtin_funcname_pattern[i].pattern;
1259
1260         return NULL;
1261 }
1262
1263 static void builtin_diff(const char *name_a,
1264                          const char *name_b,
1265                          struct diff_filespec *one,
1266                          struct diff_filespec *two,
1267                          const char *xfrm_msg,
1268                          struct diff_options *o,
1269                          int complete_rewrite)
1270 {
1271         mmfile_t mf1, mf2;
1272         const char *lbl[2];
1273         char *a_one, *b_two;
1274         const char *set = diff_get_color(o->color_diff, DIFF_METAINFO);
1275         const char *reset = diff_get_color(o->color_diff, DIFF_RESET);
1276
1277         a_one = quote_two("a/", name_a + (*name_a == '/'));
1278         b_two = quote_two("b/", name_b + (*name_b == '/'));
1279         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1280         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1281         printf("%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1282         if (lbl[0][0] == '/') {
1283                 /* /dev/null */
1284                 printf("%snew file mode %06o%s\n", set, two->mode, reset);
1285                 if (xfrm_msg && xfrm_msg[0])
1286                         printf("%s%s%s\n", set, xfrm_msg, reset);
1287         }
1288         else if (lbl[1][0] == '/') {
1289                 printf("%sdeleted file mode %06o%s\n", set, one->mode, reset);
1290                 if (xfrm_msg && xfrm_msg[0])
1291                         printf("%s%s%s\n", set, xfrm_msg, reset);
1292         }
1293         else {
1294                 if (one->mode != two->mode) {
1295                         printf("%sold mode %06o%s\n", set, one->mode, reset);
1296                         printf("%snew mode %06o%s\n", set, two->mode, reset);
1297                 }
1298                 if (xfrm_msg && xfrm_msg[0])
1299                         printf("%s%s%s\n", set, xfrm_msg, reset);
1300                 /*
1301                  * we do not run diff between different kind
1302                  * of objects.
1303                  */
1304                 if ((one->mode ^ two->mode) & S_IFMT)
1305                         goto free_ab_and_return;
1306                 if (complete_rewrite) {
1307                         emit_rewrite_diff(name_a, name_b, one, two,
1308                                         o->color_diff);
1309                         o->found_changes = 1;
1310                         goto free_ab_and_return;
1311                 }
1312         }
1313
1314         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1315                 die("unable to read files to diff");
1316
1317         if (!o->text &&
1318             (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1319                 /* Quite common confusing case */
1320                 if (mf1.size == mf2.size &&
1321                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1322                         goto free_ab_and_return;
1323                 if (o->binary)
1324                         emit_binary_diff(&mf1, &mf2);
1325                 else
1326                         printf("Binary files %s and %s differ\n",
1327                                lbl[0], lbl[1]);
1328                 o->found_changes = 1;
1329         }
1330         else {
1331                 /* Crazy xdl interfaces.. */
1332                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1333                 xpparam_t xpp;
1334                 xdemitconf_t xecfg;
1335                 xdemitcb_t ecb;
1336                 struct emit_callback ecbdata;
1337                 const char *funcname_pattern;
1338
1339                 funcname_pattern = diff_funcname_pattern(one);
1340                 if (!funcname_pattern)
1341                         funcname_pattern = diff_funcname_pattern(two);
1342
1343                 memset(&xecfg, 0, sizeof(xecfg));
1344                 memset(&ecbdata, 0, sizeof(ecbdata));
1345                 ecbdata.label_path = lbl;
1346                 ecbdata.color_diff = o->color_diff;
1347                 ecbdata.found_changesp = &o->found_changes;
1348                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1349                 xecfg.ctxlen = o->context;
1350                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1351                 if (funcname_pattern)
1352                         xdiff_set_find_func(&xecfg, funcname_pattern);
1353                 if (!diffopts)
1354                         ;
1355                 else if (!prefixcmp(diffopts, "--unified="))
1356                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1357                 else if (!prefixcmp(diffopts, "-u"))
1358                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1359                 ecb.outf = xdiff_outf;
1360                 ecb.priv = &ecbdata;
1361                 ecbdata.xm.consume = fn_out_consume;
1362                 if (o->color_diff_words)
1363                         ecbdata.diff_words =
1364                                 xcalloc(1, sizeof(struct diff_words_data));
1365                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1366                 if (o->color_diff_words)
1367                         free_diff_words_data(&ecbdata);
1368         }
1369
1370  free_ab_and_return:
1371         diff_free_filespec_data(one);
1372         diff_free_filespec_data(two);
1373         free(a_one);
1374         free(b_two);
1375         return;
1376 }
1377
1378 static void builtin_diffstat(const char *name_a, const char *name_b,
1379                              struct diff_filespec *one,
1380                              struct diff_filespec *two,
1381                              struct diffstat_t *diffstat,
1382                              struct diff_options *o,
1383                              int complete_rewrite)
1384 {
1385         mmfile_t mf1, mf2;
1386         struct diffstat_file *data;
1387
1388         data = diffstat_add(diffstat, name_a, name_b);
1389
1390         if (!one || !two) {
1391                 data->is_unmerged = 1;
1392                 return;
1393         }
1394         if (complete_rewrite) {
1395                 diff_populate_filespec(one, 0);
1396                 diff_populate_filespec(two, 0);
1397                 data->deleted = count_lines(one->data, one->size);
1398                 data->added = count_lines(two->data, two->size);
1399                 goto free_and_return;
1400         }
1401         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1402                 die("unable to read files to diff");
1403
1404         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1405                 data->is_binary = 1;
1406                 data->added = mf2.size;
1407                 data->deleted = mf1.size;
1408         } else {
1409                 /* Crazy xdl interfaces.. */
1410                 xpparam_t xpp;
1411                 xdemitconf_t xecfg;
1412                 xdemitcb_t ecb;
1413
1414                 memset(&xecfg, 0, sizeof(xecfg));
1415                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1416                 ecb.outf = xdiff_outf;
1417                 ecb.priv = diffstat;
1418                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1419         }
1420
1421  free_and_return:
1422         diff_free_filespec_data(one);
1423         diff_free_filespec_data(two);
1424 }
1425
1426 static void builtin_checkdiff(const char *name_a, const char *name_b,
1427                              struct diff_filespec *one,
1428                              struct diff_filespec *two, struct diff_options *o)
1429 {
1430         mmfile_t mf1, mf2;
1431         struct checkdiff_t data;
1432
1433         if (!two)
1434                 return;
1435
1436         memset(&data, 0, sizeof(data));
1437         data.xm.consume = checkdiff_consume;
1438         data.filename = name_b ? name_b : name_a;
1439         data.lineno = 0;
1440         data.color_diff = o->color_diff;
1441
1442         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1443                 die("unable to read files to diff");
1444
1445         if (diff_filespec_is_binary(two))
1446                 goto free_and_return;
1447         else {
1448                 /* Crazy xdl interfaces.. */
1449                 xpparam_t xpp;
1450                 xdemitconf_t xecfg;
1451                 xdemitcb_t ecb;
1452
1453                 memset(&xecfg, 0, sizeof(xecfg));
1454                 xpp.flags = XDF_NEED_MINIMAL;
1455                 ecb.outf = xdiff_outf;
1456                 ecb.priv = &data;
1457                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1458         }
1459  free_and_return:
1460         diff_free_filespec_data(one);
1461         diff_free_filespec_data(two);
1462 }
1463
1464 struct diff_filespec *alloc_filespec(const char *path)
1465 {
1466         int namelen = strlen(path);
1467         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1468
1469         memset(spec, 0, sizeof(*spec));
1470         spec->path = (char *)(spec + 1);
1471         memcpy(spec->path, path, namelen+1);
1472         return spec;
1473 }
1474
1475 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1476                    unsigned short mode)
1477 {
1478         if (mode) {
1479                 spec->mode = canon_mode(mode);
1480                 hashcpy(spec->sha1, sha1);
1481                 spec->sha1_valid = !is_null_sha1(sha1);
1482         }
1483 }
1484
1485 /*
1486  * Given a name and sha1 pair, if the index tells us the file in
1487  * the work tree has that object contents, return true, so that
1488  * prepare_temp_file() does not have to inflate and extract.
1489  */
1490 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1491 {
1492         struct cache_entry *ce;
1493         struct stat st;
1494         int pos, len;
1495
1496         /* We do not read the cache ourselves here, because the
1497          * benchmark with my previous version that always reads cache
1498          * shows that it makes things worse for diff-tree comparing
1499          * two linux-2.6 kernel trees in an already checked out work
1500          * tree.  This is because most diff-tree comparisons deal with
1501          * only a small number of files, while reading the cache is
1502          * expensive for a large project, and its cost outweighs the
1503          * savings we get by not inflating the object to a temporary
1504          * file.  Practically, this code only helps when we are used
1505          * by diff-cache --cached, which does read the cache before
1506          * calling us.
1507          */
1508         if (!active_cache)
1509                 return 0;
1510
1511         /* We want to avoid the working directory if our caller
1512          * doesn't need the data in a normal file, this system
1513          * is rather slow with its stat/open/mmap/close syscalls,
1514          * and the object is contained in a pack file.  The pack
1515          * is probably already open and will be faster to obtain
1516          * the data through than the working directory.  Loose
1517          * objects however would tend to be slower as they need
1518          * to be individually opened and inflated.
1519          */
1520         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1521                 return 0;
1522
1523         len = strlen(name);
1524         pos = cache_name_pos(name, len);
1525         if (pos < 0)
1526                 return 0;
1527         ce = active_cache[pos];
1528         if ((lstat(name, &st) < 0) ||
1529             !S_ISREG(st.st_mode) || /* careful! */
1530             ce_match_stat(ce, &st, 0) ||
1531             hashcmp(sha1, ce->sha1))
1532                 return 0;
1533         /* we return 1 only when we can stat, it is a regular file,
1534          * stat information matches, and sha1 recorded in the cache
1535          * matches.  I.e. we know the file in the work tree really is
1536          * the same as the <name, sha1> pair.
1537          */
1538         return 1;
1539 }
1540
1541 static int populate_from_stdin(struct diff_filespec *s)
1542 {
1543 #define INCREMENT 1024
1544         char *buf;
1545         unsigned long size;
1546         ssize_t got;
1547
1548         size = 0;
1549         buf = NULL;
1550         while (1) {
1551                 buf = xrealloc(buf, size + INCREMENT);
1552                 got = xread(0, buf + size, INCREMENT);
1553                 if (!got)
1554                         break; /* EOF */
1555                 if (got < 0)
1556                         return error("error while reading from stdin %s",
1557                                      strerror(errno));
1558                 size += got;
1559         }
1560         s->should_munmap = 0;
1561         s->data = buf;
1562         s->size = size;
1563         s->should_free = 1;
1564         return 0;
1565 }
1566
1567 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1568 {
1569         int len;
1570         char *data = xmalloc(100);
1571         len = snprintf(data, 100,
1572                 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1573         s->data = data;
1574         s->size = len;
1575         s->should_free = 1;
1576         if (size_only) {
1577                 s->data = NULL;
1578                 free(data);
1579         }
1580         return 0;
1581 }
1582
1583 /*
1584  * While doing rename detection and pickaxe operation, we may need to
1585  * grab the data for the blob (or file) for our own in-core comparison.
1586  * diff_filespec has data and size fields for this purpose.
1587  */
1588 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1589 {
1590         int err = 0;
1591         if (!DIFF_FILE_VALID(s))
1592                 die("internal error: asking to populate invalid file.");
1593         if (S_ISDIR(s->mode))
1594                 return -1;
1595
1596         if (s->data)
1597                 return 0;
1598
1599         if (size_only && 0 < s->size)
1600                 return 0;
1601
1602         if (S_ISGITLINK(s->mode))
1603                 return diff_populate_gitlink(s, size_only);
1604
1605         if (!s->sha1_valid ||
1606             reuse_worktree_file(s->path, s->sha1, 0)) {
1607                 struct stat st;
1608                 int fd;
1609                 char *buf;
1610                 unsigned long size;
1611
1612                 if (!strcmp(s->path, "-"))
1613                         return populate_from_stdin(s);
1614
1615                 if (lstat(s->path, &st) < 0) {
1616                         if (errno == ENOENT) {
1617                         err_empty:
1618                                 err = -1;
1619                         empty:
1620                                 s->data = (char *)"";
1621                                 s->size = 0;
1622                                 return err;
1623                         }
1624                 }
1625                 s->size = xsize_t(st.st_size);
1626                 if (!s->size)
1627                         goto empty;
1628                 if (size_only)
1629                         return 0;
1630                 if (S_ISLNK(st.st_mode)) {
1631                         int ret;
1632                         s->data = xmalloc(s->size);
1633                         s->should_free = 1;
1634                         ret = readlink(s->path, s->data, s->size);
1635                         if (ret < 0) {
1636                                 free(s->data);
1637                                 goto err_empty;
1638                         }
1639                         return 0;
1640                 }
1641                 fd = open(s->path, O_RDONLY);
1642                 if (fd < 0)
1643                         goto err_empty;
1644                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1645                 close(fd);
1646                 s->should_munmap = 1;
1647
1648                 /*
1649                  * Convert from working tree format to canonical git format
1650                  */
1651                 size = s->size;
1652                 buf = convert_to_git(s->path, s->data, &size);
1653                 if (buf) {
1654                         munmap(s->data, s->size);
1655                         s->should_munmap = 0;
1656                         s->data = buf;
1657                         s->size = size;
1658                         s->should_free = 1;
1659                 }
1660         }
1661         else {
1662                 enum object_type type;
1663                 if (size_only)
1664                         type = sha1_object_info(s->sha1, &s->size);
1665                 else {
1666                         s->data = read_sha1_file(s->sha1, &type, &s->size);
1667                         s->should_free = 1;
1668                 }
1669         }
1670         return 0;
1671 }
1672
1673 void diff_free_filespec_data(struct diff_filespec *s)
1674 {
1675         if (s->should_free)
1676                 free(s->data);
1677         else if (s->should_munmap)
1678                 munmap(s->data, s->size);
1679
1680         if (s->should_free || s->should_munmap) {
1681                 s->should_free = s->should_munmap = 0;
1682                 s->data = NULL;
1683         }
1684         free(s->cnt_data);
1685         s->cnt_data = NULL;
1686 }
1687
1688 static void prep_temp_blob(struct diff_tempfile *temp,
1689                            void *blob,
1690                            unsigned long size,
1691                            const unsigned char *sha1,
1692                            int mode)
1693 {
1694         int fd;
1695
1696         fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1697         if (fd < 0)
1698                 die("unable to create temp-file");
1699         if (write_in_full(fd, blob, size) != size)
1700                 die("unable to write temp-file");
1701         close(fd);
1702         temp->name = temp->tmp_path;
1703         strcpy(temp->hex, sha1_to_hex(sha1));
1704         temp->hex[40] = 0;
1705         sprintf(temp->mode, "%06o", mode);
1706 }
1707
1708 static void prepare_temp_file(const char *name,
1709                               struct diff_tempfile *temp,
1710                               struct diff_filespec *one)
1711 {
1712         if (!DIFF_FILE_VALID(one)) {
1713         not_a_valid_file:
1714                 /* A '-' entry produces this for file-2, and
1715                  * a '+' entry produces this for file-1.
1716                  */
1717                 temp->name = "/dev/null";
1718                 strcpy(temp->hex, ".");
1719                 strcpy(temp->mode, ".");
1720                 return;
1721         }
1722
1723         if (!one->sha1_valid ||
1724             reuse_worktree_file(name, one->sha1, 1)) {
1725                 struct stat st;
1726                 if (lstat(name, &st) < 0) {
1727                         if (errno == ENOENT)
1728                                 goto not_a_valid_file;
1729                         die("stat(%s): %s", name, strerror(errno));
1730                 }
1731                 if (S_ISLNK(st.st_mode)) {
1732                         int ret;
1733                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1734                         size_t sz = xsize_t(st.st_size);
1735                         if (sizeof(buf) <= st.st_size)
1736                                 die("symlink too long: %s", name);
1737                         ret = readlink(name, buf, sz);
1738                         if (ret < 0)
1739                                 die("readlink(%s)", name);
1740                         prep_temp_blob(temp, buf, sz,
1741                                        (one->sha1_valid ?
1742                                         one->sha1 : null_sha1),
1743                                        (one->sha1_valid ?
1744                                         one->mode : S_IFLNK));
1745                 }
1746                 else {
1747                         /* we can borrow from the file in the work tree */
1748                         temp->name = name;
1749                         if (!one->sha1_valid)
1750                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
1751                         else
1752                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
1753                         /* Even though we may sometimes borrow the
1754                          * contents from the work tree, we always want
1755                          * one->mode.  mode is trustworthy even when
1756                          * !(one->sha1_valid), as long as
1757                          * DIFF_FILE_VALID(one).
1758                          */
1759                         sprintf(temp->mode, "%06o", one->mode);
1760                 }
1761                 return;
1762         }
1763         else {
1764                 if (diff_populate_filespec(one, 0))
1765                         die("cannot read data blob for %s", one->path);
1766                 prep_temp_blob(temp, one->data, one->size,
1767                                one->sha1, one->mode);
1768         }
1769 }
1770
1771 static void remove_tempfile(void)
1772 {
1773         int i;
1774
1775         for (i = 0; i < 2; i++)
1776                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1777                         unlink(diff_temp[i].name);
1778                         diff_temp[i].name = NULL;
1779                 }
1780 }
1781
1782 static void remove_tempfile_on_signal(int signo)
1783 {
1784         remove_tempfile();
1785         signal(SIGINT, SIG_DFL);
1786         raise(signo);
1787 }
1788
1789 static int spawn_prog(const char *pgm, const char **arg)
1790 {
1791         pid_t pid;
1792         int status;
1793
1794         fflush(NULL);
1795         pid = fork();
1796         if (pid < 0)
1797                 die("unable to fork");
1798         if (!pid) {
1799                 execvp(pgm, (char *const*) arg);
1800                 exit(255);
1801         }
1802
1803         while (waitpid(pid, &status, 0) < 0) {
1804                 if (errno == EINTR)
1805                         continue;
1806                 return -1;
1807         }
1808
1809         /* Earlier we did not check the exit status because
1810          * diff exits non-zero if files are different, and
1811          * we are not interested in knowing that.  It was a
1812          * mistake which made it harder to quit a diff-*
1813          * session that uses the git-apply-patch-script as
1814          * the GIT_EXTERNAL_DIFF.  A custom GIT_EXTERNAL_DIFF
1815          * should also exit non-zero only when it wants to
1816          * abort the entire diff-* session.
1817          */
1818         if (WIFEXITED(status) && !WEXITSTATUS(status))
1819                 return 0;
1820         return -1;
1821 }
1822
1823 /* An external diff command takes:
1824  *
1825  * diff-cmd name infile1 infile1-sha1 infile1-mode \
1826  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
1827  *
1828  */
1829 static void run_external_diff(const char *pgm,
1830                               const char *name,
1831                               const char *other,
1832                               struct diff_filespec *one,
1833                               struct diff_filespec *two,
1834                               const char *xfrm_msg,
1835                               int complete_rewrite)
1836 {
1837         const char *spawn_arg[10];
1838         struct diff_tempfile *temp = diff_temp;
1839         int retval;
1840         static int atexit_asked = 0;
1841         const char *othername;
1842         const char **arg = &spawn_arg[0];
1843
1844         othername = (other? other : name);
1845         if (one && two) {
1846                 prepare_temp_file(name, &temp[0], one);
1847                 prepare_temp_file(othername, &temp[1], two);
1848                 if (! atexit_asked &&
1849                     (temp[0].name == temp[0].tmp_path ||
1850                      temp[1].name == temp[1].tmp_path)) {
1851                         atexit_asked = 1;
1852                         atexit(remove_tempfile);
1853                 }
1854                 signal(SIGINT, remove_tempfile_on_signal);
1855         }
1856
1857         if (one && two) {
1858                 *arg++ = pgm;
1859                 *arg++ = name;
1860                 *arg++ = temp[0].name;
1861                 *arg++ = temp[0].hex;
1862                 *arg++ = temp[0].mode;
1863                 *arg++ = temp[1].name;
1864                 *arg++ = temp[1].hex;
1865                 *arg++ = temp[1].mode;
1866                 if (other) {
1867                         *arg++ = other;
1868                         *arg++ = xfrm_msg;
1869                 }
1870         } else {
1871                 *arg++ = pgm;
1872                 *arg++ = name;
1873         }
1874         *arg = NULL;
1875         retval = spawn_prog(pgm, spawn_arg);
1876         remove_tempfile();
1877         if (retval) {
1878                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1879                 exit(1);
1880         }
1881 }
1882
1883 static const char *external_diff_attr(const char *name)
1884 {
1885         struct git_attr_check attr_diff_check;
1886
1887         setup_diff_attr_check(&attr_diff_check);
1888         if (!git_checkattr(name, 1, &attr_diff_check)) {
1889                 const char *value = attr_diff_check.value;
1890                 if (!ATTR_TRUE(value) &&
1891                     !ATTR_FALSE(value) &&
1892                     !ATTR_UNSET(value)) {
1893                         struct ll_diff_driver *drv;
1894
1895                         read_config_if_needed();
1896                         for (drv = user_diff; drv; drv = drv->next)
1897                                 if (!strcmp(drv->name, value))
1898                                         return drv->cmd;
1899                 }
1900         }
1901         return NULL;
1902 }
1903
1904 static void run_diff_cmd(const char *pgm,
1905                          const char *name,
1906                          const char *other,
1907                          struct diff_filespec *one,
1908                          struct diff_filespec *two,
1909                          const char *xfrm_msg,
1910                          struct diff_options *o,
1911                          int complete_rewrite)
1912 {
1913         if (!o->allow_external)
1914                 pgm = NULL;
1915         else {
1916                 const char *cmd = external_diff_attr(name);
1917                 if (cmd)
1918                         pgm = cmd;
1919         }
1920
1921         if (pgm) {
1922                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
1923                                   complete_rewrite);
1924                 return;
1925         }
1926         if (one && two)
1927                 builtin_diff(name, other ? other : name,
1928                              one, two, xfrm_msg, o, complete_rewrite);
1929         else
1930                 printf("* Unmerged path %s\n", name);
1931 }
1932
1933 static void diff_fill_sha1_info(struct diff_filespec *one)
1934 {
1935         if (DIFF_FILE_VALID(one)) {
1936                 if (!one->sha1_valid) {
1937                         struct stat st;
1938                         if (!strcmp(one->path, "-")) {
1939                                 hashcpy(one->sha1, null_sha1);
1940                                 return;
1941                         }
1942                         if (lstat(one->path, &st) < 0)
1943                                 die("stat %s", one->path);
1944                         if (index_path(one->sha1, one->path, &st, 0))
1945                                 die("cannot hash %s\n", one->path);
1946                 }
1947         }
1948         else
1949                 hashclr(one->sha1);
1950 }
1951
1952 static int similarity_index(struct diff_filepair *p)
1953 {
1954         return p->score * 100 / MAX_SCORE;
1955 }
1956
1957 static void run_diff(struct diff_filepair *p, struct diff_options *o)
1958 {
1959         const char *pgm = external_diff();
1960         char msg[PATH_MAX*2+300], *xfrm_msg;
1961         struct diff_filespec *one;
1962         struct diff_filespec *two;
1963         const char *name;
1964         const char *other;
1965         char *name_munged, *other_munged;
1966         int complete_rewrite = 0;
1967         int len;
1968
1969         if (DIFF_PAIR_UNMERGED(p)) {
1970                 /* unmerged */
1971                 run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, o, 0);
1972                 return;
1973         }
1974
1975         name = p->one->path;
1976         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1977         name_munged = quote_one(name);
1978         other_munged = quote_one(other);
1979         one = p->one; two = p->two;
1980
1981         diff_fill_sha1_info(one);
1982         diff_fill_sha1_info(two);
1983
1984         len = 0;
1985         switch (p->status) {
1986         case DIFF_STATUS_COPIED:
1987                 len += snprintf(msg + len, sizeof(msg) - len,
1988                                 "similarity index %d%%\n"
1989                                 "copy from %s\n"
1990                                 "copy to %s\n",
1991                                 similarity_index(p), name_munged, other_munged);
1992                 break;
1993         case DIFF_STATUS_RENAMED:
1994                 len += snprintf(msg + len, sizeof(msg) - len,
1995                                 "similarity index %d%%\n"
1996                                 "rename from %s\n"
1997                                 "rename to %s\n",
1998                                 similarity_index(p), name_munged, other_munged);
1999                 break;
2000         case DIFF_STATUS_MODIFIED:
2001                 if (p->score) {
2002                         len += snprintf(msg + len, sizeof(msg) - len,
2003                                         "dissimilarity index %d%%\n",
2004                                         similarity_index(p));
2005                         complete_rewrite = 1;
2006                         break;
2007                 }
2008                 /* fallthru */
2009         default:
2010                 /* nothing */
2011                 ;
2012         }
2013
2014         if (hashcmp(one->sha1, two->sha1)) {
2015                 int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
2016
2017                 if (o->binary) {
2018                         mmfile_t mf;
2019                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2020                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2021                                 abbrev = 40;
2022                 }
2023                 len += snprintf(msg + len, sizeof(msg) - len,
2024                                 "index %.*s..%.*s",
2025                                 abbrev, sha1_to_hex(one->sha1),
2026                                 abbrev, sha1_to_hex(two->sha1));
2027                 if (one->mode == two->mode)
2028                         len += snprintf(msg + len, sizeof(msg) - len,
2029                                         " %06o", one->mode);
2030                 len += snprintf(msg + len, sizeof(msg) - len, "\n");
2031         }
2032
2033         if (len)
2034                 msg[--len] = 0;
2035         xfrm_msg = len ? msg : NULL;
2036
2037         if (!pgm &&
2038             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2039             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2040                 /* a filepair that changes between file and symlink
2041                  * needs to be split into deletion and creation.
2042                  */
2043                 struct diff_filespec *null = alloc_filespec(two->path);
2044                 run_diff_cmd(NULL, name, other, one, null, xfrm_msg, o, 0);
2045                 free(null);
2046                 null = alloc_filespec(one->path);
2047                 run_diff_cmd(NULL, name, other, null, two, xfrm_msg, o, 0);
2048                 free(null);
2049         }
2050         else
2051                 run_diff_cmd(pgm, name, other, one, two, xfrm_msg, o,
2052                              complete_rewrite);
2053
2054         free(name_munged);
2055         free(other_munged);
2056 }
2057
2058 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2059                          struct diffstat_t *diffstat)
2060 {
2061         const char *name;
2062         const char *other;
2063         int complete_rewrite = 0;
2064
2065         if (DIFF_PAIR_UNMERGED(p)) {
2066                 /* unmerged */
2067                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2068                 return;
2069         }
2070
2071         name = p->one->path;
2072         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2073
2074         diff_fill_sha1_info(p->one);
2075         diff_fill_sha1_info(p->two);
2076
2077         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2078                 complete_rewrite = 1;
2079         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2080 }
2081
2082 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2083 {
2084         const char *name;
2085         const char *other;
2086
2087         if (DIFF_PAIR_UNMERGED(p)) {
2088                 /* unmerged */
2089                 return;
2090         }
2091
2092         name = p->one->path;
2093         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2094
2095         diff_fill_sha1_info(p->one);
2096         diff_fill_sha1_info(p->two);
2097
2098         builtin_checkdiff(name, other, p->one, p->two, o);
2099 }
2100
2101 void diff_setup(struct diff_options *options)
2102 {
2103         memset(options, 0, sizeof(*options));
2104         options->line_termination = '\n';
2105         options->break_opt = -1;
2106         options->rename_limit = -1;
2107         options->context = 3;
2108         options->msg_sep = "";
2109
2110         options->change = diff_change;
2111         options->add_remove = diff_addremove;
2112         options->color_diff = diff_use_color_default;
2113         options->detect_rename = diff_detect_rename_default;
2114 }
2115
2116 int diff_setup_done(struct diff_options *options)
2117 {
2118         int count = 0;
2119
2120         if (options->output_format & DIFF_FORMAT_NAME)
2121                 count++;
2122         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2123                 count++;
2124         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2125                 count++;
2126         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2127                 count++;
2128         if (count > 1)
2129                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2130
2131         if (options->find_copies_harder)
2132                 options->detect_rename = DIFF_DETECT_COPY;
2133
2134         if (options->output_format & (DIFF_FORMAT_NAME |
2135                                       DIFF_FORMAT_NAME_STATUS |
2136                                       DIFF_FORMAT_CHECKDIFF |
2137                                       DIFF_FORMAT_NO_OUTPUT))
2138                 options->output_format &= ~(DIFF_FORMAT_RAW |
2139                                             DIFF_FORMAT_NUMSTAT |
2140                                             DIFF_FORMAT_DIFFSTAT |
2141                                             DIFF_FORMAT_SHORTSTAT |
2142                                             DIFF_FORMAT_SUMMARY |
2143                                             DIFF_FORMAT_PATCH);
2144
2145         /*
2146          * These cases always need recursive; we do not drop caller-supplied
2147          * recursive bits for other formats here.
2148          */
2149         if (options->output_format & (DIFF_FORMAT_PATCH |
2150                                       DIFF_FORMAT_NUMSTAT |
2151                                       DIFF_FORMAT_DIFFSTAT |
2152                                       DIFF_FORMAT_SHORTSTAT |
2153                                       DIFF_FORMAT_SUMMARY |
2154                                       DIFF_FORMAT_CHECKDIFF))
2155                 options->recursive = 1;
2156         /*
2157          * Also pickaxe would not work very well if you do not say recursive
2158          */
2159         if (options->pickaxe)
2160                 options->recursive = 1;
2161
2162         if (options->detect_rename && options->rename_limit < 0)
2163                 options->rename_limit = diff_rename_limit_default;
2164         if (options->setup & DIFF_SETUP_USE_CACHE) {
2165                 if (!active_cache)
2166                         /* read-cache does not die even when it fails
2167                          * so it is safe for us to do this here.  Also
2168                          * it does not smudge active_cache or active_nr
2169                          * when it fails, so we do not have to worry about
2170                          * cleaning it up ourselves either.
2171                          */
2172                         read_cache();
2173         }
2174         if (options->abbrev <= 0 || 40 < options->abbrev)
2175                 options->abbrev = 40; /* full */
2176
2177         /*
2178          * It does not make sense to show the first hit we happened
2179          * to have found.  It does not make sense not to return with
2180          * exit code in such a case either.
2181          */
2182         if (options->quiet) {
2183                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2184                 options->exit_with_status = 1;
2185         }
2186
2187         /*
2188          * If we postprocess in diffcore, we cannot simply return
2189          * upon the first hit.  We need to run diff as usual.
2190          */
2191         if (options->pickaxe || options->filter)
2192                 options->quiet = 0;
2193
2194         return 0;
2195 }
2196
2197 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2198 {
2199         char c, *eq;
2200         int len;
2201
2202         if (*arg != '-')
2203                 return 0;
2204         c = *++arg;
2205         if (!c)
2206                 return 0;
2207         if (c == arg_short) {
2208                 c = *++arg;
2209                 if (!c)
2210                         return 1;
2211                 if (val && isdigit(c)) {
2212                         char *end;
2213                         int n = strtoul(arg, &end, 10);
2214                         if (*end)
2215                                 return 0;
2216                         *val = n;
2217                         return 1;
2218                 }
2219                 return 0;
2220         }
2221         if (c != '-')
2222                 return 0;
2223         arg++;
2224         eq = strchr(arg, '=');
2225         if (eq)
2226                 len = eq - arg;
2227         else
2228                 len = strlen(arg);
2229         if (!len || strncmp(arg, arg_long, len))
2230                 return 0;
2231         if (eq) {
2232                 int n;
2233                 char *end;
2234                 if (!isdigit(*++eq))
2235                         return 0;
2236                 n = strtoul(eq, &end, 10);
2237                 if (*end)
2238                         return 0;
2239                 *val = n;
2240         }
2241         return 1;
2242 }
2243
2244 static int diff_scoreopt_parse(const char *opt);
2245
2246 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2247 {
2248         const char *arg = av[0];
2249         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2250                 options->output_format |= DIFF_FORMAT_PATCH;
2251         else if (opt_arg(arg, 'U', "unified", &options->context))
2252                 options->output_format |= DIFF_FORMAT_PATCH;
2253         else if (!strcmp(arg, "--raw"))
2254                 options->output_format |= DIFF_FORMAT_RAW;
2255         else if (!strcmp(arg, "--patch-with-raw")) {
2256                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2257         }
2258         else if (!strcmp(arg, "--numstat")) {
2259                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2260         }
2261         else if (!strcmp(arg, "--shortstat")) {
2262                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2263         }
2264         else if (!prefixcmp(arg, "--stat")) {
2265                 char *end;
2266                 int width = options->stat_width;
2267                 int name_width = options->stat_name_width;
2268                 arg += 6;
2269                 end = (char *)arg;
2270
2271                 switch (*arg) {
2272                 case '-':
2273                         if (!prefixcmp(arg, "-width="))
2274                                 width = strtoul(arg + 7, &end, 10);
2275                         else if (!prefixcmp(arg, "-name-width="))
2276                                 name_width = strtoul(arg + 12, &end, 10);
2277                         break;
2278                 case '=':
2279                         width = strtoul(arg+1, &end, 10);
2280                         if (*end == ',')
2281                                 name_width = strtoul(end+1, &end, 10);
2282                 }
2283
2284                 /* Important! This checks all the error cases! */
2285                 if (*end)
2286                         return 0;
2287                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2288                 options->stat_name_width = name_width;
2289                 options->stat_width = width;
2290         }
2291         else if (!strcmp(arg, "--check"))
2292                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2293         else if (!strcmp(arg, "--summary"))
2294                 options->output_format |= DIFF_FORMAT_SUMMARY;
2295         else if (!strcmp(arg, "--patch-with-stat")) {
2296                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2297         }
2298         else if (!strcmp(arg, "-z"))
2299                 options->line_termination = 0;
2300         else if (!prefixcmp(arg, "-l"))
2301                 options->rename_limit = strtoul(arg+2, NULL, 10);
2302         else if (!strcmp(arg, "--full-index"))
2303                 options->full_index = 1;
2304         else if (!strcmp(arg, "--binary")) {
2305                 options->output_format |= DIFF_FORMAT_PATCH;
2306                 options->binary = 1;
2307         }
2308         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text")) {
2309                 options->text = 1;
2310         }
2311         else if (!strcmp(arg, "--name-only"))
2312                 options->output_format |= DIFF_FORMAT_NAME;
2313         else if (!strcmp(arg, "--name-status"))
2314                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2315         else if (!strcmp(arg, "-R"))
2316                 options->reverse_diff = 1;
2317         else if (!prefixcmp(arg, "-S"))
2318                 options->pickaxe = arg + 2;
2319         else if (!strcmp(arg, "-s")) {
2320                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2321         }
2322         else if (!prefixcmp(arg, "-O"))
2323                 options->orderfile = arg + 2;
2324         else if (!prefixcmp(arg, "--diff-filter="))
2325                 options->filter = arg + 14;
2326         else if (!strcmp(arg, "--pickaxe-all"))
2327                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2328         else if (!strcmp(arg, "--pickaxe-regex"))
2329                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2330         else if (!prefixcmp(arg, "-B")) {
2331                 if ((options->break_opt =
2332                      diff_scoreopt_parse(arg)) == -1)
2333                         return -1;
2334         }
2335         else if (!prefixcmp(arg, "-M")) {
2336                 if ((options->rename_score =
2337                      diff_scoreopt_parse(arg)) == -1)
2338                         return -1;
2339                 options->detect_rename = DIFF_DETECT_RENAME;
2340         }
2341         else if (!prefixcmp(arg, "-C")) {
2342                 if (options->detect_rename == DIFF_DETECT_COPY)
2343                         options->find_copies_harder = 1;
2344                 if ((options->rename_score =
2345                      diff_scoreopt_parse(arg)) == -1)
2346                         return -1;
2347                 options->detect_rename = DIFF_DETECT_COPY;
2348         }
2349         else if (!strcmp(arg, "--find-copies-harder"))
2350                 options->find_copies_harder = 1;
2351         else if (!strcmp(arg, "--follow"))
2352                 options->follow_renames = 1;
2353         else if (!strcmp(arg, "--abbrev"))
2354                 options->abbrev = DEFAULT_ABBREV;
2355         else if (!prefixcmp(arg, "--abbrev=")) {
2356                 options->abbrev = strtoul(arg + 9, NULL, 10);
2357                 if (options->abbrev < MINIMUM_ABBREV)
2358                         options->abbrev = MINIMUM_ABBREV;
2359                 else if (40 < options->abbrev)
2360                         options->abbrev = 40;
2361         }
2362         else if (!strcmp(arg, "--color"))
2363                 options->color_diff = 1;
2364         else if (!strcmp(arg, "--no-color"))
2365                 options->color_diff = 0;
2366         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2367                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2368         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2369                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2370         else if (!strcmp(arg, "--ignore-space-at-eol"))
2371                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2372         else if (!strcmp(arg, "--color-words"))
2373                 options->color_diff = options->color_diff_words = 1;
2374         else if (!strcmp(arg, "--no-renames"))
2375                 options->detect_rename = 0;
2376         else if (!strcmp(arg, "--exit-code"))
2377                 options->exit_with_status = 1;
2378         else if (!strcmp(arg, "--quiet"))
2379                 options->quiet = 1;
2380         else if (!strcmp(arg, "--ext-diff"))
2381                 options->allow_external = 1;
2382         else if (!strcmp(arg, "--no-ext-diff"))
2383                 options->allow_external = 0;
2384         else
2385                 return 0;
2386         return 1;
2387 }
2388
2389 static int parse_num(const char **cp_p)
2390 {
2391         unsigned long num, scale;
2392         int ch, dot;
2393         const char *cp = *cp_p;
2394
2395         num = 0;
2396         scale = 1;
2397         dot = 0;
2398         for(;;) {
2399                 ch = *cp;
2400                 if ( !dot && ch == '.' ) {
2401                         scale = 1;
2402                         dot = 1;
2403                 } else if ( ch == '%' ) {
2404                         scale = dot ? scale*100 : 100;
2405                         cp++;   /* % is always at the end */
2406                         break;
2407                 } else if ( ch >= '0' && ch <= '9' ) {
2408                         if ( scale < 100000 ) {
2409                                 scale *= 10;
2410                                 num = (num*10) + (ch-'0');
2411                         }
2412                 } else {
2413                         break;
2414                 }
2415                 cp++;
2416         }
2417         *cp_p = cp;
2418
2419         /* user says num divided by scale and we say internally that
2420          * is MAX_SCORE * num / scale.
2421          */
2422         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2423 }
2424
2425 static int diff_scoreopt_parse(const char *opt)
2426 {
2427         int opt1, opt2, cmd;
2428
2429         if (*opt++ != '-')
2430                 return -1;
2431         cmd = *opt++;
2432         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2433                 return -1; /* that is not a -M, -C nor -B option */
2434
2435         opt1 = parse_num(&opt);
2436         if (cmd != 'B')
2437                 opt2 = 0;
2438         else {
2439                 if (*opt == 0)
2440                         opt2 = 0;
2441                 else if (*opt != '/')
2442                         return -1; /* we expect -B80/99 or -B80 */
2443                 else {
2444                         opt++;
2445                         opt2 = parse_num(&opt);
2446                 }
2447         }
2448         if (*opt != 0)
2449                 return -1;
2450         return opt1 | (opt2 << 16);
2451 }
2452
2453 struct diff_queue_struct diff_queued_diff;
2454
2455 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2456 {
2457         if (queue->alloc <= queue->nr) {
2458                 queue->alloc = alloc_nr(queue->alloc);
2459                 queue->queue = xrealloc(queue->queue,
2460                                         sizeof(dp) * queue->alloc);
2461         }
2462         queue->queue[queue->nr++] = dp;
2463 }
2464
2465 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2466                                  struct diff_filespec *one,
2467                                  struct diff_filespec *two)
2468 {
2469         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2470         dp->one = one;
2471         dp->two = two;
2472         if (queue)
2473                 diff_q(queue, dp);
2474         return dp;
2475 }
2476
2477 void diff_free_filepair(struct diff_filepair *p)
2478 {
2479         diff_free_filespec_data(p->one);
2480         diff_free_filespec_data(p->two);
2481         free(p->one);
2482         free(p->two);
2483         free(p);
2484 }
2485
2486 /* This is different from find_unique_abbrev() in that
2487  * it stuffs the result with dots for alignment.
2488  */
2489 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2490 {
2491         int abblen;
2492         const char *abbrev;
2493         if (len == 40)
2494                 return sha1_to_hex(sha1);
2495
2496         abbrev = find_unique_abbrev(sha1, len);
2497         if (!abbrev)
2498                 return sha1_to_hex(sha1);
2499         abblen = strlen(abbrev);
2500         if (abblen < 37) {
2501                 static char hex[41];
2502                 if (len < abblen && abblen <= len + 2)
2503                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2504                 else
2505                         sprintf(hex, "%s...", abbrev);
2506                 return hex;
2507         }
2508         return sha1_to_hex(sha1);
2509 }
2510
2511 static void diff_flush_raw(struct diff_filepair *p,
2512                            struct diff_options *options)
2513 {
2514         int two_paths;
2515         char status[10];
2516         int abbrev = options->abbrev;
2517         const char *path_one, *path_two;
2518         int inter_name_termination = '\t';
2519         int line_termination = options->line_termination;
2520
2521         if (!line_termination)
2522                 inter_name_termination = 0;
2523
2524         path_one = p->one->path;
2525         path_two = p->two->path;
2526         if (line_termination) {
2527                 path_one = quote_one(path_one);
2528                 path_two = quote_one(path_two);
2529         }
2530
2531         if (p->score)
2532                 sprintf(status, "%c%03d", p->status, similarity_index(p));
2533         else {
2534                 status[0] = p->status;
2535                 status[1] = 0;
2536         }
2537         switch (p->status) {
2538         case DIFF_STATUS_COPIED:
2539         case DIFF_STATUS_RENAMED:
2540                 two_paths = 1;
2541                 break;
2542         case DIFF_STATUS_ADDED:
2543         case DIFF_STATUS_DELETED:
2544                 two_paths = 0;
2545                 break;
2546         default:
2547                 two_paths = 0;
2548                 break;
2549         }
2550         if (!(options->output_format & DIFF_FORMAT_NAME_STATUS)) {
2551                 printf(":%06o %06o %s ",
2552                        p->one->mode, p->two->mode,
2553                        diff_unique_abbrev(p->one->sha1, abbrev));
2554                 printf("%s ",
2555                        diff_unique_abbrev(p->two->sha1, abbrev));
2556         }
2557         printf("%s%c%s", status, inter_name_termination,
2558                         two_paths || p->one->mode ?  path_one : path_two);
2559         if (two_paths)
2560                 printf("%c%s", inter_name_termination, path_two);
2561         putchar(line_termination);
2562         if (path_one != p->one->path)
2563                 free((void*)path_one);
2564         if (path_two != p->two->path)
2565                 free((void*)path_two);
2566 }
2567
2568 static void diff_flush_name(struct diff_filepair *p, struct diff_options *opt)
2569 {
2570         char *path = p->two->path;
2571
2572         if (opt->line_termination)
2573                 path = quote_one(p->two->path);
2574         printf("%s%c", path, opt->line_termination);
2575         if (p->two->path != path)
2576                 free(path);
2577 }
2578
2579 int diff_unmodified_pair(struct diff_filepair *p)
2580 {
2581         /* This function is written stricter than necessary to support
2582          * the currently implemented transformers, but the idea is to
2583          * let transformers to produce diff_filepairs any way they want,
2584          * and filter and clean them up here before producing the output.
2585          */
2586         struct diff_filespec *one, *two;
2587
2588         if (DIFF_PAIR_UNMERGED(p))
2589                 return 0; /* unmerged is interesting */
2590
2591         one = p->one;
2592         two = p->two;
2593
2594         /* deletion, addition, mode or type change
2595          * and rename are all interesting.
2596          */
2597         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2598             DIFF_PAIR_MODE_CHANGED(p) ||
2599             strcmp(one->path, two->path))
2600                 return 0;
2601
2602         /* both are valid and point at the same path.  that is, we are
2603          * dealing with a change.
2604          */
2605         if (one->sha1_valid && two->sha1_valid &&
2606             !hashcmp(one->sha1, two->sha1))
2607                 return 1; /* no change */
2608         if (!one->sha1_valid && !two->sha1_valid)
2609                 return 1; /* both look at the same file on the filesystem. */
2610         return 0;
2611 }
2612
2613 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2614 {
2615         if (diff_unmodified_pair(p))
2616                 return;
2617
2618         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2619             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2620                 return; /* no tree diffs in patch format */
2621
2622         run_diff(p, o);
2623 }
2624
2625 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2626                             struct diffstat_t *diffstat)
2627 {
2628         if (diff_unmodified_pair(p))
2629                 return;
2630
2631         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2632             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2633                 return; /* no tree diffs in patch format */
2634
2635         run_diffstat(p, o, diffstat);
2636 }
2637
2638 static void diff_flush_checkdiff(struct diff_filepair *p,
2639                 struct diff_options *o)
2640 {
2641         if (diff_unmodified_pair(p))
2642                 return;
2643
2644         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2645             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2646                 return; /* no tree diffs in patch format */
2647
2648         run_checkdiff(p, o);
2649 }
2650
2651 int diff_queue_is_empty(void)
2652 {
2653         struct diff_queue_struct *q = &diff_queued_diff;
2654         int i;
2655         for (i = 0; i < q->nr; i++)
2656                 if (!diff_unmodified_pair(q->queue[i]))
2657                         return 0;
2658         return 1;
2659 }
2660
2661 #if DIFF_DEBUG
2662 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2663 {
2664         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2665                 x, one ? one : "",
2666                 s->path,
2667                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2668                 s->mode,
2669                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2670         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2671                 x, one ? one : "",
2672                 s->size, s->xfrm_flags);
2673 }
2674
2675 void diff_debug_filepair(const struct diff_filepair *p, int i)
2676 {
2677         diff_debug_filespec(p->one, i, "one");
2678         diff_debug_filespec(p->two, i, "two");
2679         fprintf(stderr, "score %d, status %c stays %d broken %d\n",
2680                 p->score, p->status ? p->status : '?',
2681                 p->source_stays, p->broken_pair);
2682 }
2683
2684 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2685 {
2686         int i;
2687         if (msg)
2688                 fprintf(stderr, "%s\n", msg);
2689         fprintf(stderr, "q->nr = %d\n", q->nr);
2690         for (i = 0; i < q->nr; i++) {
2691                 struct diff_filepair *p = q->queue[i];
2692                 diff_debug_filepair(p, i);
2693         }
2694 }
2695 #endif
2696
2697 static void diff_resolve_rename_copy(void)
2698 {
2699         int i, j;
2700         struct diff_filepair *p, *pp;
2701         struct diff_queue_struct *q = &diff_queued_diff;
2702
2703         diff_debug_queue("resolve-rename-copy", q);
2704
2705         for (i = 0; i < q->nr; i++) {
2706                 p = q->queue[i];
2707                 p->status = 0; /* undecided */
2708                 if (DIFF_PAIR_UNMERGED(p))
2709                         p->status = DIFF_STATUS_UNMERGED;
2710                 else if (!DIFF_FILE_VALID(p->one))
2711                         p->status = DIFF_STATUS_ADDED;
2712                 else if (!DIFF_FILE_VALID(p->two))
2713                         p->status = DIFF_STATUS_DELETED;
2714                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2715                         p->status = DIFF_STATUS_TYPE_CHANGED;
2716
2717                 /* from this point on, we are dealing with a pair
2718                  * whose both sides are valid and of the same type, i.e.
2719                  * either in-place edit or rename/copy edit.
2720                  */
2721                 else if (DIFF_PAIR_RENAME(p)) {
2722                         if (p->source_stays) {
2723                                 p->status = DIFF_STATUS_COPIED;
2724                                 continue;
2725                         }
2726                         /* See if there is some other filepair that
2727                          * copies from the same source as us.  If so
2728                          * we are a copy.  Otherwise we are either a
2729                          * copy if the path stays, or a rename if it
2730                          * does not, but we already handled "stays" case.
2731                          */
2732                         for (j = i + 1; j < q->nr; j++) {
2733                                 pp = q->queue[j];
2734                                 if (strcmp(pp->one->path, p->one->path))
2735                                         continue; /* not us */
2736                                 if (!DIFF_PAIR_RENAME(pp))
2737                                         continue; /* not a rename/copy */
2738                                 /* pp is a rename/copy from the same source */
2739                                 p->status = DIFF_STATUS_COPIED;
2740                                 break;
2741                         }
2742                         if (!p->status)
2743                                 p->status = DIFF_STATUS_RENAMED;
2744                 }
2745                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2746                          p->one->mode != p->two->mode ||
2747                          is_null_sha1(p->one->sha1))
2748                         p->status = DIFF_STATUS_MODIFIED;
2749                 else {
2750                         /* This is a "no-change" entry and should not
2751                          * happen anymore, but prepare for broken callers.
2752                          */
2753                         error("feeding unmodified %s to diffcore",
2754                               p->one->path);
2755                         p->status = DIFF_STATUS_UNKNOWN;
2756                 }
2757         }
2758         diff_debug_queue("resolve-rename-copy done", q);
2759 }
2760
2761 static int check_pair_status(struct diff_filepair *p)
2762 {
2763         switch (p->status) {
2764         case DIFF_STATUS_UNKNOWN:
2765                 return 0;
2766         case 0:
2767                 die("internal error in diff-resolve-rename-copy");
2768         default:
2769                 return 1;
2770         }
2771 }
2772
2773 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2774 {
2775         int fmt = opt->output_format;
2776
2777         if (fmt & DIFF_FORMAT_CHECKDIFF)
2778                 diff_flush_checkdiff(p, opt);
2779         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2780                 diff_flush_raw(p, opt);
2781         else if (fmt & DIFF_FORMAT_NAME)
2782                 diff_flush_name(p, opt);
2783 }
2784
2785 static void show_file_mode_name(const char *newdelete, struct diff_filespec *fs)
2786 {
2787         char *name = quote_one(fs->path);
2788         if (fs->mode)
2789                 printf(" %s mode %06o %s\n", newdelete, fs->mode, name);
2790         else
2791                 printf(" %s %s\n", newdelete, name);
2792         free(name);
2793 }
2794
2795
2796 static void show_mode_change(struct diff_filepair *p, int show_name)
2797 {
2798         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2799                 if (show_name) {
2800                         char *name = quote_one(p->two->path);
2801                         printf(" mode change %06o => %06o %s\n",
2802                                p->one->mode, p->two->mode, name);
2803                         free(name);
2804                 }
2805                 else
2806                         printf(" mode change %06o => %06o\n",
2807                                p->one->mode, p->two->mode);
2808         }
2809 }
2810
2811 static void show_rename_copy(const char *renamecopy, struct diff_filepair *p)
2812 {
2813         char *names = pprint_rename(p->one->path, p->two->path);
2814
2815         printf(" %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2816         free(names);
2817         show_mode_change(p, 0);
2818 }
2819
2820 static void diff_summary(struct diff_filepair *p)
2821 {
2822         switch(p->status) {
2823         case DIFF_STATUS_DELETED:
2824                 show_file_mode_name("delete", p->one);
2825                 break;
2826         case DIFF_STATUS_ADDED:
2827                 show_file_mode_name("create", p->two);
2828                 break;
2829         case DIFF_STATUS_COPIED:
2830                 show_rename_copy("copy", p);
2831                 break;
2832         case DIFF_STATUS_RENAMED:
2833                 show_rename_copy("rename", p);
2834                 break;
2835         default:
2836                 if (p->score) {
2837                         char *name = quote_one(p->two->path);
2838                         printf(" rewrite %s (%d%%)\n", name,
2839                                similarity_index(p));
2840                         free(name);
2841                         show_mode_change(p, 0);
2842                 } else  show_mode_change(p, 1);
2843                 break;
2844         }
2845 }
2846
2847 struct patch_id_t {
2848         struct xdiff_emit_state xm;
2849         SHA_CTX *ctx;
2850         int patchlen;
2851 };
2852
2853 static int remove_space(char *line, int len)
2854 {
2855         int i;
2856         char *dst = line;
2857         unsigned char c;
2858
2859         for (i = 0; i < len; i++)
2860                 if (!isspace((c = line[i])))
2861                         *dst++ = c;
2862
2863         return dst - line;
2864 }
2865
2866 static void patch_id_consume(void *priv, char *line, unsigned long len)
2867 {
2868         struct patch_id_t *data = priv;
2869         int new_len;
2870
2871         /* Ignore line numbers when computing the SHA1 of the patch */
2872         if (!prefixcmp(line, "@@ -"))
2873                 return;
2874
2875         new_len = remove_space(line, len);
2876
2877         SHA1_Update(data->ctx, line, new_len);
2878         data->patchlen += new_len;
2879 }
2880
2881 /* returns 0 upon success, and writes result into sha1 */
2882 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2883 {
2884         struct diff_queue_struct *q = &diff_queued_diff;
2885         int i;
2886         SHA_CTX ctx;
2887         struct patch_id_t data;
2888         char buffer[PATH_MAX * 4 + 20];
2889
2890         SHA1_Init(&ctx);
2891         memset(&data, 0, sizeof(struct patch_id_t));
2892         data.ctx = &ctx;
2893         data.xm.consume = patch_id_consume;
2894
2895         for (i = 0; i < q->nr; i++) {
2896                 xpparam_t xpp;
2897                 xdemitconf_t xecfg;
2898                 xdemitcb_t ecb;
2899                 mmfile_t mf1, mf2;
2900                 struct diff_filepair *p = q->queue[i];
2901                 int len1, len2;
2902
2903                 memset(&xecfg, 0, sizeof(xecfg));
2904                 if (p->status == 0)
2905                         return error("internal diff status error");
2906                 if (p->status == DIFF_STATUS_UNKNOWN)
2907                         continue;
2908                 if (diff_unmodified_pair(p))
2909                         continue;
2910                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2911                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2912                         continue;
2913                 if (DIFF_PAIR_UNMERGED(p))
2914                         continue;
2915
2916                 diff_fill_sha1_info(p->one);
2917                 diff_fill_sha1_info(p->two);
2918                 if (fill_mmfile(&mf1, p->one) < 0 ||
2919                                 fill_mmfile(&mf2, p->two) < 0)
2920                         return error("unable to read files to diff");
2921
2922                 /* Maybe hash p->two? into the patch id? */
2923                 if (diff_filespec_is_binary(p->two))
2924                         continue;
2925
2926                 len1 = remove_space(p->one->path, strlen(p->one->path));
2927                 len2 = remove_space(p->two->path, strlen(p->two->path));
2928                 if (p->one->mode == 0)
2929                         len1 = snprintf(buffer, sizeof(buffer),
2930                                         "diff--gita/%.*sb/%.*s"
2931                                         "newfilemode%06o"
2932                                         "---/dev/null"
2933                                         "+++b/%.*s",
2934                                         len1, p->one->path,
2935                                         len2, p->two->path,
2936                                         p->two->mode,
2937                                         len2, p->two->path);
2938                 else if (p->two->mode == 0)
2939                         len1 = snprintf(buffer, sizeof(buffer),
2940                                         "diff--gita/%.*sb/%.*s"
2941                                         "deletedfilemode%06o"
2942                                         "---a/%.*s"
2943                                         "+++/dev/null",
2944                                         len1, p->one->path,
2945                                         len2, p->two->path,
2946                                         p->one->mode,
2947                                         len1, p->one->path);
2948                 else
2949                         len1 = snprintf(buffer, sizeof(buffer),
2950                                         "diff--gita/%.*sb/%.*s"
2951                                         "---a/%.*s"
2952                                         "+++b/%.*s",
2953                                         len1, p->one->path,
2954                                         len2, p->two->path,
2955                                         len1, p->one->path,
2956                                         len2, p->two->path);
2957                 SHA1_Update(&ctx, buffer, len1);
2958
2959                 xpp.flags = XDF_NEED_MINIMAL;
2960                 xecfg.ctxlen = 3;
2961                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2962                 ecb.outf = xdiff_outf;
2963                 ecb.priv = &data;
2964                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
2965         }
2966
2967         SHA1_Final(sha1, &ctx);
2968         return 0;
2969 }
2970
2971 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
2972 {
2973         struct diff_queue_struct *q = &diff_queued_diff;
2974         int i;
2975         int result = diff_get_patch_id(options, sha1);
2976
2977         for (i = 0; i < q->nr; i++)
2978                 diff_free_filepair(q->queue[i]);
2979
2980         free(q->queue);
2981         q->queue = NULL;
2982         q->nr = q->alloc = 0;
2983
2984         return result;
2985 }
2986
2987 static int is_summary_empty(const struct diff_queue_struct *q)
2988 {
2989         int i;
2990
2991         for (i = 0; i < q->nr; i++) {
2992                 const struct diff_filepair *p = q->queue[i];
2993
2994                 switch (p->status) {
2995                 case DIFF_STATUS_DELETED:
2996                 case DIFF_STATUS_ADDED:
2997                 case DIFF_STATUS_COPIED:
2998                 case DIFF_STATUS_RENAMED:
2999                         return 0;
3000                 default:
3001                         if (p->score)
3002                                 return 0;
3003                         if (p->one->mode && p->two->mode &&
3004                             p->one->mode != p->two->mode)
3005                                 return 0;
3006                         break;
3007                 }
3008         }
3009         return 1;
3010 }
3011
3012 void diff_flush(struct diff_options *options)
3013 {
3014         struct diff_queue_struct *q = &diff_queued_diff;
3015         int i, output_format = options->output_format;
3016         int separator = 0;
3017
3018         /*
3019          * Order: raw, stat, summary, patch
3020          * or:    name/name-status/checkdiff (other bits clear)
3021          */
3022         if (!q->nr)
3023                 goto free_queue;
3024
3025         if (output_format & (DIFF_FORMAT_RAW |
3026                              DIFF_FORMAT_NAME |
3027                              DIFF_FORMAT_NAME_STATUS |
3028                              DIFF_FORMAT_CHECKDIFF)) {
3029                 for (i = 0; i < q->nr; i++) {
3030                         struct diff_filepair *p = q->queue[i];
3031                         if (check_pair_status(p))
3032                                 flush_one_pair(p, options);
3033                 }
3034                 separator++;
3035         }
3036
3037         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3038                 struct diffstat_t diffstat;
3039
3040                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3041                 diffstat.xm.consume = diffstat_consume;
3042                 for (i = 0; i < q->nr; i++) {
3043                         struct diff_filepair *p = q->queue[i];
3044                         if (check_pair_status(p))
3045                                 diff_flush_stat(p, options, &diffstat);
3046                 }
3047                 if (output_format & DIFF_FORMAT_NUMSTAT)
3048                         show_numstat(&diffstat, options);
3049                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3050                         show_stats(&diffstat, options);
3051                 else if (output_format & DIFF_FORMAT_SHORTSTAT)
3052                         show_shortstats(&diffstat);
3053                 separator++;
3054         }
3055
3056         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3057                 for (i = 0; i < q->nr; i++)
3058                         diff_summary(q->queue[i]);
3059                 separator++;
3060         }
3061
3062         if (output_format & DIFF_FORMAT_PATCH) {
3063                 if (separator) {
3064                         if (options->stat_sep) {
3065                                 /* attach patch instead of inline */
3066                                 fputs(options->stat_sep, stdout);
3067                         } else {
3068                                 putchar(options->line_termination);
3069                         }
3070                 }
3071
3072                 for (i = 0; i < q->nr; i++) {
3073                         struct diff_filepair *p = q->queue[i];
3074                         if (check_pair_status(p))
3075                                 diff_flush_patch(p, options);
3076                 }
3077         }
3078
3079         if (output_format & DIFF_FORMAT_CALLBACK)
3080                 options->format_callback(q, options, options->format_callback_data);
3081
3082         for (i = 0; i < q->nr; i++)
3083                 diff_free_filepair(q->queue[i]);
3084 free_queue:
3085         free(q->queue);
3086         q->queue = NULL;
3087         q->nr = q->alloc = 0;
3088 }
3089
3090 static void diffcore_apply_filter(const char *filter)
3091 {
3092         int i;
3093         struct diff_queue_struct *q = &diff_queued_diff;
3094         struct diff_queue_struct outq;
3095         outq.queue = NULL;
3096         outq.nr = outq.alloc = 0;
3097
3098         if (!filter)
3099                 return;
3100
3101         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3102                 int found;
3103                 for (i = found = 0; !found && i < q->nr; i++) {
3104                         struct diff_filepair *p = q->queue[i];
3105                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3106                              ((p->score &&
3107                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3108                               (!p->score &&
3109                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3110                             ((p->status != DIFF_STATUS_MODIFIED) &&
3111                              strchr(filter, p->status)))
3112                                 found++;
3113                 }
3114                 if (found)
3115                         return;
3116
3117                 /* otherwise we will clear the whole queue
3118                  * by copying the empty outq at the end of this
3119                  * function, but first clear the current entries
3120                  * in the queue.
3121                  */
3122                 for (i = 0; i < q->nr; i++)
3123                         diff_free_filepair(q->queue[i]);
3124         }
3125         else {
3126                 /* Only the matching ones */
3127                 for (i = 0; i < q->nr; i++) {
3128                         struct diff_filepair *p = q->queue[i];
3129
3130                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3131                              ((p->score &&
3132                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3133                               (!p->score &&
3134                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3135                             ((p->status != DIFF_STATUS_MODIFIED) &&
3136                              strchr(filter, p->status)))
3137                                 diff_q(&outq, p);
3138                         else
3139                                 diff_free_filepair(p);
3140                 }
3141         }
3142         free(q->queue);
3143         *q = outq;
3144 }
3145
3146 void diffcore_std(struct diff_options *options)
3147 {
3148         if (options->quiet)
3149                 return;
3150
3151         if (options->break_opt != -1)
3152                 diffcore_break(options->break_opt);
3153         if (options->detect_rename)
3154                 diffcore_rename(options);
3155         if (options->break_opt != -1)
3156                 diffcore_merge_broken();
3157         if (options->pickaxe)
3158                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3159         if (options->orderfile)
3160                 diffcore_order(options->orderfile);
3161         diff_resolve_rename_copy();
3162         diffcore_apply_filter(options->filter);
3163
3164         options->has_changes = !!diff_queued_diff.nr;
3165 }
3166
3167
3168 void diff_addremove(struct diff_options *options,
3169                     int addremove, unsigned mode,
3170                     const unsigned char *sha1,
3171                     const char *base, const char *path)
3172 {
3173         char concatpath[PATH_MAX];
3174         struct diff_filespec *one, *two;
3175
3176         /* This may look odd, but it is a preparation for
3177          * feeding "there are unchanged files which should
3178          * not produce diffs, but when you are doing copy
3179          * detection you would need them, so here they are"
3180          * entries to the diff-core.  They will be prefixed
3181          * with something like '=' or '*' (I haven't decided
3182          * which but should not make any difference).
3183          * Feeding the same new and old to diff_change()
3184          * also has the same effect.
3185          * Before the final output happens, they are pruned after
3186          * merged into rename/copy pairs as appropriate.
3187          */
3188         if (options->reverse_diff)
3189                 addremove = (addremove == '+' ? '-' :
3190                              addremove == '-' ? '+' : addremove);
3191
3192         if (!path) path = "";
3193         sprintf(concatpath, "%s%s", base, path);
3194         one = alloc_filespec(concatpath);
3195         two = alloc_filespec(concatpath);
3196
3197         if (addremove != '+')
3198                 fill_filespec(one, sha1, mode);
3199         if (addremove != '-')
3200                 fill_filespec(two, sha1, mode);
3201
3202         diff_queue(&diff_queued_diff, one, two);
3203         options->has_changes = 1;
3204 }
3205
3206 void diff_change(struct diff_options *options,
3207                  unsigned old_mode, unsigned new_mode,
3208                  const unsigned char *old_sha1,
3209                  const unsigned char *new_sha1,
3210                  const char *base, const char *path)
3211 {
3212         char concatpath[PATH_MAX];
3213         struct diff_filespec *one, *two;
3214
3215         if (options->reverse_diff) {
3216                 unsigned tmp;
3217                 const unsigned char *tmp_c;
3218                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3219                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3220         }
3221         if (!path) path = "";
3222         sprintf(concatpath, "%s%s", base, path);
3223         one = alloc_filespec(concatpath);
3224         two = alloc_filespec(concatpath);
3225         fill_filespec(one, old_sha1, old_mode);
3226         fill_filespec(two, new_sha1, new_mode);
3227
3228         diff_queue(&diff_queued_diff, one, two);
3229         options->has_changes = 1;
3230 }
3231
3232 void diff_unmerge(struct diff_options *options,
3233                   const char *path,
3234                   unsigned mode, const unsigned char *sha1)
3235 {
3236         struct diff_filespec *one, *two;
3237         one = alloc_filespec(path);
3238         two = alloc_filespec(path);
3239         fill_filespec(one, sha1, mode);
3240         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3241 }