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

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include <sys/types.h>
   5 #include <sys/wait.h>
   6 #include <signal.h>
   7 #include "cache.h"
   8 #include "quote.h"
   9 #include "diff.h"
  10 #include "diffcore.h"
  11 #include "delta.h"
  12 #include "xdiff-interface.h"
  13
  14 static int use_size_cache;
  15
  16 static int diff_detect_rename_default = 0;
  17 static int diff_rename_limit_default = -1;
  18 static int diff_use_color_default = 0;
  19
  20 /* "\033[1;38;5;2xx;48;5;2xxm\0" is 23 bytes */
  21 static char diff_colors[][24] = {
  22         "\033[m",       /* reset */
  23         "",             /* normal */
  24         "\033[1m",      /* bold */
  25         "\033[36m",     /* cyan */
  26         "\033[31m",     /* red */
  27         "\033[32m",     /* green */
  28         "\033[33m"      /* yellow */
  29 };
  30
  31 static int parse_diff_color_slot(const char *var, int ofs)
  32 {
  33         if (!strcasecmp(var+ofs, "plain"))
  34                 return DIFF_PLAIN;
  35         if (!strcasecmp(var+ofs, "meta"))
  36                 return DIFF_METAINFO;
  37         if (!strcasecmp(var+ofs, "frag"))
  38                 return DIFF_FRAGINFO;
  39         if (!strcasecmp(var+ofs, "old"))
  40                 return DIFF_FILE_OLD;
  41         if (!strcasecmp(var+ofs, "new"))
  42                 return DIFF_FILE_NEW;
  43         if (!strcasecmp(var+ofs, "commit"))
  44                 return DIFF_COMMIT;
  45         die("bad config variable '%s'", var);
  46 }
  47
  48 static int parse_color(const char *name, int len)
  49 {
  50         static const char * const color_names[] = {
  51                 "normal", "black", "red", "green", "yellow",
  52                 "blue", "magenta", "cyan", "white"
  53         };
  54         char *end;
  55         int i;
  56         for (i = 0; i < ARRAY_SIZE(color_names); i++) {
  57                 const char *str = color_names[i];
  58                 if (!strncasecmp(name, str, len) && !str[len])
  59                         return i - 1;
  60         }
  61         i = strtol(name, &end, 10);
  62         if (*name && !*end && i >= -1 && i <= 255)
  63                 return i;
  64         return -2;
  65 }
  66
  67 static int parse_attr(const char *name, int len)
  68 {
  69         static const int attr_values[] = { 1, 2, 4, 5, 7 };
  70         static const char * const attr_names[] = {
  71                 "bold", "dim", "ul", "blink", "reverse"
  72         };
  73         int i;
  74         for (i = 0; i < ARRAY_SIZE(attr_names); i++) {
  75                 const char *str = attr_names[i];
  76                 if (!strncasecmp(name, str, len) && !str[len])
  77                         return attr_values[i];
  78         }
  79         return -1;
  80 }
  81
  82 static void parse_diff_color_value(const char *value, const char *var, char *dst)
  83 {
  84         const char *ptr = value;
  85         int attr = -1;
  86         int fg = -2;
  87         int bg = -2;
  88
  89         if (!strcasecmp(value, "reset")) {
  90                 strcpy(dst, "\033[m");
  91                 return;
  92         }
  93
  94         /* [fg [bg]] [attr] */
  95         while (*ptr) {
  96                 const char *word = ptr;
  97                 int val, len = 0;
  98
  99                 while (word[len] && !isspace(word[len]))
 100                         len++;
 101
 102                 ptr = word + len;
 103                 while (*ptr && isspace(*ptr))
 104                         ptr++;
 105
 106                 val = parse_color(word, len);
 107                 if (val >= -1) {
 108                         if (fg == -2) {
 109                                 fg = val;
 110                                 continue;
 111                         }
 112                         if (bg == -2) {
 113                                 bg = val;
 114                                 continue;
 115                         }
 116                         goto bad;
 117                 }
 118                 val = parse_attr(word, len);
 119                 if (val < 0 || attr != -1)
 120                         goto bad;
 121                 attr = val;
 122         }
 123
 124         if (attr >= 0 || fg >= 0 || bg >= 0) {
 125                 int sep = 0;
 126
 127                 *dst++ = '\033';
 128                 *dst++ = '[';
 129                 if (attr >= 0) {
 130                         *dst++ = '0' + attr;
 131                         sep++;
 132                 }
 133                 if (fg >= 0) {
 134                         if (sep++)
 135                                 *dst++ = ';';
 136                         if (fg < 8) {
 137                                 *dst++ = '3';
 138                                 *dst++ = '0' + fg;
 139                         } else {
 140                                 dst += sprintf(dst, "38;5;%d", fg);
 141                         }
 142                 }
 143                 if (bg >= 0) {
 144                         if (sep++)
 145                                 *dst++ = ';';
 146                         if (bg < 8) {
 147                                 *dst++ = '4';
 148                                 *dst++ = '0' + bg;
 149                         } else {
 150                                 dst += sprintf(dst, "48;5;%d", bg);
 151                         }
 152                 }
 153                 *dst++ = 'm';
 154         }
 155         *dst = 0;
 156         return;
 157 bad:
 158         die("bad config value '%s' for variable '%s'", value, var);
 159 }
 160
 161 /*
 162  * These are to give UI layer defaults.
 163  * The core-level commands such as git-diff-files should
 164  * never be affected by the setting of diff.renames
 165  * the user happens to have in the configuration file.
 166  */
 167 int git_diff_ui_config(const char *var, const char *value)
 168 {
 169         if (!strcmp(var, "diff.renamelimit")) {
 170                 diff_rename_limit_default = git_config_int(var, value);
 171                 return 0;
 172         }
 173         if (!strcmp(var, "diff.color")) {
 174                 if (!value)
 175                         diff_use_color_default = 1; /* bool */
 176                 else if (!strcasecmp(value, "auto")) {
 177                         diff_use_color_default = 0;
 178                         if (isatty(1) || (pager_in_use && pager_use_color)) {
 179                                 char *term = getenv("TERM");
 180                                 if (term && strcmp(term, "dumb"))
 181                                         diff_use_color_default = 1;
 182                         }
 183                 }
 184                 else if (!strcasecmp(value, "never"))
 185                         diff_use_color_default = 0;
 186                 else if (!strcasecmp(value, "always"))
 187                         diff_use_color_default = 1;
 188                 else
 189                         diff_use_color_default = git_config_bool(var, value);
 190                 return 0;
 191         }
 192         if (!strcmp(var, "diff.renames")) {
 193                 if (!value)
 194                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 195                 else if (!strcasecmp(value, "copies") ||
 196                          !strcasecmp(value, "copy"))
 197                         diff_detect_rename_default = DIFF_DETECT_COPY;
 198                 else if (git_config_bool(var,value))
 199                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 200                 return 0;
 201         }
 202         if (!strncmp(var, "diff.color.", 11)) {
 203                 int slot = parse_diff_color_slot(var, 11);
 204                 parse_diff_color_value(value, var, diff_colors[slot]);
 205                 return 0;
 206         }
 207         return git_default_config(var, value);
 208 }
 209
 210 static char *quote_one(const char *str)
 211 {
 212         int needlen;
 213         char *xp;
 214
 215         if (!str)
 216                 return NULL;
 217         needlen = quote_c_style(str, NULL, NULL, 0);
 218         if (!needlen)
 219                 return strdup(str);
 220         xp = xmalloc(needlen + 1);
 221         quote_c_style(str, xp, NULL, 0);
 222         return xp;
 223 }
 224
 225 static char *quote_two(const char *one, const char *two)
 226 {
 227         int need_one = quote_c_style(one, NULL, NULL, 1);
 228         int need_two = quote_c_style(two, NULL, NULL, 1);
 229         char *xp;
 230
 231         if (need_one + need_two) {
 232                 if (!need_one) need_one = strlen(one);
 233                 if (!need_two) need_one = strlen(two);
 234
 235                 xp = xmalloc(need_one + need_two + 3);
 236                 xp[0] = '"';
 237                 quote_c_style(one, xp + 1, NULL, 1);
 238                 quote_c_style(two, xp + need_one + 1, NULL, 1);
 239                 strcpy(xp + need_one + need_two + 1, "\"");
 240                 return xp;
 241         }
 242         need_one = strlen(one);
 243         need_two = strlen(two);
 244         xp = xmalloc(need_one + need_two + 1);
 245         strcpy(xp, one);
 246         strcpy(xp + need_one, two);
 247         return xp;
 248 }
 249
 250 static const char *external_diff(void)
 251 {
 252         static const char *external_diff_cmd = NULL;
 253         static int done_preparing = 0;
 254
 255         if (done_preparing)
 256                 return external_diff_cmd;
 257         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 258         done_preparing = 1;
 259         return external_diff_cmd;
 260 }
 261
 262 #define TEMPFILE_PATH_LEN               50
 263
 264 static struct diff_tempfile {
 265         const char *name; /* filename external diff should read from */
 266         char hex[41];
 267         char mode[10];
 268         char tmp_path[TEMPFILE_PATH_LEN];
 269 } diff_temp[2];
 270
 271 static int count_lines(const char *data, int size)
 272 {
 273         int count, ch, completely_empty = 1, nl_just_seen = 0;
 274         count = 0;
 275         while (0 < size--) {
 276                 ch = *data++;
 277                 if (ch == '\n') {
 278                         count++;
 279                         nl_just_seen = 1;
 280                         completely_empty = 0;
 281                 }
 282                 else {
 283                         nl_just_seen = 0;
 284                         completely_empty = 0;
 285                 }
 286         }
 287         if (completely_empty)
 288                 return 0;
 289         if (!nl_just_seen)
 290                 count++; /* no trailing newline */
 291         return count;
 292 }
 293
 294 static void print_line_count(int count)
 295 {
 296         switch (count) {
 297         case 0:
 298                 printf("0,0");
 299                 break;
 300         case 1:
 301                 printf("1");
 302                 break;
 303         default:
 304                 printf("1,%d", count);
 305                 break;
 306         }
 307 }
 308
 309 static void copy_file(int prefix, const char *data, int size)
 310 {
 311         int ch, nl_just_seen = 1;
 312         while (0 < size--) {
 313                 ch = *data++;
 314                 if (nl_just_seen)
 315                         putchar(prefix);
 316                 putchar(ch);
 317                 if (ch == '\n')
 318                         nl_just_seen = 1;
 319                 else
 320                         nl_just_seen = 0;
 321         }
 322         if (!nl_just_seen)
 323                 printf("\n\\ No newline at end of file\n");
 324 }
 325
 326 static void emit_rewrite_diff(const char *name_a,
 327                               const char *name_b,
 328                               struct diff_filespec *one,
 329                               struct diff_filespec *two)
 330 {
 331         int lc_a, lc_b;
 332         diff_populate_filespec(one, 0);
 333         diff_populate_filespec(two, 0);
 334         lc_a = count_lines(one->data, one->size);
 335         lc_b = count_lines(two->data, two->size);
 336         printf("--- %s\n+++ %s\n@@ -", name_a, name_b);
 337         print_line_count(lc_a);
 338         printf(" +");
 339         print_line_count(lc_b);
 340         printf(" @@\n");
 341         if (lc_a)
 342                 copy_file('-', one->data, one->size);
 343         if (lc_b)
 344                 copy_file('+', two->data, two->size);
 345 }
 346
 347 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 348 {
 349         if (!DIFF_FILE_VALID(one)) {
 350                 mf->ptr = (char *)""; /* does not matter */
 351                 mf->size = 0;
 352                 return 0;
 353         }
 354         else if (diff_populate_filespec(one, 0))
 355                 return -1;
 356         mf->ptr = one->data;
 357         mf->size = one->size;
 358         return 0;
 359 }
 360
 361 struct emit_callback {
 362         struct xdiff_emit_state xm;
 363         int nparents, color_diff;
 364         const char **label_path;
 365 };
 366
 367 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 368 {
 369         if (diff_use_color)
 370                 return diff_colors[ix];
 371         return "";
 372 }
 373
 374 static void fn_out_consume(void *priv, char *line, unsigned long len)
 375 {
 376         int i;
 377         struct emit_callback *ecbdata = priv;
 378         const char *set = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 379         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 380
 381         if (ecbdata->label_path[0]) {
 382                 printf("%s--- %s%s\n", set, ecbdata->label_path[0], reset);
 383                 printf("%s+++ %s%s\n", set, ecbdata->label_path[1], reset);
 384                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 385         }
 386
 387         /* This is not really necessary for now because
 388          * this codepath only deals with two-way diffs.
 389          */
 390         for (i = 0; i < len && line[i] == '@'; i++)
 391                 ;
 392         if (2 <= i && i < len && line[i] == ' ') {
 393                 ecbdata->nparents = i - 1;
 394                 set = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
 395         }
 396         else if (len < ecbdata->nparents)
 397                 set = reset;
 398         else {
 399                 int nparents = ecbdata->nparents;
 400                 int color = DIFF_PLAIN;
 401                 for (i = 0; i < nparents && len; i++) {
 402                         if (line[i] == '-')
 403                                 color = DIFF_FILE_OLD;
 404                         else if (line[i] == '+')
 405                                 color = DIFF_FILE_NEW;
 406                 }
 407                 set = diff_get_color(ecbdata->color_diff, color);
 408         }
 409         if (len > 0 && line[len-1] == '\n')
 410                 len--;
 411         fputs (set, stdout);
 412         fwrite (line, len, 1, stdout);
 413         puts (reset);
 414 }
 415
 416 static char *pprint_rename(const char *a, const char *b)
 417 {
 418         const char *old = a;
 419         const char *new = b;
 420         char *name = NULL;
 421         int pfx_length, sfx_length;
 422         int len_a = strlen(a);
 423         int len_b = strlen(b);
 424
 425         /* Find common prefix */
 426         pfx_length = 0;
 427         while (*old && *new && *old == *new) {
 428                 if (*old == '/')
 429                         pfx_length = old - a + 1;
 430                 old++;
 431                 new++;
 432         }
 433
 434         /* Find common suffix */
 435         old = a + len_a;
 436         new = b + len_b;
 437         sfx_length = 0;
 438         while (a <= old && b <= new && *old == *new) {
 439                 if (*old == '/')
 440                         sfx_length = len_a - (old - a);
 441                 old--;
 442                 new--;
 443         }
 444
 445         /*
 446          * pfx{mid-a => mid-b}sfx
 447          * {pfx-a => pfx-b}sfx
 448          * pfx{sfx-a => sfx-b}
 449          * name-a => name-b
 450          */
 451         if (pfx_length + sfx_length) {
 452                 int a_midlen = len_a - pfx_length - sfx_length;
 453                 int b_midlen = len_b - pfx_length - sfx_length;
 454                 if (a_midlen < 0) a_midlen = 0;
 455                 if (b_midlen < 0) b_midlen = 0;
 456
 457                 name = xmalloc(pfx_length + a_midlen + b_midlen + sfx_length + 7);
 458                 sprintf(name, "%.*s{%.*s => %.*s}%s",
 459                         pfx_length, a,
 460                         a_midlen, a + pfx_length,
 461                         b_midlen, b + pfx_length,
 462                         a + len_a - sfx_length);
 463         }
 464         else {
 465                 name = xmalloc(len_a + len_b + 5);
 466                 sprintf(name, "%s => %s", a, b);
 467         }
 468         return name;
 469 }
 470
 471 struct diffstat_t {
 472         struct xdiff_emit_state xm;
 473
 474         int nr;
 475         int alloc;
 476         struct diffstat_file {
 477                 char *name;
 478                 unsigned is_unmerged:1;
 479                 unsigned is_binary:1;
 480                 unsigned is_renamed:1;
 481                 unsigned int added, deleted;
 482         } **files;
 483 };
 484
 485 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 486                                           const char *name_a,
 487                                           const char *name_b)
 488 {
 489         struct diffstat_file *x;
 490         x = xcalloc(sizeof (*x), 1);
 491         if (diffstat->nr == diffstat->alloc) {
 492                 diffstat->alloc = alloc_nr(diffstat->alloc);
 493                 diffstat->files = xrealloc(diffstat->files,
 494                                 diffstat->alloc * sizeof(x));
 495         }
 496         diffstat->files[diffstat->nr++] = x;
 497         if (name_b) {
 498                 x->name = pprint_rename(name_a, name_b);
 499                 x->is_renamed = 1;
 500         }
 501         else
 502                 x->name = strdup(name_a);
 503         return x;
 504 }
 505
 506 static void diffstat_consume(void *priv, char *line, unsigned long len)
 507 {
 508         struct diffstat_t *diffstat = priv;
 509         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 510
 511         if (line[0] == '+')
 512                 x->added++;
 513         else if (line[0] == '-')
 514                 x->deleted++;
 515 }
 516
 517 static const char pluses[] = "++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++";
 518 static const char minuses[]= "----------------------------------------------------------------------";
 519 const char mime_boundary_leader[] = "------------";
 520
 521 static void show_stats(struct diffstat_t* data)
 522 {
 523         int i, len, add, del, total, adds = 0, dels = 0;
 524         int max, max_change = 0, max_len = 0;
 525         int total_files = data->nr;
 526
 527         if (data->nr == 0)
 528                 return;
 529
 530         for (i = 0; i < data->nr; i++) {
 531                 struct diffstat_file *file = data->files[i];
 532
 533                 len = strlen(file->name);
 534                 if (max_len < len)
 535                         max_len = len;
 536
 537                 if (file->is_binary || file->is_unmerged)
 538                         continue;
 539                 if (max_change < file->added + file->deleted)
 540                         max_change = file->added + file->deleted;
 541         }
 542
 543         for (i = 0; i < data->nr; i++) {
 544                 const char *prefix = "";
 545                 char *name = data->files[i]->name;
 546                 int added = data->files[i]->added;
 547                 int deleted = data->files[i]->deleted;
 548
 549                 if (0 < (len = quote_c_style(name, NULL, NULL, 0))) {
 550                         char *qname = xmalloc(len + 1);
 551                         quote_c_style(name, qname, NULL, 0);
 552                         free(name);
 553                         data->files[i]->name = name = qname;
 554                 }
 555
 556                 /*
 557                  * "scale" the filename
 558                  */
 559                 len = strlen(name);
 560                 max = max_len;
 561                 if (max > 50)
 562                         max = 50;
 563                 if (len > max) {
 564                         char *slash;
 565                         prefix = "...";
 566                         max -= 3;
 567                         name += len - max;
 568                         slash = strchr(name, '/');
 569                         if (slash)
 570                                 name = slash;
 571                 }
 572                 len = max;
 573
 574                 /*
 575                  * scale the add/delete
 576                  */
 577                 max = max_change;
 578                 if (max + len > 70)
 579                         max = 70 - len;
 580
 581                 if (data->files[i]->is_binary) {
 582                         printf(" %s%-*s |  Bin\n", prefix, len, name);
 583                         goto free_diffstat_file;
 584                 }
 585                 else if (data->files[i]->is_unmerged) {
 586                         printf(" %s%-*s |  Unmerged\n", prefix, len, name);
 587                         goto free_diffstat_file;
 588                 }
 589                 else if (!data->files[i]->is_renamed &&
 590                          (added + deleted == 0)) {
 591                         total_files--;
 592                         goto free_diffstat_file;
 593                 }
 594
 595                 add = added;
 596                 del = deleted;
 597                 total = add + del;
 598                 adds += add;
 599                 dels += del;
 600
 601                 if (max_change > 0) {
 602                         total = (total * max + max_change / 2) / max_change;
 603                         add = (add * max + max_change / 2) / max_change;
 604                         del = total - add;
 605                 }
 606                 printf(" %s%-*s |%5d %.*s%.*s\n", prefix,
 607                                 len, name, added + deleted,
 608                                 add, pluses, del, minuses);
 609         free_diffstat_file:
 610                 free(data->files[i]->name);
 611                 free(data->files[i]);
 612         }
 613         free(data->files);
 614         printf(" %d files changed, %d insertions(+), %d deletions(-)\n",
 615                         total_files, adds, dels);
 616 }
 617
 618 struct checkdiff_t {
 619         struct xdiff_emit_state xm;
 620         const char *filename;
 621         int lineno;
 622 };
 623
 624 static void checkdiff_consume(void *priv, char *line, unsigned long len)
 625 {
 626         struct checkdiff_t *data = priv;
 627
 628         if (line[0] == '+') {
 629                 int i, spaces = 0;
 630
 631                 data->lineno++;
 632
 633                 /* check space before tab */
 634                 for (i = 1; i < len && (line[i] == ' ' || line[i] == '\t'); i++)
 635                         if (line[i] == ' ')
 636                                 spaces++;
 637                 if (line[i - 1] == '\t' && spaces)
 638                         printf("%s:%d: space before tab:%.*s\n",
 639                                 data->filename, data->lineno, (int)len, line);
 640
 641                 /* check white space at line end */
 642                 if (line[len - 1] == '\n')
 643                         len--;
 644                 if (isspace(line[len - 1]))
 645                         printf("%s:%d: white space at end: %.*s\n",
 646                                 data->filename, data->lineno, (int)len, line);
 647         } else if (line[0] == ' ')
 648                 data->lineno++;
 649         else if (line[0] == '@') {
 650                 char *plus = strchr(line, '+');
 651                 if (plus)
 652                         data->lineno = strtol(plus, NULL, 10);
 653                 else
 654                         die("invalid diff");
 655         }
 656 }
 657
 658 static unsigned char *deflate_it(char *data,
 659                                  unsigned long size,
 660                                  unsigned long *result_size)
 661 {
 662         int bound;
 663         unsigned char *deflated;
 664         z_stream stream;
 665
 666         memset(&stream, 0, sizeof(stream));
 667         deflateInit(&stream, zlib_compression_level);
 668         bound = deflateBound(&stream, size);
 669         deflated = xmalloc(bound);
 670         stream.next_out = deflated;
 671         stream.avail_out = bound;
 672
 673         stream.next_in = (unsigned char *)data;
 674         stream.avail_in = size;
 675         while (deflate(&stream, Z_FINISH) == Z_OK)
 676                 ; /* nothing */
 677         deflateEnd(&stream);
 678         *result_size = stream.total_out;
 679         return deflated;
 680 }
 681
 682 static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
 683 {
 684         void *cp;
 685         void *delta;
 686         void *deflated;
 687         void *data;
 688         unsigned long orig_size;
 689         unsigned long delta_size;
 690         unsigned long deflate_size;
 691         unsigned long data_size;
 692
 693         printf("GIT binary patch\n");
 694         /* We could do deflated delta, or we could do just deflated two,
 695          * whichever is smaller.
 696          */
 697         delta = NULL;
 698         deflated = deflate_it(two->ptr, two->size, &deflate_size);
 699         if (one->size && two->size) {
 700                 delta = diff_delta(one->ptr, one->size,
 701                                    two->ptr, two->size,
 702                                    &delta_size, deflate_size);
 703                 if (delta) {
 704                         void *to_free = delta;
 705                         orig_size = delta_size;
 706                         delta = deflate_it(delta, delta_size, &delta_size);
 707                         free(to_free);
 708                 }
 709         }
 710
 711         if (delta && delta_size < deflate_size) {
 712                 printf("delta %lu\n", orig_size);
 713                 free(deflated);
 714                 data = delta;
 715                 data_size = delta_size;
 716         }
 717         else {
 718                 printf("literal %lu\n", two->size);
 719                 free(delta);
 720                 data = deflated;
 721                 data_size = deflate_size;
 722         }
 723
 724         /* emit data encoded in base85 */
 725         cp = data;
 726         while (data_size) {
 727                 int bytes = (52 < data_size) ? 52 : data_size;
 728                 char line[70];
 729                 data_size -= bytes;
 730                 if (bytes <= 26)
 731                         line[0] = bytes + 'A' - 1;
 732                 else
 733                         line[0] = bytes - 26 + 'a' - 1;
 734                 encode_85(line + 1, cp, bytes);
 735                 cp = (char *) cp + bytes;
 736                 puts(line);
 737         }
 738         printf("\n");
 739         free(data);
 740 }
 741
 742 #define FIRST_FEW_BYTES 8000
 743 static int mmfile_is_binary(mmfile_t *mf)
 744 {
 745         long sz = mf->size;
 746         if (FIRST_FEW_BYTES < sz)
 747                 sz = FIRST_FEW_BYTES;
 748         if (memchr(mf->ptr, 0, sz))
 749                 return 1;
 750         return 0;
 751 }
 752
 753 static void builtin_diff(const char *name_a,
 754                          const char *name_b,
 755                          struct diff_filespec *one,
 756                          struct diff_filespec *two,
 757                          const char *xfrm_msg,
 758                          struct diff_options *o,
 759                          int complete_rewrite)
 760 {
 761         mmfile_t mf1, mf2;
 762         const char *lbl[2];
 763         char *a_one, *b_two;
 764         const char *set = diff_get_color(o->color_diff, DIFF_METAINFO);
 765         const char *reset = diff_get_color(o->color_diff, DIFF_RESET);
 766
 767         a_one = quote_two("a/", name_a);
 768         b_two = quote_two("b/", name_b);
 769         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
 770         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
 771         printf("%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
 772         if (lbl[0][0] == '/') {
 773                 /* /dev/null */
 774                 printf("%snew file mode %06o%s\n", set, two->mode, reset);
 775                 if (xfrm_msg && xfrm_msg[0])
 776                         printf("%s%s%s\n", set, xfrm_msg, reset);
 777         }
 778         else if (lbl[1][0] == '/') {
 779                 printf("%sdeleted file mode %06o%s\n", set, one->mode, reset);
 780                 if (xfrm_msg && xfrm_msg[0])
 781                         printf("%s%s%s\n", set, xfrm_msg, reset);
 782         }
 783         else {
 784                 if (one->mode != two->mode) {
 785                         printf("%sold mode %06o%s\n", set, one->mode, reset);
 786                         printf("%snew mode %06o%s\n", set, two->mode, reset);
 787                 }
 788                 if (xfrm_msg && xfrm_msg[0])
 789                         printf("%s%s%s\n", set, xfrm_msg, reset);
 790                 /*
 791                  * we do not run diff between different kind
 792                  * of objects.
 793                  */
 794                 if ((one->mode ^ two->mode) & S_IFMT)
 795                         goto free_ab_and_return;
 796                 if (complete_rewrite) {
 797                         emit_rewrite_diff(name_a, name_b, one, two);
 798                         goto free_ab_and_return;
 799                 }
 800         }
 801
 802         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
 803                 die("unable to read files to diff");
 804
 805         if (!o->text && (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))) {
 806                 /* Quite common confusing case */
 807                 if (mf1.size == mf2.size &&
 808                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
 809                         goto free_ab_and_return;
 810                 if (o->binary)
 811                         emit_binary_diff(&mf1, &mf2);
 812                 else
 813                         printf("Binary files %s and %s differ\n",
 814                                lbl[0], lbl[1]);
 815         }
 816         else {
 817                 /* Crazy xdl interfaces.. */
 818                 const char *diffopts = getenv("GIT_DIFF_OPTS");
 819                 xpparam_t xpp;
 820                 xdemitconf_t xecfg;
 821                 xdemitcb_t ecb;
 822                 struct emit_callback ecbdata;
 823
 824                 memset(&ecbdata, 0, sizeof(ecbdata));
 825                 ecbdata.label_path = lbl;
 826                 ecbdata.color_diff = o->color_diff;
 827                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
 828                 xecfg.ctxlen = o->context;
 829                 xecfg.flags = XDL_EMIT_FUNCNAMES;
 830                 if (!diffopts)
 831                         ;
 832                 else if (!strncmp(diffopts, "--unified=", 10))
 833                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
 834                 else if (!strncmp(diffopts, "-u", 2))
 835                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
 836                 ecb.outf = xdiff_outf;
 837                 ecb.priv = &ecbdata;
 838                 ecbdata.xm.consume = fn_out_consume;
 839                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
 840         }
 841
 842  free_ab_and_return:
 843         free(a_one);
 844         free(b_two);
 845         return;
 846 }
 847
 848 static void builtin_diffstat(const char *name_a, const char *name_b,
 849                              struct diff_filespec *one,
 850                              struct diff_filespec *two,
 851                              struct diffstat_t *diffstat,
 852                              struct diff_options *o,
 853                              int complete_rewrite)
 854 {
 855         mmfile_t mf1, mf2;
 856         struct diffstat_file *data;
 857
 858         data = diffstat_add(diffstat, name_a, name_b);
 859
 860         if (!one || !two) {
 861                 data->is_unmerged = 1;
 862                 return;
 863         }
 864         if (complete_rewrite) {
 865                 diff_populate_filespec(one, 0);
 866                 diff_populate_filespec(two, 0);
 867                 data->deleted = count_lines(one->data, one->size);
 868                 data->added = count_lines(two->data, two->size);
 869                 return;
 870         }
 871         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
 872                 die("unable to read files to diff");
 873
 874         if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))
 875                 data->is_binary = 1;
 876         else {
 877                 /* Crazy xdl interfaces.. */
 878                 xpparam_t xpp;
 879                 xdemitconf_t xecfg;
 880                 xdemitcb_t ecb;
 881
 882                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
 883                 xecfg.ctxlen = 0;
 884                 xecfg.flags = 0;
 885                 ecb.outf = xdiff_outf;
 886                 ecb.priv = diffstat;
 887                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
 888         }
 889 }
 890
 891 static void builtin_checkdiff(const char *name_a, const char *name_b,
 892                              struct diff_filespec *one,
 893                              struct diff_filespec *two)
 894 {
 895         mmfile_t mf1, mf2;
 896         struct checkdiff_t data;
 897
 898         if (!two)
 899                 return;
 900
 901         memset(&data, 0, sizeof(data));
 902         data.xm.consume = checkdiff_consume;
 903         data.filename = name_b ? name_b : name_a;
 904         data.lineno = 0;
 905
 906         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
 907                 die("unable to read files to diff");
 908
 909         if (mmfile_is_binary(&mf2))
 910                 return;
 911         else {
 912                 /* Crazy xdl interfaces.. */
 913                 xpparam_t xpp;
 914                 xdemitconf_t xecfg;
 915                 xdemitcb_t ecb;
 916
 917                 xpp.flags = XDF_NEED_MINIMAL;
 918                 xecfg.ctxlen = 0;
 919                 xecfg.flags = 0;
 920                 ecb.outf = xdiff_outf;
 921                 ecb.priv = &data;
 922                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
 923         }
 924 }
 925
 926 struct diff_filespec *alloc_filespec(const char *path)
 927 {
 928         int namelen = strlen(path);
 929         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
 930
 931         memset(spec, 0, sizeof(*spec));
 932         spec->path = (char *)(spec + 1);
 933         memcpy(spec->path, path, namelen+1);
 934         return spec;
 935 }
 936
 937 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
 938                    unsigned short mode)
 939 {
 940         if (mode) {
 941                 spec->mode = canon_mode(mode);
 942                 memcpy(spec->sha1, sha1, 20);
 943                 spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
 944         }
 945 }
 946
 947 /*
 948  * Given a name and sha1 pair, if the dircache tells us the file in
 949  * the work tree has that object contents, return true, so that
 950  * prepare_temp_file() does not have to inflate and extract.
 951  */
 952 static int work_tree_matches(const char *name, const unsigned char *sha1)
 953 {
 954         struct cache_entry *ce;
 955         struct stat st;
 956         int pos, len;
 957
 958         /* We do not read the cache ourselves here, because the
 959          * benchmark with my previous version that always reads cache
 960          * shows that it makes things worse for diff-tree comparing
 961          * two linux-2.6 kernel trees in an already checked out work
 962          * tree.  This is because most diff-tree comparisons deal with
 963          * only a small number of files, while reading the cache is
 964          * expensive for a large project, and its cost outweighs the
 965          * savings we get by not inflating the object to a temporary
 966          * file.  Practically, this code only helps when we are used
 967          * by diff-cache --cached, which does read the cache before
 968          * calling us.
 969          */
 970         if (!active_cache)
 971                 return 0;
 972
 973         len = strlen(name);
 974         pos = cache_name_pos(name, len);
 975         if (pos < 0)
 976                 return 0;
 977         ce = active_cache[pos];
 978         if ((lstat(name, &st) < 0) ||
 979             !S_ISREG(st.st_mode) || /* careful! */
 980             ce_match_stat(ce, &st, 0) ||
 981             memcmp(sha1, ce->sha1, 20))
 982                 return 0;
 983         /* we return 1 only when we can stat, it is a regular file,
 984          * stat information matches, and sha1 recorded in the cache
 985          * matches.  I.e. we know the file in the work tree really is
 986          * the same as the <name, sha1> pair.
 987          */
 988         return 1;
 989 }
 990
 991 static struct sha1_size_cache {
 992         unsigned char sha1[20];
 993         unsigned long size;
 994 } **sha1_size_cache;
 995 static int sha1_size_cache_nr, sha1_size_cache_alloc;
 996
 997 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
 998                                                  int find_only,
 999                                                  unsigned long size)
1000 {
1001         int first, last;
1002         struct sha1_size_cache *e;
1003
1004         first = 0;
1005         last = sha1_size_cache_nr;
1006         while (last > first) {
1007                 int cmp, next = (last + first) >> 1;
1008                 e = sha1_size_cache[next];
1009                 cmp = memcmp(e->sha1, sha1, 20);
1010                 if (!cmp)
1011                         return e;
1012                 if (cmp < 0) {
1013                         last = next;
1014                         continue;
1015                 }
1016                 first = next+1;
1017         }
1018         /* not found */
1019         if (find_only)
1020                 return NULL;
1021         /* insert to make it at "first" */
1022         if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
1023                 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
1024                 sha1_size_cache = xrealloc(sha1_size_cache,
1025                                            sha1_size_cache_alloc *
1026                                            sizeof(*sha1_size_cache));
1027         }
1028         sha1_size_cache_nr++;
1029         if (first < sha1_size_cache_nr)
1030                 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
1031                         (sha1_size_cache_nr - first - 1) *
1032                         sizeof(*sha1_size_cache));
1033         e = xmalloc(sizeof(struct sha1_size_cache));
1034         sha1_size_cache[first] = e;
1035         memcpy(e->sha1, sha1, 20);
1036         e->size = size;
1037         return e;
1038 }
1039
1040 /*
1041  * While doing rename detection and pickaxe operation, we may need to
1042  * grab the data for the blob (or file) for our own in-core comparison.
1043  * diff_filespec has data and size fields for this purpose.
1044  */
1045 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1046 {
1047         int err = 0;
1048         if (!DIFF_FILE_VALID(s))
1049                 die("internal error: asking to populate invalid file.");
1050         if (S_ISDIR(s->mode))
1051                 return -1;
1052
1053         if (!use_size_cache)
1054                 size_only = 0;
1055
1056         if (s->data)
1057                 return err;
1058         if (!s->sha1_valid ||
1059             work_tree_matches(s->path, s->sha1)) {
1060                 struct stat st;
1061                 int fd;
1062                 if (lstat(s->path, &st) < 0) {
1063                         if (errno == ENOENT) {
1064                         err_empty:
1065                                 err = -1;
1066                         empty:
1067                                 s->data = (char *)"";
1068                                 s->size = 0;
1069                                 return err;
1070                         }
1071                 }
1072                 s->size = st.st_size;
1073                 if (!s->size)
1074                         goto empty;
1075                 if (size_only)
1076                         return 0;
1077                 if (S_ISLNK(st.st_mode)) {
1078                         int ret;
1079                         s->data = xmalloc(s->size);
1080                         s->should_free = 1;
1081                         ret = readlink(s->path, s->data, s->size);
1082                         if (ret < 0) {
1083                                 free(s->data);
1084                                 goto err_empty;
1085                         }
1086                         return 0;
1087                 }
1088                 fd = open(s->path, O_RDONLY);
1089                 if (fd < 0)
1090                         goto err_empty;
1091                 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1092                 close(fd);
1093                 if (s->data == MAP_FAILED)
1094                         goto err_empty;
1095                 s->should_munmap = 1;
1096         }
1097         else {
1098                 char type[20];
1099                 struct sha1_size_cache *e;
1100
1101                 if (size_only) {
1102                         e = locate_size_cache(s->sha1, 1, 0);
1103                         if (e) {
1104                                 s->size = e->size;
1105                                 return 0;
1106                         }
1107                         if (!sha1_object_info(s->sha1, type, &s->size))
1108                                 locate_size_cache(s->sha1, 0, s->size);
1109                 }
1110                 else {
1111                         s->data = read_sha1_file(s->sha1, type, &s->size);
1112                         s->should_free = 1;
1113                 }
1114         }
1115         return 0;
1116 }
1117
1118 void diff_free_filespec_data(struct diff_filespec *s)
1119 {
1120         if (s->should_free)
1121                 free(s->data);
1122         else if (s->should_munmap)
1123                 munmap(s->data, s->size);
1124         s->should_free = s->should_munmap = 0;
1125         s->data = NULL;
1126         free(s->cnt_data);
1127         s->cnt_data = NULL;
1128 }
1129
1130 static void prep_temp_blob(struct diff_tempfile *temp,
1131                            void *blob,
1132                            unsigned long size,
1133                            const unsigned char *sha1,
1134                            int mode)
1135 {
1136         int fd;
1137
1138         fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
1139         if (fd < 0)
1140                 die("unable to create temp-file");
1141         if (write(fd, blob, size) != size)
1142                 die("unable to write temp-file");
1143         close(fd);
1144         temp->name = temp->tmp_path;
1145         strcpy(temp->hex, sha1_to_hex(sha1));
1146         temp->hex[40] = 0;
1147         sprintf(temp->mode, "%06o", mode);
1148 }
1149
1150 static void prepare_temp_file(const char *name,
1151                               struct diff_tempfile *temp,
1152                               struct diff_filespec *one)
1153 {
1154         if (!DIFF_FILE_VALID(one)) {
1155         not_a_valid_file:
1156                 /* A '-' entry produces this for file-2, and
1157                  * a '+' entry produces this for file-1.
1158                  */
1159                 temp->name = "/dev/null";
1160                 strcpy(temp->hex, ".");
1161                 strcpy(temp->mode, ".");
1162                 return;
1163         }
1164
1165         if (!one->sha1_valid ||
1166             work_tree_matches(name, one->sha1)) {
1167                 struct stat st;
1168                 if (lstat(name, &st) < 0) {
1169                         if (errno == ENOENT)
1170                                 goto not_a_valid_file;
1171                         die("stat(%s): %s", name, strerror(errno));
1172                 }
1173                 if (S_ISLNK(st.st_mode)) {
1174                         int ret;
1175                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1176                         if (sizeof(buf) <= st.st_size)
1177                                 die("symlink too long: %s", name);
1178                         ret = readlink(name, buf, st.st_size);
1179                         if (ret < 0)
1180                                 die("readlink(%s)", name);
1181                         prep_temp_blob(temp, buf, st.st_size,
1182                                        (one->sha1_valid ?
1183                                         one->sha1 : null_sha1),
1184                                        (one->sha1_valid ?
1185                                         one->mode : S_IFLNK));
1186                 }
1187                 else {
1188                         /* we can borrow from the file in the work tree */
1189                         temp->name = name;
1190                         if (!one->sha1_valid)
1191                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
1192                         else
1193                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
1194                         /* Even though we may sometimes borrow the
1195                          * contents from the work tree, we always want
1196                          * one->mode.  mode is trustworthy even when
1197                          * !(one->sha1_valid), as long as
1198                          * DIFF_FILE_VALID(one).
1199                          */
1200                         sprintf(temp->mode, "%06o", one->mode);
1201                 }
1202                 return;
1203         }
1204         else {
1205                 if (diff_populate_filespec(one, 0))
1206                         die("cannot read data blob for %s", one->path);
1207                 prep_temp_blob(temp, one->data, one->size,
1208                                one->sha1, one->mode);
1209         }
1210 }
1211
1212 static void remove_tempfile(void)
1213 {
1214         int i;
1215
1216         for (i = 0; i < 2; i++)
1217                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1218                         unlink(diff_temp[i].name);
1219                         diff_temp[i].name = NULL;
1220                 }
1221 }
1222
1223 static void remove_tempfile_on_signal(int signo)
1224 {
1225         remove_tempfile();
1226         signal(SIGINT, SIG_DFL);
1227         raise(signo);
1228 }
1229
1230 static int spawn_prog(const char *pgm, const char **arg)
1231 {
1232         pid_t pid;
1233         int status;
1234
1235         fflush(NULL);
1236         pid = fork();
1237         if (pid < 0)
1238                 die("unable to fork");
1239         if (!pid) {
1240                 execvp(pgm, (char *const*) arg);
1241                 exit(255);
1242         }
1243
1244         while (waitpid(pid, &status, 0) < 0) {
1245                 if (errno == EINTR)
1246                         continue;
1247                 return -1;
1248         }
1249
1250         /* Earlier we did not check the exit status because
1251          * diff exits non-zero if files are different, and
1252          * we are not interested in knowing that.  It was a
1253          * mistake which made it harder to quit a diff-*
1254          * session that uses the git-apply-patch-script as
1255          * the GIT_EXTERNAL_DIFF.  A custom GIT_EXTERNAL_DIFF
1256          * should also exit non-zero only when it wants to
1257          * abort the entire diff-* session.
1258          */
1259         if (WIFEXITED(status) && !WEXITSTATUS(status))
1260                 return 0;
1261         return -1;
1262 }
1263
1264 /* An external diff command takes:
1265  *
1266  * diff-cmd name infile1 infile1-sha1 infile1-mode \
1267  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
1268  *
1269  */
1270 static void run_external_diff(const char *pgm,
1271                               const char *name,
1272                               const char *other,
1273                               struct diff_filespec *one,
1274                               struct diff_filespec *two,
1275                               const char *xfrm_msg,
1276                               int complete_rewrite)
1277 {
1278         const char *spawn_arg[10];
1279         struct diff_tempfile *temp = diff_temp;
1280         int retval;
1281         static int atexit_asked = 0;
1282         const char *othername;
1283         const char **arg = &spawn_arg[0];
1284
1285         othername = (other? other : name);
1286         if (one && two) {
1287                 prepare_temp_file(name, &temp[0], one);
1288                 prepare_temp_file(othername, &temp[1], two);
1289                 if (! atexit_asked &&
1290                     (temp[0].name == temp[0].tmp_path ||
1291                      temp[1].name == temp[1].tmp_path)) {
1292                         atexit_asked = 1;
1293                         atexit(remove_tempfile);
1294                 }
1295                 signal(SIGINT, remove_tempfile_on_signal);
1296         }
1297
1298         if (one && two) {
1299                 *arg++ = pgm;
1300                 *arg++ = name;
1301                 *arg++ = temp[0].name;
1302                 *arg++ = temp[0].hex;
1303                 *arg++ = temp[0].mode;
1304                 *arg++ = temp[1].name;
1305                 *arg++ = temp[1].hex;
1306                 *arg++ = temp[1].mode;
1307                 if (other) {
1308                         *arg++ = other;
1309                         *arg++ = xfrm_msg;
1310                 }
1311         } else {
1312                 *arg++ = pgm;
1313                 *arg++ = name;
1314         }
1315         *arg = NULL;
1316         retval = spawn_prog(pgm, spawn_arg);
1317         remove_tempfile();
1318         if (retval) {
1319                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1320                 exit(1);
1321         }
1322 }
1323
1324 static void run_diff_cmd(const char *pgm,
1325                          const char *name,
1326                          const char *other,
1327                          struct diff_filespec *one,
1328                          struct diff_filespec *two,
1329                          const char *xfrm_msg,
1330                          struct diff_options *o,
1331                          int complete_rewrite)
1332 {
1333         if (pgm) {
1334                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
1335                                   complete_rewrite);
1336                 return;
1337         }
1338         if (one && two)
1339                 builtin_diff(name, other ? other : name,
1340                              one, two, xfrm_msg, o, complete_rewrite);
1341         else
1342                 printf("* Unmerged path %s\n", name);
1343 }
1344
1345 static void diff_fill_sha1_info(struct diff_filespec *one)
1346 {
1347         if (DIFF_FILE_VALID(one)) {
1348                 if (!one->sha1_valid) {
1349                         struct stat st;
1350                         if (lstat(one->path, &st) < 0)
1351                                 die("stat %s", one->path);
1352                         if (index_path(one->sha1, one->path, &st, 0))
1353                                 die("cannot hash %s\n", one->path);
1354                 }
1355         }
1356         else
1357                 memset(one->sha1, 0, 20);
1358 }
1359
1360 static void run_diff(struct diff_filepair *p, struct diff_options *o)
1361 {
1362         const char *pgm = external_diff();
1363         char msg[PATH_MAX*2+300], *xfrm_msg;
1364         struct diff_filespec *one;
1365         struct diff_filespec *two;
1366         const char *name;
1367         const char *other;
1368         char *name_munged, *other_munged;
1369         int complete_rewrite = 0;
1370         int len;
1371
1372         if (DIFF_PAIR_UNMERGED(p)) {
1373                 /* unmerged */
1374                 run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, o, 0);
1375                 return;
1376         }
1377
1378         name = p->one->path;
1379         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1380         name_munged = quote_one(name);
1381         other_munged = quote_one(other);
1382         one = p->one; two = p->two;
1383
1384         diff_fill_sha1_info(one);
1385         diff_fill_sha1_info(two);
1386
1387         len = 0;
1388         switch (p->status) {
1389         case DIFF_STATUS_COPIED:
1390                 len += snprintf(msg + len, sizeof(msg) - len,
1391                                 "similarity index %d%%\n"
1392                                 "copy from %s\n"
1393                                 "copy to %s\n",
1394                                 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1395                                 name_munged, other_munged);
1396                 break;
1397         case DIFF_STATUS_RENAMED:
1398                 len += snprintf(msg + len, sizeof(msg) - len,
1399                                 "similarity index %d%%\n"
1400                                 "rename from %s\n"
1401                                 "rename to %s\n",
1402                                 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1403                                 name_munged, other_munged);
1404                 break;
1405         case DIFF_STATUS_MODIFIED:
1406                 if (p->score) {
1407                         len += snprintf(msg + len, sizeof(msg) - len,
1408                                         "dissimilarity index %d%%\n",
1409                                         (int)(0.5 + p->score *
1410                                               100.0/MAX_SCORE));
1411                         complete_rewrite = 1;
1412                         break;
1413                 }
1414                 /* fallthru */
1415         default:
1416                 /* nothing */
1417                 ;
1418         }
1419
1420         if (memcmp(one->sha1, two->sha1, 20)) {
1421                 int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
1422
1423                 len += snprintf(msg + len, sizeof(msg) - len,
1424                                 "index %.*s..%.*s",
1425                                 abbrev, sha1_to_hex(one->sha1),
1426                                 abbrev, sha1_to_hex(two->sha1));
1427                 if (one->mode == two->mode)
1428                         len += snprintf(msg + len, sizeof(msg) - len,
1429                                         " %06o", one->mode);
1430                 len += snprintf(msg + len, sizeof(msg) - len, "\n");
1431         }
1432
1433         if (len)
1434                 msg[--len] = 0;
1435         xfrm_msg = len ? msg : NULL;
1436
1437         if (!pgm &&
1438             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
1439             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
1440                 /* a filepair that changes between file and symlink
1441                  * needs to be split into deletion and creation.
1442                  */
1443                 struct diff_filespec *null = alloc_filespec(two->path);
1444                 run_diff_cmd(NULL, name, other, one, null, xfrm_msg, o, 0);
1445                 free(null);
1446                 null = alloc_filespec(one->path);
1447                 run_diff_cmd(NULL, name, other, null, two, xfrm_msg, o, 0);
1448                 free(null);
1449         }
1450         else
1451                 run_diff_cmd(pgm, name, other, one, two, xfrm_msg, o,
1452                              complete_rewrite);
1453
1454         free(name_munged);
1455         free(other_munged);
1456 }
1457
1458 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
1459                          struct diffstat_t *diffstat)
1460 {
1461         const char *name;
1462         const char *other;
1463         int complete_rewrite = 0;
1464
1465         if (DIFF_PAIR_UNMERGED(p)) {
1466                 /* unmerged */
1467                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
1468                 return;
1469         }
1470
1471         name = p->one->path;
1472         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1473
1474         diff_fill_sha1_info(p->one);
1475         diff_fill_sha1_info(p->two);
1476
1477         if (p->status == DIFF_STATUS_MODIFIED && p->score)
1478                 complete_rewrite = 1;
1479         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
1480 }
1481
1482 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
1483 {
1484         const char *name;
1485         const char *other;
1486
1487         if (DIFF_PAIR_UNMERGED(p)) {
1488                 /* unmerged */
1489                 return;
1490         }
1491
1492         name = p->one->path;
1493         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1494
1495         diff_fill_sha1_info(p->one);
1496         diff_fill_sha1_info(p->two);
1497
1498         builtin_checkdiff(name, other, p->one, p->two);
1499 }
1500
1501 void diff_setup(struct diff_options *options)
1502 {
1503         memset(options, 0, sizeof(*options));
1504         options->line_termination = '\n';
1505         options->break_opt = -1;
1506         options->rename_limit = -1;
1507         options->context = 3;
1508         options->msg_sep = "";
1509
1510         options->change = diff_change;
1511         options->add_remove = diff_addremove;
1512         options->color_diff = diff_use_color_default;
1513         options->detect_rename = diff_detect_rename_default;
1514 }
1515
1516 int diff_setup_done(struct diff_options *options)
1517 {
1518         if (options->find_copies_harder)
1519                 options->detect_rename = DIFF_DETECT_COPY;
1520
1521         if (options->output_format & (DIFF_FORMAT_NAME |
1522                                       DIFF_FORMAT_NAME_STATUS |
1523                                       DIFF_FORMAT_CHECKDIFF |
1524                                       DIFF_FORMAT_NO_OUTPUT))
1525                 options->output_format &= ~(DIFF_FORMAT_RAW |
1526                                             DIFF_FORMAT_DIFFSTAT |
1527                                             DIFF_FORMAT_SUMMARY |
1528                                             DIFF_FORMAT_PATCH);
1529
1530         /*
1531          * These cases always need recursive; we do not drop caller-supplied
1532          * recursive bits for other formats here.
1533          */
1534         if (options->output_format & (DIFF_FORMAT_PATCH |
1535                                       DIFF_FORMAT_DIFFSTAT |
1536                                       DIFF_FORMAT_CHECKDIFF))
1537                 options->recursive = 1;
1538         /*
1539          * Also pickaxe would not work very well if you do not say recursive
1540          */
1541         if (options->pickaxe)
1542                 options->recursive = 1;
1543
1544         if (options->detect_rename && options->rename_limit < 0)
1545                 options->rename_limit = diff_rename_limit_default;
1546         if (options->setup & DIFF_SETUP_USE_CACHE) {
1547                 if (!active_cache)
1548                         /* read-cache does not die even when it fails
1549                          * so it is safe for us to do this here.  Also
1550                          * it does not smudge active_cache or active_nr
1551                          * when it fails, so we do not have to worry about
1552                          * cleaning it up ourselves either.
1553                          */
1554                         read_cache();
1555         }
1556         if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
1557                 use_size_cache = 1;
1558         if (options->abbrev <= 0 || 40 < options->abbrev)
1559                 options->abbrev = 40; /* full */
1560
1561         return 0;
1562 }
1563
1564 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
1565 {
1566         char c, *eq;
1567         int len;
1568
1569         if (*arg != '-')
1570                 return 0;
1571         c = *++arg;
1572         if (!c)
1573                 return 0;
1574         if (c == arg_short) {
1575                 c = *++arg;
1576                 if (!c)
1577                         return 1;
1578                 if (val && isdigit(c)) {
1579                         char *end;
1580                         int n = strtoul(arg, &end, 10);
1581                         if (*end)
1582                                 return 0;
1583                         *val = n;
1584                         return 1;
1585                 }
1586                 return 0;
1587         }
1588         if (c != '-')
1589                 return 0;
1590         arg++;
1591         eq = strchr(arg, '=');
1592         if (eq)
1593                 len = eq - arg;
1594         else
1595                 len = strlen(arg);
1596         if (!len || strncmp(arg, arg_long, len))
1597                 return 0;
1598         if (eq) {
1599                 int n;
1600                 char *end;
1601                 if (!isdigit(*++eq))
1602                         return 0;
1603                 n = strtoul(eq, &end, 10);
1604                 if (*end)
1605                         return 0;
1606                 *val = n;
1607         }
1608         return 1;
1609 }
1610
1611 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
1612 {
1613         const char *arg = av[0];
1614         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
1615                 options->output_format |= DIFF_FORMAT_PATCH;
1616         else if (opt_arg(arg, 'U', "unified", &options->context))
1617                 options->output_format |= DIFF_FORMAT_PATCH;
1618         else if (!strcmp(arg, "--raw"))
1619                 options->output_format |= DIFF_FORMAT_RAW;
1620         else if (!strcmp(arg, "--patch-with-raw")) {
1621                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
1622         }
1623         else if (!strcmp(arg, "--stat"))
1624                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
1625         else if (!strcmp(arg, "--check"))
1626                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
1627         else if (!strcmp(arg, "--summary"))
1628                 options->output_format |= DIFF_FORMAT_SUMMARY;
1629         else if (!strcmp(arg, "--patch-with-stat")) {
1630                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
1631         }
1632         else if (!strcmp(arg, "-z"))
1633                 options->line_termination = 0;
1634         else if (!strncmp(arg, "-l", 2))
1635                 options->rename_limit = strtoul(arg+2, NULL, 10);
1636         else if (!strcmp(arg, "--full-index"))
1637                 options->full_index = 1;
1638         else if (!strcmp(arg, "--binary")) {
1639                 options->output_format |= DIFF_FORMAT_PATCH;
1640                 options->full_index = options->binary = 1;
1641         }
1642         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text")) {
1643                 options->text = 1;
1644         }
1645         else if (!strcmp(arg, "--name-only"))
1646                 options->output_format |= DIFF_FORMAT_NAME;
1647         else if (!strcmp(arg, "--name-status"))
1648                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
1649         else if (!strcmp(arg, "-R"))
1650                 options->reverse_diff = 1;
1651         else if (!strncmp(arg, "-S", 2))
1652                 options->pickaxe = arg + 2;
1653         else if (!strcmp(arg, "-s")) {
1654                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
1655         }
1656         else if (!strncmp(arg, "-O", 2))
1657                 options->orderfile = arg + 2;
1658         else if (!strncmp(arg, "--diff-filter=", 14))
1659                 options->filter = arg + 14;
1660         else if (!strcmp(arg, "--pickaxe-all"))
1661                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
1662         else if (!strcmp(arg, "--pickaxe-regex"))
1663                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
1664         else if (!strncmp(arg, "-B", 2)) {
1665                 if ((options->break_opt =
1666                      diff_scoreopt_parse(arg)) == -1)
1667                         return -1;
1668         }
1669         else if (!strncmp(arg, "-M", 2)) {
1670                 if ((options->rename_score =
1671                      diff_scoreopt_parse(arg)) == -1)
1672                         return -1;
1673                 options->detect_rename = DIFF_DETECT_RENAME;
1674         }
1675         else if (!strncmp(arg, "-C", 2)) {
1676                 if ((options->rename_score =
1677                      diff_scoreopt_parse(arg)) == -1)
1678                         return -1;
1679                 options->detect_rename = DIFF_DETECT_COPY;
1680         }
1681         else if (!strcmp(arg, "--find-copies-harder"))
1682                 options->find_copies_harder = 1;
1683         else if (!strcmp(arg, "--abbrev"))
1684                 options->abbrev = DEFAULT_ABBREV;
1685         else if (!strncmp(arg, "--abbrev=", 9)) {
1686                 options->abbrev = strtoul(arg + 9, NULL, 10);
1687                 if (options->abbrev < MINIMUM_ABBREV)
1688                         options->abbrev = MINIMUM_ABBREV;
1689                 else if (40 < options->abbrev)
1690                         options->abbrev = 40;
1691         }
1692         else if (!strcmp(arg, "--color"))
1693                 options->color_diff = 1;
1694         else if (!strcmp(arg, "--no-color"))
1695                 options->color_diff = 0;
1696         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
1697                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
1698         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
1699                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
1700         else if (!strcmp(arg, "--no-renames"))
1701                 options->detect_rename = 0;
1702         else
1703                 return 0;
1704         return 1;
1705 }
1706
1707 static int parse_num(const char **cp_p)
1708 {
1709         unsigned long num, scale;
1710         int ch, dot;
1711         const char *cp = *cp_p;
1712
1713         num = 0;
1714         scale = 1;
1715         dot = 0;
1716         for(;;) {
1717                 ch = *cp;
1718                 if ( !dot && ch == '.' ) {
1719                         scale = 1;
1720                         dot = 1;
1721                 } else if ( ch == '%' ) {
1722                         scale = dot ? scale*100 : 100;
1723                         cp++;   /* % is always at the end */
1724                         break;
1725                 } else if ( ch >= '0' && ch <= '9' ) {
1726                         if ( scale < 100000 ) {
1727                                 scale *= 10;
1728                                 num = (num*10) + (ch-'0');
1729                         }
1730                 } else {
1731                         break;
1732                 }
1733                 cp++;
1734         }
1735         *cp_p = cp;
1736
1737         /* user says num divided by scale and we say internally that
1738          * is MAX_SCORE * num / scale.
1739          */
1740         return (num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale);
1741 }
1742
1743 int diff_scoreopt_parse(const char *opt)
1744 {
1745         int opt1, opt2, cmd;
1746
1747         if (*opt++ != '-')
1748                 return -1;
1749         cmd = *opt++;
1750         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
1751                 return -1; /* that is not a -M, -C nor -B option */
1752
1753         opt1 = parse_num(&opt);
1754         if (cmd != 'B')
1755                 opt2 = 0;
1756         else {
1757                 if (*opt == 0)
1758                         opt2 = 0;
1759                 else if (*opt != '/')
1760                         return -1; /* we expect -B80/99 or -B80 */
1761                 else {
1762                         opt++;
1763                         opt2 = parse_num(&opt);
1764                 }
1765         }
1766         if (*opt != 0)
1767                 return -1;
1768         return opt1 | (opt2 << 16);
1769 }
1770
1771 struct diff_queue_struct diff_queued_diff;
1772
1773 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
1774 {
1775         if (queue->alloc <= queue->nr) {
1776                 queue->alloc = alloc_nr(queue->alloc);
1777                 queue->queue = xrealloc(queue->queue,
1778                                         sizeof(dp) * queue->alloc);
1779         }
1780         queue->queue[queue->nr++] = dp;
1781 }
1782
1783 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
1784                                  struct diff_filespec *one,
1785                                  struct diff_filespec *two)
1786 {
1787         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
1788         dp->one = one;
1789         dp->two = two;
1790         if (queue)
1791                 diff_q(queue, dp);
1792         return dp;
1793 }
1794
1795 void diff_free_filepair(struct diff_filepair *p)
1796 {
1797         diff_free_filespec_data(p->one);
1798         diff_free_filespec_data(p->two);
1799         free(p->one);
1800         free(p->two);
1801         free(p);
1802 }
1803
1804 /* This is different from find_unique_abbrev() in that
1805  * it stuffs the result with dots for alignment.
1806  */
1807 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
1808 {
1809         int abblen;
1810         const char *abbrev;
1811         if (len == 40)
1812                 return sha1_to_hex(sha1);
1813
1814         abbrev = find_unique_abbrev(sha1, len);
1815         if (!abbrev)
1816                 return sha1_to_hex(sha1);
1817         abblen = strlen(abbrev);
1818         if (abblen < 37) {
1819                 static char hex[41];
1820                 if (len < abblen && abblen <= len + 2)
1821                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
1822                 else
1823                         sprintf(hex, "%s...", abbrev);
1824                 return hex;
1825         }
1826         return sha1_to_hex(sha1);
1827 }
1828
1829 static void diff_flush_raw(struct diff_filepair *p,
1830                            struct diff_options *options)
1831 {
1832         int two_paths;
1833         char status[10];
1834         int abbrev = options->abbrev;
1835         const char *path_one, *path_two;
1836         int inter_name_termination = '\t';
1837         int line_termination = options->line_termination;
1838
1839         if (!line_termination)
1840                 inter_name_termination = 0;
1841
1842         path_one = p->one->path;
1843         path_two = p->two->path;
1844         if (line_termination) {
1845                 path_one = quote_one(path_one);
1846                 path_two = quote_one(path_two);
1847         }
1848
1849         if (p->score)
1850                 sprintf(status, "%c%03d", p->status,
1851                         (int)(0.5 + p->score * 100.0/MAX_SCORE));
1852         else {
1853                 status[0] = p->status;
1854                 status[1] = 0;
1855         }
1856         switch (p->status) {
1857         case DIFF_STATUS_COPIED:
1858         case DIFF_STATUS_RENAMED:
1859                 two_paths = 1;
1860                 break;
1861         case DIFF_STATUS_ADDED:
1862         case DIFF_STATUS_DELETED:
1863                 two_paths = 0;
1864                 break;
1865         default:
1866                 two_paths = 0;
1867                 break;
1868         }
1869         if (!(options->output_format & DIFF_FORMAT_NAME_STATUS)) {
1870                 printf(":%06o %06o %s ",
1871                        p->one->mode, p->two->mode,
1872                        diff_unique_abbrev(p->one->sha1, abbrev));
1873                 printf("%s ",
1874                        diff_unique_abbrev(p->two->sha1, abbrev));
1875         }
1876         printf("%s%c%s", status, inter_name_termination, path_one);
1877         if (two_paths)
1878                 printf("%c%s", inter_name_termination, path_two);
1879         putchar(line_termination);
1880         if (path_one != p->one->path)
1881                 free((void*)path_one);
1882         if (path_two != p->two->path)
1883                 free((void*)path_two);
1884 }
1885
1886 static void diff_flush_name(struct diff_filepair *p, int line_termination)
1887 {
1888         char *path = p->two->path;
1889
1890         if (line_termination)
1891                 path = quote_one(p->two->path);
1892         printf("%s%c", path, line_termination);
1893         if (p->two->path != path)
1894                 free(path);
1895 }
1896
1897 int diff_unmodified_pair(struct diff_filepair *p)
1898 {
1899         /* This function is written stricter than necessary to support
1900          * the currently implemented transformers, but the idea is to
1901          * let transformers to produce diff_filepairs any way they want,
1902          * and filter and clean them up here before producing the output.
1903          */
1904         struct diff_filespec *one, *two;
1905
1906         if (DIFF_PAIR_UNMERGED(p))
1907                 return 0; /* unmerged is interesting */
1908
1909         one = p->one;
1910         two = p->two;
1911
1912         /* deletion, addition, mode or type change
1913          * and rename are all interesting.
1914          */
1915         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
1916             DIFF_PAIR_MODE_CHANGED(p) ||
1917             strcmp(one->path, two->path))
1918                 return 0;
1919
1920         /* both are valid and point at the same path.  that is, we are
1921          * dealing with a change.
1922          */
1923         if (one->sha1_valid && two->sha1_valid &&
1924             !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
1925                 return 1; /* no change */
1926         if (!one->sha1_valid && !two->sha1_valid)
1927                 return 1; /* both look at the same file on the filesystem. */
1928         return 0;
1929 }
1930
1931 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
1932 {
1933         if (diff_unmodified_pair(p))
1934                 return;
1935
1936         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1937             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1938                 return; /* no tree diffs in patch format */
1939
1940         run_diff(p, o);
1941 }
1942
1943 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
1944                             struct diffstat_t *diffstat)
1945 {
1946         if (diff_unmodified_pair(p))
1947                 return;
1948
1949         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1950             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1951                 return; /* no tree diffs in patch format */
1952
1953         run_diffstat(p, o, diffstat);
1954 }
1955
1956 static void diff_flush_checkdiff(struct diff_filepair *p,
1957                 struct diff_options *o)
1958 {
1959         if (diff_unmodified_pair(p))
1960                 return;
1961
1962         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1963             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1964                 return; /* no tree diffs in patch format */
1965
1966         run_checkdiff(p, o);
1967 }
1968
1969 int diff_queue_is_empty(void)
1970 {
1971         struct diff_queue_struct *q = &diff_queued_diff;
1972         int i;
1973         for (i = 0; i < q->nr; i++)
1974                 if (!diff_unmodified_pair(q->queue[i]))
1975                         return 0;
1976         return 1;
1977 }
1978
1979 #if DIFF_DEBUG
1980 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
1981 {
1982         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
1983                 x, one ? one : "",
1984                 s->path,
1985                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
1986                 s->mode,
1987                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
1988         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
1989                 x, one ? one : "",
1990                 s->size, s->xfrm_flags);
1991 }
1992
1993 void diff_debug_filepair(const struct diff_filepair *p, int i)
1994 {
1995         diff_debug_filespec(p->one, i, "one");
1996         diff_debug_filespec(p->two, i, "two");
1997         fprintf(stderr, "score %d, status %c stays %d broken %d\n",
1998                 p->score, p->status ? p->status : '?',
1999                 p->source_stays, p->broken_pair);
2000 }
2001
2002 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2003 {
2004         int i;
2005         if (msg)
2006                 fprintf(stderr, "%s\n", msg);
2007         fprintf(stderr, "q->nr = %d\n", q->nr);
2008         for (i = 0; i < q->nr; i++) {
2009                 struct diff_filepair *p = q->queue[i];
2010                 diff_debug_filepair(p, i);
2011         }
2012 }
2013 #endif
2014
2015 static void diff_resolve_rename_copy(void)
2016 {
2017         int i, j;
2018         struct diff_filepair *p, *pp;
2019         struct diff_queue_struct *q = &diff_queued_diff;
2020
2021         diff_debug_queue("resolve-rename-copy", q);
2022
2023         for (i = 0; i < q->nr; i++) {
2024                 p = q->queue[i];
2025                 p->status = 0; /* undecided */
2026                 if (DIFF_PAIR_UNMERGED(p))
2027                         p->status = DIFF_STATUS_UNMERGED;
2028                 else if (!DIFF_FILE_VALID(p->one))
2029                         p->status = DIFF_STATUS_ADDED;
2030                 else if (!DIFF_FILE_VALID(p->two))
2031                         p->status = DIFF_STATUS_DELETED;
2032                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2033                         p->status = DIFF_STATUS_TYPE_CHANGED;
2034
2035                 /* from this point on, we are dealing with a pair
2036                  * whose both sides are valid and of the same type, i.e.
2037                  * either in-place edit or rename/copy edit.
2038                  */
2039                 else if (DIFF_PAIR_RENAME(p)) {
2040                         if (p->source_stays) {
2041                                 p->status = DIFF_STATUS_COPIED;
2042                                 continue;
2043                         }
2044                         /* See if there is some other filepair that
2045                          * copies from the same source as us.  If so
2046                          * we are a copy.  Otherwise we are either a
2047                          * copy if the path stays, or a rename if it
2048                          * does not, but we already handled "stays" case.
2049                          */
2050                         for (j = i + 1; j < q->nr; j++) {
2051                                 pp = q->queue[j];
2052                                 if (strcmp(pp->one->path, p->one->path))
2053                                         continue; /* not us */
2054                                 if (!DIFF_PAIR_RENAME(pp))
2055                                         continue; /* not a rename/copy */
2056                                 /* pp is a rename/copy from the same source */
2057                                 p->status = DIFF_STATUS_COPIED;
2058                                 break;
2059                         }
2060                         if (!p->status)
2061                                 p->status = DIFF_STATUS_RENAMED;
2062                 }
2063                 else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
2064                          p->one->mode != p->two->mode)
2065                         p->status = DIFF_STATUS_MODIFIED;
2066                 else {
2067                         /* This is a "no-change" entry and should not
2068                          * happen anymore, but prepare for broken callers.
2069                          */
2070                         error("feeding unmodified %s to diffcore",
2071                               p->one->path);
2072                         p->status = DIFF_STATUS_UNKNOWN;
2073                 }
2074         }
2075         diff_debug_queue("resolve-rename-copy done", q);
2076 }
2077
2078 static int check_pair_status(struct diff_filepair *p)
2079 {
2080         switch (p->status) {
2081         case DIFF_STATUS_UNKNOWN:
2082                 return 0;
2083         case 0:
2084                 die("internal error in diff-resolve-rename-copy");
2085         default:
2086                 return 1;
2087         }
2088 }
2089
2090 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2091 {
2092         int fmt = opt->output_format;
2093
2094         if (fmt & DIFF_FORMAT_CHECKDIFF)
2095                 diff_flush_checkdiff(p, opt);
2096         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2097                 diff_flush_raw(p, opt);
2098         else if (fmt & DIFF_FORMAT_NAME)
2099                 diff_flush_name(p, opt->line_termination);
2100 }
2101
2102 static void show_file_mode_name(const char *newdelete, struct diff_filespec *fs)
2103 {
2104         if (fs->mode)
2105                 printf(" %s mode %06o %s\n", newdelete, fs->mode, fs->path);
2106         else
2107                 printf(" %s %s\n", newdelete, fs->path);
2108 }
2109
2110
2111 static void show_mode_change(struct diff_filepair *p, int show_name)
2112 {
2113         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2114                 if (show_name)
2115                         printf(" mode change %06o => %06o %s\n",
2116                                p->one->mode, p->two->mode, p->two->path);
2117                 else
2118                         printf(" mode change %06o => %06o\n",
2119                                p->one->mode, p->two->mode);
2120         }
2121 }
2122
2123 static void show_rename_copy(const char *renamecopy, struct diff_filepair *p)
2124 {
2125         const char *old, *new;
2126
2127         /* Find common prefix */
2128         old = p->one->path;
2129         new = p->two->path;
2130         while (1) {
2131                 const char *slash_old, *slash_new;
2132                 slash_old = strchr(old, '/');
2133                 slash_new = strchr(new, '/');
2134                 if (!slash_old ||
2135                     !slash_new ||
2136                     slash_old - old != slash_new - new ||
2137                     memcmp(old, new, slash_new - new))
2138                         break;
2139                 old = slash_old + 1;
2140                 new = slash_new + 1;
2141         }
2142         /* p->one->path thru old is the common prefix, and old and new
2143          * through the end of names are renames
2144          */
2145         if (old != p->one->path)
2146                 printf(" %s %.*s{%s => %s} (%d%%)\n", renamecopy,
2147                        (int)(old - p->one->path), p->one->path,
2148                        old, new, (int)(0.5 + p->score * 100.0/MAX_SCORE));
2149         else
2150                 printf(" %s %s => %s (%d%%)\n", renamecopy,
2151                        p->one->path, p->two->path,
2152                        (int)(0.5 + p->score * 100.0/MAX_SCORE));
2153         show_mode_change(p, 0);
2154 }
2155
2156 static void diff_summary(struct diff_filepair *p)
2157 {
2158         switch(p->status) {
2159         case DIFF_STATUS_DELETED:
2160                 show_file_mode_name("delete", p->one);
2161                 break;
2162         case DIFF_STATUS_ADDED:
2163                 show_file_mode_name("create", p->two);
2164                 break;
2165         case DIFF_STATUS_COPIED:
2166                 show_rename_copy("copy", p);
2167                 break;
2168         case DIFF_STATUS_RENAMED:
2169                 show_rename_copy("rename", p);
2170                 break;
2171         default:
2172                 if (p->score) {
2173                         printf(" rewrite %s (%d%%)\n", p->two->path,
2174                                 (int)(0.5 + p->score * 100.0/MAX_SCORE));
2175                         show_mode_change(p, 0);
2176                 } else  show_mode_change(p, 1);
2177                 break;
2178         }
2179 }
2180
2181 struct patch_id_t {
2182         struct xdiff_emit_state xm;
2183         SHA_CTX *ctx;
2184         int patchlen;
2185 };
2186
2187 static int remove_space(char *line, int len)
2188 {
2189         int i;
2190         char *dst = line;
2191         unsigned char c;
2192
2193         for (i = 0; i < len; i++)
2194                 if (!isspace((c = line[i])))
2195                         *dst++ = c;
2196
2197         return dst - line;
2198 }
2199
2200 static void patch_id_consume(void *priv, char *line, unsigned long len)
2201 {
2202         struct patch_id_t *data = priv;
2203         int new_len;
2204
2205         /* Ignore line numbers when computing the SHA1 of the patch */
2206         if (!strncmp(line, "@@ -", 4))
2207                 return;
2208
2209         new_len = remove_space(line, len);
2210
2211         SHA1_Update(data->ctx, line, new_len);
2212         data->patchlen += new_len;
2213 }
2214
2215 /* returns 0 upon success, and writes result into sha1 */
2216 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2217 {
2218         struct diff_queue_struct *q = &diff_queued_diff;
2219         int i;
2220         SHA_CTX ctx;
2221         struct patch_id_t data;
2222         char buffer[PATH_MAX * 4 + 20];
2223
2224         SHA1_Init(&ctx);
2225         memset(&data, 0, sizeof(struct patch_id_t));
2226         data.ctx = &ctx;
2227         data.xm.consume = patch_id_consume;
2228
2229         for (i = 0; i < q->nr; i++) {
2230                 xpparam_t xpp;
2231                 xdemitconf_t xecfg;
2232                 xdemitcb_t ecb;
2233                 mmfile_t mf1, mf2;
2234                 struct diff_filepair *p = q->queue[i];
2235                 int len1, len2;
2236
2237                 if (p->status == 0)
2238                         return error("internal diff status error");
2239                 if (p->status == DIFF_STATUS_UNKNOWN)
2240                         continue;
2241                 if (diff_unmodified_pair(p))
2242                         continue;
2243                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2244                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2245                         continue;
2246                 if (DIFF_PAIR_UNMERGED(p))
2247                         continue;
2248
2249                 diff_fill_sha1_info(p->one);
2250                 diff_fill_sha1_info(p->two);
2251                 if (fill_mmfile(&mf1, p->one) < 0 ||
2252                                 fill_mmfile(&mf2, p->two) < 0)
2253                         return error("unable to read files to diff");
2254
2255                 /* Maybe hash p->two? into the patch id? */
2256                 if (mmfile_is_binary(&mf2))
2257                         continue;
2258
2259                 len1 = remove_space(p->one->path, strlen(p->one->path));
2260                 len2 = remove_space(p->two->path, strlen(p->two->path));
2261                 if (p->one->mode == 0)
2262                         len1 = snprintf(buffer, sizeof(buffer),
2263                                         "diff--gita/%.*sb/%.*s"
2264                                         "newfilemode%06o"
2265                                         "---/dev/null"
2266                                         "+++b/%.*s",
2267                                         len1, p->one->path,
2268                                         len2, p->two->path,
2269                                         p->two->mode,
2270                                         len2, p->two->path);
2271                 else if (p->two->mode == 0)
2272                         len1 = snprintf(buffer, sizeof(buffer),
2273                                         "diff--gita/%.*sb/%.*s"
2274                                         "deletedfilemode%06o"
2275                                         "---a/%.*s"
2276                                         "+++/dev/null",
2277                                         len1, p->one->path,
2278                                         len2, p->two->path,
2279                                         p->one->mode,
2280                                         len1, p->one->path);
2281                 else
2282                         len1 = snprintf(buffer, sizeof(buffer),
2283                                         "diff--gita/%.*sb/%.*s"
2284                                         "---a/%.*s"
2285                                         "+++b/%.*s",
2286                                         len1, p->one->path,
2287                                         len2, p->two->path,
2288                                         len1, p->one->path,
2289                                         len2, p->two->path);
2290                 SHA1_Update(&ctx, buffer, len1);
2291
2292                 xpp.flags = XDF_NEED_MINIMAL;
2293                 xecfg.ctxlen = 3;
2294                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2295                 ecb.outf = xdiff_outf;
2296                 ecb.priv = &data;
2297                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
2298         }
2299
2300         SHA1_Final(sha1, &ctx);
2301         return 0;
2302 }
2303
2304 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
2305 {
2306         struct diff_queue_struct *q = &diff_queued_diff;
2307         int i;
2308         int result = diff_get_patch_id(options, sha1);
2309
2310         for (i = 0; i < q->nr; i++)
2311                 diff_free_filepair(q->queue[i]);
2312
2313         free(q->queue);
2314         q->queue = NULL;
2315         q->nr = q->alloc = 0;
2316
2317         return result;
2318 }
2319
2320 static int is_summary_empty(const struct diff_queue_struct *q)
2321 {
2322         int i;
2323
2324         for (i = 0; i < q->nr; i++) {
2325                 const struct diff_filepair *p = q->queue[i];
2326
2327                 switch (p->status) {
2328                 case DIFF_STATUS_DELETED:
2329                 case DIFF_STATUS_ADDED:
2330                 case DIFF_STATUS_COPIED:
2331                 case DIFF_STATUS_RENAMED:
2332                         return 0;
2333                 default:
2334                         if (p->score)
2335                                 return 0;
2336                         if (p->one->mode && p->two->mode &&
2337                             p->one->mode != p->two->mode)
2338                                 return 0;
2339                         break;
2340                 }
2341         }
2342         return 1;
2343 }
2344
2345 void diff_flush(struct diff_options *options)
2346 {
2347         struct diff_queue_struct *q = &diff_queued_diff;
2348         int i, output_format = options->output_format;
2349         int separator = 0;
2350
2351         /*
2352          * Order: raw, stat, summary, patch
2353          * or:    name/name-status/checkdiff (other bits clear)
2354          */
2355         if (!q->nr)
2356                 goto free_queue;
2357
2358         if (output_format & (DIFF_FORMAT_RAW |
2359                              DIFF_FORMAT_NAME |
2360                              DIFF_FORMAT_NAME_STATUS |
2361                              DIFF_FORMAT_CHECKDIFF)) {
2362                 for (i = 0; i < q->nr; i++) {
2363                         struct diff_filepair *p = q->queue[i];
2364                         if (check_pair_status(p))
2365                                 flush_one_pair(p, options);
2366                 }
2367                 separator++;
2368         }
2369
2370         if (output_format & DIFF_FORMAT_DIFFSTAT) {
2371                 struct diffstat_t diffstat;
2372
2373                 memset(&diffstat, 0, sizeof(struct diffstat_t));
2374                 diffstat.xm.consume = diffstat_consume;
2375                 for (i = 0; i < q->nr; i++) {
2376                         struct diff_filepair *p = q->queue[i];
2377                         if (check_pair_status(p))
2378                                 diff_flush_stat(p, options, &diffstat);
2379                 }
2380                 show_stats(&diffstat);
2381                 separator++;
2382         }
2383
2384         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
2385                 for (i = 0; i < q->nr; i++)
2386                         diff_summary(q->queue[i]);
2387                 separator++;
2388         }
2389
2390         if (output_format & DIFF_FORMAT_PATCH) {
2391                 if (separator) {
2392                         if (options->stat_sep) {
2393                                 /* attach patch instead of inline */
2394                                 fputs(options->stat_sep, stdout);
2395                         } else {
2396                                 putchar(options->line_termination);
2397                         }
2398                 }
2399
2400                 for (i = 0; i < q->nr; i++) {
2401                         struct diff_filepair *p = q->queue[i];
2402                         if (check_pair_status(p))
2403                                 diff_flush_patch(p, options);
2404                 }
2405         }
2406
2407         for (i = 0; i < q->nr; i++)
2408                 diff_free_filepair(q->queue[i]);
2409 free_queue:
2410         free(q->queue);
2411         q->queue = NULL;
2412         q->nr = q->alloc = 0;
2413 }
2414
2415 static void diffcore_apply_filter(const char *filter)
2416 {
2417         int i;
2418         struct diff_queue_struct *q = &diff_queued_diff;
2419         struct diff_queue_struct outq;
2420         outq.queue = NULL;
2421         outq.nr = outq.alloc = 0;
2422
2423         if (!filter)
2424                 return;
2425
2426         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
2427                 int found;
2428                 for (i = found = 0; !found && i < q->nr; i++) {
2429                         struct diff_filepair *p = q->queue[i];
2430                         if (((p->status == DIFF_STATUS_MODIFIED) &&
2431                              ((p->score &&
2432                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
2433                               (!p->score &&
2434                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
2435                             ((p->status != DIFF_STATUS_MODIFIED) &&
2436                              strchr(filter, p->status)))
2437                                 found++;
2438                 }
2439                 if (found)
2440                         return;
2441
2442                 /* otherwise we will clear the whole queue
2443                  * by copying the empty outq at the end of this
2444                  * function, but first clear the current entries
2445                  * in the queue.
2446                  */
2447                 for (i = 0; i < q->nr; i++)
2448                         diff_free_filepair(q->queue[i]);
2449         }
2450         else {
2451                 /* Only the matching ones */
2452                 for (i = 0; i < q->nr; i++) {
2453                         struct diff_filepair *p = q->queue[i];
2454
2455                         if (((p->status == DIFF_STATUS_MODIFIED) &&
2456                              ((p->score &&
2457                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
2458                               (!p->score &&
2459                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
2460                             ((p->status != DIFF_STATUS_MODIFIED) &&
2461                              strchr(filter, p->status)))
2462                                 diff_q(&outq, p);
2463                         else
2464                                 diff_free_filepair(p);
2465                 }
2466         }
2467         free(q->queue);
2468         *q = outq;
2469 }
2470
2471 void diffcore_std(struct diff_options *options)
2472 {
2473         if (options->break_opt != -1)
2474                 diffcore_break(options->break_opt);
2475         if (options->detect_rename)
2476                 diffcore_rename(options);
2477         if (options->break_opt != -1)
2478                 diffcore_merge_broken();
2479         if (options->pickaxe)
2480                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
2481         if (options->orderfile)
2482                 diffcore_order(options->orderfile);
2483         diff_resolve_rename_copy();
2484         diffcore_apply_filter(options->filter);
2485 }
2486
2487
2488 void diffcore_std_no_resolve(struct diff_options *options)
2489 {
2490         if (options->pickaxe)
2491                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
2492         if (options->orderfile)
2493                 diffcore_order(options->orderfile);
2494         diffcore_apply_filter(options->filter);
2495 }
2496
2497 void diff_addremove(struct diff_options *options,
2498                     int addremove, unsigned mode,
2499                     const unsigned char *sha1,
2500                     const char *base, const char *path)
2501 {
2502         char concatpath[PATH_MAX];
2503         struct diff_filespec *one, *two;
2504
2505         /* This may look odd, but it is a preparation for
2506          * feeding "there are unchanged files which should
2507          * not produce diffs, but when you are doing copy
2508          * detection you would need them, so here they are"
2509          * entries to the diff-core.  They will be prefixed
2510          * with something like '=' or '*' (I haven't decided
2511          * which but should not make any difference).
2512          * Feeding the same new and old to diff_change()
2513          * also has the same effect.
2514          * Before the final output happens, they are pruned after
2515          * merged into rename/copy pairs as appropriate.
2516          */
2517         if (options->reverse_diff)
2518                 addremove = (addremove == '+' ? '-' :
2519                              addremove == '-' ? '+' : addremove);
2520
2521         if (!path) path = "";
2522         sprintf(concatpath, "%s%s", base, path);
2523         one = alloc_filespec(concatpath);
2524         two = alloc_filespec(concatpath);
2525
2526         if (addremove != '+')
2527                 fill_filespec(one, sha1, mode);
2528         if (addremove != '-')
2529                 fill_filespec(two, sha1, mode);
2530
2531         diff_queue(&diff_queued_diff, one, two);
2532 }
2533
2534 void diff_change(struct diff_options *options,
2535                  unsigned old_mode, unsigned new_mode,
2536                  const unsigned char *old_sha1,
2537                  const unsigned char *new_sha1,
2538                  const char *base, const char *path)
2539 {
2540         char concatpath[PATH_MAX];
2541         struct diff_filespec *one, *two;
2542
2543         if (options->reverse_diff) {
2544                 unsigned tmp;
2545                 const unsigned char *tmp_c;
2546                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
2547                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
2548         }
2549         if (!path) path = "";
2550         sprintf(concatpath, "%s%s", base, path);
2551         one = alloc_filespec(concatpath);
2552         two = alloc_filespec(concatpath);
2553         fill_filespec(one, old_sha1, old_mode);
2554         fill_filespec(two, new_sha1, new_mode);
2555
2556         diff_queue(&diff_queued_diff, one, two);
2557 }
2558
2559 void diff_unmerge(struct diff_options *options,
2560                   const char *path)
2561 {
2562         struct diff_filespec *one, *two;
2563         one = alloc_filespec(path);
2564         two = alloc_filespec(path);
2565         diff_queue(&diff_queued_diff, one, two);
2566 }