diff.c

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