diff.c

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