diff.c

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "quote.h"
   6 #include "diff.h"
   7 #include "diffcore.h"
   8 #include "delta.h"
   9 #include "xdiff-interface.h"
  10 #include "color.h"
  11 #include "attr.h"
  12 #include "run-command.h"
  13 #include "utf8.h"
  14
  15 #ifdef NO_FAST_WORKING_DIRECTORY
  16 #define FAST_WORKING_DIRECTORY 0
  17 #else
  18 #define FAST_WORKING_DIRECTORY 1
  19 #endif
  20
  21 static int diff_detect_rename_default;
  22 static int diff_rename_limit_default = 200;
  23 int diff_use_color_default = -1;
  24 static const char *external_diff_cmd_cfg;
  25 int diff_auto_refresh_index = 1;
  26
  27 static char diff_colors[][COLOR_MAXLEN] = {
  28         "\033[m",       /* reset */
  29         "",             /* PLAIN (normal) */
  30         "\033[1m",      /* METAINFO (bold) */
  31         "\033[36m",     /* FRAGINFO (cyan) */
  32         "\033[31m",     /* OLD (red) */
  33         "\033[32m",     /* NEW (green) */
  34         "\033[33m",     /* COMMIT (yellow) */
  35         "\033[41m",     /* WHITESPACE (red background) */
  36 };
  37
  38 static int parse_diff_color_slot(const char *var, int ofs)
  39 {
  40         if (!strcasecmp(var+ofs, "plain"))
  41                 return DIFF_PLAIN;
  42         if (!strcasecmp(var+ofs, "meta"))
  43                 return DIFF_METAINFO;
  44         if (!strcasecmp(var+ofs, "frag"))
  45                 return DIFF_FRAGINFO;
  46         if (!strcasecmp(var+ofs, "old"))
  47                 return DIFF_FILE_OLD;
  48         if (!strcasecmp(var+ofs, "new"))
  49                 return DIFF_FILE_NEW;
  50         if (!strcasecmp(var+ofs, "commit"))
  51                 return DIFF_COMMIT;
  52         if (!strcasecmp(var+ofs, "whitespace"))
  53                 return DIFF_WHITESPACE;
  54         die("bad config variable '%s'", var);
  55 }
  56
  57 static struct ll_diff_driver {
  58         const char *name;
  59         struct ll_diff_driver *next;
  60         const char *cmd;
  61 } *user_diff, **user_diff_tail;
  62
  63 /*
  64  * Currently there is only "diff.<drivername>.command" variable;
  65  * because there are "diff.color.<slot>" variables, we are parsing
  66  * this in a bit convoluted way to allow low level diff driver
  67  * called "color".
  68  */
  69 static int parse_lldiff_command(const char *var, const char *ep, const char *value)
  70 {
  71         const char *name;
  72         int namelen;
  73         struct ll_diff_driver *drv;
  74
  75         name = var + 5;
  76         namelen = ep - name;
  77         for (drv = user_diff; drv; drv = drv->next)
  78                 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
  79                         break;
  80         if (!drv) {
  81                 drv = xcalloc(1, sizeof(struct ll_diff_driver));
  82                 drv->name = xmemdupz(name, namelen);
  83                 if (!user_diff_tail)
  84                         user_diff_tail = &user_diff;
  85                 *user_diff_tail = drv;
  86                 user_diff_tail = &(drv->next);
  87         }
  88
  89         return git_config_string(&(drv->cmd), var, value);
  90 }
  91
  92 /*
  93  * 'diff.<what>.funcname' attribute can be specified in the configuration
  94  * to define a customized regexp to find the beginning of a function to
  95  * be used for hunk header lines of "diff -p" style output.
  96  */
  97 static struct funcname_pattern {
  98         char *name;
  99         char *pattern;
 100         struct funcname_pattern *next;
 101 } *funcname_pattern_list;
 102
 103 static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
 104 {
 105         const char *name;
 106         int namelen;
 107         struct funcname_pattern *pp;
 108
 109         name = var + 5; /* "diff." */
 110         namelen = ep - name;
 111
 112         for (pp = funcname_pattern_list; pp; pp = pp->next)
 113                 if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
 114                         break;
 115         if (!pp) {
 116                 pp = xcalloc(1, sizeof(*pp));
 117                 pp->name = xmemdupz(name, namelen);
 118                 pp->next = funcname_pattern_list;
 119                 funcname_pattern_list = pp;
 120         }
 121         free(pp->pattern);
 122         pp->pattern = xstrdup(value);
 123         return 0;
 124 }
 125
 126 /*
 127  * These are to give UI layer defaults.
 128  * The core-level commands such as git-diff-files should
 129  * never be affected by the setting of diff.renames
 130  * the user happens to have in the configuration file.
 131  */
 132 int git_diff_ui_config(const char *var, const char *value, void *cb)
 133 {
 134         if (!strcmp(var, "diff.renamelimit")) {
 135                 diff_rename_limit_default = git_config_int(var, value);
 136                 return 0;
 137         }
 138         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 139                 diff_use_color_default = git_config_colorbool(var, value, -1);
 140                 return 0;
 141         }
 142         if (!strcmp(var, "diff.renames")) {
 143                 if (!value)
 144                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 145                 else if (!strcasecmp(value, "copies") ||
 146                          !strcasecmp(value, "copy"))
 147                         diff_detect_rename_default = DIFF_DETECT_COPY;
 148                 else if (git_config_bool(var,value))
 149                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 150                 return 0;
 151         }
 152         if (!strcmp(var, "diff.autorefreshindex")) {
 153                 diff_auto_refresh_index = git_config_bool(var, value);
 154                 return 0;
 155         }
 156         if (!strcmp(var, "diff.external")) {
 157                 if (!value)
 158                         return config_error_nonbool(var);
 159                 external_diff_cmd_cfg = xstrdup(value);
 160                 return 0;
 161         }
 162         if (!prefixcmp(var, "diff.")) {
 163                 const char *ep = strrchr(var, '.');
 164
 165                 if (ep != var + 4 && !strcmp(ep, ".command"))
 166                         return parse_lldiff_command(var, ep, value);
 167         }
 168
 169         return git_diff_basic_config(var, value, cb);
 170 }
 171
 172 int git_diff_basic_config(const char *var, const char *value, void *cb)
 173 {
 174         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 175                 int slot = parse_diff_color_slot(var, 11);
 176                 if (!value)
 177                         return config_error_nonbool(var);
 178                 color_parse(value, var, diff_colors[slot]);
 179                 return 0;
 180         }
 181
 182         if (!prefixcmp(var, "diff.")) {
 183                 const char *ep = strrchr(var, '.');
 184                 if (ep != var + 4) {
 185                         if (!strcmp(ep, ".funcname")) {
 186                                 if (!value)
 187                                         return config_error_nonbool(var);
 188                                 return parse_funcname_pattern(var, ep, value);
 189                         }
 190                 }
 191         }
 192
 193         return git_color_default_config(var, value, cb);
 194 }
 195
 196 static char *quote_two(const char *one, const char *two)
 197 {
 198         int need_one = quote_c_style(one, NULL, NULL, 1);
 199         int need_two = quote_c_style(two, NULL, NULL, 1);
 200         struct strbuf res;
 201
 202         strbuf_init(&res, 0);
 203         if (need_one + need_two) {
 204                 strbuf_addch(&res, '"');
 205                 quote_c_style(one, &res, NULL, 1);
 206                 quote_c_style(two, &res, NULL, 1);
 207                 strbuf_addch(&res, '"');
 208         } else {
 209                 strbuf_addstr(&res, one);
 210                 strbuf_addstr(&res, two);
 211         }
 212         return strbuf_detach(&res, NULL);
 213 }
 214
 215 static const char *external_diff(void)
 216 {
 217         static const char *external_diff_cmd = NULL;
 218         static int done_preparing = 0;
 219
 220         if (done_preparing)
 221                 return external_diff_cmd;
 222         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 223         if (!external_diff_cmd)
 224                 external_diff_cmd = external_diff_cmd_cfg;
 225         done_preparing = 1;
 226         return external_diff_cmd;
 227 }
 228
 229 static struct diff_tempfile {
 230         const char *name; /* filename external diff should read from */
 231         char hex[41];
 232         char mode[10];
 233         char tmp_path[PATH_MAX];
 234 } diff_temp[2];
 235
 236 static int count_lines(const char *data, int size)
 237 {
 238         int count, ch, completely_empty = 1, nl_just_seen = 0;
 239         count = 0;
 240         while (0 < size--) {
 241                 ch = *data++;
 242                 if (ch == '\n') {
 243                         count++;
 244                         nl_just_seen = 1;
 245                         completely_empty = 0;
 246                 }
 247                 else {
 248                         nl_just_seen = 0;
 249                         completely_empty = 0;
 250                 }
 251         }
 252         if (completely_empty)
 253                 return 0;
 254         if (!nl_just_seen)
 255                 count++; /* no trailing newline */
 256         return count;
 257 }
 258
 259 static void print_line_count(FILE *file, int count)
 260 {
 261         switch (count) {
 262         case 0:
 263                 fprintf(file, "0,0");
 264                 break;
 265         case 1:
 266                 fprintf(file, "1");
 267                 break;
 268         default:
 269                 fprintf(file, "1,%d", count);
 270                 break;
 271         }
 272 }
 273
 274 static void copy_file_with_prefix(FILE *file,
 275                                   int prefix, const char *data, int size,
 276                                   const char *set, const char *reset)
 277 {
 278         int ch, nl_just_seen = 1;
 279         while (0 < size--) {
 280                 ch = *data++;
 281                 if (nl_just_seen) {
 282                         fputs(set, file);
 283                         putc(prefix, file);
 284                 }
 285                 if (ch == '\n') {
 286                         nl_just_seen = 1;
 287                         fputs(reset, file);
 288                 } else
 289                         nl_just_seen = 0;
 290                 putc(ch, file);
 291         }
 292         if (!nl_just_seen)
 293                 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
 294 }
 295
 296 static void emit_rewrite_diff(const char *name_a,
 297                               const char *name_b,
 298                               struct diff_filespec *one,
 299                               struct diff_filespec *two,
 300                               struct diff_options *o)
 301 {
 302         int lc_a, lc_b;
 303         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
 304         const char *name_a_tab, *name_b_tab;
 305         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 306         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 307         const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
 308         const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
 309         const char *reset = diff_get_color(color_diff, DIFF_RESET);
 310         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 311
 312         name_a += (*name_a == '/');
 313         name_b += (*name_b == '/');
 314         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 315         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 316
 317         strbuf_reset(&a_name);
 318         strbuf_reset(&b_name);
 319         quote_two_c_style(&a_name, o->a_prefix, name_a, 0);
 320         quote_two_c_style(&b_name, o->b_prefix, name_b, 0);
 321
 322         diff_populate_filespec(one, 0);
 323         diff_populate_filespec(two, 0);
 324         lc_a = count_lines(one->data, one->size);
 325         lc_b = count_lines(two->data, two->size);
 326         fprintf(o->file,
 327                 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
 328                 metainfo, a_name.buf, name_a_tab, reset,
 329                 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
 330         print_line_count(o->file, lc_a);
 331         fprintf(o->file, " +");
 332         print_line_count(o->file, lc_b);
 333         fprintf(o->file, " @@%s\n", reset);
 334         if (lc_a)
 335                 copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
 336         if (lc_b)
 337                 copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
 338 }
 339
 340 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 341 {
 342         if (!DIFF_FILE_VALID(one)) {
 343                 mf->ptr = (char *)""; /* does not matter */
 344                 mf->size = 0;
 345                 return 0;
 346         }
 347         else if (diff_populate_filespec(one, 0))
 348                 return -1;
 349         mf->ptr = one->data;
 350         mf->size = one->size;
 351         return 0;
 352 }
 353
 354 struct diff_words_buffer {
 355         mmfile_t text;
 356         long alloc;
 357         long current; /* output pointer */
 358         int suppressed_newline;
 359 };
 360
 361 static void diff_words_append(char *line, unsigned long len,
 362                 struct diff_words_buffer *buffer)
 363 {
 364         if (buffer->text.size + len > buffer->alloc) {
 365                 buffer->alloc = (buffer->text.size + len) * 3 / 2;
 366                 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
 367         }
 368         line++;
 369         len--;
 370         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 371         buffer->text.size += len;
 372 }
 373
 374 struct diff_words_data {
 375         struct xdiff_emit_state xm;
 376         struct diff_words_buffer minus, plus;
 377         FILE *file;
 378 };
 379
 380 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
 381                 int suppress_newline)
 382 {
 383         const char *ptr;
 384         int eol = 0;
 385
 386         if (len == 0)
 387                 return;
 388
 389         ptr  = buffer->text.ptr + buffer->current;
 390         buffer->current += len;
 391
 392         if (ptr[len - 1] == '\n') {
 393                 eol = 1;
 394                 len--;
 395         }
 396
 397         fputs(diff_get_color(1, color), file);
 398         fwrite(ptr, len, 1, file);
 399         fputs(diff_get_color(1, DIFF_RESET), file);
 400
 401         if (eol) {
 402                 if (suppress_newline)
 403                         buffer->suppressed_newline = 1;
 404                 else
 405                         putc('\n', file);
 406         }
 407 }
 408
 409 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 410 {
 411         struct diff_words_data *diff_words = priv;
 412
 413         if (diff_words->minus.suppressed_newline) {
 414                 if (line[0] != '+')
 415                         putc('\n', diff_words->file);
 416                 diff_words->minus.suppressed_newline = 0;
 417         }
 418
 419         len--;
 420         switch (line[0]) {
 421                 case '-':
 422                         print_word(diff_words->file,
 423                                    &diff_words->minus, len, DIFF_FILE_OLD, 1);
 424                         break;
 425                 case '+':
 426                         print_word(diff_words->file,
 427                                    &diff_words->plus, len, DIFF_FILE_NEW, 0);
 428                         break;
 429                 case ' ':
 430                         print_word(diff_words->file,
 431                                    &diff_words->plus, len, DIFF_PLAIN, 0);
 432                         diff_words->minus.current += len;
 433                         break;
 434         }
 435 }
 436
 437 /* this executes the word diff on the accumulated buffers */
 438 static void diff_words_show(struct diff_words_data *diff_words)
 439 {
 440         xpparam_t xpp;
 441         xdemitconf_t xecfg;
 442         xdemitcb_t ecb;
 443         mmfile_t minus, plus;
 444         int i;
 445
 446         memset(&xecfg, 0, sizeof(xecfg));
 447         minus.size = diff_words->minus.text.size;
 448         minus.ptr = xmalloc(minus.size);
 449         memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
 450         for (i = 0; i < minus.size; i++)
 451                 if (isspace(minus.ptr[i]))
 452                         minus.ptr[i] = '\n';
 453         diff_words->minus.current = 0;
 454
 455         plus.size = diff_words->plus.text.size;
 456         plus.ptr = xmalloc(plus.size);
 457         memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
 458         for (i = 0; i < plus.size; i++)
 459                 if (isspace(plus.ptr[i]))
 460                         plus.ptr[i] = '\n';
 461         diff_words->plus.current = 0;
 462
 463         xpp.flags = XDF_NEED_MINIMAL;
 464         xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
 465         ecb.outf = xdiff_outf;
 466         ecb.priv = diff_words;
 467         diff_words->xm.consume = fn_out_diff_words_aux;
 468         xdi_diff(&minus, &plus, &xpp, &xecfg, &ecb);
 469
 470         free(minus.ptr);
 471         free(plus.ptr);
 472         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 473
 474         if (diff_words->minus.suppressed_newline) {
 475                 putc('\n', diff_words->file);
 476                 diff_words->minus.suppressed_newline = 0;
 477         }
 478 }
 479
 480 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 481
 482 struct emit_callback {
 483         struct xdiff_emit_state xm;
 484         int nparents, color_diff;
 485         unsigned ws_rule;
 486         sane_truncate_fn truncate;
 487         const char **label_path;
 488         struct diff_words_data *diff_words;
 489         int *found_changesp;
 490         FILE *file;
 491 };
 492
 493 static void free_diff_words_data(struct emit_callback *ecbdata)
 494 {
 495         if (ecbdata->diff_words) {
 496                 /* flush buffers */
 497                 if (ecbdata->diff_words->minus.text.size ||
 498                                 ecbdata->diff_words->plus.text.size)
 499                         diff_words_show(ecbdata->diff_words);
 500
 501                 free (ecbdata->diff_words->minus.text.ptr);
 502                 free (ecbdata->diff_words->plus.text.ptr);
 503                 free(ecbdata->diff_words);
 504                 ecbdata->diff_words = NULL;
 505         }
 506 }
 507
 508 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 509 {
 510         if (diff_use_color)
 511                 return diff_colors[ix];
 512         return "";
 513 }
 514
 515 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
 516 {
 517         int has_trailing_newline = (len > 0 && line[len-1] == '\n');
 518         if (has_trailing_newline)
 519                 len--;
 520
 521         fputs(set, file);
 522         fwrite(line, len, 1, file);
 523         fputs(reset, file);
 524         if (has_trailing_newline)
 525                 fputc('\n', file);
 526 }
 527
 528 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
 529 {
 530         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 531         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 532
 533         if (!*ws)
 534                 emit_line(ecbdata->file, set, reset, line, len);
 535         else {
 536                 /* Emit just the prefix, then the rest. */
 537                 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
 538                 (void)check_and_emit_line(line + ecbdata->nparents,
 539                     len - ecbdata->nparents, ecbdata->ws_rule,
 540                     ecbdata->file, set, reset, ws);
 541         }
 542 }
 543
 544 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
 545 {
 546         const char *cp;
 547         unsigned long allot;
 548         size_t l = len;
 549
 550         if (ecb->truncate)
 551                 return ecb->truncate(line, len);
 552         cp = line;
 553         allot = l;
 554         while (0 < l) {
 555                 (void) utf8_width(&cp, &l);
 556                 if (!cp)
 557                         break; /* truncated in the middle? */
 558         }
 559         return allot - l;
 560 }
 561
 562 static void fn_out_consume(void *priv, char *line, unsigned long len)
 563 {
 564         int i;
 565         int color;
 566         struct emit_callback *ecbdata = priv;
 567         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 568         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 569         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 570
 571         *(ecbdata->found_changesp) = 1;
 572
 573         if (ecbdata->label_path[0]) {
 574                 const char *name_a_tab, *name_b_tab;
 575
 576                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 577                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 578
 579                 fprintf(ecbdata->file, "%s--- %s%s%s\n",
 580                         meta, ecbdata->label_path[0], reset, name_a_tab);
 581                 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
 582                         meta, ecbdata->label_path[1], reset, name_b_tab);
 583                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 584         }
 585
 586         /* This is not really necessary for now because
 587          * this codepath only deals with two-way diffs.
 588          */
 589         for (i = 0; i < len && line[i] == '@'; i++)
 590                 ;
 591         if (2 <= i && i < len && line[i] == ' ') {
 592                 ecbdata->nparents = i - 1;
 593                 len = sane_truncate_line(ecbdata, line, len);
 594                 emit_line(ecbdata->file,
 595                           diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
 596                           reset, line, len);
 597                 if (line[len-1] != '\n')
 598                         putc('\n', ecbdata->file);
 599                 return;
 600         }
 601
 602         if (len < ecbdata->nparents) {
 603                 emit_line(ecbdata->file, reset, reset, line, len);
 604                 return;
 605         }
 606
 607         color = DIFF_PLAIN;
 608         if (ecbdata->diff_words && ecbdata->nparents != 1)
 609                 /* fall back to normal diff */
 610                 free_diff_words_data(ecbdata);
 611         if (ecbdata->diff_words) {
 612                 if (line[0] == '-') {
 613                         diff_words_append(line, len,
 614                                           &ecbdata->diff_words->minus);
 615                         return;
 616                 } else if (line[0] == '+') {
 617                         diff_words_append(line, len,
 618                                           &ecbdata->diff_words->plus);
 619                         return;
 620                 }
 621                 if (ecbdata->diff_words->minus.text.size ||
 622                     ecbdata->diff_words->plus.text.size)
 623                         diff_words_show(ecbdata->diff_words);
 624                 line++;
 625                 len--;
 626                 emit_line(ecbdata->file, plain, reset, line, len);
 627                 return;
 628         }
 629         for (i = 0; i < ecbdata->nparents && len; i++) {
 630                 if (line[i] == '-')
 631                         color = DIFF_FILE_OLD;
 632                 else if (line[i] == '+')
 633                         color = DIFF_FILE_NEW;
 634         }
 635
 636         if (color != DIFF_FILE_NEW) {
 637                 emit_line(ecbdata->file,
 638                           diff_get_color(ecbdata->color_diff, color),
 639                           reset, line, len);
 640                 return;
 641         }
 642         emit_add_line(reset, ecbdata, line, len);
 643 }
 644
 645 static char *pprint_rename(const char *a, const char *b)
 646 {
 647         const char *old = a;
 648         const char *new = b;
 649         struct strbuf name;
 650         int pfx_length, sfx_length;
 651         int len_a = strlen(a);
 652         int len_b = strlen(b);
 653         int a_midlen, b_midlen;
 654         int qlen_a = quote_c_style(a, NULL, NULL, 0);
 655         int qlen_b = quote_c_style(b, NULL, NULL, 0);
 656
 657         strbuf_init(&name, 0);
 658         if (qlen_a || qlen_b) {
 659                 quote_c_style(a, &name, NULL, 0);
 660                 strbuf_addstr(&name, " => ");
 661                 quote_c_style(b, &name, NULL, 0);
 662                 return strbuf_detach(&name, NULL);
 663         }
 664
 665         /* Find common prefix */
 666         pfx_length = 0;
 667         while (*old && *new && *old == *new) {
 668                 if (*old == '/')
 669                         pfx_length = old - a + 1;
 670                 old++;
 671                 new++;
 672         }
 673
 674         /* Find common suffix */
 675         old = a + len_a;
 676         new = b + len_b;
 677         sfx_length = 0;
 678         while (a <= old && b <= new && *old == *new) {
 679                 if (*old == '/')
 680                         sfx_length = len_a - (old - a);
 681                 old--;
 682                 new--;
 683         }
 684
 685         /*
 686          * pfx{mid-a => mid-b}sfx
 687          * {pfx-a => pfx-b}sfx
 688          * pfx{sfx-a => sfx-b}
 689          * name-a => name-b
 690          */
 691         a_midlen = len_a - pfx_length - sfx_length;
 692         b_midlen = len_b - pfx_length - sfx_length;
 693         if (a_midlen < 0)
 694                 a_midlen = 0;
 695         if (b_midlen < 0)
 696                 b_midlen = 0;
 697
 698         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 699         if (pfx_length + sfx_length) {
 700                 strbuf_add(&name, a, pfx_length);
 701                 strbuf_addch(&name, '{');
 702         }
 703         strbuf_add(&name, a + pfx_length, a_midlen);
 704         strbuf_addstr(&name, " => ");
 705         strbuf_add(&name, b + pfx_length, b_midlen);
 706         if (pfx_length + sfx_length) {
 707                 strbuf_addch(&name, '}');
 708                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 709         }
 710         return strbuf_detach(&name, NULL);
 711 }
 712
 713 struct diffstat_t {
 714         struct xdiff_emit_state xm;
 715
 716         int nr;
 717         int alloc;
 718         struct diffstat_file {
 719                 char *from_name;
 720                 char *name;
 721                 char *print_name;
 722                 unsigned is_unmerged:1;
 723                 unsigned is_binary:1;
 724                 unsigned is_renamed:1;
 725                 unsigned int added, deleted;
 726         } **files;
 727 };
 728
 729 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 730                                           const char *name_a,
 731                                           const char *name_b)
 732 {
 733         struct diffstat_file *x;
 734         x = xcalloc(sizeof (*x), 1);
 735         if (diffstat->nr == diffstat->alloc) {
 736                 diffstat->alloc = alloc_nr(diffstat->alloc);
 737                 diffstat->files = xrealloc(diffstat->files,
 738                                 diffstat->alloc * sizeof(x));
 739         }
 740         diffstat->files[diffstat->nr++] = x;
 741         if (name_b) {
 742                 x->from_name = xstrdup(name_a);
 743                 x->name = xstrdup(name_b);
 744                 x->is_renamed = 1;
 745         }
 746         else {
 747                 x->from_name = NULL;
 748                 x->name = xstrdup(name_a);
 749         }
 750         return x;
 751 }
 752
 753 static void diffstat_consume(void *priv, char *line, unsigned long len)
 754 {
 755         struct diffstat_t *diffstat = priv;
 756         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 757
 758         if (line[0] == '+')
 759                 x->added++;
 760         else if (line[0] == '-')
 761                 x->deleted++;
 762 }
 763
 764 const char mime_boundary_leader[] = "------------";
 765
 766 static int scale_linear(int it, int width, int max_change)
 767 {
 768         /*
 769          * make sure that at least one '-' is printed if there were deletions,
 770          * and likewise for '+'.
 771          */
 772         if (max_change < 2)
 773                 return it;
 774         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
 775 }
 776
 777 static void show_name(FILE *file,
 778                       const char *prefix, const char *name, int len,
 779                       const char *reset, const char *set)
 780 {
 781         fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
 782 }
 783
 784 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
 785 {
 786         if (cnt <= 0)
 787                 return;
 788         fprintf(file, "%s", set);
 789         while (cnt--)
 790                 putc(ch, file);
 791         fprintf(file, "%s", reset);
 792 }
 793
 794 static void fill_print_name(struct diffstat_file *file)
 795 {
 796         char *pname;
 797
 798         if (file->print_name)
 799                 return;
 800
 801         if (!file->is_renamed) {
 802                 struct strbuf buf;
 803                 strbuf_init(&buf, 0);
 804                 if (quote_c_style(file->name, &buf, NULL, 0)) {
 805                         pname = strbuf_detach(&buf, NULL);
 806                 } else {
 807                         pname = file->name;
 808                         strbuf_release(&buf);
 809                 }
 810         } else {
 811                 pname = pprint_rename(file->from_name, file->name);
 812         }
 813         file->print_name = pname;
 814 }
 815
 816 static void show_stats(struct diffstat_t* data, struct diff_options *options)
 817 {
 818         int i, len, add, del, total, adds = 0, dels = 0;
 819         int max_change = 0, max_len = 0;
 820         int total_files = data->nr;
 821         int width, name_width;
 822         const char *reset, *set, *add_c, *del_c;
 823
 824         if (data->nr == 0)
 825                 return;
 826
 827         width = options->stat_width ? options->stat_width : 80;
 828         name_width = options->stat_name_width ? options->stat_name_width : 50;
 829
 830         /* Sanity: give at least 5 columns to the graph,
 831          * but leave at least 10 columns for the name.
 832          */
 833         if (width < name_width + 15) {
 834                 if (name_width <= 25)
 835                         width = name_width + 15;
 836                 else
 837                         name_width = width - 15;
 838         }
 839
 840         /* Find the longest filename and max number of changes */
 841         reset = diff_get_color_opt(options, DIFF_RESET);
 842         set   = diff_get_color_opt(options, DIFF_PLAIN);
 843         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
 844         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
 845
 846         for (i = 0; i < data->nr; i++) {
 847                 struct diffstat_file *file = data->files[i];
 848                 int change = file->added + file->deleted;
 849                 fill_print_name(file);
 850                 len = strlen(file->print_name);
 851                 if (max_len < len)
 852                         max_len = len;
 853
 854                 if (file->is_binary || file->is_unmerged)
 855                         continue;
 856                 if (max_change < change)
 857                         max_change = change;
 858         }
 859
 860         /* Compute the width of the graph part;
 861          * 10 is for one blank at the beginning of the line plus
 862          * " | count " between the name and the graph.
 863          *
 864          * From here on, name_width is the width of the name area,
 865          * and width is the width of the graph area.
 866          */
 867         name_width = (name_width < max_len) ? name_width : max_len;
 868         if (width < (name_width + 10) + max_change)
 869                 width = width - (name_width + 10);
 870         else
 871                 width = max_change;
 872
 873         for (i = 0; i < data->nr; i++) {
 874                 const char *prefix = "";
 875                 char *name = data->files[i]->print_name;
 876                 int added = data->files[i]->added;
 877                 int deleted = data->files[i]->deleted;
 878                 int name_len;
 879
 880                 /*
 881                  * "scale" the filename
 882                  */
 883                 len = name_width;
 884                 name_len = strlen(name);
 885                 if (name_width < name_len) {
 886                         char *slash;
 887                         prefix = "...";
 888                         len -= 3;
 889                         name += name_len - len;
 890                         slash = strchr(name, '/');
 891                         if (slash)
 892                                 name = slash;
 893                 }
 894
 895                 if (data->files[i]->is_binary) {
 896                         show_name(options->file, prefix, name, len, reset, set);
 897                         fprintf(options->file, "  Bin ");
 898                         fprintf(options->file, "%s%d%s", del_c, deleted, reset);
 899                         fprintf(options->file, " -> ");
 900                         fprintf(options->file, "%s%d%s", add_c, added, reset);
 901                         fprintf(options->file, " bytes");
 902                         fprintf(options->file, "\n");
 903                         continue;
 904                 }
 905                 else if (data->files[i]->is_unmerged) {
 906                         show_name(options->file, prefix, name, len, reset, set);
 907                         fprintf(options->file, "  Unmerged\n");
 908                         continue;
 909                 }
 910                 else if (!data->files[i]->is_renamed &&
 911                          (added + deleted == 0)) {
 912                         total_files--;
 913                         continue;
 914                 }
 915
 916                 /*
 917                  * scale the add/delete
 918                  */
 919                 add = added;
 920                 del = deleted;
 921                 total = add + del;
 922                 adds += add;
 923                 dels += del;
 924
 925                 if (width <= max_change) {
 926                         add = scale_linear(add, width, max_change);
 927                         del = scale_linear(del, width, max_change);
 928                         total = add + del;
 929                 }
 930                 show_name(options->file, prefix, name, len, reset, set);
 931                 fprintf(options->file, "%5d ", added + deleted);
 932                 show_graph(options->file, '+', add, add_c, reset);
 933                 show_graph(options->file, '-', del, del_c, reset);
 934                 fprintf(options->file, "\n");
 935         }
 936         fprintf(options->file,
 937                "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
 938                set, total_files, adds, dels, reset);
 939 }
 940
 941 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
 942 {
 943         int i, adds = 0, dels = 0, total_files = data->nr;
 944
 945         if (data->nr == 0)
 946                 return;
 947
 948         for (i = 0; i < data->nr; i++) {
 949                 if (!data->files[i]->is_binary &&
 950                     !data->files[i]->is_unmerged) {
 951                         int added = data->files[i]->added;
 952                         int deleted= data->files[i]->deleted;
 953                         if (!data->files[i]->is_renamed &&
 954                             (added + deleted == 0)) {
 955                                 total_files--;
 956                         } else {
 957                                 adds += added;
 958                                 dels += deleted;
 959                         }
 960                 }
 961         }
 962         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
 963                total_files, adds, dels);
 964 }
 965
 966 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
 967 {
 968         int i;
 969
 970         if (data->nr == 0)
 971                 return;
 972
 973         for (i = 0; i < data->nr; i++) {
 974                 struct diffstat_file *file = data->files[i];
 975
 976                 if (file->is_binary)
 977                         fprintf(options->file, "-\t-\t");
 978                 else
 979                         fprintf(options->file,
 980                                 "%d\t%d\t", file->added, file->deleted);
 981                 if (options->line_termination) {
 982                         fill_print_name(file);
 983                         if (!file->is_renamed)
 984                                 write_name_quoted(file->name, options->file,
 985                                                   options->line_termination);
 986                         else {
 987                                 fputs(file->print_name, options->file);
 988                                 putc(options->line_termination, options->file);
 989                         }
 990                 } else {
 991                         if (file->is_renamed) {
 992                                 putc('\0', options->file);
 993                                 write_name_quoted(file->from_name, options->file, '\0');
 994                         }
 995                         write_name_quoted(file->name, options->file, '\0');
 996                 }
 997         }
 998 }
 999
1000 struct dirstat_file {
1001         const char *name;
1002         unsigned long changed;
1003 };
1004
1005 struct dirstat_dir {
1006         struct dirstat_file *files;
1007         int alloc, nr, percent, cumulative;
1008 };
1009
1010 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1011 {
1012         unsigned long this_dir = 0;
1013         unsigned int sources = 0;
1014
1015         while (dir->nr) {
1016                 struct dirstat_file *f = dir->files;
1017                 int namelen = strlen(f->name);
1018                 unsigned long this;
1019                 char *slash;
1020
1021                 if (namelen < baselen)
1022                         break;
1023                 if (memcmp(f->name, base, baselen))
1024                         break;
1025                 slash = strchr(f->name + baselen, '/');
1026                 if (slash) {
1027                         int newbaselen = slash + 1 - f->name;
1028                         this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1029                         sources++;
1030                 } else {
1031                         this = f->changed;
1032                         dir->files++;
1033                         dir->nr--;
1034                         sources += 2;
1035                 }
1036                 this_dir += this;
1037         }
1038
1039         /*
1040          * We don't report dirstat's for
1041          *  - the top level
1042          *  - or cases where everything came from a single directory
1043          *    under this directory (sources == 1).
1044          */
1045         if (baselen && sources != 1) {
1046                 int permille = this_dir * 1000 / changed;
1047                 if (permille) {
1048                         int percent = permille / 10;
1049                         if (percent >= dir->percent) {
1050                                 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1051                                 if (!dir->cumulative)
1052                                         return 0;
1053                         }
1054                 }
1055         }
1056         return this_dir;
1057 }
1058
1059 static void show_dirstat(struct diff_options *options)
1060 {
1061         int i;
1062         unsigned long changed;
1063         struct dirstat_dir dir;
1064         struct diff_queue_struct *q = &diff_queued_diff;
1065
1066         dir.files = NULL;
1067         dir.alloc = 0;
1068         dir.nr = 0;
1069         dir.percent = options->dirstat_percent;
1070         dir.cumulative = options->output_format & DIFF_FORMAT_CUMULATIVE;
1071
1072         changed = 0;
1073         for (i = 0; i < q->nr; i++) {
1074                 struct diff_filepair *p = q->queue[i];
1075                 const char *name;
1076                 unsigned long copied, added, damage;
1077
1078                 name = p->one->path ? p->one->path : p->two->path;
1079
1080                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1081                         diff_populate_filespec(p->one, 0);
1082                         diff_populate_filespec(p->two, 0);
1083                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1084                                                &copied, &added);
1085                         diff_free_filespec_data(p->one);
1086                         diff_free_filespec_data(p->two);
1087                 } else if (DIFF_FILE_VALID(p->one)) {
1088                         diff_populate_filespec(p->one, 1);
1089                         copied = added = 0;
1090                         diff_free_filespec_data(p->one);
1091                 } else if (DIFF_FILE_VALID(p->two)) {
1092                         diff_populate_filespec(p->two, 1);
1093                         copied = 0;
1094                         added = p->two->size;
1095                         diff_free_filespec_data(p->two);
1096                 } else
1097                         continue;
1098
1099                 /*
1100                  * Original minus copied is the removed material,
1101                  * added is the new material.  They are both damages
1102                  * made to the preimage.
1103                  */
1104                 damage = (p->one->size - copied) + added;
1105
1106                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1107                 dir.files[dir.nr].name = name;
1108                 dir.files[dir.nr].changed = damage;
1109                 changed += damage;
1110                 dir.nr++;
1111         }
1112
1113         /* This can happen even with many files, if everything was renames */
1114         if (!changed)
1115                 return;
1116
1117         /* Show all directories with more than x% of the changes */
1118         gather_dirstat(options->file, &dir, changed, "", 0);
1119 }
1120
1121 static void free_diffstat_info(struct diffstat_t *diffstat)
1122 {
1123         int i;
1124         for (i = 0; i < diffstat->nr; i++) {
1125                 struct diffstat_file *f = diffstat->files[i];
1126                 if (f->name != f->print_name)
1127                         free(f->print_name);
1128                 free(f->name);
1129                 free(f->from_name);
1130                 free(f);
1131         }
1132         free(diffstat->files);
1133 }
1134
1135 struct checkdiff_t {
1136         struct xdiff_emit_state xm;
1137         const char *filename;
1138         int lineno, color_diff;
1139         unsigned ws_rule;
1140         unsigned status;
1141         FILE *file;
1142 };
1143
1144 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1145 {
1146         struct checkdiff_t *data = priv;
1147         const char *ws = diff_get_color(data->color_diff, DIFF_WHITESPACE);
1148         const char *reset = diff_get_color(data->color_diff, DIFF_RESET);
1149         const char *set = diff_get_color(data->color_diff, DIFF_FILE_NEW);
1150         char *err;
1151
1152         if (line[0] == '+') {
1153                 unsigned bad;
1154                 data->lineno++;
1155                 bad = check_and_emit_line(line + 1, len - 1,
1156                     data->ws_rule, NULL, NULL, NULL, NULL);
1157                 if (!bad)
1158                         return;
1159                 data->status |= bad;
1160                 err = whitespace_error_string(bad);
1161                 fprintf(data->file, "%s:%d: %s.\n", data->filename, data->lineno, err);
1162                 free(err);
1163                 emit_line(data->file, set, reset, line, 1);
1164                 (void)check_and_emit_line(line + 1, len - 1, data->ws_rule,
1165                     data->file, set, reset, ws);
1166         } else if (line[0] == ' ')
1167                 data->lineno++;
1168         else if (line[0] == '@') {
1169                 char *plus = strchr(line, '+');
1170                 if (plus)
1171                         data->lineno = strtol(plus, NULL, 10) - 1;
1172                 else
1173                         die("invalid diff");
1174         }
1175 }
1176
1177 static unsigned char *deflate_it(char *data,
1178                                  unsigned long size,
1179                                  unsigned long *result_size)
1180 {
1181         int bound;
1182         unsigned char *deflated;
1183         z_stream stream;
1184
1185         memset(&stream, 0, sizeof(stream));
1186         deflateInit(&stream, zlib_compression_level);
1187         bound = deflateBound(&stream, size);
1188         deflated = xmalloc(bound);
1189         stream.next_out = deflated;
1190         stream.avail_out = bound;
1191
1192         stream.next_in = (unsigned char *)data;
1193         stream.avail_in = size;
1194         while (deflate(&stream, Z_FINISH) == Z_OK)
1195                 ; /* nothing */
1196         deflateEnd(&stream);
1197         *result_size = stream.total_out;
1198         return deflated;
1199 }
1200
1201 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1202 {
1203         void *cp;
1204         void *delta;
1205         void *deflated;
1206         void *data;
1207         unsigned long orig_size;
1208         unsigned long delta_size;
1209         unsigned long deflate_size;
1210         unsigned long data_size;
1211
1212         /* We could do deflated delta, or we could do just deflated two,
1213          * whichever is smaller.
1214          */
1215         delta = NULL;
1216         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1217         if (one->size && two->size) {
1218                 delta = diff_delta(one->ptr, one->size,
1219                                    two->ptr, two->size,
1220                                    &delta_size, deflate_size);
1221                 if (delta) {
1222                         void *to_free = delta;
1223                         orig_size = delta_size;
1224                         delta = deflate_it(delta, delta_size, &delta_size);
1225                         free(to_free);
1226                 }
1227         }
1228
1229         if (delta && delta_size < deflate_size) {
1230                 fprintf(file, "delta %lu\n", orig_size);
1231                 free(deflated);
1232                 data = delta;
1233                 data_size = delta_size;
1234         }
1235         else {
1236                 fprintf(file, "literal %lu\n", two->size);
1237                 free(delta);
1238                 data = deflated;
1239                 data_size = deflate_size;
1240         }
1241
1242         /* emit data encoded in base85 */
1243         cp = data;
1244         while (data_size) {
1245                 int bytes = (52 < data_size) ? 52 : data_size;
1246                 char line[70];
1247                 data_size -= bytes;
1248                 if (bytes <= 26)
1249                         line[0] = bytes + 'A' - 1;
1250                 else
1251                         line[0] = bytes - 26 + 'a' - 1;
1252                 encode_85(line + 1, cp, bytes);
1253                 cp = (char *) cp + bytes;
1254                 fputs(line, file);
1255                 fputc('\n', file);
1256         }
1257         fprintf(file, "\n");
1258         free(data);
1259 }
1260
1261 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1262 {
1263         fprintf(file, "GIT binary patch\n");
1264         emit_binary_diff_body(file, one, two);
1265         emit_binary_diff_body(file, two, one);
1266 }
1267
1268 static void setup_diff_attr_check(struct git_attr_check *check)
1269 {
1270         static struct git_attr *attr_diff;
1271
1272         if (!attr_diff) {
1273                 attr_diff = git_attr("diff", 4);
1274         }
1275         check[0].attr = attr_diff;
1276 }
1277
1278 static void diff_filespec_check_attr(struct diff_filespec *one)
1279 {
1280         struct git_attr_check attr_diff_check;
1281         int check_from_data = 0;
1282
1283         if (one->checked_attr)
1284                 return;
1285
1286         setup_diff_attr_check(&attr_diff_check);
1287         one->is_binary = 0;
1288         one->funcname_pattern_ident = NULL;
1289
1290         if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1291                 const char *value;
1292
1293                 /* binaryness */
1294                 value = attr_diff_check.value;
1295                 if (ATTR_TRUE(value))
1296                         ;
1297                 else if (ATTR_FALSE(value))
1298                         one->is_binary = 1;
1299                 else
1300                         check_from_data = 1;
1301
1302                 /* funcname pattern ident */
1303                 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1304                         ;
1305                 else
1306                         one->funcname_pattern_ident = value;
1307         }
1308
1309         if (check_from_data) {
1310                 if (!one->data && DIFF_FILE_VALID(one))
1311                         diff_populate_filespec(one, 0);
1312
1313                 if (one->data)
1314                         one->is_binary = buffer_is_binary(one->data, one->size);
1315         }
1316 }
1317
1318 int diff_filespec_is_binary(struct diff_filespec *one)
1319 {
1320         diff_filespec_check_attr(one);
1321         return one->is_binary;
1322 }
1323
1324 static const char *funcname_pattern(const char *ident)
1325 {
1326         struct funcname_pattern *pp;
1327
1328         for (pp = funcname_pattern_list; pp; pp = pp->next)
1329                 if (!strcmp(ident, pp->name))
1330                         return pp->pattern;
1331         return NULL;
1332 }
1333
1334 static struct builtin_funcname_pattern {
1335         const char *name;
1336         const char *pattern;
1337 } builtin_funcname_pattern[] = {
1338         { "java", "!^[  ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1339                         "new\\|return\\|switch\\|throw\\|while\\)\n"
1340                         "^[     ]*\\(\\([       ]*"
1341                         "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1342                         "[      ]*([^;]*\\)$" },
1343         { "tex", "^\\(\\\\\\(sub\\)*section{.*\\)$" },
1344 };
1345
1346 static const char *diff_funcname_pattern(struct diff_filespec *one)
1347 {
1348         const char *ident, *pattern;
1349         int i;
1350
1351         diff_filespec_check_attr(one);
1352         ident = one->funcname_pattern_ident;
1353
1354         if (!ident)
1355                 /*
1356                  * If the config file has "funcname.default" defined, that
1357                  * regexp is used; otherwise NULL is returned and xemit uses
1358                  * the built-in default.
1359                  */
1360                 return funcname_pattern("default");
1361
1362         /* Look up custom "funcname.$ident" regexp from config. */
1363         pattern = funcname_pattern(ident);
1364         if (pattern)
1365                 return pattern;
1366
1367         /*
1368          * And define built-in fallback patterns here.  Note that
1369          * these can be overridden by the user's config settings.
1370          */
1371         for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1372                 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1373                         return builtin_funcname_pattern[i].pattern;
1374
1375         return NULL;
1376 }
1377
1378 static void builtin_diff(const char *name_a,
1379                          const char *name_b,
1380                          struct diff_filespec *one,
1381                          struct diff_filespec *two,
1382                          const char *xfrm_msg,
1383                          struct diff_options *o,
1384                          int complete_rewrite)
1385 {
1386         mmfile_t mf1, mf2;
1387         const char *lbl[2];
1388         char *a_one, *b_two;
1389         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1390         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1391
1392         a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1393         b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1394         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1395         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1396         fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1397         if (lbl[0][0] == '/') {
1398                 /* /dev/null */
1399                 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1400                 if (xfrm_msg && xfrm_msg[0])
1401                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1402         }
1403         else if (lbl[1][0] == '/') {
1404                 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1405                 if (xfrm_msg && xfrm_msg[0])
1406                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1407         }
1408         else {
1409                 if (one->mode != two->mode) {
1410                         fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1411                         fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1412                 }
1413                 if (xfrm_msg && xfrm_msg[0])
1414                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1415                 /*
1416                  * we do not run diff between different kind
1417                  * of objects.
1418                  */
1419                 if ((one->mode ^ two->mode) & S_IFMT)
1420                         goto free_ab_and_return;
1421                 if (complete_rewrite) {
1422                         emit_rewrite_diff(name_a, name_b, one, two, o);
1423                         o->found_changes = 1;
1424                         goto free_ab_and_return;
1425                 }
1426         }
1427
1428         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1429                 die("unable to read files to diff");
1430
1431         if (!DIFF_OPT_TST(o, TEXT) &&
1432             (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1433                 /* Quite common confusing case */
1434                 if (mf1.size == mf2.size &&
1435                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1436                         goto free_ab_and_return;
1437                 if (DIFF_OPT_TST(o, BINARY))
1438                         emit_binary_diff(o->file, &mf1, &mf2);
1439                 else
1440                         fprintf(o->file, "Binary files %s and %s differ\n",
1441                                 lbl[0], lbl[1]);
1442                 o->found_changes = 1;
1443         }
1444         else {
1445                 /* Crazy xdl interfaces.. */
1446                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1447                 xpparam_t xpp;
1448                 xdemitconf_t xecfg;
1449                 xdemitcb_t ecb;
1450                 struct emit_callback ecbdata;
1451                 const char *funcname_pattern;
1452
1453                 funcname_pattern = diff_funcname_pattern(one);
1454                 if (!funcname_pattern)
1455                         funcname_pattern = diff_funcname_pattern(two);
1456
1457                 memset(&xecfg, 0, sizeof(xecfg));
1458                 memset(&ecbdata, 0, sizeof(ecbdata));
1459                 ecbdata.label_path = lbl;
1460                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1461                 ecbdata.found_changesp = &o->found_changes;
1462                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1463                 ecbdata.file = o->file;
1464                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1465                 xecfg.ctxlen = o->context;
1466                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1467                 if (funcname_pattern)
1468                         xdiff_set_find_func(&xecfg, funcname_pattern);
1469                 if (!diffopts)
1470                         ;
1471                 else if (!prefixcmp(diffopts, "--unified="))
1472                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1473                 else if (!prefixcmp(diffopts, "-u"))
1474                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1475                 ecb.outf = xdiff_outf;
1476                 ecb.priv = &ecbdata;
1477                 ecbdata.xm.consume = fn_out_consume;
1478                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1479                         ecbdata.diff_words =
1480                                 xcalloc(1, sizeof(struct diff_words_data));
1481                         ecbdata.diff_words->file = o->file;
1482                 }
1483                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1484                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1485                         free_diff_words_data(&ecbdata);
1486         }
1487
1488  free_ab_and_return:
1489         diff_free_filespec_data(one);
1490         diff_free_filespec_data(two);
1491         free(a_one);
1492         free(b_two);
1493         return;
1494 }
1495
1496 static void builtin_diffstat(const char *name_a, const char *name_b,
1497                              struct diff_filespec *one,
1498                              struct diff_filespec *two,
1499                              struct diffstat_t *diffstat,
1500                              struct diff_options *o,
1501                              int complete_rewrite)
1502 {
1503         mmfile_t mf1, mf2;
1504         struct diffstat_file *data;
1505
1506         data = diffstat_add(diffstat, name_a, name_b);
1507
1508         if (!one || !two) {
1509                 data->is_unmerged = 1;
1510                 return;
1511         }
1512         if (complete_rewrite) {
1513                 diff_populate_filespec(one, 0);
1514                 diff_populate_filespec(two, 0);
1515                 data->deleted = count_lines(one->data, one->size);
1516                 data->added = count_lines(two->data, two->size);
1517                 goto free_and_return;
1518         }
1519         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1520                 die("unable to read files to diff");
1521
1522         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1523                 data->is_binary = 1;
1524                 data->added = mf2.size;
1525                 data->deleted = mf1.size;
1526         } else {
1527                 /* Crazy xdl interfaces.. */
1528                 xpparam_t xpp;
1529                 xdemitconf_t xecfg;
1530                 xdemitcb_t ecb;
1531
1532                 memset(&xecfg, 0, sizeof(xecfg));
1533                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1534                 ecb.outf = xdiff_outf;
1535                 ecb.priv = diffstat;
1536                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1537         }
1538
1539  free_and_return:
1540         diff_free_filespec_data(one);
1541         diff_free_filespec_data(two);
1542 }
1543
1544 static void builtin_checkdiff(const char *name_a, const char *name_b,
1545                               const char *attr_path,
1546                              struct diff_filespec *one,
1547                              struct diff_filespec *two, struct diff_options *o)
1548 {
1549         mmfile_t mf1, mf2;
1550         struct checkdiff_t data;
1551
1552         if (!two)
1553                 return;
1554
1555         memset(&data, 0, sizeof(data));
1556         data.xm.consume = checkdiff_consume;
1557         data.filename = name_b ? name_b : name_a;
1558         data.lineno = 0;
1559         data.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1560         data.ws_rule = whitespace_rule(attr_path);
1561         data.file = o->file;
1562
1563         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1564                 die("unable to read files to diff");
1565
1566         if (diff_filespec_is_binary(two))
1567                 goto free_and_return;
1568         else {
1569                 /* Crazy xdl interfaces.. */
1570                 xpparam_t xpp;
1571                 xdemitconf_t xecfg;
1572                 xdemitcb_t ecb;
1573
1574                 memset(&xecfg, 0, sizeof(xecfg));
1575                 xpp.flags = XDF_NEED_MINIMAL;
1576                 ecb.outf = xdiff_outf;
1577                 ecb.priv = &data;
1578                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1579         }
1580  free_and_return:
1581         diff_free_filespec_data(one);
1582         diff_free_filespec_data(two);
1583         if (data.status)
1584                 DIFF_OPT_SET(o, CHECK_FAILED);
1585 }
1586
1587 struct diff_filespec *alloc_filespec(const char *path)
1588 {
1589         int namelen = strlen(path);
1590         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1591
1592         memset(spec, 0, sizeof(*spec));
1593         spec->path = (char *)(spec + 1);
1594         memcpy(spec->path, path, namelen+1);
1595         spec->count = 1;
1596         return spec;
1597 }
1598
1599 void free_filespec(struct diff_filespec *spec)
1600 {
1601         if (!--spec->count) {
1602                 diff_free_filespec_data(spec);
1603                 free(spec);
1604         }
1605 }
1606
1607 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1608                    unsigned short mode)
1609 {
1610         if (mode) {
1611                 spec->mode = canon_mode(mode);
1612                 hashcpy(spec->sha1, sha1);
1613                 spec->sha1_valid = !is_null_sha1(sha1);
1614         }
1615 }
1616
1617 /*
1618  * Given a name and sha1 pair, if the index tells us the file in
1619  * the work tree has that object contents, return true, so that
1620  * prepare_temp_file() does not have to inflate and extract.
1621  */
1622 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1623 {
1624         struct cache_entry *ce;
1625         struct stat st;
1626         int pos, len;
1627
1628         /* We do not read the cache ourselves here, because the
1629          * benchmark with my previous version that always reads cache
1630          * shows that it makes things worse for diff-tree comparing
1631          * two linux-2.6 kernel trees in an already checked out work
1632          * tree.  This is because most diff-tree comparisons deal with
1633          * only a small number of files, while reading the cache is
1634          * expensive for a large project, and its cost outweighs the
1635          * savings we get by not inflating the object to a temporary
1636          * file.  Practically, this code only helps when we are used
1637          * by diff-cache --cached, which does read the cache before
1638          * calling us.
1639          */
1640         if (!active_cache)
1641                 return 0;
1642
1643         /* We want to avoid the working directory if our caller
1644          * doesn't need the data in a normal file, this system
1645          * is rather slow with its stat/open/mmap/close syscalls,
1646          * and the object is contained in a pack file.  The pack
1647          * is probably already open and will be faster to obtain
1648          * the data through than the working directory.  Loose
1649          * objects however would tend to be slower as they need
1650          * to be individually opened and inflated.
1651          */
1652         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1653                 return 0;
1654
1655         len = strlen(name);
1656         pos = cache_name_pos(name, len);
1657         if (pos < 0)
1658                 return 0;
1659         ce = active_cache[pos];
1660
1661         /*
1662          * This is not the sha1 we are looking for, or
1663          * unreusable because it is not a regular file.
1664          */
1665         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1666                 return 0;
1667
1668         /*
1669          * If ce matches the file in the work tree, we can reuse it.
1670          */
1671         if (ce_uptodate(ce) ||
1672             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1673                 return 1;
1674
1675         return 0;
1676 }
1677
1678 static int populate_from_stdin(struct diff_filespec *s)
1679 {
1680         struct strbuf buf;
1681         size_t size = 0;
1682
1683         strbuf_init(&buf, 0);
1684         if (strbuf_read(&buf, 0, 0) < 0)
1685                 return error("error while reading from stdin %s",
1686                                      strerror(errno));
1687
1688         s->should_munmap = 0;
1689         s->data = strbuf_detach(&buf, &size);
1690         s->size = size;
1691         s->should_free = 1;
1692         return 0;
1693 }
1694
1695 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1696 {
1697         int len;
1698         char *data = xmalloc(100);
1699         len = snprintf(data, 100,
1700                 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1701         s->data = data;
1702         s->size = len;
1703         s->should_free = 1;
1704         if (size_only) {
1705                 s->data = NULL;
1706                 free(data);
1707         }
1708         return 0;
1709 }
1710
1711 /*
1712  * While doing rename detection and pickaxe operation, we may need to
1713  * grab the data for the blob (or file) for our own in-core comparison.
1714  * diff_filespec has data and size fields for this purpose.
1715  */
1716 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1717 {
1718         int err = 0;
1719         if (!DIFF_FILE_VALID(s))
1720                 die("internal error: asking to populate invalid file.");
1721         if (S_ISDIR(s->mode))
1722                 return -1;
1723
1724         if (s->data)
1725                 return 0;
1726
1727         if (size_only && 0 < s->size)
1728                 return 0;
1729
1730         if (S_ISGITLINK(s->mode))
1731                 return diff_populate_gitlink(s, size_only);
1732
1733         if (!s->sha1_valid ||
1734             reuse_worktree_file(s->path, s->sha1, 0)) {
1735                 struct strbuf buf;
1736                 struct stat st;
1737                 int fd;
1738
1739                 if (!strcmp(s->path, "-"))
1740                         return populate_from_stdin(s);
1741
1742                 if (lstat(s->path, &st) < 0) {
1743                         if (errno == ENOENT) {
1744                         err_empty:
1745                                 err = -1;
1746                         empty:
1747                                 s->data = (char *)"";
1748                                 s->size = 0;
1749                                 return err;
1750                         }
1751                 }
1752                 s->size = xsize_t(st.st_size);
1753                 if (!s->size)
1754                         goto empty;
1755                 if (size_only)
1756                         return 0;
1757                 if (S_ISLNK(st.st_mode)) {
1758                         int ret;
1759                         s->data = xmalloc(s->size);
1760                         s->should_free = 1;
1761                         ret = readlink(s->path, s->data, s->size);
1762                         if (ret < 0) {
1763                                 free(s->data);
1764                                 goto err_empty;
1765                         }
1766                         return 0;
1767                 }
1768                 fd = open(s->path, O_RDONLY);
1769                 if (fd < 0)
1770                         goto err_empty;
1771                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1772                 close(fd);
1773                 s->should_munmap = 1;
1774
1775                 /*
1776                  * Convert from working tree format to canonical git format
1777                  */
1778                 strbuf_init(&buf, 0);
1779                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1780                         size_t size = 0;
1781                         munmap(s->data, s->size);
1782                         s->should_munmap = 0;
1783                         s->data = strbuf_detach(&buf, &size);
1784                         s->size = size;
1785                         s->should_free = 1;
1786                 }
1787         }
1788         else {
1789                 enum object_type type;
1790                 if (size_only)
1791                         type = sha1_object_info(s->sha1, &s->size);
1792                 else {
1793                         s->data = read_sha1_file(s->sha1, &type, &s->size);
1794                         s->should_free = 1;
1795                 }
1796         }
1797         return 0;
1798 }
1799
1800 void diff_free_filespec_blob(struct diff_filespec *s)
1801 {
1802         if (s->should_free)
1803                 free(s->data);
1804         else if (s->should_munmap)
1805                 munmap(s->data, s->size);
1806
1807         if (s->should_free || s->should_munmap) {
1808                 s->should_free = s->should_munmap = 0;
1809                 s->data = NULL;
1810         }
1811 }
1812
1813 void diff_free_filespec_data(struct diff_filespec *s)
1814 {
1815         diff_free_filespec_blob(s);
1816         free(s->cnt_data);
1817         s->cnt_data = NULL;
1818 }
1819
1820 static void prep_temp_blob(struct diff_tempfile *temp,
1821                            void *blob,
1822                            unsigned long size,
1823                            const unsigned char *sha1,
1824                            int mode)
1825 {
1826         int fd;
1827
1828         fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1829         if (fd < 0)
1830                 die("unable to create temp-file: %s", strerror(errno));
1831         if (write_in_full(fd, blob, size) != size)
1832                 die("unable to write temp-file");
1833         close(fd);
1834         temp->name = temp->tmp_path;
1835         strcpy(temp->hex, sha1_to_hex(sha1));
1836         temp->hex[40] = 0;
1837         sprintf(temp->mode, "%06o", mode);
1838 }
1839
1840 static void prepare_temp_file(const char *name,
1841                               struct diff_tempfile *temp,
1842                               struct diff_filespec *one)
1843 {
1844         if (!DIFF_FILE_VALID(one)) {
1845         not_a_valid_file:
1846                 /* A '-' entry produces this for file-2, and
1847                  * a '+' entry produces this for file-1.
1848                  */
1849                 temp->name = "/dev/null";
1850                 strcpy(temp->hex, ".");
1851                 strcpy(temp->mode, ".");
1852                 return;
1853         }
1854
1855         if (!one->sha1_valid ||
1856             reuse_worktree_file(name, one->sha1, 1)) {
1857                 struct stat st;
1858                 if (lstat(name, &st) < 0) {
1859                         if (errno == ENOENT)
1860                                 goto not_a_valid_file;
1861                         die("stat(%s): %s", name, strerror(errno));
1862                 }
1863                 if (S_ISLNK(st.st_mode)) {
1864                         int ret;
1865                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1866                         size_t sz = xsize_t(st.st_size);
1867                         if (sizeof(buf) <= st.st_size)
1868                                 die("symlink too long: %s", name);
1869                         ret = readlink(name, buf, sz);
1870                         if (ret < 0)
1871                                 die("readlink(%s)", name);
1872                         prep_temp_blob(temp, buf, sz,
1873                                        (one->sha1_valid ?
1874                                         one->sha1 : null_sha1),
1875                                        (one->sha1_valid ?
1876                                         one->mode : S_IFLNK));
1877                 }
1878                 else {
1879                         /* we can borrow from the file in the work tree */
1880                         temp->name = name;
1881                         if (!one->sha1_valid)
1882                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
1883                         else
1884                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
1885                         /* Even though we may sometimes borrow the
1886                          * contents from the work tree, we always want
1887                          * one->mode.  mode is trustworthy even when
1888                          * !(one->sha1_valid), as long as
1889                          * DIFF_FILE_VALID(one).
1890                          */
1891                         sprintf(temp->mode, "%06o", one->mode);
1892                 }
1893                 return;
1894         }
1895         else {
1896                 if (diff_populate_filespec(one, 0))
1897                         die("cannot read data blob for %s", one->path);
1898                 prep_temp_blob(temp, one->data, one->size,
1899                                one->sha1, one->mode);
1900         }
1901 }
1902
1903 static void remove_tempfile(void)
1904 {
1905         int i;
1906
1907         for (i = 0; i < 2; i++)
1908                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1909                         unlink(diff_temp[i].name);
1910                         diff_temp[i].name = NULL;
1911                 }
1912 }
1913
1914 static void remove_tempfile_on_signal(int signo)
1915 {
1916         remove_tempfile();
1917         signal(SIGINT, SIG_DFL);
1918         raise(signo);
1919 }
1920
1921 /* An external diff command takes:
1922  *
1923  * diff-cmd name infile1 infile1-sha1 infile1-mode \
1924  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
1925  *
1926  */
1927 static void run_external_diff(const char *pgm,
1928                               const char *name,
1929                               const char *other,
1930                               struct diff_filespec *one,
1931                               struct diff_filespec *two,
1932                               const char *xfrm_msg,
1933                               int complete_rewrite)
1934 {
1935         const char *spawn_arg[10];
1936         struct diff_tempfile *temp = diff_temp;
1937         int retval;
1938         static int atexit_asked = 0;
1939         const char *othername;
1940         const char **arg = &spawn_arg[0];
1941
1942         othername = (other? other : name);
1943         if (one && two) {
1944                 prepare_temp_file(name, &temp[0], one);
1945                 prepare_temp_file(othername, &temp[1], two);
1946                 if (! atexit_asked &&
1947                     (temp[0].name == temp[0].tmp_path ||
1948                      temp[1].name == temp[1].tmp_path)) {
1949                         atexit_asked = 1;
1950                         atexit(remove_tempfile);
1951                 }
1952                 signal(SIGINT, remove_tempfile_on_signal);
1953         }
1954
1955         if (one && two) {
1956                 *arg++ = pgm;
1957                 *arg++ = name;
1958                 *arg++ = temp[0].name;
1959                 *arg++ = temp[0].hex;
1960                 *arg++ = temp[0].mode;
1961                 *arg++ = temp[1].name;
1962                 *arg++ = temp[1].hex;
1963                 *arg++ = temp[1].mode;
1964                 if (other) {
1965                         *arg++ = other;
1966                         *arg++ = xfrm_msg;
1967                 }
1968         } else {
1969                 *arg++ = pgm;
1970                 *arg++ = name;
1971         }
1972         *arg = NULL;
1973         fflush(NULL);
1974         retval = run_command_v_opt(spawn_arg, 0);
1975         remove_tempfile();
1976         if (retval) {
1977                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1978                 exit(1);
1979         }
1980 }
1981
1982 static const char *external_diff_attr(const char *name)
1983 {
1984         struct git_attr_check attr_diff_check;
1985
1986         if (!name)
1987                 return NULL;
1988
1989         setup_diff_attr_check(&attr_diff_check);
1990         if (!git_checkattr(name, 1, &attr_diff_check)) {
1991                 const char *value = attr_diff_check.value;
1992                 if (!ATTR_TRUE(value) &&
1993                     !ATTR_FALSE(value) &&
1994                     !ATTR_UNSET(value)) {
1995                         struct ll_diff_driver *drv;
1996
1997                         for (drv = user_diff; drv; drv = drv->next)
1998                                 if (!strcmp(drv->name, value))
1999                                         return drv->cmd;
2000                 }
2001         }
2002         return NULL;
2003 }
2004
2005 static void run_diff_cmd(const char *pgm,
2006                          const char *name,
2007                          const char *other,
2008                          const char *attr_path,
2009                          struct diff_filespec *one,
2010                          struct diff_filespec *two,
2011                          const char *xfrm_msg,
2012                          struct diff_options *o,
2013                          int complete_rewrite)
2014 {
2015         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2016                 pgm = NULL;
2017         else {
2018                 const char *cmd = external_diff_attr(attr_path);
2019                 if (cmd)
2020                         pgm = cmd;
2021         }
2022
2023         if (pgm) {
2024                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2025                                   complete_rewrite);
2026                 return;
2027         }
2028         if (one && two)
2029                 builtin_diff(name, other ? other : name,
2030                              one, two, xfrm_msg, o, complete_rewrite);
2031         else
2032                 fprintf(o->file, "* Unmerged path %s\n", name);
2033 }
2034
2035 static void diff_fill_sha1_info(struct diff_filespec *one)
2036 {
2037         if (DIFF_FILE_VALID(one)) {
2038                 if (!one->sha1_valid) {
2039                         struct stat st;
2040                         if (!strcmp(one->path, "-")) {
2041                                 hashcpy(one->sha1, null_sha1);
2042                                 return;
2043                         }
2044                         if (lstat(one->path, &st) < 0)
2045                                 die("stat %s", one->path);
2046                         if (index_path(one->sha1, one->path, &st, 0))
2047                                 die("cannot hash %s\n", one->path);
2048                 }
2049         }
2050         else
2051                 hashclr(one->sha1);
2052 }
2053
2054 static int similarity_index(struct diff_filepair *p)
2055 {
2056         return p->score * 100 / MAX_SCORE;
2057 }
2058
2059 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2060 {
2061         /* Strip the prefix but do not molest /dev/null and absolute paths */
2062         if (*namep && **namep != '/')
2063                 *namep += prefix_length;
2064         if (*otherp && **otherp != '/')
2065                 *otherp += prefix_length;
2066 }
2067
2068 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2069 {
2070         const char *pgm = external_diff();
2071         struct strbuf msg;
2072         char *xfrm_msg;
2073         struct diff_filespec *one = p->one;
2074         struct diff_filespec *two = p->two;
2075         const char *name;
2076         const char *other;
2077         const char *attr_path;
2078         int complete_rewrite = 0;
2079
2080         name  = p->one->path;
2081         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2082         attr_path = name;
2083         if (o->prefix_length)
2084                 strip_prefix(o->prefix_length, &name, &other);
2085
2086         if (DIFF_PAIR_UNMERGED(p)) {
2087                 run_diff_cmd(pgm, name, NULL, attr_path,
2088                              NULL, NULL, NULL, o, 0);
2089                 return;
2090         }
2091
2092         diff_fill_sha1_info(one);
2093         diff_fill_sha1_info(two);
2094
2095         strbuf_init(&msg, PATH_MAX * 2 + 300);
2096         switch (p->status) {
2097         case DIFF_STATUS_COPIED:
2098                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2099                 strbuf_addstr(&msg, "\ncopy from ");
2100                 quote_c_style(name, &msg, NULL, 0);
2101                 strbuf_addstr(&msg, "\ncopy to ");
2102                 quote_c_style(other, &msg, NULL, 0);
2103                 strbuf_addch(&msg, '\n');
2104                 break;
2105         case DIFF_STATUS_RENAMED:
2106                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2107                 strbuf_addstr(&msg, "\nrename from ");
2108                 quote_c_style(name, &msg, NULL, 0);
2109                 strbuf_addstr(&msg, "\nrename to ");
2110                 quote_c_style(other, &msg, NULL, 0);
2111                 strbuf_addch(&msg, '\n');
2112                 break;
2113         case DIFF_STATUS_MODIFIED:
2114                 if (p->score) {
2115                         strbuf_addf(&msg, "dissimilarity index %d%%\n",
2116                                         similarity_index(p));
2117                         complete_rewrite = 1;
2118                         break;
2119                 }
2120                 /* fallthru */
2121         default:
2122                 /* nothing */
2123                 ;
2124         }
2125
2126         if (hashcmp(one->sha1, two->sha1)) {
2127                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2128
2129                 if (DIFF_OPT_TST(o, BINARY)) {
2130                         mmfile_t mf;
2131                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2132                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2133                                 abbrev = 40;
2134                 }
2135                 strbuf_addf(&msg, "index %.*s..%.*s",
2136                                 abbrev, sha1_to_hex(one->sha1),
2137                                 abbrev, sha1_to_hex(two->sha1));
2138                 if (one->mode == two->mode)
2139                         strbuf_addf(&msg, " %06o", one->mode);
2140                 strbuf_addch(&msg, '\n');
2141         }
2142
2143         if (msg.len)
2144                 strbuf_setlen(&msg, msg.len - 1);
2145         xfrm_msg = msg.len ? msg.buf : NULL;
2146
2147         if (!pgm &&
2148             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2149             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2150                 /* a filepair that changes between file and symlink
2151                  * needs to be split into deletion and creation.
2152                  */
2153                 struct diff_filespec *null = alloc_filespec(two->path);
2154                 run_diff_cmd(NULL, name, other, attr_path,
2155                              one, null, xfrm_msg, o, 0);
2156                 free(null);
2157                 null = alloc_filespec(one->path);
2158                 run_diff_cmd(NULL, name, other, attr_path,
2159                              null, two, xfrm_msg, o, 0);
2160                 free(null);
2161         }
2162         else
2163                 run_diff_cmd(pgm, name, other, attr_path,
2164                              one, two, xfrm_msg, o, complete_rewrite);
2165
2166         strbuf_release(&msg);
2167 }
2168
2169 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2170                          struct diffstat_t *diffstat)
2171 {
2172         const char *name;
2173         const char *other;
2174         int complete_rewrite = 0;
2175
2176         if (DIFF_PAIR_UNMERGED(p)) {
2177                 /* unmerged */
2178                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2179                 return;
2180         }
2181
2182         name = p->one->path;
2183         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2184
2185         if (o->prefix_length)
2186                 strip_prefix(o->prefix_length, &name, &other);
2187
2188         diff_fill_sha1_info(p->one);
2189         diff_fill_sha1_info(p->two);
2190
2191         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2192                 complete_rewrite = 1;
2193         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2194 }
2195
2196 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2197 {
2198         const char *name;
2199         const char *other;
2200         const char *attr_path;
2201
2202         if (DIFF_PAIR_UNMERGED(p)) {
2203                 /* unmerged */
2204                 return;
2205         }
2206
2207         name = p->one->path;
2208         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2209         attr_path = other ? other : name;
2210
2211         if (o->prefix_length)
2212                 strip_prefix(o->prefix_length, &name, &other);
2213
2214         diff_fill_sha1_info(p->one);
2215         diff_fill_sha1_info(p->two);
2216
2217         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2218 }
2219
2220 void diff_setup(struct diff_options *options)
2221 {
2222         memset(options, 0, sizeof(*options));
2223
2224         options->file = stdout;
2225
2226         options->line_termination = '\n';
2227         options->break_opt = -1;
2228         options->rename_limit = -1;
2229         options->dirstat_percent = 3;
2230         options->context = 3;
2231
2232         options->change = diff_change;
2233         options->add_remove = diff_addremove;
2234         if (diff_use_color_default > 0)
2235                 DIFF_OPT_SET(options, COLOR_DIFF);
2236         else
2237                 DIFF_OPT_CLR(options, COLOR_DIFF);
2238         options->detect_rename = diff_detect_rename_default;
2239
2240         options->a_prefix = "a/";
2241         options->b_prefix = "b/";
2242 }
2243
2244 int diff_setup_done(struct diff_options *options)
2245 {
2246         int count = 0;
2247
2248         if (options->output_format & DIFF_FORMAT_NAME)
2249                 count++;
2250         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2251                 count++;
2252         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2253                 count++;
2254         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2255                 count++;
2256         if (count > 1)
2257                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2258
2259         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2260                 options->detect_rename = DIFF_DETECT_COPY;
2261
2262         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2263                 options->prefix = NULL;
2264         if (options->prefix)
2265                 options->prefix_length = strlen(options->prefix);
2266         else
2267                 options->prefix_length = 0;
2268
2269         if (options->output_format & (DIFF_FORMAT_NAME |
2270                                       DIFF_FORMAT_NAME_STATUS |
2271                                       DIFF_FORMAT_CHECKDIFF |
2272                                       DIFF_FORMAT_NO_OUTPUT))
2273                 options->output_format &= ~(DIFF_FORMAT_RAW |
2274                                             DIFF_FORMAT_NUMSTAT |
2275                                             DIFF_FORMAT_DIFFSTAT |
2276                                             DIFF_FORMAT_SHORTSTAT |
2277                                             DIFF_FORMAT_DIRSTAT |
2278                                             DIFF_FORMAT_SUMMARY |
2279                                             DIFF_FORMAT_PATCH);
2280
2281         /*
2282          * These cases always need recursive; we do not drop caller-supplied
2283          * recursive bits for other formats here.
2284          */
2285         if (options->output_format & (DIFF_FORMAT_PATCH |
2286                                       DIFF_FORMAT_NUMSTAT |
2287                                       DIFF_FORMAT_DIFFSTAT |
2288                                       DIFF_FORMAT_SHORTSTAT |
2289                                       DIFF_FORMAT_DIRSTAT |
2290                                       DIFF_FORMAT_SUMMARY |
2291                                       DIFF_FORMAT_CHECKDIFF))
2292                 DIFF_OPT_SET(options, RECURSIVE);
2293         /*
2294          * Also pickaxe would not work very well if you do not say recursive
2295          */
2296         if (options->pickaxe)
2297                 DIFF_OPT_SET(options, RECURSIVE);
2298
2299         if (options->detect_rename && options->rename_limit < 0)
2300                 options->rename_limit = diff_rename_limit_default;
2301         if (options->setup & DIFF_SETUP_USE_CACHE) {
2302                 if (!active_cache)
2303                         /* read-cache does not die even when it fails
2304                          * so it is safe for us to do this here.  Also
2305                          * it does not smudge active_cache or active_nr
2306                          * when it fails, so we do not have to worry about
2307                          * cleaning it up ourselves either.
2308                          */
2309                         read_cache();
2310         }
2311         if (options->abbrev <= 0 || 40 < options->abbrev)
2312                 options->abbrev = 40; /* full */
2313
2314         /*
2315          * It does not make sense to show the first hit we happened
2316          * to have found.  It does not make sense not to return with
2317          * exit code in such a case either.
2318          */
2319         if (DIFF_OPT_TST(options, QUIET)) {
2320                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2321                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2322         }
2323
2324         /*
2325          * If we postprocess in diffcore, we cannot simply return
2326          * upon the first hit.  We need to run diff as usual.
2327          */
2328         if (options->pickaxe || options->filter)
2329                 DIFF_OPT_CLR(options, QUIET);
2330
2331         return 0;
2332 }
2333
2334 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2335 {
2336         char c, *eq;
2337         int len;
2338
2339         if (*arg != '-')
2340                 return 0;
2341         c = *++arg;
2342         if (!c)
2343                 return 0;
2344         if (c == arg_short) {
2345                 c = *++arg;
2346                 if (!c)
2347                         return 1;
2348                 if (val && isdigit(c)) {
2349                         char *end;
2350                         int n = strtoul(arg, &end, 10);
2351                         if (*end)
2352                                 return 0;
2353                         *val = n;
2354                         return 1;
2355                 }
2356                 return 0;
2357         }
2358         if (c != '-')
2359                 return 0;
2360         arg++;
2361         eq = strchr(arg, '=');
2362         if (eq)
2363                 len = eq - arg;
2364         else
2365                 len = strlen(arg);
2366         if (!len || strncmp(arg, arg_long, len))
2367                 return 0;
2368         if (eq) {
2369                 int n;
2370                 char *end;
2371                 if (!isdigit(*++eq))
2372                         return 0;
2373                 n = strtoul(eq, &end, 10);
2374                 if (*end)
2375                         return 0;
2376                 *val = n;
2377         }
2378         return 1;
2379 }
2380
2381 static int diff_scoreopt_parse(const char *opt);
2382
2383 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2384 {
2385         const char *arg = av[0];
2386
2387         /* Output format options */
2388         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2389                 options->output_format |= DIFF_FORMAT_PATCH;
2390         else if (opt_arg(arg, 'U', "unified", &options->context))
2391                 options->output_format |= DIFF_FORMAT_PATCH;
2392         else if (!strcmp(arg, "--raw"))
2393                 options->output_format |= DIFF_FORMAT_RAW;
2394         else if (!strcmp(arg, "--patch-with-raw"))
2395                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2396         else if (!strcmp(arg, "--numstat"))
2397                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2398         else if (!strcmp(arg, "--shortstat"))
2399                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2400         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2401                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2402         else if (!strcmp(arg, "--cumulative"))
2403                 options->output_format |= DIFF_FORMAT_CUMULATIVE;
2404         else if (!strcmp(arg, "--check"))
2405                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2406         else if (!strcmp(arg, "--summary"))
2407                 options->output_format |= DIFF_FORMAT_SUMMARY;
2408         else if (!strcmp(arg, "--patch-with-stat"))
2409                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2410         else if (!strcmp(arg, "--name-only"))
2411                 options->output_format |= DIFF_FORMAT_NAME;
2412         else if (!strcmp(arg, "--name-status"))
2413                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2414         else if (!strcmp(arg, "-s"))
2415                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2416         else if (!prefixcmp(arg, "--stat")) {
2417                 char *end;
2418                 int width = options->stat_width;
2419                 int name_width = options->stat_name_width;
2420                 arg += 6;
2421                 end = (char *)arg;
2422
2423                 switch (*arg) {
2424                 case '-':
2425                         if (!prefixcmp(arg, "-width="))
2426                                 width = strtoul(arg + 7, &end, 10);
2427                         else if (!prefixcmp(arg, "-name-width="))
2428                                 name_width = strtoul(arg + 12, &end, 10);
2429                         break;
2430                 case '=':
2431                         width = strtoul(arg+1, &end, 10);
2432                         if (*end == ',')
2433                                 name_width = strtoul(end+1, &end, 10);
2434                 }
2435
2436                 /* Important! This checks all the error cases! */
2437                 if (*end)
2438                         return 0;
2439                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2440                 options->stat_name_width = name_width;
2441                 options->stat_width = width;
2442         }
2443
2444         /* renames options */
2445         else if (!prefixcmp(arg, "-B")) {
2446                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2447                         return -1;
2448         }
2449         else if (!prefixcmp(arg, "-M")) {
2450                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2451                         return -1;
2452                 options->detect_rename = DIFF_DETECT_RENAME;
2453         }
2454         else if (!prefixcmp(arg, "-C")) {
2455                 if (options->detect_rename == DIFF_DETECT_COPY)
2456                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2457                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2458                         return -1;
2459                 options->detect_rename = DIFF_DETECT_COPY;
2460         }
2461         else if (!strcmp(arg, "--no-renames"))
2462                 options->detect_rename = 0;
2463         else if (!strcmp(arg, "--relative"))
2464                 DIFF_OPT_SET(options, RELATIVE_NAME);
2465         else if (!prefixcmp(arg, "--relative=")) {
2466                 DIFF_OPT_SET(options, RELATIVE_NAME);
2467                 options->prefix = arg + 11;
2468         }
2469
2470         /* xdiff options */
2471         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2472                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2473         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2474                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2475         else if (!strcmp(arg, "--ignore-space-at-eol"))
2476                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2477
2478         /* flags options */
2479         else if (!strcmp(arg, "--binary")) {
2480                 options->output_format |= DIFF_FORMAT_PATCH;
2481                 DIFF_OPT_SET(options, BINARY);
2482         }
2483         else if (!strcmp(arg, "--full-index"))
2484                 DIFF_OPT_SET(options, FULL_INDEX);
2485         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2486                 DIFF_OPT_SET(options, TEXT);
2487         else if (!strcmp(arg, "-R"))
2488                 DIFF_OPT_SET(options, REVERSE_DIFF);
2489         else if (!strcmp(arg, "--find-copies-harder"))
2490                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2491         else if (!strcmp(arg, "--follow"))
2492                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2493         else if (!strcmp(arg, "--color"))
2494                 DIFF_OPT_SET(options, COLOR_DIFF);
2495         else if (!strcmp(arg, "--no-color"))
2496                 DIFF_OPT_CLR(options, COLOR_DIFF);
2497         else if (!strcmp(arg, "--color-words"))
2498                 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2499         else if (!strcmp(arg, "--exit-code"))
2500                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2501         else if (!strcmp(arg, "--quiet"))
2502                 DIFF_OPT_SET(options, QUIET);
2503         else if (!strcmp(arg, "--ext-diff"))
2504                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2505         else if (!strcmp(arg, "--no-ext-diff"))
2506                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2507         else if (!strcmp(arg, "--ignore-submodules"))
2508                 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2509
2510         /* misc options */
2511         else if (!strcmp(arg, "-z"))
2512                 options->line_termination = 0;
2513         else if (!prefixcmp(arg, "-l"))
2514                 options->rename_limit = strtoul(arg+2, NULL, 10);
2515         else if (!prefixcmp(arg, "-S"))
2516                 options->pickaxe = arg + 2;
2517         else if (!strcmp(arg, "--pickaxe-all"))
2518                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2519         else if (!strcmp(arg, "--pickaxe-regex"))
2520                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2521         else if (!prefixcmp(arg, "-O"))
2522                 options->orderfile = arg + 2;
2523         else if (!prefixcmp(arg, "--diff-filter="))
2524                 options->filter = arg + 14;
2525         else if (!strcmp(arg, "--abbrev"))
2526                 options->abbrev = DEFAULT_ABBREV;
2527         else if (!prefixcmp(arg, "--abbrev=")) {
2528                 options->abbrev = strtoul(arg + 9, NULL, 10);
2529                 if (options->abbrev < MINIMUM_ABBREV)
2530                         options->abbrev = MINIMUM_ABBREV;
2531                 else if (40 < options->abbrev)
2532                         options->abbrev = 40;
2533         }
2534         else if (!prefixcmp(arg, "--src-prefix="))
2535                 options->a_prefix = arg + 13;
2536         else if (!prefixcmp(arg, "--dst-prefix="))
2537                 options->b_prefix = arg + 13;
2538         else if (!strcmp(arg, "--no-prefix"))
2539                 options->a_prefix = options->b_prefix = "";
2540         else if (!prefixcmp(arg, "--output=")) {
2541                 options->file = fopen(arg + strlen("--output="), "w");
2542                 options->close_file = 1;
2543         } else
2544                 return 0;
2545         return 1;
2546 }
2547
2548 static int parse_num(const char **cp_p)
2549 {
2550         unsigned long num, scale;
2551         int ch, dot;
2552         const char *cp = *cp_p;
2553
2554         num = 0;
2555         scale = 1;
2556         dot = 0;
2557         for(;;) {
2558                 ch = *cp;
2559                 if ( !dot && ch == '.' ) {
2560                         scale = 1;
2561                         dot = 1;
2562                 } else if ( ch == '%' ) {
2563                         scale = dot ? scale*100 : 100;
2564                         cp++;   /* % is always at the end */
2565                         break;
2566                 } else if ( ch >= '0' && ch <= '9' ) {
2567                         if ( scale < 100000 ) {
2568                                 scale *= 10;
2569                                 num = (num*10) + (ch-'0');
2570                         }
2571                 } else {
2572                         break;
2573                 }
2574                 cp++;
2575         }
2576         *cp_p = cp;
2577
2578         /* user says num divided by scale and we say internally that
2579          * is MAX_SCORE * num / scale.
2580          */
2581         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2582 }
2583
2584 static int diff_scoreopt_parse(const char *opt)
2585 {
2586         int opt1, opt2, cmd;
2587
2588         if (*opt++ != '-')
2589                 return -1;
2590         cmd = *opt++;
2591         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2592                 return -1; /* that is not a -M, -C nor -B option */
2593
2594         opt1 = parse_num(&opt);
2595         if (cmd != 'B')
2596                 opt2 = 0;
2597         else {
2598                 if (*opt == 0)
2599                         opt2 = 0;
2600                 else if (*opt != '/')
2601                         return -1; /* we expect -B80/99 or -B80 */
2602                 else {
2603                         opt++;
2604                         opt2 = parse_num(&opt);
2605                 }
2606         }
2607         if (*opt != 0)
2608                 return -1;
2609         return opt1 | (opt2 << 16);
2610 }
2611
2612 struct diff_queue_struct diff_queued_diff;
2613
2614 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2615 {
2616         if (queue->alloc <= queue->nr) {
2617                 queue->alloc = alloc_nr(queue->alloc);
2618                 queue->queue = xrealloc(queue->queue,
2619                                         sizeof(dp) * queue->alloc);
2620         }
2621         queue->queue[queue->nr++] = dp;
2622 }
2623
2624 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2625                                  struct diff_filespec *one,
2626                                  struct diff_filespec *two)
2627 {
2628         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2629         dp->one = one;
2630         dp->two = two;
2631         if (queue)
2632                 diff_q(queue, dp);
2633         return dp;
2634 }
2635
2636 void diff_free_filepair(struct diff_filepair *p)
2637 {
2638         free_filespec(p->one);
2639         free_filespec(p->two);
2640         free(p);
2641 }
2642
2643 /* This is different from find_unique_abbrev() in that
2644  * it stuffs the result with dots for alignment.
2645  */
2646 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2647 {
2648         int abblen;
2649         const char *abbrev;
2650         if (len == 40)
2651                 return sha1_to_hex(sha1);
2652
2653         abbrev = find_unique_abbrev(sha1, len);
2654         abblen = strlen(abbrev);
2655         if (abblen < 37) {
2656                 static char hex[41];
2657                 if (len < abblen && abblen <= len + 2)
2658                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2659                 else
2660                         sprintf(hex, "%s...", abbrev);
2661                 return hex;
2662         }
2663         return sha1_to_hex(sha1);
2664 }
2665
2666 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2667 {
2668         int line_termination = opt->line_termination;
2669         int inter_name_termination = line_termination ? '\t' : '\0';
2670
2671         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2672                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2673                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
2674                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2675         }
2676         if (p->score) {
2677                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2678                         inter_name_termination);
2679         } else {
2680                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2681         }
2682
2683         if (p->status == DIFF_STATUS_COPIED ||
2684             p->status == DIFF_STATUS_RENAMED) {
2685                 const char *name_a, *name_b;
2686                 name_a = p->one->path;
2687                 name_b = p->two->path;
2688                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2689                 write_name_quoted(name_a, opt->file, inter_name_termination);
2690                 write_name_quoted(name_b, opt->file, line_termination);
2691         } else {
2692                 const char *name_a, *name_b;
2693                 name_a = p->one->mode ? p->one->path : p->two->path;
2694                 name_b = NULL;
2695                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2696                 write_name_quoted(name_a, opt->file, line_termination);
2697         }
2698 }
2699
2700 int diff_unmodified_pair(struct diff_filepair *p)
2701 {
2702         /* This function is written stricter than necessary to support
2703          * the currently implemented transformers, but the idea is to
2704          * let transformers to produce diff_filepairs any way they want,
2705          * and filter and clean them up here before producing the output.
2706          */
2707         struct diff_filespec *one = p->one, *two = p->two;
2708
2709         if (DIFF_PAIR_UNMERGED(p))
2710                 return 0; /* unmerged is interesting */
2711
2712         /* deletion, addition, mode or type change
2713          * and rename are all interesting.
2714          */
2715         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2716             DIFF_PAIR_MODE_CHANGED(p) ||
2717             strcmp(one->path, two->path))
2718                 return 0;
2719
2720         /* both are valid and point at the same path.  that is, we are
2721          * dealing with a change.
2722          */
2723         if (one->sha1_valid && two->sha1_valid &&
2724             !hashcmp(one->sha1, two->sha1))
2725                 return 1; /* no change */
2726         if (!one->sha1_valid && !two->sha1_valid)
2727                 return 1; /* both look at the same file on the filesystem. */
2728         return 0;
2729 }
2730
2731 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2732 {
2733         if (diff_unmodified_pair(p))
2734                 return;
2735
2736         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2737             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2738                 return; /* no tree diffs in patch format */
2739
2740         run_diff(p, o);
2741 }
2742
2743 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2744                             struct diffstat_t *diffstat)
2745 {
2746         if (diff_unmodified_pair(p))
2747                 return;
2748
2749         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2750             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2751                 return; /* no tree diffs in patch format */
2752
2753         run_diffstat(p, o, diffstat);
2754 }
2755
2756 static void diff_flush_checkdiff(struct diff_filepair *p,
2757                 struct diff_options *o)
2758 {
2759         if (diff_unmodified_pair(p))
2760                 return;
2761
2762         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2763             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2764                 return; /* no tree diffs in patch format */
2765
2766         run_checkdiff(p, o);
2767 }
2768
2769 int diff_queue_is_empty(void)
2770 {
2771         struct diff_queue_struct *q = &diff_queued_diff;
2772         int i;
2773         for (i = 0; i < q->nr; i++)
2774                 if (!diff_unmodified_pair(q->queue[i]))
2775                         return 0;
2776         return 1;
2777 }
2778
2779 #if DIFF_DEBUG
2780 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2781 {
2782         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2783                 x, one ? one : "",
2784                 s->path,
2785                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2786                 s->mode,
2787                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2788         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2789                 x, one ? one : "",
2790                 s->size, s->xfrm_flags);
2791 }
2792
2793 void diff_debug_filepair(const struct diff_filepair *p, int i)
2794 {
2795         diff_debug_filespec(p->one, i, "one");
2796         diff_debug_filespec(p->two, i, "two");
2797         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2798                 p->score, p->status ? p->status : '?',
2799                 p->one->rename_used, p->broken_pair);
2800 }
2801
2802 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2803 {
2804         int i;
2805         if (msg)
2806                 fprintf(stderr, "%s\n", msg);
2807         fprintf(stderr, "q->nr = %d\n", q->nr);
2808         for (i = 0; i < q->nr; i++) {
2809                 struct diff_filepair *p = q->queue[i];
2810                 diff_debug_filepair(p, i);
2811         }
2812 }
2813 #endif
2814
2815 static void diff_resolve_rename_copy(void)
2816 {
2817         int i;
2818         struct diff_filepair *p;
2819         struct diff_queue_struct *q = &diff_queued_diff;
2820
2821         diff_debug_queue("resolve-rename-copy", q);
2822
2823         for (i = 0; i < q->nr; i++) {
2824                 p = q->queue[i];
2825                 p->status = 0; /* undecided */
2826                 if (DIFF_PAIR_UNMERGED(p))
2827                         p->status = DIFF_STATUS_UNMERGED;
2828                 else if (!DIFF_FILE_VALID(p->one))
2829                         p->status = DIFF_STATUS_ADDED;
2830                 else if (!DIFF_FILE_VALID(p->two))
2831                         p->status = DIFF_STATUS_DELETED;
2832                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2833                         p->status = DIFF_STATUS_TYPE_CHANGED;
2834
2835                 /* from this point on, we are dealing with a pair
2836                  * whose both sides are valid and of the same type, i.e.
2837                  * either in-place edit or rename/copy edit.
2838                  */
2839                 else if (DIFF_PAIR_RENAME(p)) {
2840                         /*
2841                          * A rename might have re-connected a broken
2842                          * pair up, causing the pathnames to be the
2843                          * same again. If so, that's not a rename at
2844                          * all, just a modification..
2845                          *
2846                          * Otherwise, see if this source was used for
2847                          * multiple renames, in which case we decrement
2848                          * the count, and call it a copy.
2849                          */
2850                         if (!strcmp(p->one->path, p->two->path))
2851                                 p->status = DIFF_STATUS_MODIFIED;
2852                         else if (--p->one->rename_used > 0)
2853                                 p->status = DIFF_STATUS_COPIED;
2854                         else
2855                                 p->status = DIFF_STATUS_RENAMED;
2856                 }
2857                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2858                          p->one->mode != p->two->mode ||
2859                          is_null_sha1(p->one->sha1))
2860                         p->status = DIFF_STATUS_MODIFIED;
2861                 else {
2862                         /* This is a "no-change" entry and should not
2863                          * happen anymore, but prepare for broken callers.
2864                          */
2865                         error("feeding unmodified %s to diffcore",
2866                               p->one->path);
2867                         p->status = DIFF_STATUS_UNKNOWN;
2868                 }
2869         }
2870         diff_debug_queue("resolve-rename-copy done", q);
2871 }
2872
2873 static int check_pair_status(struct diff_filepair *p)
2874 {
2875         switch (p->status) {
2876         case DIFF_STATUS_UNKNOWN:
2877                 return 0;
2878         case 0:
2879                 die("internal error in diff-resolve-rename-copy");
2880         default:
2881                 return 1;
2882         }
2883 }
2884
2885 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2886 {
2887         int fmt = opt->output_format;
2888
2889         if (fmt & DIFF_FORMAT_CHECKDIFF)
2890                 diff_flush_checkdiff(p, opt);
2891         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2892                 diff_flush_raw(p, opt);
2893         else if (fmt & DIFF_FORMAT_NAME) {
2894                 const char *name_a, *name_b;
2895                 name_a = p->two->path;
2896                 name_b = NULL;
2897                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2898                 write_name_quoted(name_a, opt->file, opt->line_termination);
2899         }
2900 }
2901
2902 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2903 {
2904         if (fs->mode)
2905                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2906         else
2907                 fprintf(file, " %s ", newdelete);
2908         write_name_quoted(fs->path, file, '\n');
2909 }
2910
2911
2912 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
2913 {
2914         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2915                 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
2916                         show_name ? ' ' : '\n');
2917                 if (show_name) {
2918                         write_name_quoted(p->two->path, file, '\n');
2919                 }
2920         }
2921 }
2922
2923 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
2924 {
2925         char *names = pprint_rename(p->one->path, p->two->path);
2926
2927         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2928         free(names);
2929         show_mode_change(file, p, 0);
2930 }
2931
2932 static void diff_summary(FILE *file, struct diff_filepair *p)
2933 {
2934         switch(p->status) {
2935         case DIFF_STATUS_DELETED:
2936                 show_file_mode_name(file, "delete", p->one);
2937                 break;
2938         case DIFF_STATUS_ADDED:
2939                 show_file_mode_name(file, "create", p->two);
2940                 break;
2941         case DIFF_STATUS_COPIED:
2942                 show_rename_copy(file, "copy", p);
2943                 break;
2944         case DIFF_STATUS_RENAMED:
2945                 show_rename_copy(file, "rename", p);
2946                 break;
2947         default:
2948                 if (p->score) {
2949                         fputs(" rewrite ", file);
2950                         write_name_quoted(p->two->path, file, ' ');
2951                         fprintf(file, "(%d%%)\n", similarity_index(p));
2952                 }
2953                 show_mode_change(file, p, !p->score);
2954                 break;
2955         }
2956 }
2957
2958 struct patch_id_t {
2959         struct xdiff_emit_state xm;
2960         SHA_CTX *ctx;
2961         int patchlen;
2962 };
2963
2964 static int remove_space(char *line, int len)
2965 {
2966         int i;
2967         char *dst = line;
2968         unsigned char c;
2969
2970         for (i = 0; i < len; i++)
2971                 if (!isspace((c = line[i])))
2972                         *dst++ = c;
2973
2974         return dst - line;
2975 }
2976
2977 static void patch_id_consume(void *priv, char *line, unsigned long len)
2978 {
2979         struct patch_id_t *data = priv;
2980         int new_len;
2981
2982         /* Ignore line numbers when computing the SHA1 of the patch */
2983         if (!prefixcmp(line, "@@ -"))
2984                 return;
2985
2986         new_len = remove_space(line, len);
2987
2988         SHA1_Update(data->ctx, line, new_len);
2989         data->patchlen += new_len;
2990 }
2991
2992 /* returns 0 upon success, and writes result into sha1 */
2993 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2994 {
2995         struct diff_queue_struct *q = &diff_queued_diff;
2996         int i;
2997         SHA_CTX ctx;
2998         struct patch_id_t data;
2999         char buffer[PATH_MAX * 4 + 20];
3000
3001         SHA1_Init(&ctx);
3002         memset(&data, 0, sizeof(struct patch_id_t));
3003         data.ctx = &ctx;
3004         data.xm.consume = patch_id_consume;
3005
3006         for (i = 0; i < q->nr; i++) {
3007                 xpparam_t xpp;
3008                 xdemitconf_t xecfg;
3009                 xdemitcb_t ecb;
3010                 mmfile_t mf1, mf2;
3011                 struct diff_filepair *p = q->queue[i];
3012                 int len1, len2;
3013
3014                 memset(&xecfg, 0, sizeof(xecfg));
3015                 if (p->status == 0)
3016                         return error("internal diff status error");
3017                 if (p->status == DIFF_STATUS_UNKNOWN)
3018                         continue;
3019                 if (diff_unmodified_pair(p))
3020                         continue;
3021                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3022                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3023                         continue;
3024                 if (DIFF_PAIR_UNMERGED(p))
3025                         continue;
3026
3027                 diff_fill_sha1_info(p->one);
3028                 diff_fill_sha1_info(p->two);
3029                 if (fill_mmfile(&mf1, p->one) < 0 ||
3030                                 fill_mmfile(&mf2, p->two) < 0)
3031                         return error("unable to read files to diff");
3032
3033                 len1 = remove_space(p->one->path, strlen(p->one->path));
3034                 len2 = remove_space(p->two->path, strlen(p->two->path));
3035                 if (p->one->mode == 0)
3036                         len1 = snprintf(buffer, sizeof(buffer),
3037                                         "diff--gita/%.*sb/%.*s"
3038                                         "newfilemode%06o"
3039                                         "---/dev/null"
3040                                         "+++b/%.*s",
3041                                         len1, p->one->path,
3042                                         len2, p->two->path,
3043                                         p->two->mode,
3044                                         len2, p->two->path);
3045                 else if (p->two->mode == 0)
3046                         len1 = snprintf(buffer, sizeof(buffer),
3047                                         "diff--gita/%.*sb/%.*s"
3048                                         "deletedfilemode%06o"
3049                                         "---a/%.*s"
3050                                         "+++/dev/null",
3051                                         len1, p->one->path,
3052                                         len2, p->two->path,
3053                                         p->one->mode,
3054                                         len1, p->one->path);
3055                 else
3056                         len1 = snprintf(buffer, sizeof(buffer),
3057                                         "diff--gita/%.*sb/%.*s"
3058                                         "---a/%.*s"
3059                                         "+++b/%.*s",
3060                                         len1, p->one->path,
3061                                         len2, p->two->path,
3062                                         len1, p->one->path,
3063                                         len2, p->two->path);
3064                 SHA1_Update(&ctx, buffer, len1);
3065
3066                 xpp.flags = XDF_NEED_MINIMAL;
3067                 xecfg.ctxlen = 3;
3068                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3069                 ecb.outf = xdiff_outf;
3070                 ecb.priv = &data;
3071                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
3072         }
3073
3074         SHA1_Final(sha1, &ctx);
3075         return 0;
3076 }
3077
3078 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3079 {
3080         struct diff_queue_struct *q = &diff_queued_diff;
3081         int i;
3082         int result = diff_get_patch_id(options, sha1);
3083
3084         for (i = 0; i < q->nr; i++)
3085                 diff_free_filepair(q->queue[i]);
3086
3087         free(q->queue);
3088         q->queue = NULL;
3089         q->nr = q->alloc = 0;
3090
3091         return result;
3092 }
3093
3094 static int is_summary_empty(const struct diff_queue_struct *q)
3095 {
3096         int i;
3097
3098         for (i = 0; i < q->nr; i++) {
3099                 const struct diff_filepair *p = q->queue[i];
3100
3101                 switch (p->status) {
3102                 case DIFF_STATUS_DELETED:
3103                 case DIFF_STATUS_ADDED:
3104                 case DIFF_STATUS_COPIED:
3105                 case DIFF_STATUS_RENAMED:
3106                         return 0;
3107                 default:
3108                         if (p->score)
3109                                 return 0;
3110                         if (p->one->mode && p->two->mode &&
3111                             p->one->mode != p->two->mode)
3112                                 return 0;
3113                         break;
3114                 }
3115         }
3116         return 1;
3117 }
3118
3119 void diff_flush(struct diff_options *options)
3120 {
3121         struct diff_queue_struct *q = &diff_queued_diff;
3122         int i, output_format = options->output_format;
3123         int separator = 0;
3124
3125         /*
3126          * Order: raw, stat, summary, patch
3127          * or:    name/name-status/checkdiff (other bits clear)
3128          */
3129         if (!q->nr)
3130                 goto free_queue;
3131
3132         if (output_format & (DIFF_FORMAT_RAW |
3133                              DIFF_FORMAT_NAME |
3134                              DIFF_FORMAT_NAME_STATUS |
3135                              DIFF_FORMAT_CHECKDIFF)) {
3136                 for (i = 0; i < q->nr; i++) {
3137                         struct diff_filepair *p = q->queue[i];
3138                         if (check_pair_status(p))
3139                                 flush_one_pair(p, options);
3140                 }
3141                 separator++;
3142         }
3143
3144         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3145                 struct diffstat_t diffstat;
3146
3147                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3148                 diffstat.xm.consume = diffstat_consume;
3149                 for (i = 0; i < q->nr; i++) {
3150                         struct diff_filepair *p = q->queue[i];
3151                         if (check_pair_status(p))
3152                                 diff_flush_stat(p, options, &diffstat);
3153                 }
3154                 if (output_format & DIFF_FORMAT_NUMSTAT)
3155                         show_numstat(&diffstat, options);
3156                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3157                         show_stats(&diffstat, options);
3158                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3159                         show_shortstats(&diffstat, options);
3160                 free_diffstat_info(&diffstat);
3161                 separator++;
3162         }
3163         if (output_format & DIFF_FORMAT_DIRSTAT)
3164                 show_dirstat(options);
3165
3166         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3167                 for (i = 0; i < q->nr; i++)
3168                         diff_summary(options->file, q->queue[i]);
3169                 separator++;
3170         }
3171
3172         if (output_format & DIFF_FORMAT_PATCH) {
3173                 if (separator) {
3174                         if (options->stat_sep) {
3175                                 /* attach patch instead of inline */
3176                                 fputs(options->stat_sep, options->file);
3177                         } else {
3178                                 putc(options->line_termination, options->file);
3179                         }
3180                 }
3181
3182                 for (i = 0; i < q->nr; i++) {
3183                         struct diff_filepair *p = q->queue[i];
3184                         if (check_pair_status(p))
3185                                 diff_flush_patch(p, options);
3186                 }
3187         }
3188
3189         if (output_format & DIFF_FORMAT_CALLBACK)
3190                 options->format_callback(q, options, options->format_callback_data);
3191
3192         for (i = 0; i < q->nr; i++)
3193                 diff_free_filepair(q->queue[i]);
3194 free_queue:
3195         free(q->queue);
3196         q->queue = NULL;
3197         q->nr = q->alloc = 0;
3198         if (options->close_file)
3199                 fclose(options->file);
3200 }
3201
3202 static void diffcore_apply_filter(const char *filter)
3203 {
3204         int i;
3205         struct diff_queue_struct *q = &diff_queued_diff;
3206         struct diff_queue_struct outq;
3207         outq.queue = NULL;
3208         outq.nr = outq.alloc = 0;
3209
3210         if (!filter)
3211                 return;
3212
3213         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3214                 int found;
3215                 for (i = found = 0; !found && i < q->nr; i++) {
3216                         struct diff_filepair *p = q->queue[i];
3217                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3218                              ((p->score &&
3219                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3220                               (!p->score &&
3221                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3222                             ((p->status != DIFF_STATUS_MODIFIED) &&
3223                              strchr(filter, p->status)))
3224                                 found++;
3225                 }
3226                 if (found)
3227                         return;
3228
3229                 /* otherwise we will clear the whole queue
3230                  * by copying the empty outq at the end of this
3231                  * function, but first clear the current entries
3232                  * in the queue.
3233                  */
3234                 for (i = 0; i < q->nr; i++)
3235                         diff_free_filepair(q->queue[i]);
3236         }
3237         else {
3238                 /* Only the matching ones */
3239                 for (i = 0; i < q->nr; i++) {
3240                         struct diff_filepair *p = q->queue[i];
3241
3242                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3243                              ((p->score &&
3244                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3245                               (!p->score &&
3246                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3247                             ((p->status != DIFF_STATUS_MODIFIED) &&
3248                              strchr(filter, p->status)))
3249                                 diff_q(&outq, p);
3250                         else
3251                                 diff_free_filepair(p);
3252                 }
3253         }
3254         free(q->queue);
3255         *q = outq;
3256 }
3257
3258 /* Check whether two filespecs with the same mode and size are identical */
3259 static int diff_filespec_is_identical(struct diff_filespec *one,
3260                                       struct diff_filespec *two)
3261 {
3262         if (S_ISGITLINK(one->mode))
3263                 return 0;
3264         if (diff_populate_filespec(one, 0))
3265                 return 0;
3266         if (diff_populate_filespec(two, 0))
3267                 return 0;
3268         return !memcmp(one->data, two->data, one->size);
3269 }
3270
3271 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3272 {
3273         int i;
3274         struct diff_queue_struct *q = &diff_queued_diff;
3275         struct diff_queue_struct outq;
3276         outq.queue = NULL;
3277         outq.nr = outq.alloc = 0;
3278
3279         for (i = 0; i < q->nr; i++) {
3280                 struct diff_filepair *p = q->queue[i];
3281
3282                 /*
3283                  * 1. Entries that come from stat info dirtyness
3284                  *    always have both sides (iow, not create/delete),
3285                  *    one side of the object name is unknown, with
3286                  *    the same mode and size.  Keep the ones that
3287                  *    do not match these criteria.  They have real
3288                  *    differences.
3289                  *
3290                  * 2. At this point, the file is known to be modified,
3291                  *    with the same mode and size, and the object
3292                  *    name of one side is unknown.  Need to inspect
3293                  *    the identical contents.
3294                  */
3295                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3296                     !DIFF_FILE_VALID(p->two) ||
3297                     (p->one->sha1_valid && p->two->sha1_valid) ||
3298                     (p->one->mode != p->two->mode) ||
3299                     diff_populate_filespec(p->one, 1) ||
3300                     diff_populate_filespec(p->two, 1) ||
3301                     (p->one->size != p->two->size) ||
3302                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3303                         diff_q(&outq, p);
3304                 else {
3305                         /*
3306                          * The caller can subtract 1 from skip_stat_unmatch
3307                          * to determine how many paths were dirty only
3308                          * due to stat info mismatch.
3309                          */
3310                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3311                                 diffopt->skip_stat_unmatch++;
3312                         diff_free_filepair(p);
3313                 }
3314         }
3315         free(q->queue);
3316         *q = outq;
3317 }
3318
3319 void diffcore_std(struct diff_options *options)
3320 {
3321         if (DIFF_OPT_TST(options, QUIET))
3322                 return;
3323
3324         if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3325                 diffcore_skip_stat_unmatch(options);
3326         if (options->break_opt != -1)
3327                 diffcore_break(options->break_opt);
3328         if (options->detect_rename)
3329                 diffcore_rename(options);
3330         if (options->break_opt != -1)
3331                 diffcore_merge_broken();
3332         if (options->pickaxe)
3333                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3334         if (options->orderfile)
3335                 diffcore_order(options->orderfile);
3336         diff_resolve_rename_copy();
3337         diffcore_apply_filter(options->filter);
3338
3339         if (diff_queued_diff.nr)
3340                 DIFF_OPT_SET(options, HAS_CHANGES);
3341         else
3342                 DIFF_OPT_CLR(options, HAS_CHANGES);
3343 }
3344
3345 int diff_result_code(struct diff_options *opt, int status)
3346 {
3347         int result = 0;
3348         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3349             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3350                 return status;
3351         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3352             DIFF_OPT_TST(opt, HAS_CHANGES))
3353                 result |= 01;
3354         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3355             DIFF_OPT_TST(opt, CHECK_FAILED))
3356                 result |= 02;
3357         return result;
3358 }
3359
3360 void diff_addremove(struct diff_options *options,
3361                     int addremove, unsigned mode,
3362                     const unsigned char *sha1,
3363                     const char *base, const char *path)
3364 {
3365         char concatpath[PATH_MAX];
3366         struct diff_filespec *one, *two;
3367
3368         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3369                 return;
3370
3371         /* This may look odd, but it is a preparation for
3372          * feeding "there are unchanged files which should
3373          * not produce diffs, but when you are doing copy
3374          * detection you would need them, so here they are"
3375          * entries to the diff-core.  They will be prefixed
3376          * with something like '=' or '*' (I haven't decided
3377          * which but should not make any difference).
3378          * Feeding the same new and old to diff_change()
3379          * also has the same effect.
3380          * Before the final output happens, they are pruned after
3381          * merged into rename/copy pairs as appropriate.
3382          */
3383         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3384                 addremove = (addremove == '+' ? '-' :
3385                              addremove == '-' ? '+' : addremove);
3386
3387         if (!path) path = "";
3388         sprintf(concatpath, "%s%s", base, path);
3389
3390         if (options->prefix &&
3391             strncmp(concatpath, options->prefix, options->prefix_length))
3392                 return;
3393
3394         one = alloc_filespec(concatpath);
3395         two = alloc_filespec(concatpath);
3396
3397         if (addremove != '+')
3398                 fill_filespec(one, sha1, mode);
3399         if (addremove != '-')
3400                 fill_filespec(two, sha1, mode);
3401
3402         diff_queue(&diff_queued_diff, one, two);
3403         DIFF_OPT_SET(options, HAS_CHANGES);
3404 }
3405
3406 void diff_change(struct diff_options *options,
3407                  unsigned old_mode, unsigned new_mode,
3408                  const unsigned char *old_sha1,
3409                  const unsigned char *new_sha1,
3410                  const char *base, const char *path)
3411 {
3412         char concatpath[PATH_MAX];
3413         struct diff_filespec *one, *two;
3414
3415         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3416                         && S_ISGITLINK(new_mode))
3417                 return;
3418
3419         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3420                 unsigned tmp;
3421                 const unsigned char *tmp_c;
3422                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3423                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3424         }
3425         if (!path) path = "";
3426         sprintf(concatpath, "%s%s", base, path);
3427
3428         if (options->prefix &&
3429             strncmp(concatpath, options->prefix, options->prefix_length))
3430                 return;
3431
3432         one = alloc_filespec(concatpath);
3433         two = alloc_filespec(concatpath);
3434         fill_filespec(one, old_sha1, old_mode);
3435         fill_filespec(two, new_sha1, new_mode);
3436
3437         diff_queue(&diff_queued_diff, one, two);
3438         DIFF_OPT_SET(options, HAS_CHANGES);
3439 }
3440
3441 void diff_unmerge(struct diff_options *options,
3442                   const char *path,
3443                   unsigned mode, const unsigned char *sha1)
3444 {
3445         struct diff_filespec *one, *two;
3446
3447         if (options->prefix &&
3448             strncmp(path, options->prefix, options->prefix_length))
3449                 return;
3450
3451         one = alloc_filespec(path);
3452         two = alloc_filespec(path);
3453         fill_filespec(one, sha1, mode);
3454         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3455 }