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

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "quote.h"
   6 #include "diff.h"
   7 #include "diffcore.h"
   8 #include "delta.h"
   9 #include "xdiff-interface.h"
  10 #include "color.h"
  11 #include "attr.h"
  12 #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 struct funcname_pattern_entry {
  98         char *name;
  99         char *pattern;
 100         int cflags;
 101 };
 102 static struct funcname_pattern_list {
 103         struct funcname_pattern_list *next;
 104         struct funcname_pattern_entry e;
 105 } *funcname_pattern_list;
 106
 107 static int parse_funcname_pattern(const char *var, const char *ep, const char *value, int cflags)
 108 {
 109         const char *name;
 110         int namelen;
 111         struct funcname_pattern_list *pp;
 112
 113         name = var + 5; /* "diff." */
 114         namelen = ep - name;
 115
 116         for (pp = funcname_pattern_list; pp; pp = pp->next)
 117                 if (!strncmp(pp->e.name, name, namelen) && !pp->e.name[namelen])
 118                         break;
 119         if (!pp) {
 120                 pp = xcalloc(1, sizeof(*pp));
 121                 pp->e.name = xmemdupz(name, namelen);
 122                 pp->next = funcname_pattern_list;
 123                 funcname_pattern_list = pp;
 124         }
 125         free(pp->e.pattern);
 126         pp->e.pattern = xstrdup(value);
 127         pp->e.cflags = cflags;
 128         return 0;
 129 }
 130
 131 /*
 132  * These are to give UI layer defaults.
 133  * The core-level commands such as git-diff-files should
 134  * never be affected by the setting of diff.renames
 135  * the user happens to have in the configuration file.
 136  */
 137 int git_diff_ui_config(const char *var, const char *value, void *cb)
 138 {
 139         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 140                 diff_use_color_default = git_config_colorbool(var, value, -1);
 141                 return 0;
 142         }
 143         if (!strcmp(var, "diff.renames")) {
 144                 if (!value)
 145                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 146                 else if (!strcasecmp(value, "copies") ||
 147                          !strcasecmp(value, "copy"))
 148                         diff_detect_rename_default = DIFF_DETECT_COPY;
 149                 else if (git_config_bool(var,value))
 150                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 151                 return 0;
 152         }
 153         if (!strcmp(var, "diff.autorefreshindex")) {
 154                 diff_auto_refresh_index = git_config_bool(var, value);
 155                 return 0;
 156         }
 157         if (!strcmp(var, "diff.external"))
 158                 return git_config_string(&external_diff_cmd_cfg, var, value);
 159         if (!prefixcmp(var, "diff.")) {
 160                 const char *ep = strrchr(var, '.');
 161
 162                 if (ep != var + 4 && !strcmp(ep, ".command"))
 163                         return parse_lldiff_command(var, ep, value);
 164         }
 165
 166         return git_diff_basic_config(var, value, cb);
 167 }
 168
 169 int git_diff_basic_config(const char *var, const char *value, void *cb)
 170 {
 171         if (!strcmp(var, "diff.renamelimit")) {
 172                 diff_rename_limit_default = git_config_int(var, value);
 173                 return 0;
 174         }
 175
 176         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 177                 int slot = parse_diff_color_slot(var, 11);
 178                 if (!value)
 179                         return config_error_nonbool(var);
 180                 color_parse(value, var, diff_colors[slot]);
 181                 return 0;
 182         }
 183
 184         if (!prefixcmp(var, "diff.")) {
 185                 const char *ep = strrchr(var, '.');
 186                 if (ep != var + 4) {
 187                         if (!strcmp(ep, ".funcname")) {
 188                                 if (!value)
 189                                         return config_error_nonbool(var);
 190                                 return parse_funcname_pattern(var, ep, value,
 191                                         0);
 192                         } else if (!strcmp(ep, ".xfuncname")) {
 193                                 if (!value)
 194                                         return config_error_nonbool(var);
 195                                 return parse_funcname_pattern(var, ep, value,
 196                                         REG_EXTENDED);
 197                         }
 198                 }
 199         }
 200
 201         return git_color_default_config(var, value, cb);
 202 }
 203
 204 static char *quote_two(const char *one, const char *two)
 205 {
 206         int need_one = quote_c_style(one, NULL, NULL, 1);
 207         int need_two = quote_c_style(two, NULL, NULL, 1);
 208         struct strbuf res;
 209
 210         strbuf_init(&res, 0);
 211         if (need_one + need_two) {
 212                 strbuf_addch(&res, '"');
 213                 quote_c_style(one, &res, NULL, 1);
 214                 quote_c_style(two, &res, NULL, 1);
 215                 strbuf_addch(&res, '"');
 216         } else {
 217                 strbuf_addstr(&res, one);
 218                 strbuf_addstr(&res, two);
 219         }
 220         return strbuf_detach(&res, NULL);
 221 }
 222
 223 static const char *external_diff(void)
 224 {
 225         static const char *external_diff_cmd = NULL;
 226         static int done_preparing = 0;
 227
 228         if (done_preparing)
 229                 return external_diff_cmd;
 230         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 231         if (!external_diff_cmd)
 232                 external_diff_cmd = external_diff_cmd_cfg;
 233         done_preparing = 1;
 234         return external_diff_cmd;
 235 }
 236
 237 static struct diff_tempfile {
 238         const char *name; /* filename external diff should read from */
 239         char hex[41];
 240         char mode[10];
 241         char tmp_path[PATH_MAX];
 242 } diff_temp[2];
 243
 244 static int count_lines(const char *data, int size)
 245 {
 246         int count, ch, completely_empty = 1, nl_just_seen = 0;
 247         count = 0;
 248         while (0 < size--) {
 249                 ch = *data++;
 250                 if (ch == '\n') {
 251                         count++;
 252                         nl_just_seen = 1;
 253                         completely_empty = 0;
 254                 }
 255                 else {
 256                         nl_just_seen = 0;
 257                         completely_empty = 0;
 258                 }
 259         }
 260         if (completely_empty)
 261                 return 0;
 262         if (!nl_just_seen)
 263                 count++; /* no trailing newline */
 264         return count;
 265 }
 266
 267 static void print_line_count(FILE *file, int count)
 268 {
 269         switch (count) {
 270         case 0:
 271                 fprintf(file, "0,0");
 272                 break;
 273         case 1:
 274                 fprintf(file, "1");
 275                 break;
 276         default:
 277                 fprintf(file, "1,%d", count);
 278                 break;
 279         }
 280 }
 281
 282 static void copy_file_with_prefix(FILE *file,
 283                                   int prefix, const char *data, int size,
 284                                   const char *set, const char *reset)
 285 {
 286         int ch, nl_just_seen = 1;
 287         while (0 < size--) {
 288                 ch = *data++;
 289                 if (nl_just_seen) {
 290                         fputs(set, file);
 291                         putc(prefix, file);
 292                 }
 293                 if (ch == '\n') {
 294                         nl_just_seen = 1;
 295                         fputs(reset, file);
 296                 } else
 297                         nl_just_seen = 0;
 298                 putc(ch, file);
 299         }
 300         if (!nl_just_seen)
 301                 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
 302 }
 303
 304 static void emit_rewrite_diff(const char *name_a,
 305                               const char *name_b,
 306                               struct diff_filespec *one,
 307                               struct diff_filespec *two,
 308                               struct diff_options *o)
 309 {
 310         int lc_a, lc_b;
 311         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
 312         const char *name_a_tab, *name_b_tab;
 313         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 314         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 315         const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
 316         const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
 317         const char *reset = diff_get_color(color_diff, DIFF_RESET);
 318         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 319
 320         name_a += (*name_a == '/');
 321         name_b += (*name_b == '/');
 322         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 323         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 324
 325         strbuf_reset(&a_name);
 326         strbuf_reset(&b_name);
 327         quote_two_c_style(&a_name, o->a_prefix, name_a, 0);
 328         quote_two_c_style(&b_name, o->b_prefix, name_b, 0);
 329
 330         diff_populate_filespec(one, 0);
 331         diff_populate_filespec(two, 0);
 332         lc_a = count_lines(one->data, one->size);
 333         lc_b = count_lines(two->data, two->size);
 334         fprintf(o->file,
 335                 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
 336                 metainfo, a_name.buf, name_a_tab, reset,
 337                 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
 338         print_line_count(o->file, lc_a);
 339         fprintf(o->file, " +");
 340         print_line_count(o->file, lc_b);
 341         fprintf(o->file, " @@%s\n", reset);
 342         if (lc_a)
 343                 copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
 344         if (lc_b)
 345                 copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
 346 }
 347
 348 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 349 {
 350         if (!DIFF_FILE_VALID(one)) {
 351                 mf->ptr = (char *)""; /* does not matter */
 352                 mf->size = 0;
 353                 return 0;
 354         }
 355         else if (diff_populate_filespec(one, 0))
 356                 return -1;
 357         mf->ptr = one->data;
 358         mf->size = one->size;
 359         return 0;
 360 }
 361
 362 struct diff_words_buffer {
 363         mmfile_t text;
 364         long alloc;
 365         long current; /* output pointer */
 366         int suppressed_newline;
 367 };
 368
 369 static void diff_words_append(char *line, unsigned long len,
 370                 struct diff_words_buffer *buffer)
 371 {
 372         if (buffer->text.size + len > buffer->alloc) {
 373                 buffer->alloc = (buffer->text.size + len) * 3 / 2;
 374                 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
 375         }
 376         line++;
 377         len--;
 378         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 379         buffer->text.size += len;
 380 }
 381
 382 struct diff_words_data {
 383         struct xdiff_emit_state xm;
 384         struct diff_words_buffer minus, plus;
 385         FILE *file;
 386 };
 387
 388 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
 389                 int suppress_newline)
 390 {
 391         const char *ptr;
 392         int eol = 0;
 393
 394         if (len == 0)
 395                 return;
 396
 397         ptr  = buffer->text.ptr + buffer->current;
 398         buffer->current += len;
 399
 400         if (ptr[len - 1] == '\n') {
 401                 eol = 1;
 402                 len--;
 403         }
 404
 405         fputs(diff_get_color(1, color), file);
 406         fwrite(ptr, len, 1, file);
 407         fputs(diff_get_color(1, DIFF_RESET), file);
 408
 409         if (eol) {
 410                 if (suppress_newline)
 411                         buffer->suppressed_newline = 1;
 412                 else
 413                         putc('\n', file);
 414         }
 415 }
 416
 417 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 418 {
 419         struct diff_words_data *diff_words = priv;
 420
 421         if (diff_words->minus.suppressed_newline) {
 422                 if (line[0] != '+')
 423                         putc('\n', diff_words->file);
 424                 diff_words->minus.suppressed_newline = 0;
 425         }
 426
 427         len--;
 428         switch (line[0]) {
 429                 case '-':
 430                         print_word(diff_words->file,
 431                                    &diff_words->minus, len, DIFF_FILE_OLD, 1);
 432                         break;
 433                 case '+':
 434                         print_word(diff_words->file,
 435                                    &diff_words->plus, len, DIFF_FILE_NEW, 0);
 436                         break;
 437                 case ' ':
 438                         print_word(diff_words->file,
 439                                    &diff_words->plus, len, DIFF_PLAIN, 0);
 440                         diff_words->minus.current += len;
 441                         break;
 442         }
 443 }
 444
 445 /* this executes the word diff on the accumulated buffers */
 446 static void diff_words_show(struct diff_words_data *diff_words)
 447 {
 448         xpparam_t xpp;
 449         xdemitconf_t xecfg;
 450         xdemitcb_t ecb;
 451         mmfile_t minus, plus;
 452         int i;
 453
 454         memset(&xecfg, 0, sizeof(xecfg));
 455         minus.size = diff_words->minus.text.size;
 456         minus.ptr = xmalloc(minus.size);
 457         memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
 458         for (i = 0; i < minus.size; i++)
 459                 if (isspace(minus.ptr[i]))
 460                         minus.ptr[i] = '\n';
 461         diff_words->minus.current = 0;
 462
 463         plus.size = diff_words->plus.text.size;
 464         plus.ptr = xmalloc(plus.size);
 465         memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
 466         for (i = 0; i < plus.size; i++)
 467                 if (isspace(plus.ptr[i]))
 468                         plus.ptr[i] = '\n';
 469         diff_words->plus.current = 0;
 470
 471         xpp.flags = XDF_NEED_MINIMAL;
 472         xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
 473         ecb.outf = xdiff_outf;
 474         ecb.priv = diff_words;
 475         diff_words->xm.consume = fn_out_diff_words_aux;
 476         xdi_diff(&minus, &plus, &xpp, &xecfg, &ecb);
 477
 478         free(minus.ptr);
 479         free(plus.ptr);
 480         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 481
 482         if (diff_words->minus.suppressed_newline) {
 483                 putc('\n', diff_words->file);
 484                 diff_words->minus.suppressed_newline = 0;
 485         }
 486 }
 487
 488 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 489
 490 struct emit_callback {
 491         struct xdiff_emit_state xm;
 492         int nparents, color_diff;
 493         unsigned ws_rule;
 494         sane_truncate_fn truncate;
 495         const char **label_path;
 496         struct diff_words_data *diff_words;
 497         int *found_changesp;
 498         FILE *file;
 499 };
 500
 501 static void free_diff_words_data(struct emit_callback *ecbdata)
 502 {
 503         if (ecbdata->diff_words) {
 504                 /* flush buffers */
 505                 if (ecbdata->diff_words->minus.text.size ||
 506                                 ecbdata->diff_words->plus.text.size)
 507                         diff_words_show(ecbdata->diff_words);
 508
 509                 free (ecbdata->diff_words->minus.text.ptr);
 510                 free (ecbdata->diff_words->plus.text.ptr);
 511                 free(ecbdata->diff_words);
 512                 ecbdata->diff_words = NULL;
 513         }
 514 }
 515
 516 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 517 {
 518         if (diff_use_color)
 519                 return diff_colors[ix];
 520         return "";
 521 }
 522
 523 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
 524 {
 525         int has_trailing_newline, has_trailing_carriage_return;
 526
 527         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 528         if (has_trailing_newline)
 529                 len--;
 530         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 531         if (has_trailing_carriage_return)
 532                 len--;
 533
 534         fputs(set, file);
 535         fwrite(line, len, 1, file);
 536         fputs(reset, file);
 537         if (has_trailing_carriage_return)
 538                 fputc('\r', file);
 539         if (has_trailing_newline)
 540                 fputc('\n', file);
 541 }
 542
 543 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
 544 {
 545         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 546         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 547
 548         if (!*ws)
 549                 emit_line(ecbdata->file, set, reset, line, len);
 550         else {
 551                 /* Emit just the prefix, then the rest. */
 552                 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
 553                 ws_check_emit(line + ecbdata->nparents,
 554                               len - ecbdata->nparents, ecbdata->ws_rule,
 555                               ecbdata->file, set, reset, ws);
 556         }
 557 }
 558
 559 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
 560 {
 561         const char *cp;
 562         unsigned long allot;
 563         size_t l = len;
 564
 565         if (ecb->truncate)
 566                 return ecb->truncate(line, len);
 567         cp = line;
 568         allot = l;
 569         while (0 < l) {
 570                 (void) utf8_width(&cp, &l);
 571                 if (!cp)
 572                         break; /* truncated in the middle? */
 573         }
 574         return allot - l;
 575 }
 576
 577 static void fn_out_consume(void *priv, char *line, unsigned long len)
 578 {
 579         int i;
 580         int color;
 581         struct emit_callback *ecbdata = priv;
 582         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 583         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 584         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 585
 586         *(ecbdata->found_changesp) = 1;
 587
 588         if (ecbdata->label_path[0]) {
 589                 const char *name_a_tab, *name_b_tab;
 590
 591                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 592                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 593
 594                 fprintf(ecbdata->file, "%s--- %s%s%s\n",
 595                         meta, ecbdata->label_path[0], reset, name_a_tab);
 596                 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
 597                         meta, ecbdata->label_path[1], reset, name_b_tab);
 598                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 599         }
 600
 601         /* This is not really necessary for now because
 602          * this codepath only deals with two-way diffs.
 603          */
 604         for (i = 0; i < len && line[i] == '@'; i++)
 605                 ;
 606         if (2 <= i && i < len && line[i] == ' ') {
 607                 ecbdata->nparents = i - 1;
 608                 len = sane_truncate_line(ecbdata, line, len);
 609                 emit_line(ecbdata->file,
 610                           diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
 611                           reset, line, len);
 612                 if (line[len-1] != '\n')
 613                         putc('\n', ecbdata->file);
 614                 return;
 615         }
 616
 617         if (len < ecbdata->nparents) {
 618                 emit_line(ecbdata->file, reset, reset, line, len);
 619                 return;
 620         }
 621
 622         color = DIFF_PLAIN;
 623         if (ecbdata->diff_words && ecbdata->nparents != 1)
 624                 /* fall back to normal diff */
 625                 free_diff_words_data(ecbdata);
 626         if (ecbdata->diff_words) {
 627                 if (line[0] == '-') {
 628                         diff_words_append(line, len,
 629                                           &ecbdata->diff_words->minus);
 630                         return;
 631                 } else if (line[0] == '+') {
 632                         diff_words_append(line, len,
 633                                           &ecbdata->diff_words->plus);
 634                         return;
 635                 }
 636                 if (ecbdata->diff_words->minus.text.size ||
 637                     ecbdata->diff_words->plus.text.size)
 638                         diff_words_show(ecbdata->diff_words);
 639                 line++;
 640                 len--;
 641                 emit_line(ecbdata->file, plain, reset, line, len);
 642                 return;
 643         }
 644         for (i = 0; i < ecbdata->nparents && len; i++) {
 645                 if (line[i] == '-')
 646                         color = DIFF_FILE_OLD;
 647                 else if (line[i] == '+')
 648                         color = DIFF_FILE_NEW;
 649         }
 650
 651         if (color != DIFF_FILE_NEW) {
 652                 emit_line(ecbdata->file,
 653                           diff_get_color(ecbdata->color_diff, color),
 654                           reset, line, len);
 655                 return;
 656         }
 657         emit_add_line(reset, ecbdata, line, len);
 658 }
 659
 660 static char *pprint_rename(const char *a, const char *b)
 661 {
 662         const char *old = a;
 663         const char *new = b;
 664         struct strbuf name;
 665         int pfx_length, sfx_length;
 666         int len_a = strlen(a);
 667         int len_b = strlen(b);
 668         int a_midlen, b_midlen;
 669         int qlen_a = quote_c_style(a, NULL, NULL, 0);
 670         int qlen_b = quote_c_style(b, NULL, NULL, 0);
 671
 672         strbuf_init(&name, 0);
 673         if (qlen_a || qlen_b) {
 674                 quote_c_style(a, &name, NULL, 0);
 675                 strbuf_addstr(&name, " => ");
 676                 quote_c_style(b, &name, NULL, 0);
 677                 return strbuf_detach(&name, NULL);
 678         }
 679
 680         /* Find common prefix */
 681         pfx_length = 0;
 682         while (*old && *new && *old == *new) {
 683                 if (*old == '/')
 684                         pfx_length = old - a + 1;
 685                 old++;
 686                 new++;
 687         }
 688
 689         /* Find common suffix */
 690         old = a + len_a;
 691         new = b + len_b;
 692         sfx_length = 0;
 693         while (a <= old && b <= new && *old == *new) {
 694                 if (*old == '/')
 695                         sfx_length = len_a - (old - a);
 696                 old--;
 697                 new--;
 698         }
 699
 700         /*
 701          * pfx{mid-a => mid-b}sfx
 702          * {pfx-a => pfx-b}sfx
 703          * pfx{sfx-a => sfx-b}
 704          * name-a => name-b
 705          */
 706         a_midlen = len_a - pfx_length - sfx_length;
 707         b_midlen = len_b - pfx_length - sfx_length;
 708         if (a_midlen < 0)
 709                 a_midlen = 0;
 710         if (b_midlen < 0)
 711                 b_midlen = 0;
 712
 713         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 714         if (pfx_length + sfx_length) {
 715                 strbuf_add(&name, a, pfx_length);
 716                 strbuf_addch(&name, '{');
 717         }
 718         strbuf_add(&name, a + pfx_length, a_midlen);
 719         strbuf_addstr(&name, " => ");
 720         strbuf_add(&name, b + pfx_length, b_midlen);
 721         if (pfx_length + sfx_length) {
 722                 strbuf_addch(&name, '}');
 723                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 724         }
 725         return strbuf_detach(&name, NULL);
 726 }
 727
 728 struct diffstat_t {
 729         struct xdiff_emit_state xm;
 730
 731         int nr;
 732         int alloc;
 733         struct diffstat_file {
 734                 char *from_name;
 735                 char *name;
 736                 char *print_name;
 737                 unsigned is_unmerged:1;
 738                 unsigned is_binary:1;
 739                 unsigned is_renamed:1;
 740                 unsigned int added, deleted;
 741         } **files;
 742 };
 743
 744 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 745                                           const char *name_a,
 746                                           const char *name_b)
 747 {
 748         struct diffstat_file *x;
 749         x = xcalloc(sizeof (*x), 1);
 750         if (diffstat->nr == diffstat->alloc) {
 751                 diffstat->alloc = alloc_nr(diffstat->alloc);
 752                 diffstat->files = xrealloc(diffstat->files,
 753                                 diffstat->alloc * sizeof(x));
 754         }
 755         diffstat->files[diffstat->nr++] = x;
 756         if (name_b) {
 757                 x->from_name = xstrdup(name_a);
 758                 x->name = xstrdup(name_b);
 759                 x->is_renamed = 1;
 760         }
 761         else {
 762                 x->from_name = NULL;
 763                 x->name = xstrdup(name_a);
 764         }
 765         return x;
 766 }
 767
 768 static void diffstat_consume(void *priv, char *line, unsigned long len)
 769 {
 770         struct diffstat_t *diffstat = priv;
 771         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 772
 773         if (line[0] == '+')
 774                 x->added++;
 775         else if (line[0] == '-')
 776                 x->deleted++;
 777 }
 778
 779 const char mime_boundary_leader[] = "------------";
 780
 781 static int scale_linear(int it, int width, int max_change)
 782 {
 783         /*
 784          * make sure that at least one '-' is printed if there were deletions,
 785          * and likewise for '+'.
 786          */
 787         if (max_change < 2)
 788                 return it;
 789         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
 790 }
 791
 792 static void show_name(FILE *file,
 793                       const char *prefix, const char *name, int len,
 794                       const char *reset, const char *set)
 795 {
 796         fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
 797 }
 798
 799 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
 800 {
 801         if (cnt <= 0)
 802                 return;
 803         fprintf(file, "%s", set);
 804         while (cnt--)
 805                 putc(ch, file);
 806         fprintf(file, "%s", reset);
 807 }
 808
 809 static void fill_print_name(struct diffstat_file *file)
 810 {
 811         char *pname;
 812
 813         if (file->print_name)
 814                 return;
 815
 816         if (!file->is_renamed) {
 817                 struct strbuf buf;
 818                 strbuf_init(&buf, 0);
 819                 if (quote_c_style(file->name, &buf, NULL, 0)) {
 820                         pname = strbuf_detach(&buf, NULL);
 821                 } else {
 822                         pname = file->name;
 823                         strbuf_release(&buf);
 824                 }
 825         } else {
 826                 pname = pprint_rename(file->from_name, file->name);
 827         }
 828         file->print_name = pname;
 829 }
 830
 831 static void show_stats(struct diffstat_t* data, struct diff_options *options)
 832 {
 833         int i, len, add, del, total, adds = 0, dels = 0;
 834         int max_change = 0, max_len = 0;
 835         int total_files = data->nr;
 836         int width, name_width;
 837         const char *reset, *set, *add_c, *del_c;
 838
 839         if (data->nr == 0)
 840                 return;
 841
 842         width = options->stat_width ? options->stat_width : 80;
 843         name_width = options->stat_name_width ? options->stat_name_width : 50;
 844
 845         /* Sanity: give at least 5 columns to the graph,
 846          * but leave at least 10 columns for the name.
 847          */
 848         if (width < 25)
 849                 width = 25;
 850         if (name_width < 10)
 851                 name_width = 10;
 852         else if (width < name_width + 15)
 853                 name_width = width - 15;
 854
 855         /* Find the longest filename and max number of changes */
 856         reset = diff_get_color_opt(options, DIFF_RESET);
 857         set   = diff_get_color_opt(options, DIFF_PLAIN);
 858         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
 859         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
 860
 861         for (i = 0; i < data->nr; i++) {
 862                 struct diffstat_file *file = data->files[i];
 863                 int change = file->added + file->deleted;
 864                 fill_print_name(file);
 865                 len = strlen(file->print_name);
 866                 if (max_len < len)
 867                         max_len = len;
 868
 869                 if (file->is_binary || file->is_unmerged)
 870                         continue;
 871                 if (max_change < change)
 872                         max_change = change;
 873         }
 874
 875         /* Compute the width of the graph part;
 876          * 10 is for one blank at the beginning of the line plus
 877          * " | count " between the name and the graph.
 878          *
 879          * From here on, name_width is the width of the name area,
 880          * and width is the width of the graph area.
 881          */
 882         name_width = (name_width < max_len) ? name_width : max_len;
 883         if (width < (name_width + 10) + max_change)
 884                 width = width - (name_width + 10);
 885         else
 886                 width = max_change;
 887
 888         for (i = 0; i < data->nr; i++) {
 889                 const char *prefix = "";
 890                 char *name = data->files[i]->print_name;
 891                 int added = data->files[i]->added;
 892                 int deleted = data->files[i]->deleted;
 893                 int name_len;
 894
 895                 /*
 896                  * "scale" the filename
 897                  */
 898                 len = name_width;
 899                 name_len = strlen(name);
 900                 if (name_width < name_len) {
 901                         char *slash;
 902                         prefix = "...";
 903                         len -= 3;
 904                         name += name_len - len;
 905                         slash = strchr(name, '/');
 906                         if (slash)
 907                                 name = slash;
 908                 }
 909
 910                 if (data->files[i]->is_binary) {
 911                         show_name(options->file, prefix, name, len, reset, set);
 912                         fprintf(options->file, "  Bin ");
 913                         fprintf(options->file, "%s%d%s", del_c, deleted, reset);
 914                         fprintf(options->file, " -> ");
 915                         fprintf(options->file, "%s%d%s", add_c, added, reset);
 916                         fprintf(options->file, " bytes");
 917                         fprintf(options->file, "\n");
 918                         continue;
 919                 }
 920                 else if (data->files[i]->is_unmerged) {
 921                         show_name(options->file, prefix, name, len, reset, set);
 922                         fprintf(options->file, "  Unmerged\n");
 923                         continue;
 924                 }
 925                 else if (!data->files[i]->is_renamed &&
 926                          (added + deleted == 0)) {
 927                         total_files--;
 928                         continue;
 929                 }
 930
 931                 /*
 932                  * scale the add/delete
 933                  */
 934                 add = added;
 935                 del = deleted;
 936                 total = add + del;
 937                 adds += add;
 938                 dels += del;
 939
 940                 if (width <= max_change) {
 941                         add = scale_linear(add, width, max_change);
 942                         del = scale_linear(del, width, max_change);
 943                         total = add + del;
 944                 }
 945                 show_name(options->file, prefix, name, len, reset, set);
 946                 fprintf(options->file, "%5d%s", added + deleted,
 947                                 added + deleted ? " " : "");
 948                 show_graph(options->file, '+', add, add_c, reset);
 949                 show_graph(options->file, '-', del, del_c, reset);
 950                 fprintf(options->file, "\n");
 951         }
 952         fprintf(options->file,
 953                "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
 954                set, total_files, adds, dels, reset);
 955 }
 956
 957 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
 958 {
 959         int i, adds = 0, dels = 0, total_files = data->nr;
 960
 961         if (data->nr == 0)
 962                 return;
 963
 964         for (i = 0; i < data->nr; i++) {
 965                 if (!data->files[i]->is_binary &&
 966                     !data->files[i]->is_unmerged) {
 967                         int added = data->files[i]->added;
 968                         int deleted= data->files[i]->deleted;
 969                         if (!data->files[i]->is_renamed &&
 970                             (added + deleted == 0)) {
 971                                 total_files--;
 972                         } else {
 973                                 adds += added;
 974                                 dels += deleted;
 975                         }
 976                 }
 977         }
 978         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
 979                total_files, adds, dels);
 980 }
 981
 982 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
 983 {
 984         int i;
 985
 986         if (data->nr == 0)
 987                 return;
 988
 989         for (i = 0; i < data->nr; i++) {
 990                 struct diffstat_file *file = data->files[i];
 991
 992                 if (file->is_binary)
 993                         fprintf(options->file, "-\t-\t");
 994                 else
 995                         fprintf(options->file,
 996                                 "%d\t%d\t", file->added, file->deleted);
 997                 if (options->line_termination) {
 998                         fill_print_name(file);
 999                         if (!file->is_renamed)
1000                                 write_name_quoted(file->name, options->file,
1001                                                   options->line_termination);
1002                         else {
1003                                 fputs(file->print_name, options->file);
1004                                 putc(options->line_termination, options->file);
1005                         }
1006                 } else {
1007                         if (file->is_renamed) {
1008                                 putc('\0', options->file);
1009                                 write_name_quoted(file->from_name, options->file, '\0');
1010                         }
1011                         write_name_quoted(file->name, options->file, '\0');
1012                 }
1013         }
1014 }
1015
1016 struct dirstat_file {
1017         const char *name;
1018         unsigned long changed;
1019 };
1020
1021 struct dirstat_dir {
1022         struct dirstat_file *files;
1023         int alloc, nr, percent, cumulative;
1024 };
1025
1026 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1027 {
1028         unsigned long this_dir = 0;
1029         unsigned int sources = 0;
1030
1031         while (dir->nr) {
1032                 struct dirstat_file *f = dir->files;
1033                 int namelen = strlen(f->name);
1034                 unsigned long this;
1035                 char *slash;
1036
1037                 if (namelen < baselen)
1038                         break;
1039                 if (memcmp(f->name, base, baselen))
1040                         break;
1041                 slash = strchr(f->name + baselen, '/');
1042                 if (slash) {
1043                         int newbaselen = slash + 1 - f->name;
1044                         this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1045                         sources++;
1046                 } else {
1047                         this = f->changed;
1048                         dir->files++;
1049                         dir->nr--;
1050                         sources += 2;
1051                 }
1052                 this_dir += this;
1053         }
1054
1055         /*
1056          * We don't report dirstat's for
1057          *  - the top level
1058          *  - or cases where everything came from a single directory
1059          *    under this directory (sources == 1).
1060          */
1061         if (baselen && sources != 1) {
1062                 int permille = this_dir * 1000 / changed;
1063                 if (permille) {
1064                         int percent = permille / 10;
1065                         if (percent >= dir->percent) {
1066                                 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1067                                 if (!dir->cumulative)
1068                                         return 0;
1069                         }
1070                 }
1071         }
1072         return this_dir;
1073 }
1074
1075 static int dirstat_compare(const void *_a, const void *_b)
1076 {
1077         const struct dirstat_file *a = _a;
1078         const struct dirstat_file *b = _b;
1079         return strcmp(a->name, b->name);
1080 }
1081
1082 static void show_dirstat(struct diff_options *options)
1083 {
1084         int i;
1085         unsigned long changed;
1086         struct dirstat_dir dir;
1087         struct diff_queue_struct *q = &diff_queued_diff;
1088
1089         dir.files = NULL;
1090         dir.alloc = 0;
1091         dir.nr = 0;
1092         dir.percent = options->dirstat_percent;
1093         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1094
1095         changed = 0;
1096         for (i = 0; i < q->nr; i++) {
1097                 struct diff_filepair *p = q->queue[i];
1098                 const char *name;
1099                 unsigned long copied, added, damage;
1100
1101                 name = p->one->path ? p->one->path : p->two->path;
1102
1103                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1104                         diff_populate_filespec(p->one, 0);
1105                         diff_populate_filespec(p->two, 0);
1106                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1107                                                &copied, &added);
1108                         diff_free_filespec_data(p->one);
1109                         diff_free_filespec_data(p->two);
1110                 } else if (DIFF_FILE_VALID(p->one)) {
1111                         diff_populate_filespec(p->one, 1);
1112                         copied = added = 0;
1113                         diff_free_filespec_data(p->one);
1114                 } else if (DIFF_FILE_VALID(p->two)) {
1115                         diff_populate_filespec(p->two, 1);
1116                         copied = 0;
1117                         added = p->two->size;
1118                         diff_free_filespec_data(p->two);
1119                 } else
1120                         continue;
1121
1122                 /*
1123                  * Original minus copied is the removed material,
1124                  * added is the new material.  They are both damages
1125                  * made to the preimage.
1126                  */
1127                 damage = (p->one->size - copied) + added;
1128
1129                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1130                 dir.files[dir.nr].name = name;
1131                 dir.files[dir.nr].changed = damage;
1132                 changed += damage;
1133                 dir.nr++;
1134         }
1135
1136         /* This can happen even with many files, if everything was renames */
1137         if (!changed)
1138                 return;
1139
1140         /* Show all directories with more than x% of the changes */
1141         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1142         gather_dirstat(options->file, &dir, changed, "", 0);
1143 }
1144
1145 static void free_diffstat_info(struct diffstat_t *diffstat)
1146 {
1147         int i;
1148         for (i = 0; i < diffstat->nr; i++) {
1149                 struct diffstat_file *f = diffstat->files[i];
1150                 if (f->name != f->print_name)
1151                         free(f->print_name);
1152                 free(f->name);
1153                 free(f->from_name);
1154                 free(f);
1155         }
1156         free(diffstat->files);
1157 }
1158
1159 struct checkdiff_t {
1160         struct xdiff_emit_state xm;
1161         const char *filename;
1162         int lineno;
1163         struct diff_options *o;
1164         unsigned ws_rule;
1165         unsigned status;
1166         int trailing_blanks_start;
1167 };
1168
1169 static int is_conflict_marker(const char *line, unsigned long len)
1170 {
1171         char firstchar;
1172         int cnt;
1173
1174         if (len < 8)
1175                 return 0;
1176         firstchar = line[0];
1177         switch (firstchar) {
1178         case '=': case '>': case '<':
1179                 break;
1180         default:
1181                 return 0;
1182         }
1183         for (cnt = 1; cnt < 7; cnt++)
1184                 if (line[cnt] != firstchar)
1185                         return 0;
1186         /* line[0] thru line[6] are same as firstchar */
1187         if (firstchar == '=') {
1188                 /* divider between ours and theirs? */
1189                 if (len != 8 || line[7] != '\n')
1190                         return 0;
1191         } else if (len < 8 || !isspace(line[7])) {
1192                 /* not divider before ours nor after theirs */
1193                 return 0;
1194         }
1195         return 1;
1196 }
1197
1198 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1199 {
1200         struct checkdiff_t *data = priv;
1201         int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1202         const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1203         const char *reset = diff_get_color(color_diff, DIFF_RESET);
1204         const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1205         char *err;
1206
1207         if (line[0] == '+') {
1208                 unsigned bad;
1209                 data->lineno++;
1210                 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1211                         data->trailing_blanks_start = 0;
1212                 else if (!data->trailing_blanks_start)
1213                         data->trailing_blanks_start = data->lineno;
1214                 if (is_conflict_marker(line + 1, len - 1)) {
1215                         data->status |= 1;
1216                         fprintf(data->o->file,
1217                                 "%s:%d: leftover conflict marker\n",
1218                                 data->filename, data->lineno);
1219                 }
1220                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1221                 if (!bad)
1222                         return;
1223                 data->status |= bad;
1224                 err = whitespace_error_string(bad);
1225                 fprintf(data->o->file, "%s:%d: %s.\n",
1226                         data->filename, data->lineno, err);
1227                 free(err);
1228                 emit_line(data->o->file, set, reset, line, 1);
1229                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1230                               data->o->file, set, reset, ws);
1231         } else if (line[0] == ' ') {
1232                 data->lineno++;
1233                 data->trailing_blanks_start = 0;
1234         } else if (line[0] == '@') {
1235                 char *plus = strchr(line, '+');
1236                 if (plus)
1237                         data->lineno = strtol(plus, NULL, 10) - 1;
1238                 else
1239                         die("invalid diff");
1240                 data->trailing_blanks_start = 0;
1241         }
1242 }
1243
1244 static unsigned char *deflate_it(char *data,
1245                                  unsigned long size,
1246                                  unsigned long *result_size)
1247 {
1248         int bound;
1249         unsigned char *deflated;
1250         z_stream stream;
1251
1252         memset(&stream, 0, sizeof(stream));
1253         deflateInit(&stream, zlib_compression_level);
1254         bound = deflateBound(&stream, size);
1255         deflated = xmalloc(bound);
1256         stream.next_out = deflated;
1257         stream.avail_out = bound;
1258
1259         stream.next_in = (unsigned char *)data;
1260         stream.avail_in = size;
1261         while (deflate(&stream, Z_FINISH) == Z_OK)
1262                 ; /* nothing */
1263         deflateEnd(&stream);
1264         *result_size = stream.total_out;
1265         return deflated;
1266 }
1267
1268 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1269 {
1270         void *cp;
1271         void *delta;
1272         void *deflated;
1273         void *data;
1274         unsigned long orig_size;
1275         unsigned long delta_size;
1276         unsigned long deflate_size;
1277         unsigned long data_size;
1278
1279         /* We could do deflated delta, or we could do just deflated two,
1280          * whichever is smaller.
1281          */
1282         delta = NULL;
1283         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1284         if (one->size && two->size) {
1285                 delta = diff_delta(one->ptr, one->size,
1286                                    two->ptr, two->size,
1287                                    &delta_size, deflate_size);
1288                 if (delta) {
1289                         void *to_free = delta;
1290                         orig_size = delta_size;
1291                         delta = deflate_it(delta, delta_size, &delta_size);
1292                         free(to_free);
1293                 }
1294         }
1295
1296         if (delta && delta_size < deflate_size) {
1297                 fprintf(file, "delta %lu\n", orig_size);
1298                 free(deflated);
1299                 data = delta;
1300                 data_size = delta_size;
1301         }
1302         else {
1303                 fprintf(file, "literal %lu\n", two->size);
1304                 free(delta);
1305                 data = deflated;
1306                 data_size = deflate_size;
1307         }
1308
1309         /* emit data encoded in base85 */
1310         cp = data;
1311         while (data_size) {
1312                 int bytes = (52 < data_size) ? 52 : data_size;
1313                 char line[70];
1314                 data_size -= bytes;
1315                 if (bytes <= 26)
1316                         line[0] = bytes + 'A' - 1;
1317                 else
1318                         line[0] = bytes - 26 + 'a' - 1;
1319                 encode_85(line + 1, cp, bytes);
1320                 cp = (char *) cp + bytes;
1321                 fputs(line, file);
1322                 fputc('\n', file);
1323         }
1324         fprintf(file, "\n");
1325         free(data);
1326 }
1327
1328 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1329 {
1330         fprintf(file, "GIT binary patch\n");
1331         emit_binary_diff_body(file, one, two);
1332         emit_binary_diff_body(file, two, one);
1333 }
1334
1335 static void setup_diff_attr_check(struct git_attr_check *check)
1336 {
1337         static struct git_attr *attr_diff;
1338
1339         if (!attr_diff) {
1340                 attr_diff = git_attr("diff", 4);
1341         }
1342         check[0].attr = attr_diff;
1343 }
1344
1345 static void diff_filespec_check_attr(struct diff_filespec *one)
1346 {
1347         struct git_attr_check attr_diff_check;
1348         int check_from_data = 0;
1349
1350         if (one->checked_attr)
1351                 return;
1352
1353         setup_diff_attr_check(&attr_diff_check);
1354         one->is_binary = 0;
1355         one->funcname_pattern_ident = NULL;
1356
1357         if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1358                 const char *value;
1359
1360                 /* binaryness */
1361                 value = attr_diff_check.value;
1362                 if (ATTR_TRUE(value))
1363                         ;
1364                 else if (ATTR_FALSE(value))
1365                         one->is_binary = 1;
1366                 else
1367                         check_from_data = 1;
1368
1369                 /* funcname pattern ident */
1370                 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1371                         ;
1372                 else
1373                         one->funcname_pattern_ident = value;
1374         }
1375
1376         if (check_from_data) {
1377                 if (!one->data && DIFF_FILE_VALID(one))
1378                         diff_populate_filespec(one, 0);
1379
1380                 if (one->data)
1381                         one->is_binary = buffer_is_binary(one->data, one->size);
1382         }
1383 }
1384
1385 int diff_filespec_is_binary(struct diff_filespec *one)
1386 {
1387         diff_filespec_check_attr(one);
1388         return one->is_binary;
1389 }
1390
1391 static const struct funcname_pattern_entry *funcname_pattern(const char *ident)
1392 {
1393         struct funcname_pattern_list *pp;
1394
1395         for (pp = funcname_pattern_list; pp; pp = pp->next)
1396                 if (!strcmp(ident, pp->e.name))
1397                         return &pp->e;
1398         return NULL;
1399 }
1400
1401 static const struct funcname_pattern_entry builtin_funcname_pattern[] = {
1402         { "java",
1403           "!^[ \t]*(catch|do|for|if|instanceof|new|return|switch|throw|while)\n"
1404           "^[ \t]*(([ \t]*[A-Za-z_][A-Za-z_0-9]*){2,}[ \t]*\\([^;]*)$",
1405           REG_EXTENDED },
1406         { "pascal",
1407           "^((procedure|function|constructor|destructor|interface|"
1408                 "implementation|initialization|finalization)[ \t]*.*)$"
1409           "|"
1410           "^(.*=[ \t]*(class|record).*)$",
1411           REG_EXTENDED },
1412         { "bibtex", "(@[a-zA-Z]{1,}[ \t]*\\{{0,1}[ \t]*[^ \t\"@',\\#}{~%]*).*$",
1413           REG_EXTENDED },
1414         { "tex",
1415           "^(\\\\((sub)*section|chapter|part)\\*{0,1}\\{.*)$",
1416           REG_EXTENDED },
1417         { "ruby", "^[ \t]*((class|module|def)[ \t].*)$",
1418           REG_EXTENDED },
1419 };
1420
1421 static const struct funcname_pattern_entry *diff_funcname_pattern(struct diff_filespec *one)
1422 {
1423         const char *ident;
1424         const struct funcname_pattern_entry *pe;
1425         int i;
1426
1427         diff_filespec_check_attr(one);
1428         ident = one->funcname_pattern_ident;
1429
1430         if (!ident)
1431                 /*
1432                  * If the config file has "funcname.default" defined, that
1433                  * regexp is used; otherwise NULL is returned and xemit uses
1434                  * the built-in default.
1435                  */
1436                 return funcname_pattern("default");
1437
1438         /* Look up custom "funcname.$ident" regexp from config. */
1439         pe = funcname_pattern(ident);
1440         if (pe)
1441                 return pe;
1442
1443         /*
1444          * And define built-in fallback patterns here.  Note that
1445          * these can be overridden by the user's config settings.
1446          */
1447         for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1448                 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1449                         return &builtin_funcname_pattern[i];
1450
1451         return NULL;
1452 }
1453
1454 static void builtin_diff(const char *name_a,
1455                          const char *name_b,
1456                          struct diff_filespec *one,
1457                          struct diff_filespec *two,
1458                          const char *xfrm_msg,
1459                          struct diff_options *o,
1460                          int complete_rewrite)
1461 {
1462         mmfile_t mf1, mf2;
1463         const char *lbl[2];
1464         char *a_one, *b_two;
1465         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1466         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1467
1468         /* Never use a non-valid filename anywhere if at all possible */
1469         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1470         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1471
1472         a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1473         b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1474         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1475         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1476         fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1477         if (lbl[0][0] == '/') {
1478                 /* /dev/null */
1479                 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1480                 if (xfrm_msg && xfrm_msg[0])
1481                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1482         }
1483         else if (lbl[1][0] == '/') {
1484                 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1485                 if (xfrm_msg && xfrm_msg[0])
1486                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1487         }
1488         else {
1489                 if (one->mode != two->mode) {
1490                         fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1491                         fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1492                 }
1493                 if (xfrm_msg && xfrm_msg[0])
1494                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1495                 /*
1496                  * we do not run diff between different kind
1497                  * of objects.
1498                  */
1499                 if ((one->mode ^ two->mode) & S_IFMT)
1500                         goto free_ab_and_return;
1501                 if (complete_rewrite) {
1502                         emit_rewrite_diff(name_a, name_b, one, two, o);
1503                         o->found_changes = 1;
1504                         goto free_ab_and_return;
1505                 }
1506         }
1507
1508         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1509                 die("unable to read files to diff");
1510
1511         if (!DIFF_OPT_TST(o, TEXT) &&
1512             (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1513                 /* Quite common confusing case */
1514                 if (mf1.size == mf2.size &&
1515                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1516                         goto free_ab_and_return;
1517                 if (DIFF_OPT_TST(o, BINARY))
1518                         emit_binary_diff(o->file, &mf1, &mf2);
1519                 else
1520                         fprintf(o->file, "Binary files %s and %s differ\n",
1521                                 lbl[0], lbl[1]);
1522                 o->found_changes = 1;
1523         }
1524         else {
1525                 /* Crazy xdl interfaces.. */
1526                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1527                 xpparam_t xpp;
1528                 xdemitconf_t xecfg;
1529                 xdemitcb_t ecb;
1530                 struct emit_callback ecbdata;
1531                 const struct funcname_pattern_entry *pe;
1532
1533                 pe = diff_funcname_pattern(one);
1534                 if (!pe)
1535                         pe = diff_funcname_pattern(two);
1536
1537                 memset(&xecfg, 0, sizeof(xecfg));
1538                 memset(&ecbdata, 0, sizeof(ecbdata));
1539                 ecbdata.label_path = lbl;
1540                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1541                 ecbdata.found_changesp = &o->found_changes;
1542                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1543                 ecbdata.file = o->file;
1544                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1545                 xecfg.ctxlen = o->context;
1546                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1547                 if (pe)
1548                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1549                 if (!diffopts)
1550                         ;
1551                 else if (!prefixcmp(diffopts, "--unified="))
1552                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1553                 else if (!prefixcmp(diffopts, "-u"))
1554                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1555                 ecb.outf = xdiff_outf;
1556                 ecb.priv = &ecbdata;
1557                 ecbdata.xm.consume = fn_out_consume;
1558                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1559                         ecbdata.diff_words =
1560                                 xcalloc(1, sizeof(struct diff_words_data));
1561                         ecbdata.diff_words->file = o->file;
1562                 }
1563                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1564                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1565                         free_diff_words_data(&ecbdata);
1566         }
1567
1568  free_ab_and_return:
1569         diff_free_filespec_data(one);
1570         diff_free_filespec_data(two);
1571         free(a_one);
1572         free(b_two);
1573         return;
1574 }
1575
1576 static void builtin_diffstat(const char *name_a, const char *name_b,
1577                              struct diff_filespec *one,
1578                              struct diff_filespec *two,
1579                              struct diffstat_t *diffstat,
1580                              struct diff_options *o,
1581                              int complete_rewrite)
1582 {
1583         mmfile_t mf1, mf2;
1584         struct diffstat_file *data;
1585
1586         data = diffstat_add(diffstat, name_a, name_b);
1587
1588         if (!one || !two) {
1589                 data->is_unmerged = 1;
1590                 return;
1591         }
1592         if (complete_rewrite) {
1593                 diff_populate_filespec(one, 0);
1594                 diff_populate_filespec(two, 0);
1595                 data->deleted = count_lines(one->data, one->size);
1596                 data->added = count_lines(two->data, two->size);
1597                 goto free_and_return;
1598         }
1599         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1600                 die("unable to read files to diff");
1601
1602         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1603                 data->is_binary = 1;
1604                 data->added = mf2.size;
1605                 data->deleted = mf1.size;
1606         } else {
1607                 /* Crazy xdl interfaces.. */
1608                 xpparam_t xpp;
1609                 xdemitconf_t xecfg;
1610                 xdemitcb_t ecb;
1611
1612                 memset(&xecfg, 0, sizeof(xecfg));
1613                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1614                 ecb.outf = xdiff_outf;
1615                 ecb.priv = diffstat;
1616                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1617         }
1618
1619  free_and_return:
1620         diff_free_filespec_data(one);
1621         diff_free_filespec_data(two);
1622 }
1623
1624 static void builtin_checkdiff(const char *name_a, const char *name_b,
1625                               const char *attr_path,
1626                               struct diff_filespec *one,
1627                               struct diff_filespec *two,
1628                               struct diff_options *o)
1629 {
1630         mmfile_t mf1, mf2;
1631         struct checkdiff_t data;
1632
1633         if (!two)
1634                 return;
1635
1636         memset(&data, 0, sizeof(data));
1637         data.xm.consume = checkdiff_consume;
1638         data.filename = name_b ? name_b : name_a;
1639         data.lineno = 0;
1640         data.o = o;
1641         data.ws_rule = whitespace_rule(attr_path);
1642
1643         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1644                 die("unable to read files to diff");
1645
1646         /*
1647          * All the other codepaths check both sides, but not checking
1648          * the "old" side here is deliberate.  We are checking the newly
1649          * introduced changes, and as long as the "new" side is text, we
1650          * can and should check what it introduces.
1651          */
1652         if (diff_filespec_is_binary(two))
1653                 goto free_and_return;
1654         else {
1655                 /* Crazy xdl interfaces.. */
1656                 xpparam_t xpp;
1657                 xdemitconf_t xecfg;
1658                 xdemitcb_t ecb;
1659
1660                 memset(&xecfg, 0, sizeof(xecfg));
1661                 xecfg.ctxlen = 1; /* at least one context line */
1662                 xpp.flags = XDF_NEED_MINIMAL;
1663                 ecb.outf = xdiff_outf;
1664                 ecb.priv = &data;
1665                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1666
1667                 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1668                     data.trailing_blanks_start) {
1669                         fprintf(o->file, "%s:%d: ends with blank lines.\n",
1670                                 data.filename, data.trailing_blanks_start);
1671                         data.status = 1; /* report errors */
1672                 }
1673         }
1674  free_and_return:
1675         diff_free_filespec_data(one);
1676         diff_free_filespec_data(two);
1677         if (data.status)
1678                 DIFF_OPT_SET(o, CHECK_FAILED);
1679 }
1680
1681 struct diff_filespec *alloc_filespec(const char *path)
1682 {
1683         int namelen = strlen(path);
1684         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1685
1686         memset(spec, 0, sizeof(*spec));
1687         spec->path = (char *)(spec + 1);
1688         memcpy(spec->path, path, namelen+1);
1689         spec->count = 1;
1690         return spec;
1691 }
1692
1693 void free_filespec(struct diff_filespec *spec)
1694 {
1695         if (!--spec->count) {
1696                 diff_free_filespec_data(spec);
1697                 free(spec);
1698         }
1699 }
1700
1701 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1702                    unsigned short mode)
1703 {
1704         if (mode) {
1705                 spec->mode = canon_mode(mode);
1706                 hashcpy(spec->sha1, sha1);
1707                 spec->sha1_valid = !is_null_sha1(sha1);
1708         }
1709 }
1710
1711 /*
1712  * Given a name and sha1 pair, if the index tells us the file in
1713  * the work tree has that object contents, return true, so that
1714  * prepare_temp_file() does not have to inflate and extract.
1715  */
1716 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1717 {
1718         struct cache_entry *ce;
1719         struct stat st;
1720         int pos, len;
1721
1722         /* We do not read the cache ourselves here, because the
1723          * benchmark with my previous version that always reads cache
1724          * shows that it makes things worse for diff-tree comparing
1725          * two linux-2.6 kernel trees in an already checked out work
1726          * tree.  This is because most diff-tree comparisons deal with
1727          * only a small number of files, while reading the cache is
1728          * expensive for a large project, and its cost outweighs the
1729          * savings we get by not inflating the object to a temporary
1730          * file.  Practically, this code only helps when we are used
1731          * by diff-cache --cached, which does read the cache before
1732          * calling us.
1733          */
1734         if (!active_cache)
1735                 return 0;
1736
1737         /* We want to avoid the working directory if our caller
1738          * doesn't need the data in a normal file, this system
1739          * is rather slow with its stat/open/mmap/close syscalls,
1740          * and the object is contained in a pack file.  The pack
1741          * is probably already open and will be faster to obtain
1742          * the data through than the working directory.  Loose
1743          * objects however would tend to be slower as they need
1744          * to be individually opened and inflated.
1745          */
1746         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1747                 return 0;
1748
1749         len = strlen(name);
1750         pos = cache_name_pos(name, len);
1751         if (pos < 0)
1752                 return 0;
1753         ce = active_cache[pos];
1754
1755         /*
1756          * This is not the sha1 we are looking for, or
1757          * unreusable because it is not a regular file.
1758          */
1759         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1760                 return 0;
1761
1762         /*
1763          * If ce matches the file in the work tree, we can reuse it.
1764          */
1765         if (ce_uptodate(ce) ||
1766             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1767                 return 1;
1768
1769         return 0;
1770 }
1771
1772 static int populate_from_stdin(struct diff_filespec *s)
1773 {
1774         struct strbuf buf;
1775         size_t size = 0;
1776
1777         strbuf_init(&buf, 0);
1778         if (strbuf_read(&buf, 0, 0) < 0)
1779                 return error("error while reading from stdin %s",
1780                                      strerror(errno));
1781
1782         s->should_munmap = 0;
1783         s->data = strbuf_detach(&buf, &size);
1784         s->size = size;
1785         s->should_free = 1;
1786         return 0;
1787 }
1788
1789 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1790 {
1791         int len;
1792         char *data = xmalloc(100);
1793         len = snprintf(data, 100,
1794                 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1795         s->data = data;
1796         s->size = len;
1797         s->should_free = 1;
1798         if (size_only) {
1799                 s->data = NULL;
1800                 free(data);
1801         }
1802         return 0;
1803 }
1804
1805 /*
1806  * While doing rename detection and pickaxe operation, we may need to
1807  * grab the data for the blob (or file) for our own in-core comparison.
1808  * diff_filespec has data and size fields for this purpose.
1809  */
1810 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1811 {
1812         int err = 0;
1813         if (!DIFF_FILE_VALID(s))
1814                 die("internal error: asking to populate invalid file.");
1815         if (S_ISDIR(s->mode))
1816                 return -1;
1817
1818         if (s->data)
1819                 return 0;
1820
1821         if (size_only && 0 < s->size)
1822                 return 0;
1823
1824         if (S_ISGITLINK(s->mode))
1825                 return diff_populate_gitlink(s, size_only);
1826
1827         if (!s->sha1_valid ||
1828             reuse_worktree_file(s->path, s->sha1, 0)) {
1829                 struct strbuf buf;
1830                 struct stat st;
1831                 int fd;
1832
1833                 if (!strcmp(s->path, "-"))
1834                         return populate_from_stdin(s);
1835
1836                 if (lstat(s->path, &st) < 0) {
1837                         if (errno == ENOENT) {
1838                         err_empty:
1839                                 err = -1;
1840                         empty:
1841                                 s->data = (char *)"";
1842                                 s->size = 0;
1843                                 return err;
1844                         }
1845                 }
1846                 s->size = xsize_t(st.st_size);
1847                 if (!s->size)
1848                         goto empty;
1849                 if (size_only)
1850                         return 0;
1851                 if (S_ISLNK(st.st_mode)) {
1852                         int ret;
1853                         s->data = xmalloc(s->size);
1854                         s->should_free = 1;
1855                         ret = readlink(s->path, s->data, s->size);
1856                         if (ret < 0) {
1857                                 free(s->data);
1858                                 goto err_empty;
1859                         }
1860                         return 0;
1861                 }
1862                 fd = open(s->path, O_RDONLY);
1863                 if (fd < 0)
1864                         goto err_empty;
1865                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1866                 close(fd);
1867                 s->should_munmap = 1;
1868
1869                 /*
1870                  * Convert from working tree format to canonical git format
1871                  */
1872                 strbuf_init(&buf, 0);
1873                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1874                         size_t size = 0;
1875                         munmap(s->data, s->size);
1876                         s->should_munmap = 0;
1877                         s->data = strbuf_detach(&buf, &size);
1878                         s->size = size;
1879                         s->should_free = 1;
1880                 }
1881         }
1882         else {
1883                 enum object_type type;
1884                 if (size_only)
1885                         type = sha1_object_info(s->sha1, &s->size);
1886                 else {
1887                         s->data = read_sha1_file(s->sha1, &type, &s->size);
1888                         s->should_free = 1;
1889                 }
1890         }
1891         return 0;
1892 }
1893
1894 void diff_free_filespec_blob(struct diff_filespec *s)
1895 {
1896         if (s->should_free)
1897                 free(s->data);
1898         else if (s->should_munmap)
1899                 munmap(s->data, s->size);
1900
1901         if (s->should_free || s->should_munmap) {
1902                 s->should_free = s->should_munmap = 0;
1903                 s->data = NULL;
1904         }
1905 }
1906
1907 void diff_free_filespec_data(struct diff_filespec *s)
1908 {
1909         diff_free_filespec_blob(s);
1910         free(s->cnt_data);
1911         s->cnt_data = NULL;
1912 }
1913
1914 static void prep_temp_blob(struct diff_tempfile *temp,
1915                            void *blob,
1916                            unsigned long size,
1917                            const unsigned char *sha1,
1918                            int mode)
1919 {
1920         int fd;
1921
1922         fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1923         if (fd < 0)
1924                 die("unable to create temp-file: %s", strerror(errno));
1925         if (write_in_full(fd, blob, size) != size)
1926                 die("unable to write temp-file");
1927         close(fd);
1928         temp->name = temp->tmp_path;
1929         strcpy(temp->hex, sha1_to_hex(sha1));
1930         temp->hex[40] = 0;
1931         sprintf(temp->mode, "%06o", mode);
1932 }
1933
1934 static void prepare_temp_file(const char *name,
1935                               struct diff_tempfile *temp,
1936                               struct diff_filespec *one)
1937 {
1938         if (!DIFF_FILE_VALID(one)) {
1939         not_a_valid_file:
1940                 /* A '-' entry produces this for file-2, and
1941                  * a '+' entry produces this for file-1.
1942                  */
1943                 temp->name = "/dev/null";
1944                 strcpy(temp->hex, ".");
1945                 strcpy(temp->mode, ".");
1946                 return;
1947         }
1948
1949         if (!one->sha1_valid ||
1950             reuse_worktree_file(name, one->sha1, 1)) {
1951                 struct stat st;
1952                 if (lstat(name, &st) < 0) {
1953                         if (errno == ENOENT)
1954                                 goto not_a_valid_file;
1955                         die("stat(%s): %s", name, strerror(errno));
1956                 }
1957                 if (S_ISLNK(st.st_mode)) {
1958                         int ret;
1959                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1960                         size_t sz = xsize_t(st.st_size);
1961                         if (sizeof(buf) <= st.st_size)
1962                                 die("symlink too long: %s", name);
1963                         ret = readlink(name, buf, sz);
1964                         if (ret < 0)
1965                                 die("readlink(%s)", name);
1966                         prep_temp_blob(temp, buf, sz,
1967                                        (one->sha1_valid ?
1968                                         one->sha1 : null_sha1),
1969                                        (one->sha1_valid ?
1970                                         one->mode : S_IFLNK));
1971                 }
1972                 else {
1973                         /* we can borrow from the file in the work tree */
1974                         temp->name = name;
1975                         if (!one->sha1_valid)
1976                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
1977                         else
1978                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
1979                         /* Even though we may sometimes borrow the
1980                          * contents from the work tree, we always want
1981                          * one->mode.  mode is trustworthy even when
1982                          * !(one->sha1_valid), as long as
1983                          * DIFF_FILE_VALID(one).
1984                          */
1985                         sprintf(temp->mode, "%06o", one->mode);
1986                 }
1987                 return;
1988         }
1989         else {
1990                 if (diff_populate_filespec(one, 0))
1991                         die("cannot read data blob for %s", one->path);
1992                 prep_temp_blob(temp, one->data, one->size,
1993                                one->sha1, one->mode);
1994         }
1995 }
1996
1997 static void remove_tempfile(void)
1998 {
1999         int i;
2000
2001         for (i = 0; i < 2; i++)
2002                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
2003                         unlink(diff_temp[i].name);
2004                         diff_temp[i].name = NULL;
2005                 }
2006 }
2007
2008 static void remove_tempfile_on_signal(int signo)
2009 {
2010         remove_tempfile();
2011         signal(SIGINT, SIG_DFL);
2012         raise(signo);
2013 }
2014
2015 /* An external diff command takes:
2016  *
2017  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2018  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2019  *
2020  */
2021 static void run_external_diff(const char *pgm,
2022                               const char *name,
2023                               const char *other,
2024                               struct diff_filespec *one,
2025                               struct diff_filespec *two,
2026                               const char *xfrm_msg,
2027                               int complete_rewrite)
2028 {
2029         const char *spawn_arg[10];
2030         struct diff_tempfile *temp = diff_temp;
2031         int retval;
2032         static int atexit_asked = 0;
2033         const char *othername;
2034         const char **arg = &spawn_arg[0];
2035
2036         othername = (other? other : name);
2037         if (one && two) {
2038                 prepare_temp_file(name, &temp[0], one);
2039                 prepare_temp_file(othername, &temp[1], two);
2040                 if (! atexit_asked &&
2041                     (temp[0].name == temp[0].tmp_path ||
2042                      temp[1].name == temp[1].tmp_path)) {
2043                         atexit_asked = 1;
2044                         atexit(remove_tempfile);
2045                 }
2046                 signal(SIGINT, remove_tempfile_on_signal);
2047         }
2048
2049         if (one && two) {
2050                 *arg++ = pgm;
2051                 *arg++ = name;
2052                 *arg++ = temp[0].name;
2053                 *arg++ = temp[0].hex;
2054                 *arg++ = temp[0].mode;
2055                 *arg++ = temp[1].name;
2056                 *arg++ = temp[1].hex;
2057                 *arg++ = temp[1].mode;
2058                 if (other) {
2059                         *arg++ = other;
2060                         *arg++ = xfrm_msg;
2061                 }
2062         } else {
2063                 *arg++ = pgm;
2064                 *arg++ = name;
2065         }
2066         *arg = NULL;
2067         fflush(NULL);
2068         retval = run_command_v_opt(spawn_arg, 0);
2069         remove_tempfile();
2070         if (retval) {
2071                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2072                 exit(1);
2073         }
2074 }
2075
2076 static const char *external_diff_attr(const char *name)
2077 {
2078         struct git_attr_check attr_diff_check;
2079
2080         if (!name)
2081                 return NULL;
2082
2083         setup_diff_attr_check(&attr_diff_check);
2084         if (!git_checkattr(name, 1, &attr_diff_check)) {
2085                 const char *value = attr_diff_check.value;
2086                 if (!ATTR_TRUE(value) &&
2087                     !ATTR_FALSE(value) &&
2088                     !ATTR_UNSET(value)) {
2089                         struct ll_diff_driver *drv;
2090
2091                         for (drv = user_diff; drv; drv = drv->next)
2092                                 if (!strcmp(drv->name, value))
2093                                         return drv->cmd;
2094                 }
2095         }
2096         return NULL;
2097 }
2098
2099 static void run_diff_cmd(const char *pgm,
2100                          const char *name,
2101                          const char *other,
2102                          const char *attr_path,
2103                          struct diff_filespec *one,
2104                          struct diff_filespec *two,
2105                          const char *xfrm_msg,
2106                          struct diff_options *o,
2107                          int complete_rewrite)
2108 {
2109         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2110                 pgm = NULL;
2111         else {
2112                 const char *cmd = external_diff_attr(attr_path);
2113                 if (cmd)
2114                         pgm = cmd;
2115         }
2116
2117         if (pgm) {
2118                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2119                                   complete_rewrite);
2120                 return;
2121         }
2122         if (one && two)
2123                 builtin_diff(name, other ? other : name,
2124                              one, two, xfrm_msg, o, complete_rewrite);
2125         else
2126                 fprintf(o->file, "* Unmerged path %s\n", name);
2127 }
2128
2129 static void diff_fill_sha1_info(struct diff_filespec *one)
2130 {
2131         if (DIFF_FILE_VALID(one)) {
2132                 if (!one->sha1_valid) {
2133                         struct stat st;
2134                         if (!strcmp(one->path, "-")) {
2135                                 hashcpy(one->sha1, null_sha1);
2136                                 return;
2137                         }
2138                         if (lstat(one->path, &st) < 0)
2139                                 die("stat %s", one->path);
2140                         if (index_path(one->sha1, one->path, &st, 0))
2141                                 die("cannot hash %s\n", one->path);
2142                 }
2143         }
2144         else
2145                 hashclr(one->sha1);
2146 }
2147
2148 static int similarity_index(struct diff_filepair *p)
2149 {
2150         return p->score * 100 / MAX_SCORE;
2151 }
2152
2153 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2154 {
2155         /* Strip the prefix but do not molest /dev/null and absolute paths */
2156         if (*namep && **namep != '/')
2157                 *namep += prefix_length;
2158         if (*otherp && **otherp != '/')
2159                 *otherp += prefix_length;
2160 }
2161
2162 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2163 {
2164         const char *pgm = external_diff();
2165         struct strbuf msg;
2166         char *xfrm_msg;
2167         struct diff_filespec *one = p->one;
2168         struct diff_filespec *two = p->two;
2169         const char *name;
2170         const char *other;
2171         const char *attr_path;
2172         int complete_rewrite = 0;
2173
2174         name  = p->one->path;
2175         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2176         attr_path = name;
2177         if (o->prefix_length)
2178                 strip_prefix(o->prefix_length, &name, &other);
2179
2180         if (DIFF_PAIR_UNMERGED(p)) {
2181                 run_diff_cmd(pgm, name, NULL, attr_path,
2182                              NULL, NULL, NULL, o, 0);
2183                 return;
2184         }
2185
2186         diff_fill_sha1_info(one);
2187         diff_fill_sha1_info(two);
2188
2189         strbuf_init(&msg, PATH_MAX * 2 + 300);
2190         switch (p->status) {
2191         case DIFF_STATUS_COPIED:
2192                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2193                 strbuf_addstr(&msg, "\ncopy from ");
2194                 quote_c_style(name, &msg, NULL, 0);
2195                 strbuf_addstr(&msg, "\ncopy to ");
2196                 quote_c_style(other, &msg, NULL, 0);
2197                 strbuf_addch(&msg, '\n');
2198                 break;
2199         case DIFF_STATUS_RENAMED:
2200                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2201                 strbuf_addstr(&msg, "\nrename from ");
2202                 quote_c_style(name, &msg, NULL, 0);
2203                 strbuf_addstr(&msg, "\nrename to ");
2204                 quote_c_style(other, &msg, NULL, 0);
2205                 strbuf_addch(&msg, '\n');
2206                 break;
2207         case DIFF_STATUS_MODIFIED:
2208                 if (p->score) {
2209                         strbuf_addf(&msg, "dissimilarity index %d%%\n",
2210                                         similarity_index(p));
2211                         complete_rewrite = 1;
2212                         break;
2213                 }
2214                 /* fallthru */
2215         default:
2216                 /* nothing */
2217                 ;
2218         }
2219
2220         if (hashcmp(one->sha1, two->sha1)) {
2221                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2222
2223                 if (DIFF_OPT_TST(o, BINARY)) {
2224                         mmfile_t mf;
2225                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2226                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2227                                 abbrev = 40;
2228                 }
2229                 strbuf_addf(&msg, "index %.*s..%.*s",
2230                                 abbrev, sha1_to_hex(one->sha1),
2231                                 abbrev, sha1_to_hex(two->sha1));
2232                 if (one->mode == two->mode)
2233                         strbuf_addf(&msg, " %06o", one->mode);
2234                 strbuf_addch(&msg, '\n');
2235         }
2236
2237         if (msg.len)
2238                 strbuf_setlen(&msg, msg.len - 1);
2239         xfrm_msg = msg.len ? msg.buf : NULL;
2240
2241         if (!pgm &&
2242             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2243             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2244                 /* a filepair that changes between file and symlink
2245                  * needs to be split into deletion and creation.
2246                  */
2247                 struct diff_filespec *null = alloc_filespec(two->path);
2248                 run_diff_cmd(NULL, name, other, attr_path,
2249                              one, null, xfrm_msg, o, 0);
2250                 free(null);
2251                 null = alloc_filespec(one->path);
2252                 run_diff_cmd(NULL, name, other, attr_path,
2253                              null, two, xfrm_msg, o, 0);
2254                 free(null);
2255         }
2256         else
2257                 run_diff_cmd(pgm, name, other, attr_path,
2258                              one, two, xfrm_msg, o, complete_rewrite);
2259
2260         strbuf_release(&msg);
2261 }
2262
2263 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2264                          struct diffstat_t *diffstat)
2265 {
2266         const char *name;
2267         const char *other;
2268         int complete_rewrite = 0;
2269
2270         if (DIFF_PAIR_UNMERGED(p)) {
2271                 /* unmerged */
2272                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2273                 return;
2274         }
2275
2276         name = p->one->path;
2277         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2278
2279         if (o->prefix_length)
2280                 strip_prefix(o->prefix_length, &name, &other);
2281
2282         diff_fill_sha1_info(p->one);
2283         diff_fill_sha1_info(p->two);
2284
2285         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2286                 complete_rewrite = 1;
2287         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2288 }
2289
2290 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2291 {
2292         const char *name;
2293         const char *other;
2294         const char *attr_path;
2295
2296         if (DIFF_PAIR_UNMERGED(p)) {
2297                 /* unmerged */
2298                 return;
2299         }
2300
2301         name = p->one->path;
2302         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2303         attr_path = other ? other : name;
2304
2305         if (o->prefix_length)
2306                 strip_prefix(o->prefix_length, &name, &other);
2307
2308         diff_fill_sha1_info(p->one);
2309         diff_fill_sha1_info(p->two);
2310
2311         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2312 }
2313
2314 void diff_setup(struct diff_options *options)
2315 {
2316         memset(options, 0, sizeof(*options));
2317
2318         options->file = stdout;
2319
2320         options->line_termination = '\n';
2321         options->break_opt = -1;
2322         options->rename_limit = -1;
2323         options->dirstat_percent = 3;
2324         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
2325         options->context = 3;
2326
2327         options->change = diff_change;
2328         options->add_remove = diff_addremove;
2329         if (diff_use_color_default > 0)
2330                 DIFF_OPT_SET(options, COLOR_DIFF);
2331         else
2332                 DIFF_OPT_CLR(options, COLOR_DIFF);
2333         options->detect_rename = diff_detect_rename_default;
2334
2335         options->a_prefix = "a/";
2336         options->b_prefix = "b/";
2337 }
2338
2339 int diff_setup_done(struct diff_options *options)
2340 {
2341         int count = 0;
2342
2343         if (options->output_format & DIFF_FORMAT_NAME)
2344                 count++;
2345         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2346                 count++;
2347         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2348                 count++;
2349         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2350                 count++;
2351         if (count > 1)
2352                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2353
2354         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2355                 options->detect_rename = DIFF_DETECT_COPY;
2356
2357         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2358                 options->prefix = NULL;
2359         if (options->prefix)
2360                 options->prefix_length = strlen(options->prefix);
2361         else
2362                 options->prefix_length = 0;
2363
2364         if (options->output_format & (DIFF_FORMAT_NAME |
2365                                       DIFF_FORMAT_NAME_STATUS |
2366                                       DIFF_FORMAT_CHECKDIFF |
2367                                       DIFF_FORMAT_NO_OUTPUT))
2368                 options->output_format &= ~(DIFF_FORMAT_RAW |
2369                                             DIFF_FORMAT_NUMSTAT |
2370                                             DIFF_FORMAT_DIFFSTAT |
2371                                             DIFF_FORMAT_SHORTSTAT |
2372                                             DIFF_FORMAT_DIRSTAT |
2373                                             DIFF_FORMAT_SUMMARY |
2374                                             DIFF_FORMAT_PATCH);
2375
2376         /*
2377          * These cases always need recursive; we do not drop caller-supplied
2378          * recursive bits for other formats here.
2379          */
2380         if (options->output_format & (DIFF_FORMAT_PATCH |
2381                                       DIFF_FORMAT_NUMSTAT |
2382                                       DIFF_FORMAT_DIFFSTAT |
2383                                       DIFF_FORMAT_SHORTSTAT |
2384                                       DIFF_FORMAT_DIRSTAT |
2385                                       DIFF_FORMAT_SUMMARY |
2386                                       DIFF_FORMAT_CHECKDIFF))
2387                 DIFF_OPT_SET(options, RECURSIVE);
2388         /*
2389          * Also pickaxe would not work very well if you do not say recursive
2390          */
2391         if (options->pickaxe)
2392                 DIFF_OPT_SET(options, RECURSIVE);
2393
2394         if (options->detect_rename && options->rename_limit < 0)
2395                 options->rename_limit = diff_rename_limit_default;
2396         if (options->setup & DIFF_SETUP_USE_CACHE) {
2397                 if (!active_cache)
2398                         /* read-cache does not die even when it fails
2399                          * so it is safe for us to do this here.  Also
2400                          * it does not smudge active_cache or active_nr
2401                          * when it fails, so we do not have to worry about
2402                          * cleaning it up ourselves either.
2403                          */
2404                         read_cache();
2405         }
2406         if (options->abbrev <= 0 || 40 < options->abbrev)
2407                 options->abbrev = 40; /* full */
2408
2409         /*
2410          * It does not make sense to show the first hit we happened
2411          * to have found.  It does not make sense not to return with
2412          * exit code in such a case either.
2413          */
2414         if (DIFF_OPT_TST(options, QUIET)) {
2415                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2416                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2417         }
2418
2419         return 0;
2420 }
2421
2422 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2423 {
2424         char c, *eq;
2425         int len;
2426
2427         if (*arg != '-')
2428                 return 0;
2429         c = *++arg;
2430         if (!c)
2431                 return 0;
2432         if (c == arg_short) {
2433                 c = *++arg;
2434                 if (!c)
2435                         return 1;
2436                 if (val && isdigit(c)) {
2437                         char *end;
2438                         int n = strtoul(arg, &end, 10);
2439                         if (*end)
2440                                 return 0;
2441                         *val = n;
2442                         return 1;
2443                 }
2444                 return 0;
2445         }
2446         if (c != '-')
2447                 return 0;
2448         arg++;
2449         eq = strchr(arg, '=');
2450         if (eq)
2451                 len = eq - arg;
2452         else
2453                 len = strlen(arg);
2454         if (!len || strncmp(arg, arg_long, len))
2455                 return 0;
2456         if (eq) {
2457                 int n;
2458                 char *end;
2459                 if (!isdigit(*++eq))
2460                         return 0;
2461                 n = strtoul(eq, &end, 10);
2462                 if (*end)
2463                         return 0;
2464                 *val = n;
2465         }
2466         return 1;
2467 }
2468
2469 static int diff_scoreopt_parse(const char *opt);
2470
2471 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2472 {
2473         const char *arg = av[0];
2474
2475         /* Output format options */
2476         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2477                 options->output_format |= DIFF_FORMAT_PATCH;
2478         else if (opt_arg(arg, 'U', "unified", &options->context))
2479                 options->output_format |= DIFF_FORMAT_PATCH;
2480         else if (!strcmp(arg, "--raw"))
2481                 options->output_format |= DIFF_FORMAT_RAW;
2482         else if (!strcmp(arg, "--patch-with-raw"))
2483                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2484         else if (!strcmp(arg, "--numstat"))
2485                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2486         else if (!strcmp(arg, "--shortstat"))
2487                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2488         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2489                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2490         else if (!strcmp(arg, "--cumulative")) {
2491                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2492                 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2493         }
2494         else if (!strcmp(arg, "--check"))
2495                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2496         else if (!strcmp(arg, "--summary"))
2497                 options->output_format |= DIFF_FORMAT_SUMMARY;
2498         else if (!strcmp(arg, "--patch-with-stat"))
2499                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2500         else if (!strcmp(arg, "--name-only"))
2501                 options->output_format |= DIFF_FORMAT_NAME;
2502         else if (!strcmp(arg, "--name-status"))
2503                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2504         else if (!strcmp(arg, "-s"))
2505                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2506         else if (!prefixcmp(arg, "--stat")) {
2507                 char *end;
2508                 int width = options->stat_width;
2509                 int name_width = options->stat_name_width;
2510                 arg += 6;
2511                 end = (char *)arg;
2512
2513                 switch (*arg) {
2514                 case '-':
2515                         if (!prefixcmp(arg, "-width="))
2516                                 width = strtoul(arg + 7, &end, 10);
2517                         else if (!prefixcmp(arg, "-name-width="))
2518                                 name_width = strtoul(arg + 12, &end, 10);
2519                         break;
2520                 case '=':
2521                         width = strtoul(arg+1, &end, 10);
2522                         if (*end == ',')
2523                                 name_width = strtoul(end+1, &end, 10);
2524                 }
2525
2526                 /* Important! This checks all the error cases! */
2527                 if (*end)
2528                         return 0;
2529                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2530                 options->stat_name_width = name_width;
2531                 options->stat_width = width;
2532         }
2533
2534         /* renames options */
2535         else if (!prefixcmp(arg, "-B")) {
2536                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2537                         return -1;
2538         }
2539         else if (!prefixcmp(arg, "-M")) {
2540                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2541                         return -1;
2542                 options->detect_rename = DIFF_DETECT_RENAME;
2543         }
2544         else if (!prefixcmp(arg, "-C")) {
2545                 if (options->detect_rename == DIFF_DETECT_COPY)
2546                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2547                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2548                         return -1;
2549                 options->detect_rename = DIFF_DETECT_COPY;
2550         }
2551         else if (!strcmp(arg, "--no-renames"))
2552                 options->detect_rename = 0;
2553         else if (!strcmp(arg, "--relative"))
2554                 DIFF_OPT_SET(options, RELATIVE_NAME);
2555         else if (!prefixcmp(arg, "--relative=")) {
2556                 DIFF_OPT_SET(options, RELATIVE_NAME);
2557                 options->prefix = arg + 11;
2558         }
2559
2560         /* xdiff options */
2561         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2562                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2563         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2564                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2565         else if (!strcmp(arg, "--ignore-space-at-eol"))
2566                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2567
2568         /* flags options */
2569         else if (!strcmp(arg, "--binary")) {
2570                 options->output_format |= DIFF_FORMAT_PATCH;
2571                 DIFF_OPT_SET(options, BINARY);
2572         }
2573         else if (!strcmp(arg, "--full-index"))
2574                 DIFF_OPT_SET(options, FULL_INDEX);
2575         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2576                 DIFF_OPT_SET(options, TEXT);
2577         else if (!strcmp(arg, "-R"))
2578                 DIFF_OPT_SET(options, REVERSE_DIFF);
2579         else if (!strcmp(arg, "--find-copies-harder"))
2580                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2581         else if (!strcmp(arg, "--follow"))
2582                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2583         else if (!strcmp(arg, "--color"))
2584                 DIFF_OPT_SET(options, COLOR_DIFF);
2585         else if (!strcmp(arg, "--no-color"))
2586                 DIFF_OPT_CLR(options, COLOR_DIFF);
2587         else if (!strcmp(arg, "--color-words"))
2588                 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2589         else if (!strcmp(arg, "--exit-code"))
2590                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2591         else if (!strcmp(arg, "--quiet"))
2592                 DIFF_OPT_SET(options, QUIET);
2593         else if (!strcmp(arg, "--ext-diff"))
2594                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2595         else if (!strcmp(arg, "--no-ext-diff"))
2596                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2597         else if (!strcmp(arg, "--ignore-submodules"))
2598                 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2599
2600         /* misc options */
2601         else if (!strcmp(arg, "-z"))
2602                 options->line_termination = 0;
2603         else if (!prefixcmp(arg, "-l"))
2604                 options->rename_limit = strtoul(arg+2, NULL, 10);
2605         else if (!prefixcmp(arg, "-S"))
2606                 options->pickaxe = arg + 2;
2607         else if (!strcmp(arg, "--pickaxe-all"))
2608                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2609         else if (!strcmp(arg, "--pickaxe-regex"))
2610                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2611         else if (!prefixcmp(arg, "-O"))
2612                 options->orderfile = arg + 2;
2613         else if (!prefixcmp(arg, "--diff-filter="))
2614                 options->filter = arg + 14;
2615         else if (!strcmp(arg, "--abbrev"))
2616                 options->abbrev = DEFAULT_ABBREV;
2617         else if (!prefixcmp(arg, "--abbrev=")) {
2618                 options->abbrev = strtoul(arg + 9, NULL, 10);
2619                 if (options->abbrev < MINIMUM_ABBREV)
2620                         options->abbrev = MINIMUM_ABBREV;
2621                 else if (40 < options->abbrev)
2622                         options->abbrev = 40;
2623         }
2624         else if (!prefixcmp(arg, "--src-prefix="))
2625                 options->a_prefix = arg + 13;
2626         else if (!prefixcmp(arg, "--dst-prefix="))
2627                 options->b_prefix = arg + 13;
2628         else if (!strcmp(arg, "--no-prefix"))
2629                 options->a_prefix = options->b_prefix = "";
2630         else if (!prefixcmp(arg, "--output=")) {
2631                 options->file = fopen(arg + strlen("--output="), "w");
2632                 options->close_file = 1;
2633         } else
2634                 return 0;
2635         return 1;
2636 }
2637
2638 static int parse_num(const char **cp_p)
2639 {
2640         unsigned long num, scale;
2641         int ch, dot;
2642         const char *cp = *cp_p;
2643
2644         num = 0;
2645         scale = 1;
2646         dot = 0;
2647         for(;;) {
2648                 ch = *cp;
2649                 if ( !dot && ch == '.' ) {
2650                         scale = 1;
2651                         dot = 1;
2652                 } else if ( ch == '%' ) {
2653                         scale = dot ? scale*100 : 100;
2654                         cp++;   /* % is always at the end */
2655                         break;
2656                 } else if ( ch >= '0' && ch <= '9' ) {
2657                         if ( scale < 100000 ) {
2658                                 scale *= 10;
2659                                 num = (num*10) + (ch-'0');
2660                         }
2661                 } else {
2662                         break;
2663                 }
2664                 cp++;
2665         }
2666         *cp_p = cp;
2667
2668         /* user says num divided by scale and we say internally that
2669          * is MAX_SCORE * num / scale.
2670          */
2671         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2672 }
2673
2674 static int diff_scoreopt_parse(const char *opt)
2675 {
2676         int opt1, opt2, cmd;
2677
2678         if (*opt++ != '-')
2679                 return -1;
2680         cmd = *opt++;
2681         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2682                 return -1; /* that is not a -M, -C nor -B option */
2683
2684         opt1 = parse_num(&opt);
2685         if (cmd != 'B')
2686                 opt2 = 0;
2687         else {
2688                 if (*opt == 0)
2689                         opt2 = 0;
2690                 else if (*opt != '/')
2691                         return -1; /* we expect -B80/99 or -B80 */
2692                 else {
2693                         opt++;
2694                         opt2 = parse_num(&opt);
2695                 }
2696         }
2697         if (*opt != 0)
2698                 return -1;
2699         return opt1 | (opt2 << 16);
2700 }
2701
2702 struct diff_queue_struct diff_queued_diff;
2703
2704 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2705 {
2706         if (queue->alloc <= queue->nr) {
2707                 queue->alloc = alloc_nr(queue->alloc);
2708                 queue->queue = xrealloc(queue->queue,
2709                                         sizeof(dp) * queue->alloc);
2710         }
2711         queue->queue[queue->nr++] = dp;
2712 }
2713
2714 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2715                                  struct diff_filespec *one,
2716                                  struct diff_filespec *two)
2717 {
2718         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2719         dp->one = one;
2720         dp->two = two;
2721         if (queue)
2722                 diff_q(queue, dp);
2723         return dp;
2724 }
2725
2726 void diff_free_filepair(struct diff_filepair *p)
2727 {
2728         free_filespec(p->one);
2729         free_filespec(p->two);
2730         free(p);
2731 }
2732
2733 /* This is different from find_unique_abbrev() in that
2734  * it stuffs the result with dots for alignment.
2735  */
2736 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2737 {
2738         int abblen;
2739         const char *abbrev;
2740         if (len == 40)
2741                 return sha1_to_hex(sha1);
2742
2743         abbrev = find_unique_abbrev(sha1, len);
2744         abblen = strlen(abbrev);
2745         if (abblen < 37) {
2746                 static char hex[41];
2747                 if (len < abblen && abblen <= len + 2)
2748                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2749                 else
2750                         sprintf(hex, "%s...", abbrev);
2751                 return hex;
2752         }
2753         return sha1_to_hex(sha1);
2754 }
2755
2756 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2757 {
2758         int line_termination = opt->line_termination;
2759         int inter_name_termination = line_termination ? '\t' : '\0';
2760
2761         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2762                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2763                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
2764                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2765         }
2766         if (p->score) {
2767                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2768                         inter_name_termination);
2769         } else {
2770                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2771         }
2772
2773         if (p->status == DIFF_STATUS_COPIED ||
2774             p->status == DIFF_STATUS_RENAMED) {
2775                 const char *name_a, *name_b;
2776                 name_a = p->one->path;
2777                 name_b = p->two->path;
2778                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2779                 write_name_quoted(name_a, opt->file, inter_name_termination);
2780                 write_name_quoted(name_b, opt->file, line_termination);
2781         } else {
2782                 const char *name_a, *name_b;
2783                 name_a = p->one->mode ? p->one->path : p->two->path;
2784                 name_b = NULL;
2785                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2786                 write_name_quoted(name_a, opt->file, line_termination);
2787         }
2788 }
2789
2790 int diff_unmodified_pair(struct diff_filepair *p)
2791 {
2792         /* This function is written stricter than necessary to support
2793          * the currently implemented transformers, but the idea is to
2794          * let transformers to produce diff_filepairs any way they want,
2795          * and filter and clean them up here before producing the output.
2796          */
2797         struct diff_filespec *one = p->one, *two = p->two;
2798
2799         if (DIFF_PAIR_UNMERGED(p))
2800                 return 0; /* unmerged is interesting */
2801
2802         /* deletion, addition, mode or type change
2803          * and rename are all interesting.
2804          */
2805         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2806             DIFF_PAIR_MODE_CHANGED(p) ||
2807             strcmp(one->path, two->path))
2808                 return 0;
2809
2810         /* both are valid and point at the same path.  that is, we are
2811          * dealing with a change.
2812          */
2813         if (one->sha1_valid && two->sha1_valid &&
2814             !hashcmp(one->sha1, two->sha1))
2815                 return 1; /* no change */
2816         if (!one->sha1_valid && !two->sha1_valid)
2817                 return 1; /* both look at the same file on the filesystem. */
2818         return 0;
2819 }
2820
2821 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2822 {
2823         if (diff_unmodified_pair(p))
2824                 return;
2825
2826         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2827             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2828                 return; /* no tree diffs in patch format */
2829
2830         run_diff(p, o);
2831 }
2832
2833 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2834                             struct diffstat_t *diffstat)
2835 {
2836         if (diff_unmodified_pair(p))
2837                 return;
2838
2839         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2840             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2841                 return; /* no tree diffs in patch format */
2842
2843         run_diffstat(p, o, diffstat);
2844 }
2845
2846 static void diff_flush_checkdiff(struct diff_filepair *p,
2847                 struct diff_options *o)
2848 {
2849         if (diff_unmodified_pair(p))
2850                 return;
2851
2852         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2853             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2854                 return; /* no tree diffs in patch format */
2855
2856         run_checkdiff(p, o);
2857 }
2858
2859 int diff_queue_is_empty(void)
2860 {
2861         struct diff_queue_struct *q = &diff_queued_diff;
2862         int i;
2863         for (i = 0; i < q->nr; i++)
2864                 if (!diff_unmodified_pair(q->queue[i]))
2865                         return 0;
2866         return 1;
2867 }
2868
2869 #if DIFF_DEBUG
2870 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2871 {
2872         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2873                 x, one ? one : "",
2874                 s->path,
2875                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2876                 s->mode,
2877                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2878         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2879                 x, one ? one : "",
2880                 s->size, s->xfrm_flags);
2881 }
2882
2883 void diff_debug_filepair(const struct diff_filepair *p, int i)
2884 {
2885         diff_debug_filespec(p->one, i, "one");
2886         diff_debug_filespec(p->two, i, "two");
2887         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2888                 p->score, p->status ? p->status : '?',
2889                 p->one->rename_used, p->broken_pair);
2890 }
2891
2892 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2893 {
2894         int i;
2895         if (msg)
2896                 fprintf(stderr, "%s\n", msg);
2897         fprintf(stderr, "q->nr = %d\n", q->nr);
2898         for (i = 0; i < q->nr; i++) {
2899                 struct diff_filepair *p = q->queue[i];
2900                 diff_debug_filepair(p, i);
2901         }
2902 }
2903 #endif
2904
2905 static void diff_resolve_rename_copy(void)
2906 {
2907         int i;
2908         struct diff_filepair *p;
2909         struct diff_queue_struct *q = &diff_queued_diff;
2910
2911         diff_debug_queue("resolve-rename-copy", q);
2912
2913         for (i = 0; i < q->nr; i++) {
2914                 p = q->queue[i];
2915                 p->status = 0; /* undecided */
2916                 if (DIFF_PAIR_UNMERGED(p))
2917                         p->status = DIFF_STATUS_UNMERGED;
2918                 else if (!DIFF_FILE_VALID(p->one))
2919                         p->status = DIFF_STATUS_ADDED;
2920                 else if (!DIFF_FILE_VALID(p->two))
2921                         p->status = DIFF_STATUS_DELETED;
2922                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2923                         p->status = DIFF_STATUS_TYPE_CHANGED;
2924
2925                 /* from this point on, we are dealing with a pair
2926                  * whose both sides are valid and of the same type, i.e.
2927                  * either in-place edit or rename/copy edit.
2928                  */
2929                 else if (DIFF_PAIR_RENAME(p)) {
2930                         /*
2931                          * A rename might have re-connected a broken
2932                          * pair up, causing the pathnames to be the
2933                          * same again. If so, that's not a rename at
2934                          * all, just a modification..
2935                          *
2936                          * Otherwise, see if this source was used for
2937                          * multiple renames, in which case we decrement
2938                          * the count, and call it a copy.
2939                          */
2940                         if (!strcmp(p->one->path, p->two->path))
2941                                 p->status = DIFF_STATUS_MODIFIED;
2942                         else if (--p->one->rename_used > 0)
2943                                 p->status = DIFF_STATUS_COPIED;
2944                         else
2945                                 p->status = DIFF_STATUS_RENAMED;
2946                 }
2947                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2948                          p->one->mode != p->two->mode ||
2949                          is_null_sha1(p->one->sha1))
2950                         p->status = DIFF_STATUS_MODIFIED;
2951                 else {
2952                         /* This is a "no-change" entry and should not
2953                          * happen anymore, but prepare for broken callers.
2954                          */
2955                         error("feeding unmodified %s to diffcore",
2956                               p->one->path);
2957                         p->status = DIFF_STATUS_UNKNOWN;
2958                 }
2959         }
2960         diff_debug_queue("resolve-rename-copy done", q);
2961 }
2962
2963 static int check_pair_status(struct diff_filepair *p)
2964 {
2965         switch (p->status) {
2966         case DIFF_STATUS_UNKNOWN:
2967                 return 0;
2968         case 0:
2969                 die("internal error in diff-resolve-rename-copy");
2970         default:
2971                 return 1;
2972         }
2973 }
2974
2975 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2976 {
2977         int fmt = opt->output_format;
2978
2979         if (fmt & DIFF_FORMAT_CHECKDIFF)
2980                 diff_flush_checkdiff(p, opt);
2981         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2982                 diff_flush_raw(p, opt);
2983         else if (fmt & DIFF_FORMAT_NAME) {
2984                 const char *name_a, *name_b;
2985                 name_a = p->two->path;
2986                 name_b = NULL;
2987                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2988                 write_name_quoted(name_a, opt->file, opt->line_termination);
2989         }
2990 }
2991
2992 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2993 {
2994         if (fs->mode)
2995                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2996         else
2997                 fprintf(file, " %s ", newdelete);
2998         write_name_quoted(fs->path, file, '\n');
2999 }
3000
3001
3002 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
3003 {
3004         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3005                 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
3006                         show_name ? ' ' : '\n');
3007                 if (show_name) {
3008                         write_name_quoted(p->two->path, file, '\n');
3009                 }
3010         }
3011 }
3012
3013 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3014 {
3015         char *names = pprint_rename(p->one->path, p->two->path);
3016
3017         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3018         free(names);
3019         show_mode_change(file, p, 0);
3020 }
3021
3022 static void diff_summary(FILE *file, struct diff_filepair *p)
3023 {
3024         switch(p->status) {
3025         case DIFF_STATUS_DELETED:
3026                 show_file_mode_name(file, "delete", p->one);
3027                 break;
3028         case DIFF_STATUS_ADDED:
3029                 show_file_mode_name(file, "create", p->two);
3030                 break;
3031         case DIFF_STATUS_COPIED:
3032                 show_rename_copy(file, "copy", p);
3033                 break;
3034         case DIFF_STATUS_RENAMED:
3035                 show_rename_copy(file, "rename", p);
3036                 break;
3037         default:
3038                 if (p->score) {
3039                         fputs(" rewrite ", file);
3040                         write_name_quoted(p->two->path, file, ' ');
3041                         fprintf(file, "(%d%%)\n", similarity_index(p));
3042                 }
3043                 show_mode_change(file, p, !p->score);
3044                 break;
3045         }
3046 }
3047
3048 struct patch_id_t {
3049         struct xdiff_emit_state xm;
3050         SHA_CTX *ctx;
3051         int patchlen;
3052 };
3053
3054 static int remove_space(char *line, int len)
3055 {
3056         int i;
3057         char *dst = line;
3058         unsigned char c;
3059
3060         for (i = 0; i < len; i++)
3061                 if (!isspace((c = line[i])))
3062                         *dst++ = c;
3063
3064         return dst - line;
3065 }
3066
3067 static void patch_id_consume(void *priv, char *line, unsigned long len)
3068 {
3069         struct patch_id_t *data = priv;
3070         int new_len;
3071
3072         /* Ignore line numbers when computing the SHA1 of the patch */
3073         if (!prefixcmp(line, "@@ -"))
3074                 return;
3075
3076         new_len = remove_space(line, len);
3077
3078         SHA1_Update(data->ctx, line, new_len);
3079         data->patchlen += new_len;
3080 }
3081
3082 /* returns 0 upon success, and writes result into sha1 */
3083 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3084 {
3085         struct diff_queue_struct *q = &diff_queued_diff;
3086         int i;
3087         SHA_CTX ctx;
3088         struct patch_id_t data;
3089         char buffer[PATH_MAX * 4 + 20];
3090
3091         SHA1_Init(&ctx);
3092         memset(&data, 0, sizeof(struct patch_id_t));
3093         data.ctx = &ctx;
3094         data.xm.consume = patch_id_consume;
3095
3096         for (i = 0; i < q->nr; i++) {
3097                 xpparam_t xpp;
3098                 xdemitconf_t xecfg;
3099                 xdemitcb_t ecb;
3100                 mmfile_t mf1, mf2;
3101                 struct diff_filepair *p = q->queue[i];
3102                 int len1, len2;
3103
3104                 memset(&xecfg, 0, sizeof(xecfg));
3105                 if (p->status == 0)
3106                         return error("internal diff status error");
3107                 if (p->status == DIFF_STATUS_UNKNOWN)
3108                         continue;
3109                 if (diff_unmodified_pair(p))
3110                         continue;
3111                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3112                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3113                         continue;
3114                 if (DIFF_PAIR_UNMERGED(p))
3115                         continue;
3116
3117                 diff_fill_sha1_info(p->one);
3118                 diff_fill_sha1_info(p->two);
3119                 if (fill_mmfile(&mf1, p->one) < 0 ||
3120                                 fill_mmfile(&mf2, p->two) < 0)
3121                         return error("unable to read files to diff");
3122
3123                 len1 = remove_space(p->one->path, strlen(p->one->path));
3124                 len2 = remove_space(p->two->path, strlen(p->two->path));
3125                 if (p->one->mode == 0)
3126                         len1 = snprintf(buffer, sizeof(buffer),
3127                                         "diff--gita/%.*sb/%.*s"
3128                                         "newfilemode%06o"
3129                                         "---/dev/null"
3130                                         "+++b/%.*s",
3131                                         len1, p->one->path,
3132                                         len2, p->two->path,
3133                                         p->two->mode,
3134                                         len2, p->two->path);
3135                 else if (p->two->mode == 0)
3136                         len1 = snprintf(buffer, sizeof(buffer),
3137                                         "diff--gita/%.*sb/%.*s"
3138                                         "deletedfilemode%06o"
3139                                         "---a/%.*s"
3140                                         "+++/dev/null",
3141                                         len1, p->one->path,
3142                                         len2, p->two->path,
3143                                         p->one->mode,
3144                                         len1, p->one->path);
3145                 else
3146                         len1 = snprintf(buffer, sizeof(buffer),
3147                                         "diff--gita/%.*sb/%.*s"
3148                                         "---a/%.*s"
3149                                         "+++b/%.*s",
3150                                         len1, p->one->path,
3151                                         len2, p->two->path,
3152                                         len1, p->one->path,
3153                                         len2, p->two->path);
3154                 SHA1_Update(&ctx, buffer, len1);
3155
3156                 xpp.flags = XDF_NEED_MINIMAL;
3157                 xecfg.ctxlen = 3;
3158                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3159                 ecb.outf = xdiff_outf;
3160                 ecb.priv = &data;
3161                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
3162         }
3163
3164         SHA1_Final(sha1, &ctx);
3165         return 0;
3166 }
3167
3168 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3169 {
3170         struct diff_queue_struct *q = &diff_queued_diff;
3171         int i;
3172         int result = diff_get_patch_id(options, sha1);
3173
3174         for (i = 0; i < q->nr; i++)
3175                 diff_free_filepair(q->queue[i]);
3176
3177         free(q->queue);
3178         q->queue = NULL;
3179         q->nr = q->alloc = 0;
3180
3181         return result;
3182 }
3183
3184 static int is_summary_empty(const struct diff_queue_struct *q)
3185 {
3186         int i;
3187
3188         for (i = 0; i < q->nr; i++) {
3189                 const struct diff_filepair *p = q->queue[i];
3190
3191                 switch (p->status) {
3192                 case DIFF_STATUS_DELETED:
3193                 case DIFF_STATUS_ADDED:
3194                 case DIFF_STATUS_COPIED:
3195                 case DIFF_STATUS_RENAMED:
3196                         return 0;
3197                 default:
3198                         if (p->score)
3199                                 return 0;
3200                         if (p->one->mode && p->two->mode &&
3201                             p->one->mode != p->two->mode)
3202                                 return 0;
3203                         break;
3204                 }
3205         }
3206         return 1;
3207 }
3208
3209 void diff_flush(struct diff_options *options)
3210 {
3211         struct diff_queue_struct *q = &diff_queued_diff;
3212         int i, output_format = options->output_format;
3213         int separator = 0;
3214
3215         /*
3216          * Order: raw, stat, summary, patch
3217          * or:    name/name-status/checkdiff (other bits clear)
3218          */
3219         if (!q->nr)
3220                 goto free_queue;
3221
3222         if (output_format & (DIFF_FORMAT_RAW |
3223                              DIFF_FORMAT_NAME |
3224                              DIFF_FORMAT_NAME_STATUS |
3225                              DIFF_FORMAT_CHECKDIFF)) {
3226                 for (i = 0; i < q->nr; i++) {
3227                         struct diff_filepair *p = q->queue[i];
3228                         if (check_pair_status(p))
3229                                 flush_one_pair(p, options);
3230                 }
3231                 separator++;
3232         }
3233
3234         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3235                 struct diffstat_t diffstat;
3236
3237                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3238                 diffstat.xm.consume = diffstat_consume;
3239                 for (i = 0; i < q->nr; i++) {
3240                         struct diff_filepair *p = q->queue[i];
3241                         if (check_pair_status(p))
3242                                 diff_flush_stat(p, options, &diffstat);
3243                 }
3244                 if (output_format & DIFF_FORMAT_NUMSTAT)
3245                         show_numstat(&diffstat, options);
3246                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3247                         show_stats(&diffstat, options);
3248                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3249                         show_shortstats(&diffstat, options);
3250                 free_diffstat_info(&diffstat);
3251                 separator++;
3252         }
3253         if (output_format & DIFF_FORMAT_DIRSTAT)
3254                 show_dirstat(options);
3255
3256         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3257                 for (i = 0; i < q->nr; i++)
3258                         diff_summary(options->file, q->queue[i]);
3259                 separator++;
3260         }
3261
3262         if (output_format & DIFF_FORMAT_PATCH) {
3263                 if (separator) {
3264                         putc(options->line_termination, options->file);
3265                         if (options->stat_sep) {
3266                                 /* attach patch instead of inline */
3267                                 fputs(options->stat_sep, options->file);
3268                         }
3269                 }
3270
3271                 for (i = 0; i < q->nr; i++) {
3272                         struct diff_filepair *p = q->queue[i];
3273                         if (check_pair_status(p))
3274                                 diff_flush_patch(p, options);
3275                 }
3276         }
3277
3278         if (output_format & DIFF_FORMAT_CALLBACK)
3279                 options->format_callback(q, options, options->format_callback_data);
3280
3281         for (i = 0; i < q->nr; i++)
3282                 diff_free_filepair(q->queue[i]);
3283 free_queue:
3284         free(q->queue);
3285         q->queue = NULL;
3286         q->nr = q->alloc = 0;
3287         if (options->close_file)
3288                 fclose(options->file);
3289 }
3290
3291 static void diffcore_apply_filter(const char *filter)
3292 {
3293         int i;
3294         struct diff_queue_struct *q = &diff_queued_diff;
3295         struct diff_queue_struct outq;
3296         outq.queue = NULL;
3297         outq.nr = outq.alloc = 0;
3298
3299         if (!filter)
3300                 return;
3301
3302         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3303                 int found;
3304                 for (i = found = 0; !found && i < q->nr; i++) {
3305                         struct diff_filepair *p = q->queue[i];
3306                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3307                              ((p->score &&
3308                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3309                               (!p->score &&
3310                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3311                             ((p->status != DIFF_STATUS_MODIFIED) &&
3312                              strchr(filter, p->status)))
3313                                 found++;
3314                 }
3315                 if (found)
3316                         return;
3317
3318                 /* otherwise we will clear the whole queue
3319                  * by copying the empty outq at the end of this
3320                  * function, but first clear the current entries
3321                  * in the queue.
3322                  */
3323                 for (i = 0; i < q->nr; i++)
3324                         diff_free_filepair(q->queue[i]);
3325         }
3326         else {
3327                 /* Only the matching ones */
3328                 for (i = 0; i < q->nr; i++) {
3329                         struct diff_filepair *p = q->queue[i];
3330
3331                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3332                              ((p->score &&
3333                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3334                               (!p->score &&
3335                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3336                             ((p->status != DIFF_STATUS_MODIFIED) &&
3337                              strchr(filter, p->status)))
3338                                 diff_q(&outq, p);
3339                         else
3340                                 diff_free_filepair(p);
3341                 }
3342         }
3343         free(q->queue);
3344         *q = outq;
3345 }
3346
3347 /* Check whether two filespecs with the same mode and size are identical */
3348 static int diff_filespec_is_identical(struct diff_filespec *one,
3349                                       struct diff_filespec *two)
3350 {
3351         if (S_ISGITLINK(one->mode))
3352                 return 0;
3353         if (diff_populate_filespec(one, 0))
3354                 return 0;
3355         if (diff_populate_filespec(two, 0))
3356                 return 0;
3357         return !memcmp(one->data, two->data, one->size);
3358 }
3359
3360 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3361 {
3362         int i;
3363         struct diff_queue_struct *q = &diff_queued_diff;
3364         struct diff_queue_struct outq;
3365         outq.queue = NULL;
3366         outq.nr = outq.alloc = 0;
3367
3368         for (i = 0; i < q->nr; i++) {
3369                 struct diff_filepair *p = q->queue[i];
3370
3371                 /*
3372                  * 1. Entries that come from stat info dirtyness
3373                  *    always have both sides (iow, not create/delete),
3374                  *    one side of the object name is unknown, with
3375                  *    the same mode and size.  Keep the ones that
3376                  *    do not match these criteria.  They have real
3377                  *    differences.
3378                  *
3379                  * 2. At this point, the file is known to be modified,
3380                  *    with the same mode and size, and the object
3381                  *    name of one side is unknown.  Need to inspect
3382                  *    the identical contents.
3383                  */
3384                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3385                     !DIFF_FILE_VALID(p->two) ||
3386                     (p->one->sha1_valid && p->two->sha1_valid) ||
3387                     (p->one->mode != p->two->mode) ||
3388                     diff_populate_filespec(p->one, 1) ||
3389                     diff_populate_filespec(p->two, 1) ||
3390                     (p->one->size != p->two->size) ||
3391                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3392                         diff_q(&outq, p);
3393                 else {
3394                         /*
3395                          * The caller can subtract 1 from skip_stat_unmatch
3396                          * to determine how many paths were dirty only
3397                          * due to stat info mismatch.
3398                          */
3399                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3400                                 diffopt->skip_stat_unmatch++;
3401                         diff_free_filepair(p);
3402                 }
3403         }
3404         free(q->queue);
3405         *q = outq;
3406 }
3407
3408 void diffcore_std(struct diff_options *options)
3409 {
3410         if (options->skip_stat_unmatch)
3411                 diffcore_skip_stat_unmatch(options);
3412         if (options->break_opt != -1)
3413                 diffcore_break(options->break_opt);
3414         if (options->detect_rename)
3415                 diffcore_rename(options);
3416         if (options->break_opt != -1)
3417                 diffcore_merge_broken();
3418         if (options->pickaxe)
3419                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3420         if (options->orderfile)
3421                 diffcore_order(options->orderfile);
3422         diff_resolve_rename_copy();
3423         diffcore_apply_filter(options->filter);
3424
3425         if (diff_queued_diff.nr)
3426                 DIFF_OPT_SET(options, HAS_CHANGES);
3427         else
3428                 DIFF_OPT_CLR(options, HAS_CHANGES);
3429 }
3430
3431 int diff_result_code(struct diff_options *opt, int status)
3432 {
3433         int result = 0;
3434         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3435             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3436                 return status;
3437         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3438             DIFF_OPT_TST(opt, HAS_CHANGES))
3439                 result |= 01;
3440         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3441             DIFF_OPT_TST(opt, CHECK_FAILED))
3442                 result |= 02;
3443         return result;
3444 }
3445
3446 void diff_addremove(struct diff_options *options,
3447                     int addremove, unsigned mode,
3448                     const unsigned char *sha1,
3449                     const char *concatpath)
3450 {
3451         struct diff_filespec *one, *two;
3452
3453         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3454                 return;
3455
3456         /* This may look odd, but it is a preparation for
3457          * feeding "there are unchanged files which should
3458          * not produce diffs, but when you are doing copy
3459          * detection you would need them, so here they are"
3460          * entries to the diff-core.  They will be prefixed
3461          * with something like '=' or '*' (I haven't decided
3462          * which but should not make any difference).
3463          * Feeding the same new and old to diff_change()
3464          * also has the same effect.
3465          * Before the final output happens, they are pruned after
3466          * merged into rename/copy pairs as appropriate.
3467          */
3468         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3469                 addremove = (addremove == '+' ? '-' :
3470                              addremove == '-' ? '+' : addremove);
3471
3472         if (options->prefix &&
3473             strncmp(concatpath, options->prefix, options->prefix_length))
3474                 return;
3475
3476         one = alloc_filespec(concatpath);
3477         two = alloc_filespec(concatpath);
3478
3479         if (addremove != '+')
3480                 fill_filespec(one, sha1, mode);
3481         if (addremove != '-')
3482                 fill_filespec(two, sha1, mode);
3483
3484         diff_queue(&diff_queued_diff, one, two);
3485         DIFF_OPT_SET(options, HAS_CHANGES);
3486 }
3487
3488 void diff_change(struct diff_options *options,
3489                  unsigned old_mode, unsigned new_mode,
3490                  const unsigned char *old_sha1,
3491                  const unsigned char *new_sha1,
3492                  const char *concatpath)
3493 {
3494         struct diff_filespec *one, *two;
3495
3496         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3497                         && S_ISGITLINK(new_mode))
3498                 return;
3499
3500         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3501                 unsigned tmp;
3502                 const unsigned char *tmp_c;
3503                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3504                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3505         }
3506
3507         if (options->prefix &&
3508             strncmp(concatpath, options->prefix, options->prefix_length))
3509                 return;
3510
3511         one = alloc_filespec(concatpath);
3512         two = alloc_filespec(concatpath);
3513         fill_filespec(one, old_sha1, old_mode);
3514         fill_filespec(two, new_sha1, new_mode);
3515
3516         diff_queue(&diff_queued_diff, one, two);
3517         DIFF_OPT_SET(options, HAS_CHANGES);
3518 }
3519
3520 void diff_unmerge(struct diff_options *options,
3521                   const char *path,
3522                   unsigned mode, const unsigned char *sha1)
3523 {
3524         struct diff_filespec *one, *two;
3525
3526         if (options->prefix &&
3527             strncmp(path, options->prefix, options->prefix_length))
3528                 return;
3529
3530         one = alloc_filespec(path);
3531         two = alloc_filespec(path);
3532         fill_filespec(one, sha1, mode);
3533         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3534 }