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 #include "userdiff.h"
  15 #include "sigchain.h"
  16 #include "submodule.h"
  17 #include "ll-merge.h"
  18
  19 #ifdef NO_FAST_WORKING_DIRECTORY
  20 #define FAST_WORKING_DIRECTORY 0
  21 #else
  22 #define FAST_WORKING_DIRECTORY 1
  23 #endif
  24
  25 static int diff_detect_rename_default;
  26 static int diff_rename_limit_default = 200;
  27 static int diff_suppress_blank_empty;
  28 int diff_use_color_default = -1;
  29 static const char *diff_word_regex_cfg;
  30 static const char *external_diff_cmd_cfg;
  31 int diff_auto_refresh_index = 1;
  32 static int diff_mnemonic_prefix;
  33
  34 static char diff_colors[][COLOR_MAXLEN] = {
  35         GIT_COLOR_RESET,
  36         GIT_COLOR_NORMAL,       /* PLAIN */
  37         GIT_COLOR_BOLD,         /* METAINFO */
  38         GIT_COLOR_CYAN,         /* FRAGINFO */
  39         GIT_COLOR_RED,          /* OLD */
  40         GIT_COLOR_GREEN,        /* NEW */
  41         GIT_COLOR_YELLOW,       /* COMMIT */
  42         GIT_COLOR_BG_RED,       /* WHITESPACE */
  43         GIT_COLOR_NORMAL,       /* FUNCINFO */
  44 };
  45
  46 static void diff_filespec_load_driver(struct diff_filespec *one);
  47 static size_t fill_textconv(struct userdiff_driver *driver,
  48                             struct diff_filespec *df, char **outbuf);
  49
  50 static int parse_diff_color_slot(const char *var, int ofs)
  51 {
  52         if (!strcasecmp(var+ofs, "plain"))
  53                 return DIFF_PLAIN;
  54         if (!strcasecmp(var+ofs, "meta"))
  55                 return DIFF_METAINFO;
  56         if (!strcasecmp(var+ofs, "frag"))
  57                 return DIFF_FRAGINFO;
  58         if (!strcasecmp(var+ofs, "old"))
  59                 return DIFF_FILE_OLD;
  60         if (!strcasecmp(var+ofs, "new"))
  61                 return DIFF_FILE_NEW;
  62         if (!strcasecmp(var+ofs, "commit"))
  63                 return DIFF_COMMIT;
  64         if (!strcasecmp(var+ofs, "whitespace"))
  65                 return DIFF_WHITESPACE;
  66         if (!strcasecmp(var+ofs, "func"))
  67                 return DIFF_FUNCINFO;
  68         return -1;
  69 }
  70
  71 static int git_config_rename(const char *var, const char *value)
  72 {
  73         if (!value)
  74                 return DIFF_DETECT_RENAME;
  75         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
  76                 return  DIFF_DETECT_COPY;
  77         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
  78 }
  79
  80 /*
  81  * These are to give UI layer defaults.
  82  * The core-level commands such as git-diff-files should
  83  * never be affected by the setting of diff.renames
  84  * the user happens to have in the configuration file.
  85  */
  86 int git_diff_ui_config(const char *var, const char *value, void *cb)
  87 {
  88         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
  89                 diff_use_color_default = git_config_colorbool(var, value, -1);
  90                 return 0;
  91         }
  92         if (!strcmp(var, "diff.renames")) {
  93                 diff_detect_rename_default = git_config_rename(var, value);
  94                 return 0;
  95         }
  96         if (!strcmp(var, "diff.autorefreshindex")) {
  97                 diff_auto_refresh_index = git_config_bool(var, value);
  98                 return 0;
  99         }
 100         if (!strcmp(var, "diff.mnemonicprefix")) {
 101                 diff_mnemonic_prefix = git_config_bool(var, value);
 102                 return 0;
 103         }
 104         if (!strcmp(var, "diff.external"))
 105                 return git_config_string(&external_diff_cmd_cfg, var, value);
 106         if (!strcmp(var, "diff.wordregex"))
 107                 return git_config_string(&diff_word_regex_cfg, var, value);
 108
 109         return git_diff_basic_config(var, value, cb);
 110 }
 111
 112 int git_diff_basic_config(const char *var, const char *value, void *cb)
 113 {
 114         if (!strcmp(var, "diff.renamelimit")) {
 115                 diff_rename_limit_default = git_config_int(var, value);
 116                 return 0;
 117         }
 118
 119         switch (userdiff_config(var, value)) {
 120                 case 0: break;
 121                 case -1: return -1;
 122                 default: return 0;
 123         }
 124
 125         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 126                 int slot = parse_diff_color_slot(var, 11);
 127                 if (slot < 0)
 128                         return 0;
 129                 if (!value)
 130                         return config_error_nonbool(var);
 131                 color_parse(value, var, diff_colors[slot]);
 132                 return 0;
 133         }
 134
 135         /* like GNU diff's --suppress-blank-empty option  */
 136         if (!strcmp(var, "diff.suppressblankempty") ||
 137                         /* for backwards compatibility */
 138                         !strcmp(var, "diff.suppress-blank-empty")) {
 139                 diff_suppress_blank_empty = git_config_bool(var, value);
 140                 return 0;
 141         }
 142
 143         return git_color_default_config(var, value, cb);
 144 }
 145
 146 static char *quote_two(const char *one, const char *two)
 147 {
 148         int need_one = quote_c_style(one, NULL, NULL, 1);
 149         int need_two = quote_c_style(two, NULL, NULL, 1);
 150         struct strbuf res = STRBUF_INIT;
 151
 152         if (need_one + need_two) {
 153                 strbuf_addch(&res, '"');
 154                 quote_c_style(one, &res, NULL, 1);
 155                 quote_c_style(two, &res, NULL, 1);
 156                 strbuf_addch(&res, '"');
 157         } else {
 158                 strbuf_addstr(&res, one);
 159                 strbuf_addstr(&res, two);
 160         }
 161         return strbuf_detach(&res, NULL);
 162 }
 163
 164 static const char *external_diff(void)
 165 {
 166         static const char *external_diff_cmd = NULL;
 167         static int done_preparing = 0;
 168
 169         if (done_preparing)
 170                 return external_diff_cmd;
 171         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 172         if (!external_diff_cmd)
 173                 external_diff_cmd = external_diff_cmd_cfg;
 174         done_preparing = 1;
 175         return external_diff_cmd;
 176 }
 177
 178 static struct diff_tempfile {
 179         const char *name; /* filename external diff should read from */
 180         char hex[41];
 181         char mode[10];
 182         char tmp_path[PATH_MAX];
 183 } diff_temp[2];
 184
 185 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 186
 187 struct emit_callback {
 188         int color_diff;
 189         unsigned ws_rule;
 190         int blank_at_eof_in_preimage;
 191         int blank_at_eof_in_postimage;
 192         int lno_in_preimage;
 193         int lno_in_postimage;
 194         sane_truncate_fn truncate;
 195         const char **label_path;
 196         struct diff_words_data *diff_words;
 197         int *found_changesp;
 198         FILE *file;
 199         struct strbuf *header;
 200 };
 201
 202 static int count_lines(const char *data, int size)
 203 {
 204         int count, ch, completely_empty = 1, nl_just_seen = 0;
 205         count = 0;
 206         while (0 < size--) {
 207                 ch = *data++;
 208                 if (ch == '\n') {
 209                         count++;
 210                         nl_just_seen = 1;
 211                         completely_empty = 0;
 212                 }
 213                 else {
 214                         nl_just_seen = 0;
 215                         completely_empty = 0;
 216                 }
 217         }
 218         if (completely_empty)
 219                 return 0;
 220         if (!nl_just_seen)
 221                 count++; /* no trailing newline */
 222         return count;
 223 }
 224
 225 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 226 {
 227         if (!DIFF_FILE_VALID(one)) {
 228                 mf->ptr = (char *)""; /* does not matter */
 229                 mf->size = 0;
 230                 return 0;
 231         }
 232         else if (diff_populate_filespec(one, 0))
 233                 return -1;
 234
 235         mf->ptr = one->data;
 236         mf->size = one->size;
 237         return 0;
 238 }
 239
 240 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 241 {
 242         char *ptr = mf->ptr;
 243         long size = mf->size;
 244         int cnt = 0;
 245
 246         if (!size)
 247                 return cnt;
 248         ptr += size - 1; /* pointing at the very end */
 249         if (*ptr != '\n')
 250                 ; /* incomplete line */
 251         else
 252                 ptr--; /* skip the last LF */
 253         while (mf->ptr < ptr) {
 254                 char *prev_eol;
 255                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 256                         if (*prev_eol == '\n')
 257                                 break;
 258                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 259                         break;
 260                 cnt++;
 261                 ptr = prev_eol - 1;
 262         }
 263         return cnt;
 264 }
 265
 266 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 267                                struct emit_callback *ecbdata)
 268 {
 269         int l1, l2, at;
 270         unsigned ws_rule = ecbdata->ws_rule;
 271         l1 = count_trailing_blank(mf1, ws_rule);
 272         l2 = count_trailing_blank(mf2, ws_rule);
 273         if (l2 <= l1) {
 274                 ecbdata->blank_at_eof_in_preimage = 0;
 275                 ecbdata->blank_at_eof_in_postimage = 0;
 276                 return;
 277         }
 278         at = count_lines(mf1->ptr, mf1->size);
 279         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 280
 281         at = count_lines(mf2->ptr, mf2->size);
 282         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 283 }
 284
 285 static void emit_line_0(FILE *file, const char *set, const char *reset,
 286                         int first, const char *line, int len)
 287 {
 288         int has_trailing_newline, has_trailing_carriage_return;
 289         int nofirst;
 290
 291         if (len == 0) {
 292                 has_trailing_newline = (first == '\n');
 293                 has_trailing_carriage_return = (!has_trailing_newline &&
 294                                                 (first == '\r'));
 295                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 296         } else {
 297                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 298                 if (has_trailing_newline)
 299                         len--;
 300                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 301                 if (has_trailing_carriage_return)
 302                         len--;
 303                 nofirst = 0;
 304         }
 305
 306         if (len || !nofirst) {
 307                 fputs(set, file);
 308                 if (!nofirst)
 309                         fputc(first, file);
 310                 fwrite(line, len, 1, file);
 311                 fputs(reset, file);
 312         }
 313         if (has_trailing_carriage_return)
 314                 fputc('\r', file);
 315         if (has_trailing_newline)
 316                 fputc('\n', file);
 317 }
 318
 319 static void emit_line(FILE *file, const char *set, const char *reset,
 320                       const char *line, int len)
 321 {
 322         emit_line_0(file, set, reset, line[0], line+1, len-1);
 323 }
 324
 325 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
 326 {
 327         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
 328               ecbdata->blank_at_eof_in_preimage &&
 329               ecbdata->blank_at_eof_in_postimage &&
 330               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
 331               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
 332                 return 0;
 333         return ws_blank_line(line, len, ecbdata->ws_rule);
 334 }
 335
 336 static void emit_add_line(const char *reset,
 337                           struct emit_callback *ecbdata,
 338                           const char *line, int len)
 339 {
 340         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 341         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 342
 343         if (!*ws)
 344                 emit_line_0(ecbdata->file, set, reset, '+', line, len);
 345         else if (new_blank_line_at_eof(ecbdata, line, len))
 346                 /* Blank line at EOF - paint '+' as well */
 347                 emit_line_0(ecbdata->file, ws, reset, '+', line, len);
 348         else {
 349                 /* Emit just the prefix, then the rest. */
 350                 emit_line_0(ecbdata->file, set, reset, '+', "", 0);
 351                 ws_check_emit(line, len, ecbdata->ws_rule,
 352                               ecbdata->file, set, reset, ws);
 353         }
 354 }
 355
 356 static void emit_hunk_header(struct emit_callback *ecbdata,
 357                              const char *line, int len)
 358 {
 359         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 360         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
 361         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
 362         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 363         static const char atat[2] = { '@', '@' };
 364         const char *cp, *ep;
 365
 366         /*
 367          * As a hunk header must begin with "@@ -<old>, +<new> @@",
 368          * it always is at least 10 bytes long.
 369          */
 370         if (len < 10 ||
 371             memcmp(line, atat, 2) ||
 372             !(ep = memmem(line + 2, len - 2, atat, 2))) {
 373                 emit_line(ecbdata->file, plain, reset, line, len);
 374                 return;
 375         }
 376         ep += 2; /* skip over @@ */
 377
 378         /* The hunk header in fraginfo color */
 379         emit_line(ecbdata->file, frag, reset, line, ep - line);
 380
 381         /* blank before the func header */
 382         for (cp = ep; ep - line < len; ep++)
 383                 if (*ep != ' ' && *ep != '\t')
 384                         break;
 385         if (ep != cp)
 386                 emit_line(ecbdata->file, plain, reset, cp, ep - cp);
 387
 388         if (ep < line + len)
 389                 emit_line(ecbdata->file, func, reset, ep, line + len - ep);
 390 }
 391
 392 static struct diff_tempfile *claim_diff_tempfile(void) {
 393         int i;
 394         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
 395                 if (!diff_temp[i].name)
 396                         return diff_temp + i;
 397         die("BUG: diff is failing to clean up its tempfiles");
 398 }
 399
 400 static int remove_tempfile_installed;
 401
 402 static void remove_tempfile(void)
 403 {
 404         int i;
 405         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
 406                 if (diff_temp[i].name == diff_temp[i].tmp_path)
 407                         unlink_or_warn(diff_temp[i].name);
 408                 diff_temp[i].name = NULL;
 409         }
 410 }
 411
 412 static void remove_tempfile_on_signal(int signo)
 413 {
 414         remove_tempfile();
 415         sigchain_pop(signo);
 416         raise(signo);
 417 }
 418
 419 static void print_line_count(FILE *file, int count)
 420 {
 421         switch (count) {
 422         case 0:
 423                 fprintf(file, "0,0");
 424                 break;
 425         case 1:
 426                 fprintf(file, "1");
 427                 break;
 428         default:
 429                 fprintf(file, "1,%d", count);
 430                 break;
 431         }
 432 }
 433
 434 static void emit_rewrite_lines(struct emit_callback *ecb,
 435                                int prefix, const char *data, int size)
 436 {
 437         const char *endp = NULL;
 438         static const char *nneof = " No newline at end of file\n";
 439         const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
 440         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
 441
 442         while (0 < size) {
 443                 int len;
 444
 445                 endp = memchr(data, '\n', size);
 446                 len = endp ? (endp - data + 1) : size;
 447                 if (prefix != '+') {
 448                         ecb->lno_in_preimage++;
 449                         emit_line_0(ecb->file, old, reset, '-',
 450                                     data, len);
 451                 } else {
 452                         ecb->lno_in_postimage++;
 453                         emit_add_line(reset, ecb, data, len);
 454                 }
 455                 size -= len;
 456                 data += len;
 457         }
 458         if (!endp) {
 459                 const char *plain = diff_get_color(ecb->color_diff,
 460                                                    DIFF_PLAIN);
 461                 emit_line_0(ecb->file, plain, reset, '\\',
 462                             nneof, strlen(nneof));
 463         }
 464 }
 465
 466 static void emit_rewrite_diff(const char *name_a,
 467                               const char *name_b,
 468                               struct diff_filespec *one,
 469                               struct diff_filespec *two,
 470                               struct userdiff_driver *textconv_one,
 471                               struct userdiff_driver *textconv_two,
 472                               struct diff_options *o)
 473 {
 474         int lc_a, lc_b;
 475         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
 476         const char *name_a_tab, *name_b_tab;
 477         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 478         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 479         const char *reset = diff_get_color(color_diff, DIFF_RESET);
 480         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 481         const char *a_prefix, *b_prefix;
 482         char *data_one, *data_two;
 483         size_t size_one, size_two;
 484         struct emit_callback ecbdata;
 485
 486         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 487                 a_prefix = o->b_prefix;
 488                 b_prefix = o->a_prefix;
 489         } else {
 490                 a_prefix = o->a_prefix;
 491                 b_prefix = o->b_prefix;
 492         }
 493
 494         name_a += (*name_a == '/');
 495         name_b += (*name_b == '/');
 496         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 497         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 498
 499         strbuf_reset(&a_name);
 500         strbuf_reset(&b_name);
 501         quote_two_c_style(&a_name, a_prefix, name_a, 0);
 502         quote_two_c_style(&b_name, b_prefix, name_b, 0);
 503
 504         size_one = fill_textconv(textconv_one, one, &data_one);
 505         size_two = fill_textconv(textconv_two, two, &data_two);
 506
 507         memset(&ecbdata, 0, sizeof(ecbdata));
 508         ecbdata.color_diff = color_diff;
 509         ecbdata.found_changesp = &o->found_changes;
 510         ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
 511         ecbdata.file = o->file;
 512         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
 513                 mmfile_t mf1, mf2;
 514                 mf1.ptr = (char *)data_one;
 515                 mf2.ptr = (char *)data_two;
 516                 mf1.size = size_one;
 517                 mf2.size = size_two;
 518                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
 519         }
 520         ecbdata.lno_in_preimage = 1;
 521         ecbdata.lno_in_postimage = 1;
 522
 523         lc_a = count_lines(data_one, size_one);
 524         lc_b = count_lines(data_two, size_two);
 525         fprintf(o->file,
 526                 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
 527                 metainfo, a_name.buf, name_a_tab, reset,
 528                 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
 529         print_line_count(o->file, lc_a);
 530         fprintf(o->file, " +");
 531         print_line_count(o->file, lc_b);
 532         fprintf(o->file, " @@%s\n", reset);
 533         if (lc_a)
 534                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
 535         if (lc_b)
 536                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
 537         if (textconv_one)
 538                 free((char *)data_one);
 539         if (textconv_two)
 540                 free((char *)data_two);
 541 }
 542
 543 struct diff_words_buffer {
 544         mmfile_t text;
 545         long alloc;
 546         struct diff_words_orig {
 547                 const char *begin, *end;
 548         } *orig;
 549         int orig_nr, orig_alloc;
 550 };
 551
 552 static void diff_words_append(char *line, unsigned long len,
 553                 struct diff_words_buffer *buffer)
 554 {
 555         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
 556         line++;
 557         len--;
 558         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 559         buffer->text.size += len;
 560         buffer->text.ptr[buffer->text.size] = '\0';
 561 }
 562
 563 struct diff_words_data {
 564         struct diff_words_buffer minus, plus;
 565         const char *current_plus;
 566         FILE *file;
 567         regex_t *word_regex;
 568 };
 569
 570 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 571 {
 572         struct diff_words_data *diff_words = priv;
 573         int minus_first, minus_len, plus_first, plus_len;
 574         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 575
 576         if (line[0] != '@' || parse_hunk_header(line, len,
 577                         &minus_first, &minus_len, &plus_first, &plus_len))
 578                 return;
 579
 580         /* POSIX requires that first be decremented by one if len == 0... */
 581         if (minus_len) {
 582                 minus_begin = diff_words->minus.orig[minus_first].begin;
 583                 minus_end =
 584                         diff_words->minus.orig[minus_first + minus_len - 1].end;
 585         } else
 586                 minus_begin = minus_end =
 587                         diff_words->minus.orig[minus_first].end;
 588
 589         if (plus_len) {
 590                 plus_begin = diff_words->plus.orig[plus_first].begin;
 591                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
 592         } else
 593                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
 594
 595         if (diff_words->current_plus != plus_begin)
 596                 fwrite(diff_words->current_plus,
 597                                 plus_begin - diff_words->current_plus, 1,
 598                                 diff_words->file);
 599         if (minus_begin != minus_end)
 600                 color_fwrite_lines(diff_words->file,
 601                                 diff_get_color(1, DIFF_FILE_OLD),
 602                                 minus_end - minus_begin, minus_begin);
 603         if (plus_begin != plus_end)
 604                 color_fwrite_lines(diff_words->file,
 605                                 diff_get_color(1, DIFF_FILE_NEW),
 606                                 plus_end - plus_begin, plus_begin);
 607
 608         diff_words->current_plus = plus_end;
 609 }
 610
 611 /* This function starts looking at *begin, and returns 0 iff a word was found. */
 612 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
 613                 int *begin, int *end)
 614 {
 615         if (word_regex && *begin < buffer->size) {
 616                 regmatch_t match[1];
 617                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
 618                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
 619                                         '\n', match[0].rm_eo - match[0].rm_so);
 620                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
 621                         *begin += match[0].rm_so;
 622                         return *begin >= *end;
 623                 }
 624                 return -1;
 625         }
 626
 627         /* find the next word */
 628         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
 629                 (*begin)++;
 630         if (*begin >= buffer->size)
 631                 return -1;
 632
 633         /* find the end of the word */
 634         *end = *begin + 1;
 635         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
 636                 (*end)++;
 637
 638         return 0;
 639 }
 640
 641 /*
 642  * This function splits the words in buffer->text, stores the list with
 643  * newline separator into out, and saves the offsets of the original words
 644  * in buffer->orig.
 645  */
 646 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
 647                 regex_t *word_regex)
 648 {
 649         int i, j;
 650         long alloc = 0;
 651
 652         out->size = 0;
 653         out->ptr = NULL;
 654
 655         /* fake an empty "0th" word */
 656         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
 657         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
 658         buffer->orig_nr = 1;
 659
 660         for (i = 0; i < buffer->text.size; i++) {
 661                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
 662                         return;
 663
 664                 /* store original boundaries */
 665                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
 666                                 buffer->orig_alloc);
 667                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
 668                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
 669                 buffer->orig_nr++;
 670
 671                 /* store one word */
 672                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
 673                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
 674                 out->ptr[out->size + j - i] = '\n';
 675                 out->size += j - i + 1;
 676
 677                 i = j - 1;
 678         }
 679 }
 680
 681 /* this executes the word diff on the accumulated buffers */
 682 static void diff_words_show(struct diff_words_data *diff_words)
 683 {
 684         xpparam_t xpp;
 685         xdemitconf_t xecfg;
 686         mmfile_t minus, plus;
 687
 688         /* special case: only removal */
 689         if (!diff_words->plus.text.size) {
 690                 color_fwrite_lines(diff_words->file,
 691                         diff_get_color(1, DIFF_FILE_OLD),
 692                         diff_words->minus.text.size, diff_words->minus.text.ptr);
 693                 diff_words->minus.text.size = 0;
 694                 return;
 695         }
 696
 697         diff_words->current_plus = diff_words->plus.text.ptr;
 698
 699         memset(&xpp, 0, sizeof(xpp));
 700         memset(&xecfg, 0, sizeof(xecfg));
 701         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
 702         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
 703         xpp.flags = XDF_NEED_MINIMAL;
 704         /* as only the hunk header will be parsed, we need a 0-context */
 705         xecfg.ctxlen = 0;
 706         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
 707                       &xpp, &xecfg);
 708         free(minus.ptr);
 709         free(plus.ptr);
 710         if (diff_words->current_plus != diff_words->plus.text.ptr +
 711                         diff_words->plus.text.size)
 712                 fwrite(diff_words->current_plus,
 713                         diff_words->plus.text.ptr + diff_words->plus.text.size
 714                         - diff_words->current_plus, 1,
 715                         diff_words->file);
 716         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 717 }
 718
 719 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
 720 static void diff_words_flush(struct emit_callback *ecbdata)
 721 {
 722         if (ecbdata->diff_words->minus.text.size ||
 723             ecbdata->diff_words->plus.text.size)
 724                 diff_words_show(ecbdata->diff_words);
 725 }
 726
 727 static void free_diff_words_data(struct emit_callback *ecbdata)
 728 {
 729         if (ecbdata->diff_words) {
 730                 diff_words_flush(ecbdata);
 731                 free (ecbdata->diff_words->minus.text.ptr);
 732                 free (ecbdata->diff_words->minus.orig);
 733                 free (ecbdata->diff_words->plus.text.ptr);
 734                 free (ecbdata->diff_words->plus.orig);
 735                 free(ecbdata->diff_words->word_regex);
 736                 free(ecbdata->diff_words);
 737                 ecbdata->diff_words = NULL;
 738         }
 739 }
 740
 741 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 742 {
 743         if (diff_use_color)
 744                 return diff_colors[ix];
 745         return "";
 746 }
 747
 748 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
 749 {
 750         const char *cp;
 751         unsigned long allot;
 752         size_t l = len;
 753
 754         if (ecb->truncate)
 755                 return ecb->truncate(line, len);
 756         cp = line;
 757         allot = l;
 758         while (0 < l) {
 759                 (void) utf8_width(&cp, &l);
 760                 if (!cp)
 761                         break; /* truncated in the middle? */
 762         }
 763         return allot - l;
 764 }
 765
 766 static void find_lno(const char *line, struct emit_callback *ecbdata)
 767 {
 768         const char *p;
 769         ecbdata->lno_in_preimage = 0;
 770         ecbdata->lno_in_postimage = 0;
 771         p = strchr(line, '-');
 772         if (!p)
 773                 return; /* cannot happen */
 774         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
 775         p = strchr(p, '+');
 776         if (!p)
 777                 return; /* cannot happen */
 778         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
 779 }
 780
 781 static void fn_out_consume(void *priv, char *line, unsigned long len)
 782 {
 783         struct emit_callback *ecbdata = priv;
 784         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 785         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 786         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 787
 788         if (ecbdata->header) {
 789                 fprintf(ecbdata->file, "%s", ecbdata->header->buf);
 790                 strbuf_reset(ecbdata->header);
 791                 ecbdata->header = NULL;
 792         }
 793         *(ecbdata->found_changesp) = 1;
 794
 795         if (ecbdata->label_path[0]) {
 796                 const char *name_a_tab, *name_b_tab;
 797
 798                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 799                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 800
 801                 fprintf(ecbdata->file, "%s--- %s%s%s\n",
 802                         meta, ecbdata->label_path[0], reset, name_a_tab);
 803                 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
 804                         meta, ecbdata->label_path[1], reset, name_b_tab);
 805                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 806         }
 807
 808         if (diff_suppress_blank_empty
 809             && len == 2 && line[0] == ' ' && line[1] == '\n') {
 810                 line[0] = '\n';
 811                 len = 1;
 812         }
 813
 814         if (line[0] == '@') {
 815                 if (ecbdata->diff_words)
 816                         diff_words_flush(ecbdata);
 817                 len = sane_truncate_line(ecbdata, line, len);
 818                 find_lno(line, ecbdata);
 819                 emit_hunk_header(ecbdata, line, len);
 820                 if (line[len-1] != '\n')
 821                         putc('\n', ecbdata->file);
 822                 return;
 823         }
 824
 825         if (len < 1) {
 826                 emit_line(ecbdata->file, reset, reset, line, len);
 827                 return;
 828         }
 829
 830         if (ecbdata->diff_words) {
 831                 if (line[0] == '-') {
 832                         diff_words_append(line, len,
 833                                           &ecbdata->diff_words->minus);
 834                         return;
 835                 } else if (line[0] == '+') {
 836                         diff_words_append(line, len,
 837                                           &ecbdata->diff_words->plus);
 838                         return;
 839                 }
 840                 diff_words_flush(ecbdata);
 841                 line++;
 842                 len--;
 843                 emit_line(ecbdata->file, plain, reset, line, len);
 844                 return;
 845         }
 846
 847         if (line[0] != '+') {
 848                 const char *color =
 849                         diff_get_color(ecbdata->color_diff,
 850                                        line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
 851                 ecbdata->lno_in_preimage++;
 852                 if (line[0] == ' ')
 853                         ecbdata->lno_in_postimage++;
 854                 emit_line(ecbdata->file, color, reset, line, len);
 855         } else {
 856                 ecbdata->lno_in_postimage++;
 857                 emit_add_line(reset, ecbdata, line + 1, len - 1);
 858         }
 859 }
 860
 861 static char *pprint_rename(const char *a, const char *b)
 862 {
 863         const char *old = a;
 864         const char *new = b;
 865         struct strbuf name = STRBUF_INIT;
 866         int pfx_length, sfx_length;
 867         int len_a = strlen(a);
 868         int len_b = strlen(b);
 869         int a_midlen, b_midlen;
 870         int qlen_a = quote_c_style(a, NULL, NULL, 0);
 871         int qlen_b = quote_c_style(b, NULL, NULL, 0);
 872
 873         if (qlen_a || qlen_b) {
 874                 quote_c_style(a, &name, NULL, 0);
 875                 strbuf_addstr(&name, " => ");
 876                 quote_c_style(b, &name, NULL, 0);
 877                 return strbuf_detach(&name, NULL);
 878         }
 879
 880         /* Find common prefix */
 881         pfx_length = 0;
 882         while (*old && *new && *old == *new) {
 883                 if (*old == '/')
 884                         pfx_length = old - a + 1;
 885                 old++;
 886                 new++;
 887         }
 888
 889         /* Find common suffix */
 890         old = a + len_a;
 891         new = b + len_b;
 892         sfx_length = 0;
 893         while (a <= old && b <= new && *old == *new) {
 894                 if (*old == '/')
 895                         sfx_length = len_a - (old - a);
 896                 old--;
 897                 new--;
 898         }
 899
 900         /*
 901          * pfx{mid-a => mid-b}sfx
 902          * {pfx-a => pfx-b}sfx
 903          * pfx{sfx-a => sfx-b}
 904          * name-a => name-b
 905          */
 906         a_midlen = len_a - pfx_length - sfx_length;
 907         b_midlen = len_b - pfx_length - sfx_length;
 908         if (a_midlen < 0)
 909                 a_midlen = 0;
 910         if (b_midlen < 0)
 911                 b_midlen = 0;
 912
 913         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 914         if (pfx_length + sfx_length) {
 915                 strbuf_add(&name, a, pfx_length);
 916                 strbuf_addch(&name, '{');
 917         }
 918         strbuf_add(&name, a + pfx_length, a_midlen);
 919         strbuf_addstr(&name, " => ");
 920         strbuf_add(&name, b + pfx_length, b_midlen);
 921         if (pfx_length + sfx_length) {
 922                 strbuf_addch(&name, '}');
 923                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 924         }
 925         return strbuf_detach(&name, NULL);
 926 }
 927
 928 struct diffstat_t {
 929         int nr;
 930         int alloc;
 931         struct diffstat_file {
 932                 char *from_name;
 933                 char *name;
 934                 char *print_name;
 935                 unsigned is_unmerged:1;
 936                 unsigned is_binary:1;
 937                 unsigned is_renamed:1;
 938                 uintmax_t added, deleted;
 939         } **files;
 940 };
 941
 942 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 943                                           const char *name_a,
 944                                           const char *name_b)
 945 {
 946         struct diffstat_file *x;
 947         x = xcalloc(sizeof (*x), 1);
 948         if (diffstat->nr == diffstat->alloc) {
 949                 diffstat->alloc = alloc_nr(diffstat->alloc);
 950                 diffstat->files = xrealloc(diffstat->files,
 951                                 diffstat->alloc * sizeof(x));
 952         }
 953         diffstat->files[diffstat->nr++] = x;
 954         if (name_b) {
 955                 x->from_name = xstrdup(name_a);
 956                 x->name = xstrdup(name_b);
 957                 x->is_renamed = 1;
 958         }
 959         else {
 960                 x->from_name = NULL;
 961                 x->name = xstrdup(name_a);
 962         }
 963         return x;
 964 }
 965
 966 static void diffstat_consume(void *priv, char *line, unsigned long len)
 967 {
 968         struct diffstat_t *diffstat = priv;
 969         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 970
 971         if (line[0] == '+')
 972                 x->added++;
 973         else if (line[0] == '-')
 974                 x->deleted++;
 975 }
 976
 977 const char mime_boundary_leader[] = "------------";
 978
 979 static int scale_linear(int it, int width, int max_change)
 980 {
 981         /*
 982          * make sure that at least one '-' is printed if there were deletions,
 983          * and likewise for '+'.
 984          */
 985         if (max_change < 2)
 986                 return it;
 987         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
 988 }
 989
 990 static void show_name(FILE *file,
 991                       const char *prefix, const char *name, int len)
 992 {
 993         fprintf(file, " %s%-*s |", prefix, len, name);
 994 }
 995
 996 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
 997 {
 998         if (cnt <= 0)
 999                 return;
1000         fprintf(file, "%s", set);
1001         while (cnt--)
1002                 putc(ch, file);
1003         fprintf(file, "%s", reset);
1004 }
1005
1006 static void fill_print_name(struct diffstat_file *file)
1007 {
1008         char *pname;
1009
1010         if (file->print_name)
1011                 return;
1012
1013         if (!file->is_renamed) {
1014                 struct strbuf buf = STRBUF_INIT;
1015                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1016                         pname = strbuf_detach(&buf, NULL);
1017                 } else {
1018                         pname = file->name;
1019                         strbuf_release(&buf);
1020                 }
1021         } else {
1022                 pname = pprint_rename(file->from_name, file->name);
1023         }
1024         file->print_name = pname;
1025 }
1026
1027 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1028 {
1029         int i, len, add, del, adds = 0, dels = 0;
1030         uintmax_t max_change = 0, max_len = 0;
1031         int total_files = data->nr;
1032         int width, name_width;
1033         const char *reset, *set, *add_c, *del_c;
1034
1035         if (data->nr == 0)
1036                 return;
1037
1038         width = options->stat_width ? options->stat_width : 80;
1039         name_width = options->stat_name_width ? options->stat_name_width : 50;
1040
1041         /* Sanity: give at least 5 columns to the graph,
1042          * but leave at least 10 columns for the name.
1043          */
1044         if (width < 25)
1045                 width = 25;
1046         if (name_width < 10)
1047                 name_width = 10;
1048         else if (width < name_width + 15)
1049                 name_width = width - 15;
1050
1051         /* Find the longest filename and max number of changes */
1052         reset = diff_get_color_opt(options, DIFF_RESET);
1053         set   = diff_get_color_opt(options, DIFF_PLAIN);
1054         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1055         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1056
1057         for (i = 0; i < data->nr; i++) {
1058                 struct diffstat_file *file = data->files[i];
1059                 uintmax_t change = file->added + file->deleted;
1060                 fill_print_name(file);
1061                 len = strlen(file->print_name);
1062                 if (max_len < len)
1063                         max_len = len;
1064
1065                 if (file->is_binary || file->is_unmerged)
1066                         continue;
1067                 if (max_change < change)
1068                         max_change = change;
1069         }
1070
1071         /* Compute the width of the graph part;
1072          * 10 is for one blank at the beginning of the line plus
1073          * " | count " between the name and the graph.
1074          *
1075          * From here on, name_width is the width of the name area,
1076          * and width is the width of the graph area.
1077          */
1078         name_width = (name_width < max_len) ? name_width : max_len;
1079         if (width < (name_width + 10) + max_change)
1080                 width = width - (name_width + 10);
1081         else
1082                 width = max_change;
1083
1084         for (i = 0; i < data->nr; i++) {
1085                 const char *prefix = "";
1086                 char *name = data->files[i]->print_name;
1087                 uintmax_t added = data->files[i]->added;
1088                 uintmax_t deleted = data->files[i]->deleted;
1089                 int name_len;
1090
1091                 /*
1092                  * "scale" the filename
1093                  */
1094                 len = name_width;
1095                 name_len = strlen(name);
1096                 if (name_width < name_len) {
1097                         char *slash;
1098                         prefix = "...";
1099                         len -= 3;
1100                         name += name_len - len;
1101                         slash = strchr(name, '/');
1102                         if (slash)
1103                                 name = slash;
1104                 }
1105
1106                 if (data->files[i]->is_binary) {
1107                         show_name(options->file, prefix, name, len);
1108                         fprintf(options->file, "  Bin ");
1109                         fprintf(options->file, "%s%"PRIuMAX"%s",
1110                                 del_c, deleted, reset);
1111                         fprintf(options->file, " -> ");
1112                         fprintf(options->file, "%s%"PRIuMAX"%s",
1113                                 add_c, added, reset);
1114                         fprintf(options->file, " bytes");
1115                         fprintf(options->file, "\n");
1116                         continue;
1117                 }
1118                 else if (data->files[i]->is_unmerged) {
1119                         show_name(options->file, prefix, name, len);
1120                         fprintf(options->file, "  Unmerged\n");
1121                         continue;
1122                 }
1123                 else if (!data->files[i]->is_renamed &&
1124                          (added + deleted == 0)) {
1125                         total_files--;
1126                         continue;
1127                 }
1128
1129                 /*
1130                  * scale the add/delete
1131                  */
1132                 add = added;
1133                 del = deleted;
1134                 adds += add;
1135                 dels += del;
1136
1137                 if (width <= max_change) {
1138                         add = scale_linear(add, width, max_change);
1139                         del = scale_linear(del, width, max_change);
1140                 }
1141                 show_name(options->file, prefix, name, len);
1142                 fprintf(options->file, "%5"PRIuMAX"%s", added + deleted,
1143                                 added + deleted ? " " : "");
1144                 show_graph(options->file, '+', add, add_c, reset);
1145                 show_graph(options->file, '-', del, del_c, reset);
1146                 fprintf(options->file, "\n");
1147         }
1148         fprintf(options->file,
1149                " %d files changed, %d insertions(+), %d deletions(-)\n",
1150                total_files, adds, dels);
1151 }
1152
1153 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1154 {
1155         int i, adds = 0, dels = 0, total_files = data->nr;
1156
1157         if (data->nr == 0)
1158                 return;
1159
1160         for (i = 0; i < data->nr; i++) {
1161                 if (!data->files[i]->is_binary &&
1162                     !data->files[i]->is_unmerged) {
1163                         int added = data->files[i]->added;
1164                         int deleted= data->files[i]->deleted;
1165                         if (!data->files[i]->is_renamed &&
1166                             (added + deleted == 0)) {
1167                                 total_files--;
1168                         } else {
1169                                 adds += added;
1170                                 dels += deleted;
1171                         }
1172                 }
1173         }
1174         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
1175                total_files, adds, dels);
1176 }
1177
1178 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1179 {
1180         int i;
1181
1182         if (data->nr == 0)
1183                 return;
1184
1185         for (i = 0; i < data->nr; i++) {
1186                 struct diffstat_file *file = data->files[i];
1187
1188                 if (file->is_binary)
1189                         fprintf(options->file, "-\t-\t");
1190                 else
1191                         fprintf(options->file,
1192                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1193                                 file->added, file->deleted);
1194                 if (options->line_termination) {
1195                         fill_print_name(file);
1196                         if (!file->is_renamed)
1197                                 write_name_quoted(file->name, options->file,
1198                                                   options->line_termination);
1199                         else {
1200                                 fputs(file->print_name, options->file);
1201                                 putc(options->line_termination, options->file);
1202                         }
1203                 } else {
1204                         if (file->is_renamed) {
1205                                 putc('\0', options->file);
1206                                 write_name_quoted(file->from_name, options->file, '\0');
1207                         }
1208                         write_name_quoted(file->name, options->file, '\0');
1209                 }
1210         }
1211 }
1212
1213 struct dirstat_file {
1214         const char *name;
1215         unsigned long changed;
1216 };
1217
1218 struct dirstat_dir {
1219         struct dirstat_file *files;
1220         int alloc, nr, percent, cumulative;
1221 };
1222
1223 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1224 {
1225         unsigned long this_dir = 0;
1226         unsigned int sources = 0;
1227
1228         while (dir->nr) {
1229                 struct dirstat_file *f = dir->files;
1230                 int namelen = strlen(f->name);
1231                 unsigned long this;
1232                 char *slash;
1233
1234                 if (namelen < baselen)
1235                         break;
1236                 if (memcmp(f->name, base, baselen))
1237                         break;
1238                 slash = strchr(f->name + baselen, '/');
1239                 if (slash) {
1240                         int newbaselen = slash + 1 - f->name;
1241                         this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1242                         sources++;
1243                 } else {
1244                         this = f->changed;
1245                         dir->files++;
1246                         dir->nr--;
1247                         sources += 2;
1248                 }
1249                 this_dir += this;
1250         }
1251
1252         /*
1253          * We don't report dirstat's for
1254          *  - the top level
1255          *  - or cases where everything came from a single directory
1256          *    under this directory (sources == 1).
1257          */
1258         if (baselen && sources != 1) {
1259                 int permille = this_dir * 1000 / changed;
1260                 if (permille) {
1261                         int percent = permille / 10;
1262                         if (percent >= dir->percent) {
1263                                 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1264                                 if (!dir->cumulative)
1265                                         return 0;
1266                         }
1267                 }
1268         }
1269         return this_dir;
1270 }
1271
1272 static int dirstat_compare(const void *_a, const void *_b)
1273 {
1274         const struct dirstat_file *a = _a;
1275         const struct dirstat_file *b = _b;
1276         return strcmp(a->name, b->name);
1277 }
1278
1279 static void show_dirstat(struct diff_options *options)
1280 {
1281         int i;
1282         unsigned long changed;
1283         struct dirstat_dir dir;
1284         struct diff_queue_struct *q = &diff_queued_diff;
1285
1286         dir.files = NULL;
1287         dir.alloc = 0;
1288         dir.nr = 0;
1289         dir.percent = options->dirstat_percent;
1290         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1291
1292         changed = 0;
1293         for (i = 0; i < q->nr; i++) {
1294                 struct diff_filepair *p = q->queue[i];
1295                 const char *name;
1296                 unsigned long copied, added, damage;
1297
1298                 name = p->one->path ? p->one->path : p->two->path;
1299
1300                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1301                         diff_populate_filespec(p->one, 0);
1302                         diff_populate_filespec(p->two, 0);
1303                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1304                                                &copied, &added);
1305                         diff_free_filespec_data(p->one);
1306                         diff_free_filespec_data(p->two);
1307                 } else if (DIFF_FILE_VALID(p->one)) {
1308                         diff_populate_filespec(p->one, 1);
1309                         copied = added = 0;
1310                         diff_free_filespec_data(p->one);
1311                 } else if (DIFF_FILE_VALID(p->two)) {
1312                         diff_populate_filespec(p->two, 1);
1313                         copied = 0;
1314                         added = p->two->size;
1315                         diff_free_filespec_data(p->two);
1316                 } else
1317                         continue;
1318
1319                 /*
1320                  * Original minus copied is the removed material,
1321                  * added is the new material.  They are both damages
1322                  * made to the preimage. In --dirstat-by-file mode, count
1323                  * damaged files, not damaged lines. This is done by
1324                  * counting only a single damaged line per file.
1325                  */
1326                 damage = (p->one->size - copied) + added;
1327                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1328                         damage = 1;
1329
1330                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1331                 dir.files[dir.nr].name = name;
1332                 dir.files[dir.nr].changed = damage;
1333                 changed += damage;
1334                 dir.nr++;
1335         }
1336
1337         /* This can happen even with many files, if everything was renames */
1338         if (!changed)
1339                 return;
1340
1341         /* Show all directories with more than x% of the changes */
1342         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1343         gather_dirstat(options->file, &dir, changed, "", 0);
1344 }
1345
1346 static void free_diffstat_info(struct diffstat_t *diffstat)
1347 {
1348         int i;
1349         for (i = 0; i < diffstat->nr; i++) {
1350                 struct diffstat_file *f = diffstat->files[i];
1351                 if (f->name != f->print_name)
1352                         free(f->print_name);
1353                 free(f->name);
1354                 free(f->from_name);
1355                 free(f);
1356         }
1357         free(diffstat->files);
1358 }
1359
1360 struct checkdiff_t {
1361         const char *filename;
1362         int lineno;
1363         int conflict_marker_size;
1364         struct diff_options *o;
1365         unsigned ws_rule;
1366         unsigned status;
1367 };
1368
1369 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
1370 {
1371         char firstchar;
1372         int cnt;
1373
1374         if (len < marker_size + 1)
1375                 return 0;
1376         firstchar = line[0];
1377         switch (firstchar) {
1378         case '=': case '>': case '<': case '|':
1379                 break;
1380         default:
1381                 return 0;
1382         }
1383         for (cnt = 1; cnt < marker_size; cnt++)
1384                 if (line[cnt] != firstchar)
1385                         return 0;
1386         /* line[1] thru line[marker_size-1] are same as firstchar */
1387         if (len < marker_size + 1 || !isspace(line[marker_size]))
1388                 return 0;
1389         return 1;
1390 }
1391
1392 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1393 {
1394         struct checkdiff_t *data = priv;
1395         int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1396         int marker_size = data->conflict_marker_size;
1397         const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1398         const char *reset = diff_get_color(color_diff, DIFF_RESET);
1399         const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1400         char *err;
1401
1402         if (line[0] == '+') {
1403                 unsigned bad;
1404                 data->lineno++;
1405                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
1406                         data->status |= 1;
1407                         fprintf(data->o->file,
1408                                 "%s:%d: leftover conflict marker\n",
1409                                 data->filename, data->lineno);
1410                 }
1411                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1412                 if (!bad)
1413                         return;
1414                 data->status |= bad;
1415                 err = whitespace_error_string(bad);
1416                 fprintf(data->o->file, "%s:%d: %s.\n",
1417                         data->filename, data->lineno, err);
1418                 free(err);
1419                 emit_line(data->o->file, set, reset, line, 1);
1420                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1421                               data->o->file, set, reset, ws);
1422         } else if (line[0] == ' ') {
1423                 data->lineno++;
1424         } else if (line[0] == '@') {
1425                 char *plus = strchr(line, '+');
1426                 if (plus)
1427                         data->lineno = strtol(plus, NULL, 10) - 1;
1428                 else
1429                         die("invalid diff");
1430         }
1431 }
1432
1433 static unsigned char *deflate_it(char *data,
1434                                  unsigned long size,
1435                                  unsigned long *result_size)
1436 {
1437         int bound;
1438         unsigned char *deflated;
1439         z_stream stream;
1440
1441         memset(&stream, 0, sizeof(stream));
1442         deflateInit(&stream, zlib_compression_level);
1443         bound = deflateBound(&stream, size);
1444         deflated = xmalloc(bound);
1445         stream.next_out = deflated;
1446         stream.avail_out = bound;
1447
1448         stream.next_in = (unsigned char *)data;
1449         stream.avail_in = size;
1450         while (deflate(&stream, Z_FINISH) == Z_OK)
1451                 ; /* nothing */
1452         deflateEnd(&stream);
1453         *result_size = stream.total_out;
1454         return deflated;
1455 }
1456
1457 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1458 {
1459         void *cp;
1460         void *delta;
1461         void *deflated;
1462         void *data;
1463         unsigned long orig_size;
1464         unsigned long delta_size;
1465         unsigned long deflate_size;
1466         unsigned long data_size;
1467
1468         /* We could do deflated delta, or we could do just deflated two,
1469          * whichever is smaller.
1470          */
1471         delta = NULL;
1472         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1473         if (one->size && two->size) {
1474                 delta = diff_delta(one->ptr, one->size,
1475                                    two->ptr, two->size,
1476                                    &delta_size, deflate_size);
1477                 if (delta) {
1478                         void *to_free = delta;
1479                         orig_size = delta_size;
1480                         delta = deflate_it(delta, delta_size, &delta_size);
1481                         free(to_free);
1482                 }
1483         }
1484
1485         if (delta && delta_size < deflate_size) {
1486                 fprintf(file, "delta %lu\n", orig_size);
1487                 free(deflated);
1488                 data = delta;
1489                 data_size = delta_size;
1490         }
1491         else {
1492                 fprintf(file, "literal %lu\n", two->size);
1493                 free(delta);
1494                 data = deflated;
1495                 data_size = deflate_size;
1496         }
1497
1498         /* emit data encoded in base85 */
1499         cp = data;
1500         while (data_size) {
1501                 int bytes = (52 < data_size) ? 52 : data_size;
1502                 char line[70];
1503                 data_size -= bytes;
1504                 if (bytes <= 26)
1505                         line[0] = bytes + 'A' - 1;
1506                 else
1507                         line[0] = bytes - 26 + 'a' - 1;
1508                 encode_85(line + 1, cp, bytes);
1509                 cp = (char *) cp + bytes;
1510                 fputs(line, file);
1511                 fputc('\n', file);
1512         }
1513         fprintf(file, "\n");
1514         free(data);
1515 }
1516
1517 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1518 {
1519         fprintf(file, "GIT binary patch\n");
1520         emit_binary_diff_body(file, one, two);
1521         emit_binary_diff_body(file, two, one);
1522 }
1523
1524 static void diff_filespec_load_driver(struct diff_filespec *one)
1525 {
1526         if (!one->driver)
1527                 one->driver = userdiff_find_by_path(one->path);
1528         if (!one->driver)
1529                 one->driver = userdiff_find_by_name("default");
1530 }
1531
1532 int diff_filespec_is_binary(struct diff_filespec *one)
1533 {
1534         if (one->is_binary == -1) {
1535                 diff_filespec_load_driver(one);
1536                 if (one->driver->binary != -1)
1537                         one->is_binary = one->driver->binary;
1538                 else {
1539                         if (!one->data && DIFF_FILE_VALID(one))
1540                                 diff_populate_filespec(one, 0);
1541                         if (one->data)
1542                                 one->is_binary = buffer_is_binary(one->data,
1543                                                 one->size);
1544                         if (one->is_binary == -1)
1545                                 one->is_binary = 0;
1546                 }
1547         }
1548         return one->is_binary;
1549 }
1550
1551 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1552 {
1553         diff_filespec_load_driver(one);
1554         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1555 }
1556
1557 static const char *userdiff_word_regex(struct diff_filespec *one)
1558 {
1559         diff_filespec_load_driver(one);
1560         return one->driver->word_regex;
1561 }
1562
1563 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1564 {
1565         if (!options->a_prefix)
1566                 options->a_prefix = a;
1567         if (!options->b_prefix)
1568                 options->b_prefix = b;
1569 }
1570
1571 static struct userdiff_driver *get_textconv(struct diff_filespec *one)
1572 {
1573         if (!DIFF_FILE_VALID(one))
1574                 return NULL;
1575         if (!S_ISREG(one->mode))
1576                 return NULL;
1577         diff_filespec_load_driver(one);
1578         if (!one->driver->textconv)
1579                 return NULL;
1580
1581         if (one->driver->textconv_want_cache && !one->driver->textconv_cache) {
1582                 struct notes_cache *c = xmalloc(sizeof(*c));
1583                 struct strbuf name = STRBUF_INIT;
1584
1585                 strbuf_addf(&name, "textconv/%s", one->driver->name);
1586                 notes_cache_init(c, name.buf, one->driver->textconv);
1587                 one->driver->textconv_cache = c;
1588         }
1589
1590         return one->driver;
1591 }
1592
1593 static void builtin_diff(const char *name_a,
1594                          const char *name_b,
1595                          struct diff_filespec *one,
1596                          struct diff_filespec *two,
1597                          const char *xfrm_msg,
1598                          struct diff_options *o,
1599                          int complete_rewrite)
1600 {
1601         mmfile_t mf1, mf2;
1602         const char *lbl[2];
1603         char *a_one, *b_two;
1604         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1605         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1606         const char *a_prefix, *b_prefix;
1607         struct userdiff_driver *textconv_one = NULL;
1608         struct userdiff_driver *textconv_two = NULL;
1609         struct strbuf header = STRBUF_INIT;
1610
1611         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
1612                         (!one->mode || S_ISGITLINK(one->mode)) &&
1613                         (!two->mode || S_ISGITLINK(two->mode))) {
1614                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
1615                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
1616                 show_submodule_summary(o->file, one ? one->path : two->path,
1617                                 one->sha1, two->sha1, two->dirty_submodule,
1618                                 del, add, reset);
1619                 return;
1620         }
1621
1622         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
1623                 textconv_one = get_textconv(one);
1624                 textconv_two = get_textconv(two);
1625         }
1626
1627         diff_set_mnemonic_prefix(o, "a/", "b/");
1628         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1629                 a_prefix = o->b_prefix;
1630                 b_prefix = o->a_prefix;
1631         } else {
1632                 a_prefix = o->a_prefix;
1633                 b_prefix = o->b_prefix;
1634         }
1635
1636         /* Never use a non-valid filename anywhere if at all possible */
1637         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1638         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1639
1640         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1641         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1642         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1643         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1644         strbuf_addf(&header, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1645         if (lbl[0][0] == '/') {
1646                 /* /dev/null */
1647                 strbuf_addf(&header, "%snew file mode %06o%s\n", set, two->mode, reset);
1648                 if (xfrm_msg && xfrm_msg[0])
1649                         strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1650         }
1651         else if (lbl[1][0] == '/') {
1652                 strbuf_addf(&header, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1653                 if (xfrm_msg && xfrm_msg[0])
1654                         strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1655         }
1656         else {
1657                 if (one->mode != two->mode) {
1658                         strbuf_addf(&header, "%sold mode %06o%s\n", set, one->mode, reset);
1659                         strbuf_addf(&header, "%snew mode %06o%s\n", set, two->mode, reset);
1660                 }
1661                 if (xfrm_msg && xfrm_msg[0])
1662                         strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1663
1664                 /*
1665                  * we do not run diff between different kind
1666                  * of objects.
1667                  */
1668                 if ((one->mode ^ two->mode) & S_IFMT)
1669                         goto free_ab_and_return;
1670                 if (complete_rewrite &&
1671                     (textconv_one || !diff_filespec_is_binary(one)) &&
1672                     (textconv_two || !diff_filespec_is_binary(two))) {
1673                         fprintf(o->file, "%s", header.buf);
1674                         strbuf_reset(&header);
1675                         emit_rewrite_diff(name_a, name_b, one, two,
1676                                                 textconv_one, textconv_two, o);
1677                         o->found_changes = 1;
1678                         goto free_ab_and_return;
1679                 }
1680         }
1681
1682         if (!DIFF_OPT_TST(o, TEXT) &&
1683             ( (!textconv_one && diff_filespec_is_binary(one)) ||
1684               (!textconv_two && diff_filespec_is_binary(two)) )) {
1685                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1686                         die("unable to read files to diff");
1687                 /* Quite common confusing case */
1688                 if (mf1.size == mf2.size &&
1689                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1690                         goto free_ab_and_return;
1691                 fprintf(o->file, "%s", header.buf);
1692                 strbuf_reset(&header);
1693                 if (DIFF_OPT_TST(o, BINARY))
1694                         emit_binary_diff(o->file, &mf1, &mf2);
1695                 else
1696                         fprintf(o->file, "Binary files %s and %s differ\n",
1697                                 lbl[0], lbl[1]);
1698                 o->found_changes = 1;
1699         }
1700         else {
1701                 /* Crazy xdl interfaces.. */
1702                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1703                 xpparam_t xpp;
1704                 xdemitconf_t xecfg;
1705                 struct emit_callback ecbdata;
1706                 const struct userdiff_funcname *pe;
1707
1708                 if (!DIFF_XDL_TST(o, WHITESPACE_FLAGS)) {
1709                         fprintf(o->file, "%s", header.buf);
1710                         strbuf_reset(&header);
1711                 }
1712
1713                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
1714                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
1715
1716                 pe = diff_funcname_pattern(one);
1717                 if (!pe)
1718                         pe = diff_funcname_pattern(two);
1719
1720                 memset(&xpp, 0, sizeof(xpp));
1721                 memset(&xecfg, 0, sizeof(xecfg));
1722                 memset(&ecbdata, 0, sizeof(ecbdata));
1723                 ecbdata.label_path = lbl;
1724                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1725                 ecbdata.found_changesp = &o->found_changes;
1726                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1727                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
1728                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
1729                 ecbdata.file = o->file;
1730                 ecbdata.header = header.len ? &header : NULL;
1731                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1732                 xecfg.ctxlen = o->context;
1733                 xecfg.interhunkctxlen = o->interhunkcontext;
1734                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1735                 if (pe)
1736                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1737                 if (!diffopts)
1738                         ;
1739                 else if (!prefixcmp(diffopts, "--unified="))
1740                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1741                 else if (!prefixcmp(diffopts, "-u"))
1742                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1743                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1744                         ecbdata.diff_words =
1745                                 xcalloc(1, sizeof(struct diff_words_data));
1746                         ecbdata.diff_words->file = o->file;
1747                         if (!o->word_regex)
1748                                 o->word_regex = userdiff_word_regex(one);
1749                         if (!o->word_regex)
1750                                 o->word_regex = userdiff_word_regex(two);
1751                         if (!o->word_regex)
1752                                 o->word_regex = diff_word_regex_cfg;
1753                         if (o->word_regex) {
1754                                 ecbdata.diff_words->word_regex = (regex_t *)
1755                                         xmalloc(sizeof(regex_t));
1756                                 if (regcomp(ecbdata.diff_words->word_regex,
1757                                                 o->word_regex,
1758                                                 REG_EXTENDED | REG_NEWLINE))
1759                                         die ("Invalid regular expression: %s",
1760                                                         o->word_regex);
1761                         }
1762                 }
1763                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1764                               &xpp, &xecfg);
1765                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1766                         free_diff_words_data(&ecbdata);
1767                 if (textconv_one)
1768                         free(mf1.ptr);
1769                 if (textconv_two)
1770                         free(mf2.ptr);
1771                 xdiff_clear_find_func(&xecfg);
1772         }
1773
1774  free_ab_and_return:
1775         strbuf_release(&header);
1776         diff_free_filespec_data(one);
1777         diff_free_filespec_data(two);
1778         free(a_one);
1779         free(b_two);
1780         return;
1781 }
1782
1783 static void builtin_diffstat(const char *name_a, const char *name_b,
1784                              struct diff_filespec *one,
1785                              struct diff_filespec *two,
1786                              struct diffstat_t *diffstat,
1787                              struct diff_options *o,
1788                              int complete_rewrite)
1789 {
1790         mmfile_t mf1, mf2;
1791         struct diffstat_file *data;
1792
1793         data = diffstat_add(diffstat, name_a, name_b);
1794
1795         if (!one || !two) {
1796                 data->is_unmerged = 1;
1797                 return;
1798         }
1799         if (complete_rewrite) {
1800                 diff_populate_filespec(one, 0);
1801                 diff_populate_filespec(two, 0);
1802                 data->deleted = count_lines(one->data, one->size);
1803                 data->added = count_lines(two->data, two->size);
1804                 goto free_and_return;
1805         }
1806         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1807                 die("unable to read files to diff");
1808
1809         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1810                 data->is_binary = 1;
1811                 data->added = mf2.size;
1812                 data->deleted = mf1.size;
1813         } else {
1814                 /* Crazy xdl interfaces.. */
1815                 xpparam_t xpp;
1816                 xdemitconf_t xecfg;
1817
1818                 memset(&xpp, 0, sizeof(xpp));
1819                 memset(&xecfg, 0, sizeof(xecfg));
1820                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1821                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1822                               &xpp, &xecfg);
1823         }
1824
1825  free_and_return:
1826         diff_free_filespec_data(one);
1827         diff_free_filespec_data(two);
1828 }
1829
1830 static void builtin_checkdiff(const char *name_a, const char *name_b,
1831                               const char *attr_path,
1832                               struct diff_filespec *one,
1833                               struct diff_filespec *two,
1834                               struct diff_options *o)
1835 {
1836         mmfile_t mf1, mf2;
1837         struct checkdiff_t data;
1838
1839         if (!two)
1840                 return;
1841
1842         memset(&data, 0, sizeof(data));
1843         data.filename = name_b ? name_b : name_a;
1844         data.lineno = 0;
1845         data.o = o;
1846         data.ws_rule = whitespace_rule(attr_path);
1847         data.conflict_marker_size = ll_merge_marker_size(attr_path);
1848
1849         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1850                 die("unable to read files to diff");
1851
1852         /*
1853          * All the other codepaths check both sides, but not checking
1854          * the "old" side here is deliberate.  We are checking the newly
1855          * introduced changes, and as long as the "new" side is text, we
1856          * can and should check what it introduces.
1857          */
1858         if (diff_filespec_is_binary(two))
1859                 goto free_and_return;
1860         else {
1861                 /* Crazy xdl interfaces.. */
1862                 xpparam_t xpp;
1863                 xdemitconf_t xecfg;
1864
1865                 memset(&xpp, 0, sizeof(xpp));
1866                 memset(&xecfg, 0, sizeof(xecfg));
1867                 xecfg.ctxlen = 1; /* at least one context line */
1868                 xpp.flags = XDF_NEED_MINIMAL;
1869                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1870                               &xpp, &xecfg);
1871
1872                 if (data.ws_rule & WS_BLANK_AT_EOF) {
1873                         struct emit_callback ecbdata;
1874                         int blank_at_eof;
1875
1876                         ecbdata.ws_rule = data.ws_rule;
1877                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
1878                         blank_at_eof = ecbdata.blank_at_eof_in_preimage;
1879
1880                         if (blank_at_eof) {
1881                                 static char *err;
1882                                 if (!err)
1883                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
1884                                 fprintf(o->file, "%s:%d: %s.\n",
1885                                         data.filename, blank_at_eof, err);
1886                                 data.status = 1; /* report errors */
1887                         }
1888                 }
1889         }
1890  free_and_return:
1891         diff_free_filespec_data(one);
1892         diff_free_filespec_data(two);
1893         if (data.status)
1894                 DIFF_OPT_SET(o, CHECK_FAILED);
1895 }
1896
1897 struct diff_filespec *alloc_filespec(const char *path)
1898 {
1899         int namelen = strlen(path);
1900         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1901
1902         memset(spec, 0, sizeof(*spec));
1903         spec->path = (char *)(spec + 1);
1904         memcpy(spec->path, path, namelen+1);
1905         spec->count = 1;
1906         spec->is_binary = -1;
1907         return spec;
1908 }
1909
1910 void free_filespec(struct diff_filespec *spec)
1911 {
1912         if (!--spec->count) {
1913                 diff_free_filespec_data(spec);
1914                 free(spec);
1915         }
1916 }
1917
1918 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1919                    unsigned short mode)
1920 {
1921         if (mode) {
1922                 spec->mode = canon_mode(mode);
1923                 hashcpy(spec->sha1, sha1);
1924                 spec->sha1_valid = !is_null_sha1(sha1);
1925         }
1926 }
1927
1928 /*
1929  * Given a name and sha1 pair, if the index tells us the file in
1930  * the work tree has that object contents, return true, so that
1931  * prepare_temp_file() does not have to inflate and extract.
1932  */
1933 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1934 {
1935         struct cache_entry *ce;
1936         struct stat st;
1937         int pos, len;
1938
1939         /*
1940          * We do not read the cache ourselves here, because the
1941          * benchmark with my previous version that always reads cache
1942          * shows that it makes things worse for diff-tree comparing
1943          * two linux-2.6 kernel trees in an already checked out work
1944          * tree.  This is because most diff-tree comparisons deal with
1945          * only a small number of files, while reading the cache is
1946          * expensive for a large project, and its cost outweighs the
1947          * savings we get by not inflating the object to a temporary
1948          * file.  Practically, this code only helps when we are used
1949          * by diff-cache --cached, which does read the cache before
1950          * calling us.
1951          */
1952         if (!active_cache)
1953                 return 0;
1954
1955         /* We want to avoid the working directory if our caller
1956          * doesn't need the data in a normal file, this system
1957          * is rather slow with its stat/open/mmap/close syscalls,
1958          * and the object is contained in a pack file.  The pack
1959          * is probably already open and will be faster to obtain
1960          * the data through than the working directory.  Loose
1961          * objects however would tend to be slower as they need
1962          * to be individually opened and inflated.
1963          */
1964         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
1965                 return 0;
1966
1967         len = strlen(name);
1968         pos = cache_name_pos(name, len);
1969         if (pos < 0)
1970                 return 0;
1971         ce = active_cache[pos];
1972
1973         /*
1974          * This is not the sha1 we are looking for, or
1975          * unreusable because it is not a regular file.
1976          */
1977         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1978                 return 0;
1979
1980         /*
1981          * If ce is marked as "assume unchanged", there is no
1982          * guarantee that work tree matches what we are looking for.
1983          */
1984         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
1985                 return 0;
1986
1987         /*
1988          * If ce matches the file in the work tree, we can reuse it.
1989          */
1990         if (ce_uptodate(ce) ||
1991             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1992                 return 1;
1993
1994         return 0;
1995 }
1996
1997 static int populate_from_stdin(struct diff_filespec *s)
1998 {
1999         struct strbuf buf = STRBUF_INIT;
2000         size_t size = 0;
2001
2002         if (strbuf_read(&buf, 0, 0) < 0)
2003                 return error("error while reading from stdin %s",
2004                                      strerror(errno));
2005
2006         s->should_munmap = 0;
2007         s->data = strbuf_detach(&buf, &size);
2008         s->size = size;
2009         s->should_free = 1;
2010         return 0;
2011 }
2012
2013 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2014 {
2015         int len;
2016         char *data = xmalloc(100), *dirty = "";
2017
2018         /* Are we looking at the work tree? */
2019         if (s->dirty_submodule)
2020                 dirty = "-dirty";
2021
2022         len = snprintf(data, 100,
2023                        "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2024         s->data = data;
2025         s->size = len;
2026         s->should_free = 1;
2027         if (size_only) {
2028                 s->data = NULL;
2029                 free(data);
2030         }
2031         return 0;
2032 }
2033
2034 /*
2035  * While doing rename detection and pickaxe operation, we may need to
2036  * grab the data for the blob (or file) for our own in-core comparison.
2037  * diff_filespec has data and size fields for this purpose.
2038  */
2039 int diff_populate_filespec(struct diff_filespec *s, int size_only)
2040 {
2041         int err = 0;
2042         if (!DIFF_FILE_VALID(s))
2043                 die("internal error: asking to populate invalid file.");
2044         if (S_ISDIR(s->mode))
2045                 return -1;
2046
2047         if (s->data)
2048                 return 0;
2049
2050         if (size_only && 0 < s->size)
2051                 return 0;
2052
2053         if (S_ISGITLINK(s->mode))
2054                 return diff_populate_gitlink(s, size_only);
2055
2056         if (!s->sha1_valid ||
2057             reuse_worktree_file(s->path, s->sha1, 0)) {
2058                 struct strbuf buf = STRBUF_INIT;
2059                 struct stat st;
2060                 int fd;
2061
2062                 if (!strcmp(s->path, "-"))
2063                         return populate_from_stdin(s);
2064
2065                 if (lstat(s->path, &st) < 0) {
2066                         if (errno == ENOENT) {
2067                         err_empty:
2068                                 err = -1;
2069                         empty:
2070                                 s->data = (char *)"";
2071                                 s->size = 0;
2072                                 return err;
2073                         }
2074                 }
2075                 s->size = xsize_t(st.st_size);
2076                 if (!s->size)
2077                         goto empty;
2078                 if (S_ISLNK(st.st_mode)) {
2079                         struct strbuf sb = STRBUF_INIT;
2080
2081                         if (strbuf_readlink(&sb, s->path, s->size))
2082                                 goto err_empty;
2083                         s->size = sb.len;
2084                         s->data = strbuf_detach(&sb, NULL);
2085                         s->should_free = 1;
2086                         return 0;
2087                 }
2088                 if (size_only)
2089                         return 0;
2090                 fd = open(s->path, O_RDONLY);
2091                 if (fd < 0)
2092                         goto err_empty;
2093                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2094                 close(fd);
2095                 s->should_munmap = 1;
2096
2097                 /*
2098                  * Convert from working tree format to canonical git format
2099                  */
2100                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2101                         size_t size = 0;
2102                         munmap(s->data, s->size);
2103                         s->should_munmap = 0;
2104                         s->data = strbuf_detach(&buf, &size);
2105                         s->size = size;
2106                         s->should_free = 1;
2107                 }
2108         }
2109         else {
2110                 enum object_type type;
2111                 if (size_only)
2112                         type = sha1_object_info(s->sha1, &s->size);
2113                 else {
2114                         s->data = read_sha1_file(s->sha1, &type, &s->size);
2115                         s->should_free = 1;
2116                 }
2117         }
2118         return 0;
2119 }
2120
2121 void diff_free_filespec_blob(struct diff_filespec *s)
2122 {
2123         if (s->should_free)
2124                 free(s->data);
2125         else if (s->should_munmap)
2126                 munmap(s->data, s->size);
2127
2128         if (s->should_free || s->should_munmap) {
2129                 s->should_free = s->should_munmap = 0;
2130                 s->data = NULL;
2131         }
2132 }
2133
2134 void diff_free_filespec_data(struct diff_filespec *s)
2135 {
2136         diff_free_filespec_blob(s);
2137         free(s->cnt_data);
2138         s->cnt_data = NULL;
2139 }
2140
2141 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2142                            void *blob,
2143                            unsigned long size,
2144                            const unsigned char *sha1,
2145                            int mode)
2146 {
2147         int fd;
2148         struct strbuf buf = STRBUF_INIT;
2149         struct strbuf template = STRBUF_INIT;
2150         char *path_dup = xstrdup(path);
2151         const char *base = basename(path_dup);
2152
2153         /* Generate "XXXXXX_basename.ext" */
2154         strbuf_addstr(&template, "XXXXXX_");
2155         strbuf_addstr(&template, base);
2156
2157         fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2158                         strlen(base) + 1);
2159         if (fd < 0)
2160                 die_errno("unable to create temp-file");
2161         if (convert_to_working_tree(path,
2162                         (const char *)blob, (size_t)size, &buf)) {
2163                 blob = buf.buf;
2164                 size = buf.len;
2165         }
2166         if (write_in_full(fd, blob, size) != size)
2167                 die_errno("unable to write temp-file");
2168         close(fd);
2169         temp->name = temp->tmp_path;
2170         strcpy(temp->hex, sha1_to_hex(sha1));
2171         temp->hex[40] = 0;
2172         sprintf(temp->mode, "%06o", mode);
2173         strbuf_release(&buf);
2174         strbuf_release(&template);
2175         free(path_dup);
2176 }
2177
2178 static struct diff_tempfile *prepare_temp_file(const char *name,
2179                 struct diff_filespec *one)
2180 {
2181         struct diff_tempfile *temp = claim_diff_tempfile();
2182
2183         if (!DIFF_FILE_VALID(one)) {
2184         not_a_valid_file:
2185                 /* A '-' entry produces this for file-2, and
2186                  * a '+' entry produces this for file-1.
2187                  */
2188                 temp->name = "/dev/null";
2189                 strcpy(temp->hex, ".");
2190                 strcpy(temp->mode, ".");
2191                 return temp;
2192         }
2193
2194         if (!remove_tempfile_installed) {
2195                 atexit(remove_tempfile);
2196                 sigchain_push_common(remove_tempfile_on_signal);
2197                 remove_tempfile_installed = 1;
2198         }
2199
2200         if (!one->sha1_valid ||
2201             reuse_worktree_file(name, one->sha1, 1)) {
2202                 struct stat st;
2203                 if (lstat(name, &st) < 0) {
2204                         if (errno == ENOENT)
2205                                 goto not_a_valid_file;
2206                         die_errno("stat(%s)", name);
2207                 }
2208                 if (S_ISLNK(st.st_mode)) {
2209                         struct strbuf sb = STRBUF_INIT;
2210                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2211                                 die_errno("readlink(%s)", name);
2212                         prep_temp_blob(name, temp, sb.buf, sb.len,
2213                                        (one->sha1_valid ?
2214                                         one->sha1 : null_sha1),
2215                                        (one->sha1_valid ?
2216                                         one->mode : S_IFLNK));
2217                         strbuf_release(&sb);
2218                 }
2219                 else {
2220                         /* we can borrow from the file in the work tree */
2221                         temp->name = name;
2222                         if (!one->sha1_valid)
2223                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2224                         else
2225                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2226                         /* Even though we may sometimes borrow the
2227                          * contents from the work tree, we always want
2228                          * one->mode.  mode is trustworthy even when
2229                          * !(one->sha1_valid), as long as
2230                          * DIFF_FILE_VALID(one).
2231                          */
2232                         sprintf(temp->mode, "%06o", one->mode);
2233                 }
2234                 return temp;
2235         }
2236         else {
2237                 if (diff_populate_filespec(one, 0))
2238                         die("cannot read data blob for %s", one->path);
2239                 prep_temp_blob(name, temp, one->data, one->size,
2240                                one->sha1, one->mode);
2241         }
2242         return temp;
2243 }
2244
2245 /* An external diff command takes:
2246  *
2247  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2248  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2249  *
2250  */
2251 static void run_external_diff(const char *pgm,
2252                               const char *name,
2253                               const char *other,
2254                               struct diff_filespec *one,
2255                               struct diff_filespec *two,
2256                               const char *xfrm_msg,
2257                               int complete_rewrite)
2258 {
2259         const char *spawn_arg[10];
2260         int retval;
2261         const char **arg = &spawn_arg[0];
2262
2263         if (one && two) {
2264                 struct diff_tempfile *temp_one, *temp_two;
2265                 const char *othername = (other ? other : name);
2266                 temp_one = prepare_temp_file(name, one);
2267                 temp_two = prepare_temp_file(othername, two);
2268                 *arg++ = pgm;
2269                 *arg++ = name;
2270                 *arg++ = temp_one->name;
2271                 *arg++ = temp_one->hex;
2272                 *arg++ = temp_one->mode;
2273                 *arg++ = temp_two->name;
2274                 *arg++ = temp_two->hex;
2275                 *arg++ = temp_two->mode;
2276                 if (other) {
2277                         *arg++ = other;
2278                         *arg++ = xfrm_msg;
2279                 }
2280         } else {
2281                 *arg++ = pgm;
2282                 *arg++ = name;
2283         }
2284         *arg = NULL;
2285         fflush(NULL);
2286         retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2287         remove_tempfile();
2288         if (retval) {
2289                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2290                 exit(1);
2291         }
2292 }
2293
2294 static int similarity_index(struct diff_filepair *p)
2295 {
2296         return p->score * 100 / MAX_SCORE;
2297 }
2298
2299 static void fill_metainfo(struct strbuf *msg,
2300                           const char *name,
2301                           const char *other,
2302                           struct diff_filespec *one,
2303                           struct diff_filespec *two,
2304                           struct diff_options *o,
2305                           struct diff_filepair *p)
2306 {
2307         strbuf_init(msg, PATH_MAX * 2 + 300);
2308         switch (p->status) {
2309         case DIFF_STATUS_COPIED:
2310                 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2311                 strbuf_addstr(msg, "\ncopy from ");
2312                 quote_c_style(name, msg, NULL, 0);
2313                 strbuf_addstr(msg, "\ncopy to ");
2314                 quote_c_style(other, msg, NULL, 0);
2315                 strbuf_addch(msg, '\n');
2316                 break;
2317         case DIFF_STATUS_RENAMED:
2318                 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2319                 strbuf_addstr(msg, "\nrename from ");
2320                 quote_c_style(name, msg, NULL, 0);
2321                 strbuf_addstr(msg, "\nrename to ");
2322                 quote_c_style(other, msg, NULL, 0);
2323                 strbuf_addch(msg, '\n');
2324                 break;
2325         case DIFF_STATUS_MODIFIED:
2326                 if (p->score) {
2327                         strbuf_addf(msg, "dissimilarity index %d%%\n",
2328                                     similarity_index(p));
2329                         break;
2330                 }
2331                 /* fallthru */
2332         default:
2333                 /* nothing */
2334                 ;
2335         }
2336         if (one && two && hashcmp(one->sha1, two->sha1)) {
2337                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2338
2339                 if (DIFF_OPT_TST(o, BINARY)) {
2340                         mmfile_t mf;
2341                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2342                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2343                                 abbrev = 40;
2344                 }
2345                 strbuf_addf(msg, "index %.*s..%.*s",
2346                             abbrev, sha1_to_hex(one->sha1),
2347                             abbrev, sha1_to_hex(two->sha1));
2348                 if (one->mode == two->mode)
2349                         strbuf_addf(msg, " %06o", one->mode);
2350                 strbuf_addch(msg, '\n');
2351         }
2352         if (msg->len)
2353                 strbuf_setlen(msg, msg->len - 1);
2354 }
2355
2356 static void run_diff_cmd(const char *pgm,
2357                          const char *name,
2358                          const char *other,
2359                          const char *attr_path,
2360                          struct diff_filespec *one,
2361                          struct diff_filespec *two,
2362                          struct strbuf *msg,
2363                          struct diff_options *o,
2364                          struct diff_filepair *p)
2365 {
2366         const char *xfrm_msg = NULL;
2367         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2368
2369         if (msg) {
2370                 fill_metainfo(msg, name, other, one, two, o, p);
2371                 xfrm_msg = msg->len ? msg->buf : NULL;
2372         }
2373
2374         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2375                 pgm = NULL;
2376         else {
2377                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2378                 if (drv && drv->external)
2379                         pgm = drv->external;
2380         }
2381
2382         if (pgm) {
2383                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2384                                   complete_rewrite);
2385                 return;
2386         }
2387         if (one && two)
2388                 builtin_diff(name, other ? other : name,
2389                              one, two, xfrm_msg, o, complete_rewrite);
2390         else
2391                 fprintf(o->file, "* Unmerged path %s\n", name);
2392 }
2393
2394 static void diff_fill_sha1_info(struct diff_filespec *one)
2395 {
2396         if (DIFF_FILE_VALID(one)) {
2397                 if (!one->sha1_valid) {
2398                         struct stat st;
2399                         if (!strcmp(one->path, "-")) {
2400                                 hashcpy(one->sha1, null_sha1);
2401                                 return;
2402                         }
2403                         if (lstat(one->path, &st) < 0)
2404                                 die_errno("stat '%s'", one->path);
2405                         if (index_path(one->sha1, one->path, &st, 0))
2406                                 die("cannot hash %s", one->path);
2407                 }
2408         }
2409         else
2410                 hashclr(one->sha1);
2411 }
2412
2413 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2414 {
2415         /* Strip the prefix but do not molest /dev/null and absolute paths */
2416         if (*namep && **namep != '/')
2417                 *namep += prefix_length;
2418         if (*otherp && **otherp != '/')
2419                 *otherp += prefix_length;
2420 }
2421
2422 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2423 {
2424         const char *pgm = external_diff();
2425         struct strbuf msg;
2426         struct diff_filespec *one = p->one;
2427         struct diff_filespec *two = p->two;
2428         const char *name;
2429         const char *other;
2430         const char *attr_path;
2431
2432         name  = p->one->path;
2433         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2434         attr_path = name;
2435         if (o->prefix_length)
2436                 strip_prefix(o->prefix_length, &name, &other);
2437
2438         if (DIFF_PAIR_UNMERGED(p)) {
2439                 run_diff_cmd(pgm, name, NULL, attr_path,
2440                              NULL, NULL, NULL, o, p);
2441                 return;
2442         }
2443
2444         diff_fill_sha1_info(one);
2445         diff_fill_sha1_info(two);
2446
2447         if (!pgm &&
2448             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2449             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2450                 /*
2451                  * a filepair that changes between file and symlink
2452                  * needs to be split into deletion and creation.
2453                  */
2454                 struct diff_filespec *null = alloc_filespec(two->path);
2455                 run_diff_cmd(NULL, name, other, attr_path,
2456                              one, null, &msg, o, p);
2457                 free(null);
2458                 strbuf_release(&msg);
2459
2460                 null = alloc_filespec(one->path);
2461                 run_diff_cmd(NULL, name, other, attr_path,
2462                              null, two, &msg, o, p);
2463                 free(null);
2464         }
2465         else
2466                 run_diff_cmd(pgm, name, other, attr_path,
2467                              one, two, &msg, o, p);
2468
2469         strbuf_release(&msg);
2470 }
2471
2472 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2473                          struct diffstat_t *diffstat)
2474 {
2475         const char *name;
2476         const char *other;
2477         int complete_rewrite = 0;
2478
2479         if (DIFF_PAIR_UNMERGED(p)) {
2480                 /* unmerged */
2481                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2482                 return;
2483         }
2484
2485         name = p->one->path;
2486         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2487
2488         if (o->prefix_length)
2489                 strip_prefix(o->prefix_length, &name, &other);
2490
2491         diff_fill_sha1_info(p->one);
2492         diff_fill_sha1_info(p->two);
2493
2494         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2495                 complete_rewrite = 1;
2496         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2497 }
2498
2499 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2500 {
2501         const char *name;
2502         const char *other;
2503         const char *attr_path;
2504
2505         if (DIFF_PAIR_UNMERGED(p)) {
2506                 /* unmerged */
2507                 return;
2508         }
2509
2510         name = p->one->path;
2511         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2512         attr_path = other ? other : name;
2513
2514         if (o->prefix_length)
2515                 strip_prefix(o->prefix_length, &name, &other);
2516
2517         diff_fill_sha1_info(p->one);
2518         diff_fill_sha1_info(p->two);
2519
2520         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2521 }
2522
2523 void diff_setup(struct diff_options *options)
2524 {
2525         memset(options, 0, sizeof(*options));
2526
2527         options->file = stdout;
2528
2529         options->line_termination = '\n';
2530         options->break_opt = -1;
2531         options->rename_limit = -1;
2532         options->dirstat_percent = 3;
2533         options->context = 3;
2534
2535         options->change = diff_change;
2536         options->add_remove = diff_addremove;
2537         if (diff_use_color_default > 0)
2538                 DIFF_OPT_SET(options, COLOR_DIFF);
2539         options->detect_rename = diff_detect_rename_default;
2540
2541         if (!diff_mnemonic_prefix) {
2542                 options->a_prefix = "a/";
2543                 options->b_prefix = "b/";
2544         }
2545 }
2546
2547 int diff_setup_done(struct diff_options *options)
2548 {
2549         int count = 0;
2550
2551         if (options->output_format & DIFF_FORMAT_NAME)
2552                 count++;
2553         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2554                 count++;
2555         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2556                 count++;
2557         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2558                 count++;
2559         if (count > 1)
2560                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2561
2562         /*
2563          * Most of the time we can say "there are changes"
2564          * only by checking if there are changed paths, but
2565          * --ignore-whitespace* options force us to look
2566          * inside contents.
2567          */
2568
2569         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
2570             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
2571             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
2572                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
2573         else
2574                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
2575
2576         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2577                 options->detect_rename = DIFF_DETECT_COPY;
2578
2579         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2580                 options->prefix = NULL;
2581         if (options->prefix)
2582                 options->prefix_length = strlen(options->prefix);
2583         else
2584                 options->prefix_length = 0;
2585
2586         if (options->output_format & (DIFF_FORMAT_NAME |
2587                                       DIFF_FORMAT_NAME_STATUS |
2588                                       DIFF_FORMAT_CHECKDIFF |
2589                                       DIFF_FORMAT_NO_OUTPUT))
2590                 options->output_format &= ~(DIFF_FORMAT_RAW |
2591                                             DIFF_FORMAT_NUMSTAT |
2592                                             DIFF_FORMAT_DIFFSTAT |
2593                                             DIFF_FORMAT_SHORTSTAT |
2594                                             DIFF_FORMAT_DIRSTAT |
2595                                             DIFF_FORMAT_SUMMARY |
2596                                             DIFF_FORMAT_PATCH);
2597
2598         /*
2599          * These cases always need recursive; we do not drop caller-supplied
2600          * recursive bits for other formats here.
2601          */
2602         if (options->output_format & (DIFF_FORMAT_PATCH |
2603                                       DIFF_FORMAT_NUMSTAT |
2604                                       DIFF_FORMAT_DIFFSTAT |
2605                                       DIFF_FORMAT_SHORTSTAT |
2606                                       DIFF_FORMAT_DIRSTAT |
2607                                       DIFF_FORMAT_SUMMARY |
2608                                       DIFF_FORMAT_CHECKDIFF))
2609                 DIFF_OPT_SET(options, RECURSIVE);
2610         /*
2611          * Also pickaxe would not work very well if you do not say recursive
2612          */
2613         if (options->pickaxe)
2614                 DIFF_OPT_SET(options, RECURSIVE);
2615         /*
2616          * When patches are generated, submodules diffed against the work tree
2617          * must be checked for dirtiness too so it can be shown in the output
2618          */
2619         if (options->output_format & DIFF_FORMAT_PATCH)
2620                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
2621
2622         if (options->detect_rename && options->rename_limit < 0)
2623                 options->rename_limit = diff_rename_limit_default;
2624         if (options->setup & DIFF_SETUP_USE_CACHE) {
2625                 if (!active_cache)
2626                         /* read-cache does not die even when it fails
2627                          * so it is safe for us to do this here.  Also
2628                          * it does not smudge active_cache or active_nr
2629                          * when it fails, so we do not have to worry about
2630                          * cleaning it up ourselves either.
2631                          */
2632                         read_cache();
2633         }
2634         if (options->abbrev <= 0 || 40 < options->abbrev)
2635                 options->abbrev = 40; /* full */
2636
2637         /*
2638          * It does not make sense to show the first hit we happened
2639          * to have found.  It does not make sense not to return with
2640          * exit code in such a case either.
2641          */
2642         if (DIFF_OPT_TST(options, QUICK)) {
2643                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2644                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2645         }
2646
2647         return 0;
2648 }
2649
2650 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2651 {
2652         char c, *eq;
2653         int len;
2654
2655         if (*arg != '-')
2656                 return 0;
2657         c = *++arg;
2658         if (!c)
2659                 return 0;
2660         if (c == arg_short) {
2661                 c = *++arg;
2662                 if (!c)
2663                         return 1;
2664                 if (val && isdigit(c)) {
2665                         char *end;
2666                         int n = strtoul(arg, &end, 10);
2667                         if (*end)
2668                                 return 0;
2669                         *val = n;
2670                         return 1;
2671                 }
2672                 return 0;
2673         }
2674         if (c != '-')
2675                 return 0;
2676         arg++;
2677         eq = strchr(arg, '=');
2678         if (eq)
2679                 len = eq - arg;
2680         else
2681                 len = strlen(arg);
2682         if (!len || strncmp(arg, arg_long, len))
2683                 return 0;
2684         if (eq) {
2685                 int n;
2686                 char *end;
2687                 if (!isdigit(*++eq))
2688                         return 0;
2689                 n = strtoul(eq, &end, 10);
2690                 if (*end)
2691                         return 0;
2692                 *val = n;
2693         }
2694         return 1;
2695 }
2696
2697 static int diff_scoreopt_parse(const char *opt);
2698
2699 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2700 {
2701         const char *arg = av[0];
2702
2703         /* Output format options */
2704         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch"))
2705                 options->output_format |= DIFF_FORMAT_PATCH;
2706         else if (opt_arg(arg, 'U', "unified", &options->context))
2707                 options->output_format |= DIFF_FORMAT_PATCH;
2708         else if (!strcmp(arg, "--raw"))
2709                 options->output_format |= DIFF_FORMAT_RAW;
2710         else if (!strcmp(arg, "--patch-with-raw"))
2711                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2712         else if (!strcmp(arg, "--numstat"))
2713                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2714         else if (!strcmp(arg, "--shortstat"))
2715                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2716         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2717                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2718         else if (!strcmp(arg, "--cumulative")) {
2719                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2720                 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2721         } else if (opt_arg(arg, 0, "dirstat-by-file",
2722                            &options->dirstat_percent)) {
2723                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2724                 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
2725         }
2726         else if (!strcmp(arg, "--check"))
2727                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2728         else if (!strcmp(arg, "--summary"))
2729                 options->output_format |= DIFF_FORMAT_SUMMARY;
2730         else if (!strcmp(arg, "--patch-with-stat"))
2731                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2732         else if (!strcmp(arg, "--name-only"))
2733                 options->output_format |= DIFF_FORMAT_NAME;
2734         else if (!strcmp(arg, "--name-status"))
2735                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2736         else if (!strcmp(arg, "-s"))
2737                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2738         else if (!prefixcmp(arg, "--stat")) {
2739                 char *end;
2740                 int width = options->stat_width;
2741                 int name_width = options->stat_name_width;
2742                 arg += 6;
2743                 end = (char *)arg;
2744
2745                 switch (*arg) {
2746                 case '-':
2747                         if (!prefixcmp(arg, "-width="))
2748                                 width = strtoul(arg + 7, &end, 10);
2749                         else if (!prefixcmp(arg, "-name-width="))
2750                                 name_width = strtoul(arg + 12, &end, 10);
2751                         break;
2752                 case '=':
2753                         width = strtoul(arg+1, &end, 10);
2754                         if (*end == ',')
2755                                 name_width = strtoul(end+1, &end, 10);
2756                 }
2757
2758                 /* Important! This checks all the error cases! */
2759                 if (*end)
2760                         return 0;
2761                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2762                 options->stat_name_width = name_width;
2763                 options->stat_width = width;
2764         }
2765
2766         /* renames options */
2767         else if (!prefixcmp(arg, "-B")) {
2768                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2769                         return -1;
2770         }
2771         else if (!prefixcmp(arg, "-M")) {
2772                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2773                         return -1;
2774                 options->detect_rename = DIFF_DETECT_RENAME;
2775         }
2776         else if (!prefixcmp(arg, "-C")) {
2777                 if (options->detect_rename == DIFF_DETECT_COPY)
2778                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2779                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2780                         return -1;
2781                 options->detect_rename = DIFF_DETECT_COPY;
2782         }
2783         else if (!strcmp(arg, "--no-renames"))
2784                 options->detect_rename = 0;
2785         else if (!strcmp(arg, "--relative"))
2786                 DIFF_OPT_SET(options, RELATIVE_NAME);
2787         else if (!prefixcmp(arg, "--relative=")) {
2788                 DIFF_OPT_SET(options, RELATIVE_NAME);
2789                 options->prefix = arg + 11;
2790         }
2791
2792         /* xdiff options */
2793         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2794                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
2795         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2796                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
2797         else if (!strcmp(arg, "--ignore-space-at-eol"))
2798                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
2799         else if (!strcmp(arg, "--patience"))
2800                 DIFF_XDL_SET(options, PATIENCE_DIFF);
2801
2802         /* flags options */
2803         else if (!strcmp(arg, "--binary")) {
2804                 options->output_format |= DIFF_FORMAT_PATCH;
2805                 DIFF_OPT_SET(options, BINARY);
2806         }
2807         else if (!strcmp(arg, "--full-index"))
2808                 DIFF_OPT_SET(options, FULL_INDEX);
2809         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2810                 DIFF_OPT_SET(options, TEXT);
2811         else if (!strcmp(arg, "-R"))
2812                 DIFF_OPT_SET(options, REVERSE_DIFF);
2813         else if (!strcmp(arg, "--find-copies-harder"))
2814                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2815         else if (!strcmp(arg, "--follow"))
2816                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2817         else if (!strcmp(arg, "--color"))
2818                 DIFF_OPT_SET(options, COLOR_DIFF);
2819         else if (!prefixcmp(arg, "--color=")) {
2820                 int value = git_config_colorbool(NULL, arg+8, -1);
2821                 if (value == 0)
2822                         DIFF_OPT_CLR(options, COLOR_DIFF);
2823                 else if (value > 0)
2824                         DIFF_OPT_SET(options, COLOR_DIFF);
2825                 else
2826                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
2827         }
2828         else if (!strcmp(arg, "--no-color"))
2829                 DIFF_OPT_CLR(options, COLOR_DIFF);
2830         else if (!strcmp(arg, "--color-words")) {
2831                 DIFF_OPT_SET(options, COLOR_DIFF);
2832                 DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2833         }
2834         else if (!prefixcmp(arg, "--color-words=")) {
2835                 DIFF_OPT_SET(options, COLOR_DIFF);
2836                 DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2837                 options->word_regex = arg + 14;
2838         }
2839         else if (!strcmp(arg, "--exit-code"))
2840                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2841         else if (!strcmp(arg, "--quiet"))
2842                 DIFF_OPT_SET(options, QUICK);
2843         else if (!strcmp(arg, "--ext-diff"))
2844                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2845         else if (!strcmp(arg, "--no-ext-diff"))
2846                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2847         else if (!strcmp(arg, "--textconv"))
2848                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
2849         else if (!strcmp(arg, "--no-textconv"))
2850                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
2851         else if (!strcmp(arg, "--ignore-submodules"))
2852                 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2853         else if (!strcmp(arg, "--submodule"))
2854                 DIFF_OPT_SET(options, SUBMODULE_LOG);
2855         else if (!prefixcmp(arg, "--submodule=")) {
2856                 if (!strcmp(arg + 12, "log"))
2857                         DIFF_OPT_SET(options, SUBMODULE_LOG);
2858         }
2859
2860         /* misc options */
2861         else if (!strcmp(arg, "-z"))
2862                 options->line_termination = 0;
2863         else if (!prefixcmp(arg, "-l"))
2864                 options->rename_limit = strtoul(arg+2, NULL, 10);
2865         else if (!prefixcmp(arg, "-S"))
2866                 options->pickaxe = arg + 2;
2867         else if (!strcmp(arg, "--pickaxe-all"))
2868                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2869         else if (!strcmp(arg, "--pickaxe-regex"))
2870                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2871         else if (!prefixcmp(arg, "-O"))
2872                 options->orderfile = arg + 2;
2873         else if (!prefixcmp(arg, "--diff-filter="))
2874                 options->filter = arg + 14;
2875         else if (!strcmp(arg, "--abbrev"))
2876                 options->abbrev = DEFAULT_ABBREV;
2877         else if (!prefixcmp(arg, "--abbrev=")) {
2878                 options->abbrev = strtoul(arg + 9, NULL, 10);
2879                 if (options->abbrev < MINIMUM_ABBREV)
2880                         options->abbrev = MINIMUM_ABBREV;
2881                 else if (40 < options->abbrev)
2882                         options->abbrev = 40;
2883         }
2884         else if (!prefixcmp(arg, "--src-prefix="))
2885                 options->a_prefix = arg + 13;
2886         else if (!prefixcmp(arg, "--dst-prefix="))
2887                 options->b_prefix = arg + 13;
2888         else if (!strcmp(arg, "--no-prefix"))
2889                 options->a_prefix = options->b_prefix = "";
2890         else if (opt_arg(arg, '\0', "inter-hunk-context",
2891                          &options->interhunkcontext))
2892                 ;
2893         else if (!prefixcmp(arg, "--output=")) {
2894                 options->file = fopen(arg + strlen("--output="), "w");
2895                 if (!options->file)
2896                         die_errno("Could not open '%s'", arg + strlen("--output="));
2897                 options->close_file = 1;
2898         } else
2899                 return 0;
2900         return 1;
2901 }
2902
2903 static int parse_num(const char **cp_p)
2904 {
2905         unsigned long num, scale;
2906         int ch, dot;
2907         const char *cp = *cp_p;
2908
2909         num = 0;
2910         scale = 1;
2911         dot = 0;
2912         for (;;) {
2913                 ch = *cp;
2914                 if ( !dot && ch == '.' ) {
2915                         scale = 1;
2916                         dot = 1;
2917                 } else if ( ch == '%' ) {
2918                         scale = dot ? scale*100 : 100;
2919                         cp++;   /* % is always at the end */
2920                         break;
2921                 } else if ( ch >= '0' && ch <= '9' ) {
2922                         if ( scale < 100000 ) {
2923                                 scale *= 10;
2924                                 num = (num*10) + (ch-'0');
2925                         }
2926                 } else {
2927                         break;
2928                 }
2929                 cp++;
2930         }
2931         *cp_p = cp;
2932
2933         /* user says num divided by scale and we say internally that
2934          * is MAX_SCORE * num / scale.
2935          */
2936         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2937 }
2938
2939 static int diff_scoreopt_parse(const char *opt)
2940 {
2941         int opt1, opt2, cmd;
2942
2943         if (*opt++ != '-')
2944                 return -1;
2945         cmd = *opt++;
2946         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2947                 return -1; /* that is not a -M, -C nor -B option */
2948
2949         opt1 = parse_num(&opt);
2950         if (cmd != 'B')
2951                 opt2 = 0;
2952         else {
2953                 if (*opt == 0)
2954                         opt2 = 0;
2955                 else if (*opt != '/')
2956                         return -1; /* we expect -B80/99 or -B80 */
2957                 else {
2958                         opt++;
2959                         opt2 = parse_num(&opt);
2960                 }
2961         }
2962         if (*opt != 0)
2963                 return -1;
2964         return opt1 | (opt2 << 16);
2965 }
2966
2967 struct diff_queue_struct diff_queued_diff;
2968
2969 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2970 {
2971         if (queue->alloc <= queue->nr) {
2972                 queue->alloc = alloc_nr(queue->alloc);
2973                 queue->queue = xrealloc(queue->queue,
2974                                         sizeof(dp) * queue->alloc);
2975         }
2976         queue->queue[queue->nr++] = dp;
2977 }
2978
2979 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2980                                  struct diff_filespec *one,
2981                                  struct diff_filespec *two)
2982 {
2983         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2984         dp->one = one;
2985         dp->two = two;
2986         if (queue)
2987                 diff_q(queue, dp);
2988         return dp;
2989 }
2990
2991 void diff_free_filepair(struct diff_filepair *p)
2992 {
2993         free_filespec(p->one);
2994         free_filespec(p->two);
2995         free(p);
2996 }
2997
2998 /* This is different from find_unique_abbrev() in that
2999  * it stuffs the result with dots for alignment.
3000  */
3001 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3002 {
3003         int abblen;
3004         const char *abbrev;
3005         if (len == 40)
3006                 return sha1_to_hex(sha1);
3007
3008         abbrev = find_unique_abbrev(sha1, len);
3009         abblen = strlen(abbrev);
3010         if (abblen < 37) {
3011                 static char hex[41];
3012                 if (len < abblen && abblen <= len + 2)
3013                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3014                 else
3015                         sprintf(hex, "%s...", abbrev);
3016                 return hex;
3017         }
3018         return sha1_to_hex(sha1);
3019 }
3020
3021 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3022 {
3023         int line_termination = opt->line_termination;
3024         int inter_name_termination = line_termination ? '\t' : '\0';
3025
3026         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3027                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3028                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
3029                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3030         }
3031         if (p->score) {
3032                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3033                         inter_name_termination);
3034         } else {
3035                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
3036         }
3037
3038         if (p->status == DIFF_STATUS_COPIED ||
3039             p->status == DIFF_STATUS_RENAMED) {
3040                 const char *name_a, *name_b;
3041                 name_a = p->one->path;
3042                 name_b = p->two->path;
3043                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3044                 write_name_quoted(name_a, opt->file, inter_name_termination);
3045                 write_name_quoted(name_b, opt->file, line_termination);
3046         } else {
3047                 const char *name_a, *name_b;
3048                 name_a = p->one->mode ? p->one->path : p->two->path;
3049                 name_b = NULL;
3050                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3051                 write_name_quoted(name_a, opt->file, line_termination);
3052         }
3053 }
3054
3055 int diff_unmodified_pair(struct diff_filepair *p)
3056 {
3057         /* This function is written stricter than necessary to support
3058          * the currently implemented transformers, but the idea is to
3059          * let transformers to produce diff_filepairs any way they want,
3060          * and filter and clean them up here before producing the output.
3061          */
3062         struct diff_filespec *one = p->one, *two = p->two;
3063
3064         if (DIFF_PAIR_UNMERGED(p))
3065                 return 0; /* unmerged is interesting */
3066
3067         /* deletion, addition, mode or type change
3068          * and rename are all interesting.
3069          */
3070         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
3071             DIFF_PAIR_MODE_CHANGED(p) ||
3072             strcmp(one->path, two->path))
3073                 return 0;
3074
3075         /* both are valid and point at the same path.  that is, we are
3076          * dealing with a change.
3077          */
3078         if (one->sha1_valid && two->sha1_valid &&
3079             !hashcmp(one->sha1, two->sha1) &&
3080             !one->dirty_submodule && !two->dirty_submodule)
3081                 return 1; /* no change */
3082         if (!one->sha1_valid && !two->sha1_valid)
3083                 return 1; /* both look at the same file on the filesystem. */
3084         return 0;
3085 }
3086
3087 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
3088 {
3089         if (diff_unmodified_pair(p))
3090                 return;
3091
3092         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3093             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3094                 return; /* no tree diffs in patch format */
3095
3096         run_diff(p, o);
3097 }
3098
3099 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
3100                             struct diffstat_t *diffstat)
3101 {
3102         if (diff_unmodified_pair(p))
3103                 return;
3104
3105         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3106             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3107                 return; /* no tree diffs in patch format */
3108
3109         run_diffstat(p, o, diffstat);
3110 }
3111
3112 static void diff_flush_checkdiff(struct diff_filepair *p,
3113                 struct diff_options *o)
3114 {
3115         if (diff_unmodified_pair(p))
3116                 return;
3117
3118         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3119             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3120                 return; /* no tree diffs in patch format */
3121
3122         run_checkdiff(p, o);
3123 }
3124
3125 int diff_queue_is_empty(void)
3126 {
3127         struct diff_queue_struct *q = &diff_queued_diff;
3128         int i;
3129         for (i = 0; i < q->nr; i++)
3130                 if (!diff_unmodified_pair(q->queue[i]))
3131                         return 0;
3132         return 1;
3133 }
3134
3135 #if DIFF_DEBUG
3136 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
3137 {
3138         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
3139                 x, one ? one : "",
3140                 s->path,
3141                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
3142                 s->mode,
3143                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
3144         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
3145                 x, one ? one : "",
3146                 s->size, s->xfrm_flags);
3147 }
3148
3149 void diff_debug_filepair(const struct diff_filepair *p, int i)
3150 {
3151         diff_debug_filespec(p->one, i, "one");
3152         diff_debug_filespec(p->two, i, "two");
3153         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
3154                 p->score, p->status ? p->status : '?',
3155                 p->one->rename_used, p->broken_pair);
3156 }
3157
3158 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
3159 {
3160         int i;
3161         if (msg)
3162                 fprintf(stderr, "%s\n", msg);
3163         fprintf(stderr, "q->nr = %d\n", q->nr);
3164         for (i = 0; i < q->nr; i++) {
3165                 struct diff_filepair *p = q->queue[i];
3166                 diff_debug_filepair(p, i);
3167         }
3168 }
3169 #endif
3170
3171 static void diff_resolve_rename_copy(void)
3172 {
3173         int i;
3174         struct diff_filepair *p;
3175         struct diff_queue_struct *q = &diff_queued_diff;
3176
3177         diff_debug_queue("resolve-rename-copy", q);
3178
3179         for (i = 0; i < q->nr; i++) {
3180                 p = q->queue[i];
3181                 p->status = 0; /* undecided */
3182                 if (DIFF_PAIR_UNMERGED(p))
3183                         p->status = DIFF_STATUS_UNMERGED;
3184                 else if (!DIFF_FILE_VALID(p->one))
3185                         p->status = DIFF_STATUS_ADDED;
3186                 else if (!DIFF_FILE_VALID(p->two))
3187                         p->status = DIFF_STATUS_DELETED;
3188                 else if (DIFF_PAIR_TYPE_CHANGED(p))
3189                         p->status = DIFF_STATUS_TYPE_CHANGED;
3190
3191                 /* from this point on, we are dealing with a pair
3192                  * whose both sides are valid and of the same type, i.e.
3193                  * either in-place edit or rename/copy edit.
3194                  */
3195                 else if (DIFF_PAIR_RENAME(p)) {
3196                         /*
3197                          * A rename might have re-connected a broken
3198                          * pair up, causing the pathnames to be the
3199                          * same again. If so, that's not a rename at
3200                          * all, just a modification..
3201                          *
3202                          * Otherwise, see if this source was used for
3203                          * multiple renames, in which case we decrement
3204                          * the count, and call it a copy.
3205                          */
3206                         if (!strcmp(p->one->path, p->two->path))
3207                                 p->status = DIFF_STATUS_MODIFIED;
3208                         else if (--p->one->rename_used > 0)
3209                                 p->status = DIFF_STATUS_COPIED;
3210                         else
3211                                 p->status = DIFF_STATUS_RENAMED;
3212                 }
3213                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
3214                          p->one->mode != p->two->mode ||
3215                          p->one->dirty_submodule ||
3216                          p->two->dirty_submodule ||
3217                          is_null_sha1(p->one->sha1))
3218                         p->status = DIFF_STATUS_MODIFIED;
3219                 else {
3220                         /* This is a "no-change" entry and should not
3221                          * happen anymore, but prepare for broken callers.
3222                          */
3223                         error("feeding unmodified %s to diffcore",
3224                               p->one->path);
3225                         p->status = DIFF_STATUS_UNKNOWN;
3226                 }
3227         }
3228         diff_debug_queue("resolve-rename-copy done", q);
3229 }
3230
3231 static int check_pair_status(struct diff_filepair *p)
3232 {
3233         switch (p->status) {
3234         case DIFF_STATUS_UNKNOWN:
3235                 return 0;
3236         case 0:
3237                 die("internal error in diff-resolve-rename-copy");
3238         default:
3239                 return 1;
3240         }
3241 }
3242
3243 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3244 {
3245         int fmt = opt->output_format;
3246
3247         if (fmt & DIFF_FORMAT_CHECKDIFF)
3248                 diff_flush_checkdiff(p, opt);
3249         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3250                 diff_flush_raw(p, opt);
3251         else if (fmt & DIFF_FORMAT_NAME) {
3252                 const char *name_a, *name_b;
3253                 name_a = p->two->path;
3254                 name_b = NULL;
3255                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3256                 write_name_quoted(name_a, opt->file, opt->line_termination);
3257         }
3258 }
3259
3260 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3261 {
3262         if (fs->mode)
3263                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3264         else
3265                 fprintf(file, " %s ", newdelete);
3266         write_name_quoted(fs->path, file, '\n');
3267 }
3268
3269
3270 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
3271 {
3272         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3273                 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
3274                         show_name ? ' ' : '\n');
3275                 if (show_name) {
3276                         write_name_quoted(p->two->path, file, '\n');
3277                 }
3278         }
3279 }
3280
3281 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3282 {
3283         char *names = pprint_rename(p->one->path, p->two->path);
3284
3285         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3286         free(names);
3287         show_mode_change(file, p, 0);
3288 }
3289
3290 static void diff_summary(FILE *file, struct diff_filepair *p)
3291 {
3292         switch(p->status) {
3293         case DIFF_STATUS_DELETED:
3294                 show_file_mode_name(file, "delete", p->one);
3295                 break;
3296         case DIFF_STATUS_ADDED:
3297                 show_file_mode_name(file, "create", p->two);
3298                 break;
3299         case DIFF_STATUS_COPIED:
3300                 show_rename_copy(file, "copy", p);
3301                 break;
3302         case DIFF_STATUS_RENAMED:
3303                 show_rename_copy(file, "rename", p);
3304                 break;
3305         default:
3306                 if (p->score) {
3307                         fputs(" rewrite ", file);
3308                         write_name_quoted(p->two->path, file, ' ');
3309                         fprintf(file, "(%d%%)\n", similarity_index(p));
3310                 }
3311                 show_mode_change(file, p, !p->score);
3312                 break;
3313         }
3314 }
3315
3316 struct patch_id_t {
3317         git_SHA_CTX *ctx;
3318         int patchlen;
3319 };
3320
3321 static int remove_space(char *line, int len)
3322 {
3323         int i;
3324         char *dst = line;
3325         unsigned char c;
3326
3327         for (i = 0; i < len; i++)
3328                 if (!isspace((c = line[i])))
3329                         *dst++ = c;
3330
3331         return dst - line;
3332 }
3333
3334 static void patch_id_consume(void *priv, char *line, unsigned long len)
3335 {
3336         struct patch_id_t *data = priv;
3337         int new_len;
3338
3339         /* Ignore line numbers when computing the SHA1 of the patch */
3340         if (!prefixcmp(line, "@@ -"))
3341                 return;
3342
3343         new_len = remove_space(line, len);
3344
3345         git_SHA1_Update(data->ctx, line, new_len);
3346         data->patchlen += new_len;
3347 }
3348
3349 /* returns 0 upon success, and writes result into sha1 */
3350 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3351 {
3352         struct diff_queue_struct *q = &diff_queued_diff;
3353         int i;
3354         git_SHA_CTX ctx;
3355         struct patch_id_t data;
3356         char buffer[PATH_MAX * 4 + 20];
3357
3358         git_SHA1_Init(&ctx);
3359         memset(&data, 0, sizeof(struct patch_id_t));
3360         data.ctx = &ctx;
3361
3362         for (i = 0; i < q->nr; i++) {
3363                 xpparam_t xpp;
3364                 xdemitconf_t xecfg;
3365                 mmfile_t mf1, mf2;
3366                 struct diff_filepair *p = q->queue[i];
3367                 int len1, len2;
3368
3369                 memset(&xpp, 0, sizeof(xpp));
3370                 memset(&xecfg, 0, sizeof(xecfg));
3371                 if (p->status == 0)
3372                         return error("internal diff status error");
3373                 if (p->status == DIFF_STATUS_UNKNOWN)
3374                         continue;
3375                 if (diff_unmodified_pair(p))
3376                         continue;
3377                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3378                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3379                         continue;
3380                 if (DIFF_PAIR_UNMERGED(p))
3381                         continue;
3382
3383                 diff_fill_sha1_info(p->one);
3384                 diff_fill_sha1_info(p->two);
3385                 if (fill_mmfile(&mf1, p->one) < 0 ||
3386                                 fill_mmfile(&mf2, p->two) < 0)
3387                         return error("unable to read files to diff");
3388
3389                 len1 = remove_space(p->one->path, strlen(p->one->path));
3390                 len2 = remove_space(p->two->path, strlen(p->two->path));
3391                 if (p->one->mode == 0)
3392                         len1 = snprintf(buffer, sizeof(buffer),
3393                                         "diff--gita/%.*sb/%.*s"
3394                                         "newfilemode%06o"
3395                                         "---/dev/null"
3396                                         "+++b/%.*s",
3397                                         len1, p->one->path,
3398                                         len2, p->two->path,
3399                                         p->two->mode,
3400                                         len2, p->two->path);
3401                 else if (p->two->mode == 0)
3402                         len1 = snprintf(buffer, sizeof(buffer),
3403                                         "diff--gita/%.*sb/%.*s"
3404                                         "deletedfilemode%06o"
3405                                         "---a/%.*s"
3406                                         "+++/dev/null",
3407                                         len1, p->one->path,
3408                                         len2, p->two->path,
3409                                         p->one->mode,
3410                                         len1, p->one->path);
3411                 else
3412                         len1 = snprintf(buffer, sizeof(buffer),
3413                                         "diff--gita/%.*sb/%.*s"
3414                                         "---a/%.*s"
3415                                         "+++b/%.*s",
3416                                         len1, p->one->path,
3417                                         len2, p->two->path,
3418                                         len1, p->one->path,
3419                                         len2, p->two->path);
3420                 git_SHA1_Update(&ctx, buffer, len1);
3421
3422                 xpp.flags = XDF_NEED_MINIMAL;
3423                 xecfg.ctxlen = 3;
3424                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3425                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3426                               &xpp, &xecfg);
3427         }
3428
3429         git_SHA1_Final(sha1, &ctx);
3430         return 0;
3431 }
3432
3433 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3434 {
3435         struct diff_queue_struct *q = &diff_queued_diff;
3436         int i;
3437         int result = diff_get_patch_id(options, sha1);
3438
3439         for (i = 0; i < q->nr; i++)
3440                 diff_free_filepair(q->queue[i]);
3441
3442         free(q->queue);
3443         q->queue = NULL;
3444         q->nr = q->alloc = 0;
3445
3446         return result;
3447 }
3448
3449 static int is_summary_empty(const struct diff_queue_struct *q)
3450 {
3451         int i;
3452
3453         for (i = 0; i < q->nr; i++) {
3454                 const struct diff_filepair *p = q->queue[i];
3455
3456                 switch (p->status) {
3457                 case DIFF_STATUS_DELETED:
3458                 case DIFF_STATUS_ADDED:
3459                 case DIFF_STATUS_COPIED:
3460                 case DIFF_STATUS_RENAMED:
3461                         return 0;
3462                 default:
3463                         if (p->score)
3464                                 return 0;
3465                         if (p->one->mode && p->two->mode &&
3466                             p->one->mode != p->two->mode)
3467                                 return 0;
3468                         break;
3469                 }
3470         }
3471         return 1;
3472 }
3473
3474 void diff_flush(struct diff_options *options)
3475 {
3476         struct diff_queue_struct *q = &diff_queued_diff;
3477         int i, output_format = options->output_format;
3478         int separator = 0;
3479
3480         /*
3481          * Order: raw, stat, summary, patch
3482          * or:    name/name-status/checkdiff (other bits clear)
3483          */
3484         if (!q->nr)
3485                 goto free_queue;
3486
3487         if (output_format & (DIFF_FORMAT_RAW |
3488                              DIFF_FORMAT_NAME |
3489                              DIFF_FORMAT_NAME_STATUS |
3490                              DIFF_FORMAT_CHECKDIFF)) {
3491                 for (i = 0; i < q->nr; i++) {
3492                         struct diff_filepair *p = q->queue[i];
3493                         if (check_pair_status(p))
3494                                 flush_one_pair(p, options);
3495                 }
3496                 separator++;
3497         }
3498
3499         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3500                 struct diffstat_t diffstat;
3501
3502                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3503                 for (i = 0; i < q->nr; i++) {
3504                         struct diff_filepair *p = q->queue[i];
3505                         if (check_pair_status(p))
3506                                 diff_flush_stat(p, options, &diffstat);
3507                 }
3508                 if (output_format & DIFF_FORMAT_NUMSTAT)
3509                         show_numstat(&diffstat, options);
3510                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3511                         show_stats(&diffstat, options);
3512                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3513                         show_shortstats(&diffstat, options);
3514                 free_diffstat_info(&diffstat);
3515                 separator++;
3516         }
3517         if (output_format & DIFF_FORMAT_DIRSTAT)
3518                 show_dirstat(options);
3519
3520         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3521                 for (i = 0; i < q->nr; i++)
3522                         diff_summary(options->file, q->queue[i]);
3523                 separator++;
3524         }
3525
3526         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
3527             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
3528             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
3529                 /*
3530                  * run diff_flush_patch for the exit status. setting
3531                  * options->file to /dev/null should be safe, becaue we
3532                  * aren't supposed to produce any output anyway.
3533                  */
3534                 if (options->close_file)
3535                         fclose(options->file);
3536                 options->file = fopen("/dev/null", "w");
3537                 if (!options->file)
3538                         die_errno("Could not open /dev/null");
3539                 options->close_file = 1;
3540                 for (i = 0; i < q->nr; i++) {
3541                         struct diff_filepair *p = q->queue[i];
3542                         if (check_pair_status(p))
3543                                 diff_flush_patch(p, options);
3544                         if (options->found_changes)
3545                                 break;
3546                 }
3547         }
3548
3549         if (output_format & DIFF_FORMAT_PATCH) {
3550                 if (separator) {
3551                         putc(options->line_termination, options->file);
3552                         if (options->stat_sep) {
3553                                 /* attach patch instead of inline */
3554                                 fputs(options->stat_sep, options->file);
3555                         }
3556                 }
3557
3558                 for (i = 0; i < q->nr; i++) {
3559                         struct diff_filepair *p = q->queue[i];
3560                         if (check_pair_status(p))
3561                                 diff_flush_patch(p, options);
3562                 }
3563         }
3564
3565         if (output_format & DIFF_FORMAT_CALLBACK)
3566                 options->format_callback(q, options, options->format_callback_data);
3567
3568         for (i = 0; i < q->nr; i++)
3569                 diff_free_filepair(q->queue[i]);
3570 free_queue:
3571         free(q->queue);
3572         q->queue = NULL;
3573         q->nr = q->alloc = 0;
3574         if (options->close_file)
3575                 fclose(options->file);
3576
3577         /*
3578          * Report the content-level differences with HAS_CHANGES;
3579          * diff_addremove/diff_change does not set the bit when
3580          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
3581          */
3582         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
3583                 if (options->found_changes)
3584                         DIFF_OPT_SET(options, HAS_CHANGES);
3585                 else
3586                         DIFF_OPT_CLR(options, HAS_CHANGES);
3587         }
3588 }
3589
3590 static void diffcore_apply_filter(const char *filter)
3591 {
3592         int i;
3593         struct diff_queue_struct *q = &diff_queued_diff;
3594         struct diff_queue_struct outq;
3595         outq.queue = NULL;
3596         outq.nr = outq.alloc = 0;
3597
3598         if (!filter)
3599                 return;
3600
3601         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3602                 int found;
3603                 for (i = found = 0; !found && i < q->nr; i++) {
3604                         struct diff_filepair *p = q->queue[i];
3605                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3606                              ((p->score &&
3607                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3608                               (!p->score &&
3609                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3610                             ((p->status != DIFF_STATUS_MODIFIED) &&
3611                              strchr(filter, p->status)))
3612                                 found++;
3613                 }
3614                 if (found)
3615                         return;
3616
3617                 /* otherwise we will clear the whole queue
3618                  * by copying the empty outq at the end of this
3619                  * function, but first clear the current entries
3620                  * in the queue.
3621                  */
3622                 for (i = 0; i < q->nr; i++)
3623                         diff_free_filepair(q->queue[i]);
3624         }
3625         else {
3626                 /* Only the matching ones */
3627                 for (i = 0; i < q->nr; i++) {
3628                         struct diff_filepair *p = q->queue[i];
3629
3630                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3631                              ((p->score &&
3632                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3633                               (!p->score &&
3634                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3635                             ((p->status != DIFF_STATUS_MODIFIED) &&
3636                              strchr(filter, p->status)))
3637                                 diff_q(&outq, p);
3638                         else
3639                                 diff_free_filepair(p);
3640                 }
3641         }
3642         free(q->queue);
3643         *q = outq;
3644 }
3645
3646 /* Check whether two filespecs with the same mode and size are identical */
3647 static int diff_filespec_is_identical(struct diff_filespec *one,
3648                                       struct diff_filespec *two)
3649 {
3650         if (S_ISGITLINK(one->mode))
3651                 return 0;
3652         if (diff_populate_filespec(one, 0))
3653                 return 0;
3654         if (diff_populate_filespec(two, 0))
3655                 return 0;
3656         return !memcmp(one->data, two->data, one->size);
3657 }
3658
3659 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3660 {
3661         int i;
3662         struct diff_queue_struct *q = &diff_queued_diff;
3663         struct diff_queue_struct outq;
3664         outq.queue = NULL;
3665         outq.nr = outq.alloc = 0;
3666
3667         for (i = 0; i < q->nr; i++) {
3668                 struct diff_filepair *p = q->queue[i];
3669
3670                 /*
3671                  * 1. Entries that come from stat info dirtiness
3672                  *    always have both sides (iow, not create/delete),
3673                  *    one side of the object name is unknown, with
3674                  *    the same mode and size.  Keep the ones that
3675                  *    do not match these criteria.  They have real
3676                  *    differences.
3677                  *
3678                  * 2. At this point, the file is known to be modified,
3679                  *    with the same mode and size, and the object
3680                  *    name of one side is unknown.  Need to inspect
3681                  *    the identical contents.
3682                  */
3683                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3684                     !DIFF_FILE_VALID(p->two) ||
3685                     (p->one->sha1_valid && p->two->sha1_valid) ||
3686                     (p->one->mode != p->two->mode) ||
3687                     diff_populate_filespec(p->one, 1) ||
3688                     diff_populate_filespec(p->two, 1) ||
3689                     (p->one->size != p->two->size) ||
3690                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3691                         diff_q(&outq, p);
3692                 else {
3693                         /*
3694                          * The caller can subtract 1 from skip_stat_unmatch
3695                          * to determine how many paths were dirty only
3696                          * due to stat info mismatch.
3697                          */
3698                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3699                                 diffopt->skip_stat_unmatch++;
3700                         diff_free_filepair(p);
3701                 }
3702         }
3703         free(q->queue);
3704         *q = outq;
3705 }
3706
3707 static int diffnamecmp(const void *a_, const void *b_)
3708 {
3709         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
3710         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
3711         const char *name_a, *name_b;
3712
3713         name_a = a->one ? a->one->path : a->two->path;
3714         name_b = b->one ? b->one->path : b->two->path;
3715         return strcmp(name_a, name_b);
3716 }
3717
3718 void diffcore_fix_diff_index(struct diff_options *options)
3719 {
3720         struct diff_queue_struct *q = &diff_queued_diff;
3721         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
3722 }
3723
3724 void diffcore_std(struct diff_options *options)
3725 {
3726         if (options->skip_stat_unmatch)
3727                 diffcore_skip_stat_unmatch(options);
3728         if (options->break_opt != -1)
3729                 diffcore_break(options->break_opt);
3730         if (options->detect_rename)
3731                 diffcore_rename(options);
3732         if (options->break_opt != -1)
3733                 diffcore_merge_broken();
3734         if (options->pickaxe)
3735                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3736         if (options->orderfile)
3737                 diffcore_order(options->orderfile);
3738         diff_resolve_rename_copy();
3739         diffcore_apply_filter(options->filter);
3740
3741         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3742                 DIFF_OPT_SET(options, HAS_CHANGES);
3743         else
3744                 DIFF_OPT_CLR(options, HAS_CHANGES);
3745 }
3746
3747 int diff_result_code(struct diff_options *opt, int status)
3748 {
3749         int result = 0;
3750         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3751             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3752                 return status;
3753         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3754             DIFF_OPT_TST(opt, HAS_CHANGES))
3755                 result |= 01;
3756         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3757             DIFF_OPT_TST(opt, CHECK_FAILED))
3758                 result |= 02;
3759         return result;
3760 }
3761
3762 void diff_addremove(struct diff_options *options,
3763                     int addremove, unsigned mode,
3764                     const unsigned char *sha1,
3765                     const char *concatpath, unsigned dirty_submodule)
3766 {
3767         struct diff_filespec *one, *two;
3768
3769         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3770                 return;
3771
3772         /* This may look odd, but it is a preparation for
3773          * feeding "there are unchanged files which should
3774          * not produce diffs, but when you are doing copy
3775          * detection you would need them, so here they are"
3776          * entries to the diff-core.  They will be prefixed
3777          * with something like '=' or '*' (I haven't decided
3778          * which but should not make any difference).
3779          * Feeding the same new and old to diff_change()
3780          * also has the same effect.
3781          * Before the final output happens, they are pruned after
3782          * merged into rename/copy pairs as appropriate.
3783          */
3784         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3785                 addremove = (addremove == '+' ? '-' :
3786                              addremove == '-' ? '+' : addremove);
3787
3788         if (options->prefix &&
3789             strncmp(concatpath, options->prefix, options->prefix_length))
3790                 return;
3791
3792         one = alloc_filespec(concatpath);
3793         two = alloc_filespec(concatpath);
3794
3795         if (addremove != '+')
3796                 fill_filespec(one, sha1, mode);
3797         if (addremove != '-') {
3798                 fill_filespec(two, sha1, mode);
3799                 two->dirty_submodule = dirty_submodule;
3800         }
3801
3802         diff_queue(&diff_queued_diff, one, two);
3803         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3804                 DIFF_OPT_SET(options, HAS_CHANGES);
3805 }
3806
3807 void diff_change(struct diff_options *options,
3808                  unsigned old_mode, unsigned new_mode,
3809                  const unsigned char *old_sha1,
3810                  const unsigned char *new_sha1,
3811                  const char *concatpath,
3812                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
3813 {
3814         struct diff_filespec *one, *two;
3815
3816         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3817                         && S_ISGITLINK(new_mode))
3818                 return;
3819
3820         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3821                 unsigned tmp;
3822                 const unsigned char *tmp_c;
3823                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3824                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3825                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
3826                         new_dirty_submodule = tmp;
3827         }
3828
3829         if (options->prefix &&
3830             strncmp(concatpath, options->prefix, options->prefix_length))
3831                 return;
3832
3833         one = alloc_filespec(concatpath);
3834         two = alloc_filespec(concatpath);
3835         fill_filespec(one, old_sha1, old_mode);
3836         fill_filespec(two, new_sha1, new_mode);
3837         one->dirty_submodule = old_dirty_submodule;
3838         two->dirty_submodule = new_dirty_submodule;
3839
3840         diff_queue(&diff_queued_diff, one, two);
3841         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3842                 DIFF_OPT_SET(options, HAS_CHANGES);
3843 }
3844
3845 void diff_unmerge(struct diff_options *options,
3846                   const char *path,
3847                   unsigned mode, const unsigned char *sha1)
3848 {
3849         struct diff_filespec *one, *two;
3850
3851         if (options->prefix &&
3852             strncmp(path, options->prefix, options->prefix_length))
3853                 return;
3854
3855         one = alloc_filespec(path);
3856         two = alloc_filespec(path);
3857         fill_filespec(one, sha1, mode);
3858         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3859 }
3860
3861 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
3862                 size_t *outsize)
3863 {
3864         struct diff_tempfile *temp;
3865         const char *argv[3];
3866         const char **arg = argv;
3867         struct child_process child;
3868         struct strbuf buf = STRBUF_INIT;
3869         int err = 0;
3870
3871         temp = prepare_temp_file(spec->path, spec);
3872         *arg++ = pgm;
3873         *arg++ = temp->name;
3874         *arg = NULL;
3875
3876         memset(&child, 0, sizeof(child));
3877         child.use_shell = 1;
3878         child.argv = argv;
3879         child.out = -1;
3880         if (start_command(&child)) {
3881                 remove_tempfile();
3882                 return NULL;
3883         }
3884
3885         if (strbuf_read(&buf, child.out, 0) < 0)
3886                 err = error("error reading from textconv command '%s'", pgm);
3887         close(child.out);
3888
3889         if (finish_command(&child) || err) {
3890                 strbuf_release(&buf);
3891                 remove_tempfile();
3892                 return NULL;
3893         }
3894         remove_tempfile();
3895
3896         return strbuf_detach(&buf, outsize);
3897 }
3898
3899 static size_t fill_textconv(struct userdiff_driver *driver,
3900                             struct diff_filespec *df,
3901                             char **outbuf)
3902 {
3903         size_t size;
3904
3905         if (!driver || !driver->textconv) {
3906                 if (!DIFF_FILE_VALID(df)) {
3907                         *outbuf = "";
3908                         return 0;
3909                 }
3910                 if (diff_populate_filespec(df, 0))
3911                         die("unable to read files to diff");
3912                 *outbuf = df->data;
3913                 return df->size;
3914         }
3915
3916         if (driver->textconv_cache) {
3917                 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
3918                                           &size);
3919                 if (*outbuf)
3920                         return size;
3921         }
3922
3923         *outbuf = run_textconv(driver->textconv, df, &size);
3924         if (!*outbuf)
3925                 die("unable to read files to diff");
3926
3927         if (driver->textconv_cache) {
3928                 /* ignore errors, as we might be in a readonly repository */
3929                 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
3930                                 size);
3931                 /*
3932                  * we could save up changes and flush them all at the end,
3933                  * but we would need an extra call after all diffing is done.
3934                  * Since generating a cache entry is the slow path anyway,
3935                  * this extra overhead probably isn't a big deal.
3936                  */
3937                 notes_cache_write(driver->textconv_cache);
3938         }
3939
3940         return size;
3941 }