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

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "quote.h"
   6 #include "diff.h"
   7 #include "diffcore.h"
   8 #include "delta.h"
   9 #include "xdiff-interface.h"
  10 #include "color.h"
  11 #include "attr.h"
  12 #include "run-command.h"
  13 #include "utf8.h"
  14 #include "userdiff.h"
  15 #include "sigchain.h"
  16 #include "submodule.h"
  17
  18 #ifdef NO_FAST_WORKING_DIRECTORY
  19 #define FAST_WORKING_DIRECTORY 0
  20 #else
  21 #define FAST_WORKING_DIRECTORY 1
  22 #endif
  23
  24 static int diff_detect_rename_default;
  25 static int diff_rename_limit_default = 200;
  26 static int diff_suppress_blank_empty;
  27 int diff_use_color_default = -1;
  28 static const char *diff_word_regex_cfg;
  29 static const char *external_diff_cmd_cfg;
  30 int diff_auto_refresh_index = 1;
  31 static int diff_mnemonic_prefix;
  32
  33 static char diff_colors[][COLOR_MAXLEN] = {
  34         GIT_COLOR_RESET,
  35         GIT_COLOR_NORMAL,       /* PLAIN */
  36         GIT_COLOR_BOLD,         /* METAINFO */
  37         GIT_COLOR_CYAN,         /* FRAGINFO */
  38         GIT_COLOR_RED,          /* OLD */
  39         GIT_COLOR_GREEN,        /* NEW */
  40         GIT_COLOR_YELLOW,       /* COMMIT */
  41         GIT_COLOR_BG_RED,       /* WHITESPACE */
  42         GIT_COLOR_NORMAL,       /* FUNCINFO */
  43 };
  44
  45 static void diff_filespec_load_driver(struct diff_filespec *one);
  46 static size_t fill_textconv(struct userdiff_driver *driver,
  47                             struct diff_filespec *df, char **outbuf);
  48
  49 static int parse_diff_color_slot(const char *var, int ofs)
  50 {
  51         if (!strcasecmp(var+ofs, "plain"))
  52                 return DIFF_PLAIN;
  53         if (!strcasecmp(var+ofs, "meta"))
  54                 return DIFF_METAINFO;
  55         if (!strcasecmp(var+ofs, "frag"))
  56                 return DIFF_FRAGINFO;
  57         if (!strcasecmp(var+ofs, "old"))
  58                 return DIFF_FILE_OLD;
  59         if (!strcasecmp(var+ofs, "new"))
  60                 return DIFF_FILE_NEW;
  61         if (!strcasecmp(var+ofs, "commit"))
  62                 return DIFF_COMMIT;
  63         if (!strcasecmp(var+ofs, "whitespace"))
  64                 return DIFF_WHITESPACE;
  65         if (!strcasecmp(var+ofs, "func"))
  66                 return DIFF_FUNCINFO;
  67         return -1;
  68 }
  69
  70 static int git_config_rename(const char *var, const char *value)
  71 {
  72         if (!value)
  73                 return DIFF_DETECT_RENAME;
  74         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
  75                 return  DIFF_DETECT_COPY;
  76         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
  77 }
  78
  79 /*
  80  * These are to give UI layer defaults.
  81  * The core-level commands such as git-diff-files should
  82  * never be affected by the setting of diff.renames
  83  * the user happens to have in the configuration file.
  84  */
  85 int git_diff_ui_config(const char *var, const char *value, void *cb)
  86 {
  87         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
  88                 diff_use_color_default = git_config_colorbool(var, value, -1);
  89                 return 0;
  90         }
  91         if (!strcmp(var, "diff.renames")) {
  92                 diff_detect_rename_default = git_config_rename(var, value);
  93                 return 0;
  94         }
  95         if (!strcmp(var, "diff.autorefreshindex")) {
  96                 diff_auto_refresh_index = git_config_bool(var, value);
  97                 return 0;
  98         }
  99         if (!strcmp(var, "diff.mnemonicprefix")) {
 100                 diff_mnemonic_prefix = git_config_bool(var, value);
 101                 return 0;
 102         }
 103         if (!strcmp(var, "diff.external"))
 104                 return git_config_string(&external_diff_cmd_cfg, var, value);
 105         if (!strcmp(var, "diff.wordregex"))
 106                 return git_config_string(&diff_word_regex_cfg, var, value);
 107
 108         return git_diff_basic_config(var, value, cb);
 109 }
 110
 111 int git_diff_basic_config(const char *var, const char *value, void *cb)
 112 {
 113         if (!strcmp(var, "diff.renamelimit")) {
 114                 diff_rename_limit_default = git_config_int(var, value);
 115                 return 0;
 116         }
 117
 118         switch (userdiff_config(var, value)) {
 119                 case 0: break;
 120                 case -1: return -1;
 121                 default: return 0;
 122         }
 123
 124         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 125                 int slot = parse_diff_color_slot(var, 11);
 126                 if (slot < 0)
 127                         return 0;
 128                 if (!value)
 129                         return config_error_nonbool(var);
 130                 color_parse(value, var, diff_colors[slot]);
 131                 return 0;
 132         }
 133
 134         /* like GNU diff's --suppress-blank-empty option  */
 135         if (!strcmp(var, "diff.suppressblankempty") ||
 136                         /* for backwards compatibility */
 137                         !strcmp(var, "diff.suppress-blank-empty")) {
 138                 diff_suppress_blank_empty = git_config_bool(var, value);
 139                 return 0;
 140         }
 141
 142         return git_color_default_config(var, value, cb);
 143 }
 144
 145 static char *quote_two(const char *one, const char *two)
 146 {
 147         int need_one = quote_c_style(one, NULL, NULL, 1);
 148         int need_two = quote_c_style(two, NULL, NULL, 1);
 149         struct strbuf res = STRBUF_INIT;
 150
 151         if (need_one + need_two) {
 152                 strbuf_addch(&res, '"');
 153                 quote_c_style(one, &res, NULL, 1);
 154                 quote_c_style(two, &res, NULL, 1);
 155                 strbuf_addch(&res, '"');
 156         } else {
 157                 strbuf_addstr(&res, one);
 158                 strbuf_addstr(&res, two);
 159         }
 160         return strbuf_detach(&res, NULL);
 161 }
 162
 163 static const char *external_diff(void)
 164 {
 165         static const char *external_diff_cmd = NULL;
 166         static int done_preparing = 0;
 167
 168         if (done_preparing)
 169                 return external_diff_cmd;
 170         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 171         if (!external_diff_cmd)
 172                 external_diff_cmd = external_diff_cmd_cfg;
 173         done_preparing = 1;
 174         return external_diff_cmd;
 175 }
 176
 177 static struct diff_tempfile {
 178         const char *name; /* filename external diff should read from */
 179         char hex[41];
 180         char mode[10];
 181         char tmp_path[PATH_MAX];
 182 } diff_temp[2];
 183
 184 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 185
 186 struct emit_callback {
 187         int color_diff;
 188         unsigned ws_rule;
 189         int blank_at_eof_in_preimage;
 190         int blank_at_eof_in_postimage;
 191         int lno_in_preimage;
 192         int lno_in_postimage;
 193         sane_truncate_fn truncate;
 194         const char **label_path;
 195         struct diff_words_data *diff_words;
 196         int *found_changesp;
 197         FILE *file;
 198         struct strbuf *header;
 199 };
 200
 201 static int count_lines(const char *data, int size)
 202 {
 203         int count, ch, completely_empty = 1, nl_just_seen = 0;
 204         count = 0;
 205         while (0 < size--) {
 206                 ch = *data++;
 207                 if (ch == '\n') {
 208                         count++;
 209                         nl_just_seen = 1;
 210                         completely_empty = 0;
 211                 }
 212                 else {
 213                         nl_just_seen = 0;
 214                         completely_empty = 0;
 215                 }
 216         }
 217         if (completely_empty)
 218                 return 0;
 219         if (!nl_just_seen)
 220                 count++; /* no trailing newline */
 221         return count;
 222 }
 223
 224 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 225 {
 226         if (!DIFF_FILE_VALID(one)) {
 227                 mf->ptr = (char *)""; /* does not matter */
 228                 mf->size = 0;
 229                 return 0;
 230         }
 231         else if (diff_populate_filespec(one, 0))
 232                 return -1;
 233
 234         mf->ptr = one->data;
 235         mf->size = one->size;
 236         return 0;
 237 }
 238
 239 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 240 {
 241         char *ptr = mf->ptr;
 242         long size = mf->size;
 243         int cnt = 0;
 244
 245         if (!size)
 246                 return cnt;
 247         ptr += size - 1; /* pointing at the very end */
 248         if (*ptr != '\n')
 249                 ; /* incomplete line */
 250         else
 251                 ptr--; /* skip the last LF */
 252         while (mf->ptr < ptr) {
 253                 char *prev_eol;
 254                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 255                         if (*prev_eol == '\n')
 256                                 break;
 257                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 258                         break;
 259                 cnt++;
 260                 ptr = prev_eol - 1;
 261         }
 262         return cnt;
 263 }
 264
 265 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 266                                struct emit_callback *ecbdata)
 267 {
 268         int l1, l2, at;
 269         unsigned ws_rule = ecbdata->ws_rule;
 270         l1 = count_trailing_blank(mf1, ws_rule);
 271         l2 = count_trailing_blank(mf2, ws_rule);
 272         if (l2 <= l1) {
 273                 ecbdata->blank_at_eof_in_preimage = 0;
 274                 ecbdata->blank_at_eof_in_postimage = 0;
 275                 return;
 276         }
 277         at = count_lines(mf1->ptr, mf1->size);
 278         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 279
 280         at = count_lines(mf2->ptr, mf2->size);
 281         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 282 }
 283
 284 static void emit_line_0(FILE *file, const char *set, const char *reset,
 285                         int first, const char *line, int len)
 286 {
 287         int has_trailing_newline, has_trailing_carriage_return;
 288         int nofirst;
 289
 290         if (len == 0) {
 291                 has_trailing_newline = (first == '\n');
 292                 has_trailing_carriage_return = (!has_trailing_newline &&
 293                                                 (first == '\r'));
 294                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 295         } else {
 296                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 297                 if (has_trailing_newline)
 298                         len--;
 299                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 300                 if (has_trailing_carriage_return)
 301                         len--;
 302                 nofirst = 0;
 303         }
 304
 305         if (len || !nofirst) {
 306                 fputs(set, file);
 307                 if (!nofirst)
 308                         fputc(first, file);
 309                 fwrite(line, len, 1, file);
 310                 fputs(reset, file);
 311         }
 312         if (has_trailing_carriage_return)
 313                 fputc('\r', file);
 314         if (has_trailing_newline)
 315                 fputc('\n', file);
 316 }
 317
 318 static void emit_line(FILE *file, const char *set, const char *reset,
 319                       const char *line, int len)
 320 {
 321         emit_line_0(file, set, reset, line[0], line+1, len-1);
 322 }
 323
 324 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
 325 {
 326         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
 327               ecbdata->blank_at_eof_in_preimage &&
 328               ecbdata->blank_at_eof_in_postimage &&
 329               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
 330               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
 331                 return 0;
 332         return ws_blank_line(line, len, ecbdata->ws_rule);
 333 }
 334
 335 static void emit_add_line(const char *reset,
 336                           struct emit_callback *ecbdata,
 337                           const char *line, int len)
 338 {
 339         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 340         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 341
 342         if (!*ws)
 343                 emit_line_0(ecbdata->file, set, reset, '+', line, len);
 344         else if (new_blank_line_at_eof(ecbdata, line, len))
 345                 /* Blank line at EOF - paint '+' as well */
 346                 emit_line_0(ecbdata->file, ws, reset, '+', line, len);
 347         else {
 348                 /* Emit just the prefix, then the rest. */
 349                 emit_line_0(ecbdata->file, set, reset, '+', "", 0);
 350                 ws_check_emit(line, len, ecbdata->ws_rule,
 351                               ecbdata->file, set, reset, ws);
 352         }
 353 }
 354
 355 static void emit_hunk_header(struct emit_callback *ecbdata,
 356                              const char *line, int len)
 357 {
 358         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 359         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
 360         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
 361         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 362         static const char atat[2] = { '@', '@' };
 363         const char *cp, *ep;
 364
 365         /*
 366          * As a hunk header must begin with "@@ -<old>, +<new> @@",
 367          * it always is at least 10 bytes long.
 368          */
 369         if (len < 10 ||
 370             memcmp(line, atat, 2) ||
 371             !(ep = memmem(line + 2, len - 2, atat, 2))) {
 372                 emit_line(ecbdata->file, plain, reset, line, len);
 373                 return;
 374         }
 375         ep += 2; /* skip over @@ */
 376
 377         /* The hunk header in fraginfo color */
 378         emit_line(ecbdata->file, frag, reset, line, ep - line);
 379
 380         /* blank before the func header */
 381         for (cp = ep; ep - line < len; ep++)
 382                 if (*ep != ' ' && *ep != '\t')
 383                         break;
 384         if (ep != cp)
 385                 emit_line(ecbdata->file, plain, reset, cp, ep - cp);
 386
 387         if (ep < line + len)
 388                 emit_line(ecbdata->file, func, reset, ep, line + len - ep);
 389 }
 390
 391 static struct diff_tempfile *claim_diff_tempfile(void) {
 392         int i;
 393         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
 394                 if (!diff_temp[i].name)
 395                         return diff_temp + i;
 396         die("BUG: diff is failing to clean up its tempfiles");
 397 }
 398
 399 static int remove_tempfile_installed;
 400
 401 static void remove_tempfile(void)
 402 {
 403         int i;
 404         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
 405                 if (diff_temp[i].name == diff_temp[i].tmp_path)
 406                         unlink_or_warn(diff_temp[i].name);
 407                 diff_temp[i].name = NULL;
 408         }
 409 }
 410
 411 static void remove_tempfile_on_signal(int signo)
 412 {
 413         remove_tempfile();
 414         sigchain_pop(signo);
 415         raise(signo);
 416 }
 417
 418 static void print_line_count(FILE *file, int count)
 419 {
 420         switch (count) {
 421         case 0:
 422                 fprintf(file, "0,0");
 423                 break;
 424         case 1:
 425                 fprintf(file, "1");
 426                 break;
 427         default:
 428                 fprintf(file, "1,%d", count);
 429                 break;
 430         }
 431 }
 432
 433 static void emit_rewrite_lines(struct emit_callback *ecb,
 434                                int prefix, const char *data, int size)
 435 {
 436         const char *endp = NULL;
 437         static const char *nneof = " No newline at end of file\n";
 438         const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
 439         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
 440
 441         while (0 < size) {
 442                 int len;
 443
 444                 endp = memchr(data, '\n', size);
 445                 len = endp ? (endp - data + 1) : size;
 446                 if (prefix != '+') {
 447                         ecb->lno_in_preimage++;
 448                         emit_line_0(ecb->file, old, reset, '-',
 449                                     data, len);
 450                 } else {
 451                         ecb->lno_in_postimage++;
 452                         emit_add_line(reset, ecb, data, len);
 453                 }
 454                 size -= len;
 455                 data += len;
 456         }
 457         if (!endp) {
 458                 const char *plain = diff_get_color(ecb->color_diff,
 459                                                    DIFF_PLAIN);
 460                 emit_line_0(ecb->file, plain, reset, '\\',
 461                             nneof, strlen(nneof));
 462         }
 463 }
 464
 465 static void emit_rewrite_diff(const char *name_a,
 466                               const char *name_b,
 467                               struct diff_filespec *one,
 468                               struct diff_filespec *two,
 469                               struct userdiff_driver *textconv_one,
 470                               struct userdiff_driver *textconv_two,
 471                               struct diff_options *o)
 472 {
 473         int lc_a, lc_b;
 474         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
 475         const char *name_a_tab, *name_b_tab;
 476         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 477         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 478         const char *reset = diff_get_color(color_diff, DIFF_RESET);
 479         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 480         const char *a_prefix, *b_prefix;
 481         char *data_one, *data_two;
 482         size_t size_one, size_two;
 483         struct emit_callback ecbdata;
 484
 485         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 486                 a_prefix = o->b_prefix;
 487                 b_prefix = o->a_prefix;
 488         } else {
 489                 a_prefix = o->a_prefix;
 490                 b_prefix = o->b_prefix;
 491         }
 492
 493         name_a += (*name_a == '/');
 494         name_b += (*name_b == '/');
 495         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 496         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 497
 498         strbuf_reset(&a_name);
 499         strbuf_reset(&b_name);
 500         quote_two_c_style(&a_name, a_prefix, name_a, 0);
 501         quote_two_c_style(&b_name, b_prefix, name_b, 0);
 502
 503         size_one = fill_textconv(textconv_one, one, &data_one);
 504         size_two = fill_textconv(textconv_two, two, &data_two);
 505
 506         memset(&ecbdata, 0, sizeof(ecbdata));
 507         ecbdata.color_diff = color_diff;
 508         ecbdata.found_changesp = &o->found_changes;
 509         ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
 510         ecbdata.file = o->file;
 511         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
 512                 mmfile_t mf1, mf2;
 513                 mf1.ptr = (char *)data_one;
 514                 mf2.ptr = (char *)data_two;
 515                 mf1.size = size_one;
 516                 mf2.size = size_two;
 517                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
 518         }
 519         ecbdata.lno_in_preimage = 1;
 520         ecbdata.lno_in_postimage = 1;
 521
 522         lc_a = count_lines(data_one, size_one);
 523         lc_b = count_lines(data_two, size_two);
 524         fprintf(o->file,
 525                 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
 526                 metainfo, a_name.buf, name_a_tab, reset,
 527                 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
 528         print_line_count(o->file, lc_a);
 529         fprintf(o->file, " +");
 530         print_line_count(o->file, lc_b);
 531         fprintf(o->file, " @@%s\n", reset);
 532         if (lc_a)
 533                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
 534         if (lc_b)
 535                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
 536         if (textconv_one)
 537                 free(data_one);
 538         if (textconv_two)
 539                 free(data_two);
 540 }
 541
 542 struct diff_words_buffer {
 543         mmfile_t text;
 544         long alloc;
 545         struct diff_words_orig {
 546                 const char *begin, *end;
 547         } *orig;
 548         int orig_nr, orig_alloc;
 549 };
 550
 551 static void diff_words_append(char *line, unsigned long len,
 552                 struct diff_words_buffer *buffer)
 553 {
 554         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
 555         line++;
 556         len--;
 557         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 558         buffer->text.size += len;
 559         buffer->text.ptr[buffer->text.size] = '\0';
 560 }
 561
 562 struct diff_words_data {
 563         struct diff_words_buffer minus, plus;
 564         const char *current_plus;
 565         FILE *file;
 566         regex_t *word_regex;
 567 };
 568
 569 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 570 {
 571         struct diff_words_data *diff_words = priv;
 572         int minus_first, minus_len, plus_first, plus_len;
 573         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 574
 575         if (line[0] != '@' || parse_hunk_header(line, len,
 576                         &minus_first, &minus_len, &plus_first, &plus_len))
 577                 return;
 578
 579         /* POSIX requires that first be decremented by one if len == 0... */
 580         if (minus_len) {
 581                 minus_begin = diff_words->minus.orig[minus_first].begin;
 582                 minus_end =
 583                         diff_words->minus.orig[minus_first + minus_len - 1].end;
 584         } else
 585                 minus_begin = minus_end =
 586                         diff_words->minus.orig[minus_first].end;
 587
 588         if (plus_len) {
 589                 plus_begin = diff_words->plus.orig[plus_first].begin;
 590                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
 591         } else
 592                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
 593
 594         if (diff_words->current_plus != plus_begin)
 595                 fwrite(diff_words->current_plus,
 596                                 plus_begin - diff_words->current_plus, 1,
 597                                 diff_words->file);
 598         if (minus_begin != minus_end)
 599                 color_fwrite_lines(diff_words->file,
 600                                 diff_get_color(1, DIFF_FILE_OLD),
 601                                 minus_end - minus_begin, minus_begin);
 602         if (plus_begin != plus_end)
 603                 color_fwrite_lines(diff_words->file,
 604                                 diff_get_color(1, DIFF_FILE_NEW),
 605                                 plus_end - plus_begin, plus_begin);
 606
 607         diff_words->current_plus = plus_end;
 608 }
 609
 610 /* This function starts looking at *begin, and returns 0 iff a word was found. */
 611 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
 612                 int *begin, int *end)
 613 {
 614         if (word_regex && *begin < buffer->size) {
 615                 regmatch_t match[1];
 616                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
 617                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
 618                                         '\n', match[0].rm_eo - match[0].rm_so);
 619                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
 620                         *begin += match[0].rm_so;
 621                         return *begin >= *end;
 622                 }
 623                 return -1;
 624         }
 625
 626         /* find the next word */
 627         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
 628                 (*begin)++;
 629         if (*begin >= buffer->size)
 630                 return -1;
 631
 632         /* find the end of the word */
 633         *end = *begin + 1;
 634         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
 635                 (*end)++;
 636
 637         return 0;
 638 }
 639
 640 /*
 641  * This function splits the words in buffer->text, stores the list with
 642  * newline separator into out, and saves the offsets of the original words
 643  * in buffer->orig.
 644  */
 645 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
 646                 regex_t *word_regex)
 647 {
 648         int i, j;
 649         long alloc = 0;
 650
 651         out->size = 0;
 652         out->ptr = NULL;
 653
 654         /* fake an empty "0th" word */
 655         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
 656         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
 657         buffer->orig_nr = 1;
 658
 659         for (i = 0; i < buffer->text.size; i++) {
 660                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
 661                         return;
 662
 663                 /* store original boundaries */
 664                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
 665                                 buffer->orig_alloc);
 666                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
 667                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
 668                 buffer->orig_nr++;
 669
 670                 /* store one word */
 671                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
 672                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
 673                 out->ptr[out->size + j - i] = '\n';
 674                 out->size += j - i + 1;
 675
 676                 i = j - 1;
 677         }
 678 }
 679
 680 /* this executes the word diff on the accumulated buffers */
 681 static void diff_words_show(struct diff_words_data *diff_words)
 682 {
 683         xpparam_t xpp;
 684         xdemitconf_t xecfg;
 685         xdemitcb_t ecb;
 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, &ecb);
 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                 unsigned int 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         int 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                 int 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                 int added = data->files[i]->added;
1088                 int 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%d%s", del_c, deleted, reset);
1110                         fprintf(options->file, " -> ");
1111                         fprintf(options->file, "%s%d%s", add_c, added, reset);
1112                         fprintf(options->file, " bytes");
1113                         fprintf(options->file, "\n");
1114                         continue;
1115                 }
1116                 else if (data->files[i]->is_unmerged) {
1117                         show_name(options->file, prefix, name, len);
1118                         fprintf(options->file, "  Unmerged\n");
1119                         continue;
1120                 }
1121                 else if (!data->files[i]->is_renamed &&
1122                          (added + deleted == 0)) {
1123                         total_files--;
1124                         continue;
1125                 }
1126
1127                 /*
1128                  * scale the add/delete
1129                  */
1130                 add = added;
1131                 del = deleted;
1132                 adds += add;
1133                 dels += del;
1134
1135                 if (width <= max_change) {
1136                         add = scale_linear(add, width, max_change);
1137                         del = scale_linear(del, width, max_change);
1138                 }
1139                 show_name(options->file, prefix, name, len);
1140                 fprintf(options->file, "%5d%s", added + deleted,
1141                                 added + deleted ? " " : "");
1142                 show_graph(options->file, '+', add, add_c, reset);
1143                 show_graph(options->file, '-', del, del_c, reset);
1144                 fprintf(options->file, "\n");
1145         }
1146         fprintf(options->file,
1147                " %d files changed, %d insertions(+), %d deletions(-)\n",
1148                total_files, adds, dels);
1149 }
1150
1151 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1152 {
1153         int i, adds = 0, dels = 0, total_files = data->nr;
1154
1155         if (data->nr == 0)
1156                 return;
1157
1158         for (i = 0; i < data->nr; i++) {
1159                 if (!data->files[i]->is_binary &&
1160                     !data->files[i]->is_unmerged) {
1161                         int added = data->files[i]->added;
1162                         int deleted= data->files[i]->deleted;
1163                         if (!data->files[i]->is_renamed &&
1164                             (added + deleted == 0)) {
1165                                 total_files--;
1166                         } else {
1167                                 adds += added;
1168                                 dels += deleted;
1169                         }
1170                 }
1171         }
1172         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
1173                total_files, adds, dels);
1174 }
1175
1176 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1177 {
1178         int i;
1179
1180         if (data->nr == 0)
1181                 return;
1182
1183         for (i = 0; i < data->nr; i++) {
1184                 struct diffstat_file *file = data->files[i];
1185
1186                 if (file->is_binary)
1187                         fprintf(options->file, "-\t-\t");
1188                 else
1189                         fprintf(options->file,
1190                                 "%d\t%d\t", file->added, file->deleted);
1191                 if (options->line_termination) {
1192                         fill_print_name(file);
1193                         if (!file->is_renamed)
1194                                 write_name_quoted(file->name, options->file,
1195                                                   options->line_termination);
1196                         else {
1197                                 fputs(file->print_name, options->file);
1198                                 putc(options->line_termination, options->file);
1199                         }
1200                 } else {
1201                         if (file->is_renamed) {
1202                                 putc('\0', options->file);
1203                                 write_name_quoted(file->from_name, options->file, '\0');
1204                         }
1205                         write_name_quoted(file->name, options->file, '\0');
1206                 }
1207         }
1208 }
1209
1210 struct dirstat_file {
1211         const char *name;
1212         unsigned long changed;
1213 };
1214
1215 struct dirstat_dir {
1216         struct dirstat_file *files;
1217         int alloc, nr, percent, cumulative;
1218 };
1219
1220 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1221 {
1222         unsigned long this_dir = 0;
1223         unsigned int sources = 0;
1224
1225         while (dir->nr) {
1226                 struct dirstat_file *f = dir->files;
1227                 int namelen = strlen(f->name);
1228                 unsigned long this;
1229                 char *slash;
1230
1231                 if (namelen < baselen)
1232                         break;
1233                 if (memcmp(f->name, base, baselen))
1234                         break;
1235                 slash = strchr(f->name + baselen, '/');
1236                 if (slash) {
1237                         int newbaselen = slash + 1 - f->name;
1238                         this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1239                         sources++;
1240                 } else {
1241                         this = f->changed;
1242                         dir->files++;
1243                         dir->nr--;
1244                         sources += 2;
1245                 }
1246                 this_dir += this;
1247         }
1248
1249         /*
1250          * We don't report dirstat's for
1251          *  - the top level
1252          *  - or cases where everything came from a single directory
1253          *    under this directory (sources == 1).
1254          */
1255         if (baselen && sources != 1) {
1256                 int permille = this_dir * 1000 / changed;
1257                 if (permille) {
1258                         int percent = permille / 10;
1259                         if (percent >= dir->percent) {
1260                                 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1261                                 if (!dir->cumulative)
1262                                         return 0;
1263                         }
1264                 }
1265         }
1266         return this_dir;
1267 }
1268
1269 static int dirstat_compare(const void *_a, const void *_b)
1270 {
1271         const struct dirstat_file *a = _a;
1272         const struct dirstat_file *b = _b;
1273         return strcmp(a->name, b->name);
1274 }
1275
1276 static void show_dirstat(struct diff_options *options)
1277 {
1278         int i;
1279         unsigned long changed;
1280         struct dirstat_dir dir;
1281         struct diff_queue_struct *q = &diff_queued_diff;
1282
1283         dir.files = NULL;
1284         dir.alloc = 0;
1285         dir.nr = 0;
1286         dir.percent = options->dirstat_percent;
1287         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1288
1289         changed = 0;
1290         for (i = 0; i < q->nr; i++) {
1291                 struct diff_filepair *p = q->queue[i];
1292                 const char *name;
1293                 unsigned long copied, added, damage;
1294
1295                 name = p->one->path ? p->one->path : p->two->path;
1296
1297                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1298                         diff_populate_filespec(p->one, 0);
1299                         diff_populate_filespec(p->two, 0);
1300                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1301                                                &copied, &added);
1302                         diff_free_filespec_data(p->one);
1303                         diff_free_filespec_data(p->two);
1304                 } else if (DIFF_FILE_VALID(p->one)) {
1305                         diff_populate_filespec(p->one, 1);
1306                         copied = added = 0;
1307                         diff_free_filespec_data(p->one);
1308                 } else if (DIFF_FILE_VALID(p->two)) {
1309                         diff_populate_filespec(p->two, 1);
1310                         copied = 0;
1311                         added = p->two->size;
1312                         diff_free_filespec_data(p->two);
1313                 } else
1314                         continue;
1315
1316                 /*
1317                  * Original minus copied is the removed material,
1318                  * added is the new material.  They are both damages
1319                  * made to the preimage. In --dirstat-by-file mode, count
1320                  * damaged files, not damaged lines. This is done by
1321                  * counting only a single damaged line per file.
1322                  */
1323                 damage = (p->one->size - copied) + added;
1324                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1325                         damage = 1;
1326
1327                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1328                 dir.files[dir.nr].name = name;
1329                 dir.files[dir.nr].changed = damage;
1330                 changed += damage;
1331                 dir.nr++;
1332         }
1333
1334         /* This can happen even with many files, if everything was renames */
1335         if (!changed)
1336                 return;
1337
1338         /* Show all directories with more than x% of the changes */
1339         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1340         gather_dirstat(options->file, &dir, changed, "", 0);
1341 }
1342
1343 static void free_diffstat_info(struct diffstat_t *diffstat)
1344 {
1345         int i;
1346         for (i = 0; i < diffstat->nr; i++) {
1347                 struct diffstat_file *f = diffstat->files[i];
1348                 if (f->name != f->print_name)
1349                         free(f->print_name);
1350                 free(f->name);
1351                 free(f->from_name);
1352                 free(f);
1353         }
1354         free(diffstat->files);
1355 }
1356
1357 struct checkdiff_t {
1358         const char *filename;
1359         int lineno;
1360         struct diff_options *o;
1361         unsigned ws_rule;
1362         unsigned status;
1363 };
1364
1365 static int is_conflict_marker(const char *line, unsigned long len)
1366 {
1367         char firstchar;
1368         int cnt;
1369
1370         if (len < 8)
1371                 return 0;
1372         firstchar = line[0];
1373         switch (firstchar) {
1374         case '=': case '>': case '<':
1375                 break;
1376         default:
1377                 return 0;
1378         }
1379         for (cnt = 1; cnt < 7; cnt++)
1380                 if (line[cnt] != firstchar)
1381                         return 0;
1382         /* line[0] thru line[6] are same as firstchar */
1383         if (firstchar == '=') {
1384                 /* divider between ours and theirs? */
1385                 if (len != 8 || line[7] != '\n')
1386                         return 0;
1387         } else if (len < 8 || !isspace(line[7])) {
1388                 /* not divider before ours nor after theirs */
1389                 return 0;
1390         }
1391         return 1;
1392 }
1393
1394 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1395 {
1396         struct checkdiff_t *data = priv;
1397         int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1398         const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1399         const char *reset = diff_get_color(color_diff, DIFF_RESET);
1400         const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1401         char *err;
1402
1403         if (line[0] == '+') {
1404                 unsigned bad;
1405                 data->lineno++;
1406                 if (is_conflict_marker(line + 1, len - 1)) {
1407                         data->status |= 1;
1408                         fprintf(data->o->file,
1409                                 "%s:%d: leftover conflict marker\n",
1410                                 data->filename, data->lineno);
1411                 }
1412                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1413                 if (!bad)
1414                         return;
1415                 data->status |= bad;
1416                 err = whitespace_error_string(bad);
1417                 fprintf(data->o->file, "%s:%d: %s.\n",
1418                         data->filename, data->lineno, err);
1419                 free(err);
1420                 emit_line(data->o->file, set, reset, line, 1);
1421                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1422                               data->o->file, set, reset, ws);
1423         } else if (line[0] == ' ') {
1424                 data->lineno++;
1425         } else if (line[0] == '@') {
1426                 char *plus = strchr(line, '+');
1427                 if (plus)
1428                         data->lineno = strtol(plus, NULL, 10) - 1;
1429                 else
1430                         die("invalid diff");
1431         }
1432 }
1433
1434 static unsigned char *deflate_it(char *data,
1435                                  unsigned long size,
1436                                  unsigned long *result_size)
1437 {
1438         int bound;
1439         unsigned char *deflated;
1440         z_stream stream;
1441
1442         memset(&stream, 0, sizeof(stream));
1443         deflateInit(&stream, zlib_compression_level);
1444         bound = deflateBound(&stream, size);
1445         deflated = xmalloc(bound);
1446         stream.next_out = deflated;
1447         stream.avail_out = bound;
1448
1449         stream.next_in = (unsigned char *)data;
1450         stream.avail_in = size;
1451         while (deflate(&stream, Z_FINISH) == Z_OK)
1452                 ; /* nothing */
1453         deflateEnd(&stream);
1454         *result_size = stream.total_out;
1455         return deflated;
1456 }
1457
1458 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1459 {
1460         void *cp;
1461         void *delta;
1462         void *deflated;
1463         void *data;
1464         unsigned long orig_size;
1465         unsigned long delta_size;
1466         unsigned long deflate_size;
1467         unsigned long data_size;
1468
1469         /* We could do deflated delta, or we could do just deflated two,
1470          * whichever is smaller.
1471          */
1472         delta = NULL;
1473         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1474         if (one->size && two->size) {
1475                 delta = diff_delta(one->ptr, one->size,
1476                                    two->ptr, two->size,
1477                                    &delta_size, deflate_size);
1478                 if (delta) {
1479                         void *to_free = delta;
1480                         orig_size = delta_size;
1481                         delta = deflate_it(delta, delta_size, &delta_size);
1482                         free(to_free);
1483                 }
1484         }
1485
1486         if (delta && delta_size < deflate_size) {
1487                 fprintf(file, "delta %lu\n", orig_size);
1488                 free(deflated);
1489                 data = delta;
1490                 data_size = delta_size;
1491         }
1492         else {
1493                 fprintf(file, "literal %lu\n", two->size);
1494                 free(delta);
1495                 data = deflated;
1496                 data_size = deflate_size;
1497         }
1498
1499         /* emit data encoded in base85 */
1500         cp = data;
1501         while (data_size) {
1502                 int bytes = (52 < data_size) ? 52 : data_size;
1503                 char line[70];
1504                 data_size -= bytes;
1505                 if (bytes <= 26)
1506                         line[0] = bytes + 'A' - 1;
1507                 else
1508                         line[0] = bytes - 26 + 'a' - 1;
1509                 encode_85(line + 1, cp, bytes);
1510                 cp = (char *) cp + bytes;
1511                 fputs(line, file);
1512                 fputc('\n', file);
1513         }
1514         fprintf(file, "\n");
1515         free(data);
1516 }
1517
1518 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1519 {
1520         fprintf(file, "GIT binary patch\n");
1521         emit_binary_diff_body(file, one, two);
1522         emit_binary_diff_body(file, two, one);
1523 }
1524
1525 static void diff_filespec_load_driver(struct diff_filespec *one)
1526 {
1527         if (!one->driver)
1528                 one->driver = userdiff_find_by_path(one->path);
1529         if (!one->driver)
1530                 one->driver = userdiff_find_by_name("default");
1531 }
1532
1533 int diff_filespec_is_binary(struct diff_filespec *one)
1534 {
1535         if (one->is_binary == -1) {
1536                 diff_filespec_load_driver(one);
1537                 if (one->driver->binary != -1)
1538                         one->is_binary = one->driver->binary;
1539                 else {
1540                         if (!one->data && DIFF_FILE_VALID(one))
1541                                 diff_populate_filespec(one, 0);
1542                         if (one->data)
1543                                 one->is_binary = buffer_is_binary(one->data,
1544                                                 one->size);
1545                         if (one->is_binary == -1)
1546                                 one->is_binary = 0;
1547                 }
1548         }
1549         return one->is_binary;
1550 }
1551
1552 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1553 {
1554         diff_filespec_load_driver(one);
1555         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1556 }
1557
1558 static const char *userdiff_word_regex(struct diff_filespec *one)
1559 {
1560         diff_filespec_load_driver(one);
1561         return one->driver->word_regex;
1562 }
1563
1564 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1565 {
1566         if (!options->a_prefix)
1567                 options->a_prefix = a;
1568         if (!options->b_prefix)
1569                 options->b_prefix = b;
1570 }
1571
1572 static struct userdiff_driver *get_textconv(struct diff_filespec *one)
1573 {
1574         if (!DIFF_FILE_VALID(one))
1575                 return NULL;
1576         if (!S_ISREG(one->mode))
1577                 return NULL;
1578         diff_filespec_load_driver(one);
1579         if (!one->driver->textconv)
1580                 return NULL;
1581
1582         if (one->driver->textconv_want_cache && !one->driver->textconv_cache) {
1583                 struct notes_cache *c = xmalloc(sizeof(*c));
1584                 struct strbuf name = STRBUF_INIT;
1585
1586                 strbuf_addf(&name, "textconv/%s", one->driver->name);
1587                 notes_cache_init(c, name.buf, one->driver->textconv);
1588                 one->driver->textconv_cache = c;
1589         }
1590
1591         return one->driver;
1592 }
1593
1594 static void builtin_diff(const char *name_a,
1595                          const char *name_b,
1596                          struct diff_filespec *one,
1597                          struct diff_filespec *two,
1598                          const char *xfrm_msg,
1599                          struct diff_options *o,
1600                          int complete_rewrite)
1601 {
1602         mmfile_t mf1, mf2;
1603         const char *lbl[2];
1604         char *a_one, *b_two;
1605         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1606         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1607         const char *a_prefix, *b_prefix;
1608         struct userdiff_driver *textconv_one = NULL;
1609         struct userdiff_driver *textconv_two = NULL;
1610         struct strbuf header = STRBUF_INIT;
1611
1612         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
1613                         (!one->mode || S_ISGITLINK(one->mode)) &&
1614                         (!two->mode || S_ISGITLINK(two->mode))) {
1615                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
1616                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
1617                 show_submodule_summary(o->file, one ? one->path : two->path,
1618                                 one->sha1, two->sha1, two->dirty_submodule,
1619                                 del, add, reset);
1620                 return;
1621         }
1622
1623         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
1624                 textconv_one = get_textconv(one);
1625                 textconv_two = get_textconv(two);
1626         }
1627
1628         diff_set_mnemonic_prefix(o, "a/", "b/");
1629         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1630                 a_prefix = o->b_prefix;
1631                 b_prefix = o->a_prefix;
1632         } else {
1633                 a_prefix = o->a_prefix;
1634                 b_prefix = o->b_prefix;
1635         }
1636
1637         /* Never use a non-valid filename anywhere if at all possible */
1638         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1639         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1640
1641         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1642         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1643         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1644         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1645         strbuf_addf(&header, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1646         if (lbl[0][0] == '/') {
1647                 /* /dev/null */
1648                 strbuf_addf(&header, "%snew file mode %06o%s\n", set, two->mode, reset);
1649                 if (xfrm_msg && xfrm_msg[0])
1650                         strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1651         }
1652         else if (lbl[1][0] == '/') {
1653                 strbuf_addf(&header, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1654                 if (xfrm_msg && xfrm_msg[0])
1655                         strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1656         }
1657         else {
1658                 if (one->mode != two->mode) {
1659                         strbuf_addf(&header, "%sold mode %06o%s\n", set, one->mode, reset);
1660                         strbuf_addf(&header, "%snew mode %06o%s\n", set, two->mode, reset);
1661                 }
1662                 if (xfrm_msg && xfrm_msg[0])
1663                         strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1664
1665                 /*
1666                  * we do not run diff between different kind
1667                  * of objects.
1668                  */
1669                 if ((one->mode ^ two->mode) & S_IFMT)
1670                         goto free_ab_and_return;
1671                 if (complete_rewrite &&
1672                     (textconv_one || !diff_filespec_is_binary(one)) &&
1673                     (textconv_two || !diff_filespec_is_binary(two))) {
1674                         fprintf(o->file, "%s", header.buf);
1675                         strbuf_reset(&header);
1676                         emit_rewrite_diff(name_a, name_b, one, two,
1677                                                 textconv_one, textconv_two, o);
1678                         o->found_changes = 1;
1679                         goto free_ab_and_return;
1680                 }
1681         }
1682
1683         if (!DIFF_OPT_TST(o, TEXT) &&
1684             ( (!textconv_one && diff_filespec_is_binary(one)) ||
1685               (!textconv_two && diff_filespec_is_binary(two)) )) {
1686                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1687                         die("unable to read files to diff");
1688                 /* Quite common confusing case */
1689                 if (mf1.size == mf2.size &&
1690                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1691                         goto free_ab_and_return;
1692                 fprintf(o->file, "%s", header.buf);
1693                 strbuf_reset(&header);
1694                 if (DIFF_OPT_TST(o, BINARY))
1695                         emit_binary_diff(o->file, &mf1, &mf2);
1696                 else
1697                         fprintf(o->file, "Binary files %s and %s differ\n",
1698                                 lbl[0], lbl[1]);
1699                 o->found_changes = 1;
1700         }
1701         else {
1702                 /* Crazy xdl interfaces.. */
1703                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1704                 xpparam_t xpp;
1705                 xdemitconf_t xecfg;
1706                 xdemitcb_t ecb;
1707                 struct emit_callback ecbdata;
1708                 const struct userdiff_funcname *pe;
1709
1710                 if (!DIFF_XDL_TST(o, WHITESPACE_FLAGS)) {
1711                         fprintf(o->file, "%s", header.buf);
1712                         strbuf_reset(&header);
1713                 }
1714
1715                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
1716                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
1717
1718                 pe = diff_funcname_pattern(one);
1719                 if (!pe)
1720                         pe = diff_funcname_pattern(two);
1721
1722                 memset(&xpp, 0, sizeof(xpp));
1723                 memset(&xecfg, 0, sizeof(xecfg));
1724                 memset(&ecbdata, 0, sizeof(ecbdata));
1725                 ecbdata.label_path = lbl;
1726                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1727                 ecbdata.found_changesp = &o->found_changes;
1728                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1729                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
1730                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
1731                 ecbdata.file = o->file;
1732                 ecbdata.header = header.len ? &header : NULL;
1733                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1734                 xecfg.ctxlen = o->context;
1735                 xecfg.interhunkctxlen = o->interhunkcontext;
1736                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1737                 if (pe)
1738                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1739                 if (!diffopts)
1740                         ;
1741                 else if (!prefixcmp(diffopts, "--unified="))
1742                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1743                 else if (!prefixcmp(diffopts, "-u"))
1744                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1745                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1746                         ecbdata.diff_words =
1747                                 xcalloc(1, sizeof(struct diff_words_data));
1748                         ecbdata.diff_words->file = o->file;
1749                         if (!o->word_regex)
1750                                 o->word_regex = userdiff_word_regex(one);
1751                         if (!o->word_regex)
1752                                 o->word_regex = userdiff_word_regex(two);
1753                         if (!o->word_regex)
1754                                 o->word_regex = diff_word_regex_cfg;
1755                         if (o->word_regex) {
1756                                 ecbdata.diff_words->word_regex = (regex_t *)
1757                                         xmalloc(sizeof(regex_t));
1758                                 if (regcomp(ecbdata.diff_words->word_regex,
1759                                                 o->word_regex,
1760                                                 REG_EXTENDED | REG_NEWLINE))
1761                                         die ("Invalid regular expression: %s",
1762                                                         o->word_regex);
1763                         }
1764                 }
1765                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1766                               &xpp, &xecfg, &ecb);
1767                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1768                         free_diff_words_data(&ecbdata);
1769                 if (textconv_one)
1770                         free(mf1.ptr);
1771                 if (textconv_two)
1772                         free(mf2.ptr);
1773                 xdiff_clear_find_func(&xecfg);
1774         }
1775
1776  free_ab_and_return:
1777         strbuf_release(&header);
1778         diff_free_filespec_data(one);
1779         diff_free_filespec_data(two);
1780         free(a_one);
1781         free(b_two);
1782         return;
1783 }
1784
1785 static void builtin_diffstat(const char *name_a, const char *name_b,
1786                              struct diff_filespec *one,
1787                              struct diff_filespec *two,
1788                              struct diffstat_t *diffstat,
1789                              struct diff_options *o,
1790                              int complete_rewrite)
1791 {
1792         mmfile_t mf1, mf2;
1793         struct diffstat_file *data;
1794
1795         data = diffstat_add(diffstat, name_a, name_b);
1796
1797         if (!one || !two) {
1798                 data->is_unmerged = 1;
1799                 return;
1800         }
1801         if (complete_rewrite) {
1802                 diff_populate_filespec(one, 0);
1803                 diff_populate_filespec(two, 0);
1804                 data->deleted = count_lines(one->data, one->size);
1805                 data->added = count_lines(two->data, two->size);
1806                 goto free_and_return;
1807         }
1808         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1809                 die("unable to read files to diff");
1810
1811         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1812                 data->is_binary = 1;
1813                 data->added = mf2.size;
1814                 data->deleted = mf1.size;
1815         } else {
1816                 /* Crazy xdl interfaces.. */
1817                 xpparam_t xpp;
1818                 xdemitconf_t xecfg;
1819                 xdemitcb_t ecb;
1820
1821                 memset(&xpp, 0, sizeof(xpp));
1822                 memset(&xecfg, 0, sizeof(xecfg));
1823                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1824                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1825                               &xpp, &xecfg, &ecb);
1826         }
1827
1828  free_and_return:
1829         diff_free_filespec_data(one);
1830         diff_free_filespec_data(two);
1831 }
1832
1833 static void builtin_checkdiff(const char *name_a, const char *name_b,
1834                               const char *attr_path,
1835                               struct diff_filespec *one,
1836                               struct diff_filespec *two,
1837                               struct diff_options *o)
1838 {
1839         mmfile_t mf1, mf2;
1840         struct checkdiff_t data;
1841
1842         if (!two)
1843                 return;
1844
1845         memset(&data, 0, sizeof(data));
1846         data.filename = name_b ? name_b : name_a;
1847         data.lineno = 0;
1848         data.o = o;
1849         data.ws_rule = whitespace_rule(attr_path);
1850
1851         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1852                 die("unable to read files to diff");
1853
1854         /*
1855          * All the other codepaths check both sides, but not checking
1856          * the "old" side here is deliberate.  We are checking the newly
1857          * introduced changes, and as long as the "new" side is text, we
1858          * can and should check what it introduces.
1859          */
1860         if (diff_filespec_is_binary(two))
1861                 goto free_and_return;
1862         else {
1863                 /* Crazy xdl interfaces.. */
1864                 xpparam_t xpp;
1865                 xdemitconf_t xecfg;
1866                 xdemitcb_t ecb;
1867
1868                 memset(&xpp, 0, sizeof(xpp));
1869                 memset(&xecfg, 0, sizeof(xecfg));
1870                 xecfg.ctxlen = 1; /* at least one context line */
1871                 xpp.flags = XDF_NEED_MINIMAL;
1872                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1873                               &xpp, &xecfg, &ecb);
1874
1875                 if (data.ws_rule & WS_BLANK_AT_EOF) {
1876                         struct emit_callback ecbdata;
1877                         int blank_at_eof;
1878
1879                         ecbdata.ws_rule = data.ws_rule;
1880                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
1881                         blank_at_eof = ecbdata.blank_at_eof_in_preimage;
1882
1883                         if (blank_at_eof) {
1884                                 static char *err;
1885                                 if (!err)
1886                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
1887                                 fprintf(o->file, "%s:%d: %s.\n",
1888                                         data.filename, blank_at_eof, err);
1889                                 data.status = 1; /* report errors */
1890                         }
1891                 }
1892         }
1893  free_and_return:
1894         diff_free_filespec_data(one);
1895         diff_free_filespec_data(two);
1896         if (data.status)
1897                 DIFF_OPT_SET(o, CHECK_FAILED);
1898 }
1899
1900 struct diff_filespec *alloc_filespec(const char *path)
1901 {
1902         int namelen = strlen(path);
1903         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1904
1905         memset(spec, 0, sizeof(*spec));
1906         spec->path = (char *)(spec + 1);
1907         memcpy(spec->path, path, namelen+1);
1908         spec->count = 1;
1909         spec->is_binary = -1;
1910         return spec;
1911 }
1912
1913 void free_filespec(struct diff_filespec *spec)
1914 {
1915         if (!--spec->count) {
1916                 diff_free_filespec_data(spec);
1917                 free(spec);
1918         }
1919 }
1920
1921 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1922                    unsigned short mode)
1923 {
1924         if (mode) {
1925                 spec->mode = canon_mode(mode);
1926                 hashcpy(spec->sha1, sha1);
1927                 spec->sha1_valid = !is_null_sha1(sha1);
1928         }
1929 }
1930
1931 /*
1932  * Given a name and sha1 pair, if the index tells us the file in
1933  * the work tree has that object contents, return true, so that
1934  * prepare_temp_file() does not have to inflate and extract.
1935  */
1936 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1937 {
1938         struct cache_entry *ce;
1939         struct stat st;
1940         int pos, len;
1941
1942         /*
1943          * We do not read the cache ourselves here, because the
1944          * benchmark with my previous version that always reads cache
1945          * shows that it makes things worse for diff-tree comparing
1946          * two linux-2.6 kernel trees in an already checked out work
1947          * tree.  This is because most diff-tree comparisons deal with
1948          * only a small number of files, while reading the cache is
1949          * expensive for a large project, and its cost outweighs the
1950          * savings we get by not inflating the object to a temporary
1951          * file.  Practically, this code only helps when we are used
1952          * by diff-cache --cached, which does read the cache before
1953          * calling us.
1954          */
1955         if (!active_cache)
1956                 return 0;
1957
1958         /* We want to avoid the working directory if our caller
1959          * doesn't need the data in a normal file, this system
1960          * is rather slow with its stat/open/mmap/close syscalls,
1961          * and the object is contained in a pack file.  The pack
1962          * is probably already open and will be faster to obtain
1963          * the data through than the working directory.  Loose
1964          * objects however would tend to be slower as they need
1965          * to be individually opened and inflated.
1966          */
1967         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
1968                 return 0;
1969
1970         len = strlen(name);
1971         pos = cache_name_pos(name, len);
1972         if (pos < 0)
1973                 return 0;
1974         ce = active_cache[pos];
1975
1976         /*
1977          * This is not the sha1 we are looking for, or
1978          * unreusable because it is not a regular file.
1979          */
1980         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1981                 return 0;
1982
1983         /*
1984          * If ce is marked as "assume unchanged", there is no
1985          * guarantee that work tree matches what we are looking for.
1986          */
1987         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
1988                 return 0;
1989
1990         /*
1991          * If ce matches the file in the work tree, we can reuse it.
1992          */
1993         if (ce_uptodate(ce) ||
1994             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1995                 return 1;
1996
1997         return 0;
1998 }
1999
2000 static int populate_from_stdin(struct diff_filespec *s)
2001 {
2002         struct strbuf buf = STRBUF_INIT;
2003         size_t size = 0;
2004
2005         if (strbuf_read(&buf, 0, 0) < 0)
2006                 return error("error while reading from stdin %s",
2007                                      strerror(errno));
2008
2009         s->should_munmap = 0;
2010         s->data = strbuf_detach(&buf, &size);
2011         s->size = size;
2012         s->should_free = 1;
2013         return 0;
2014 }
2015
2016 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2017 {
2018         int len;
2019         char *data = xmalloc(100), *dirty = "";
2020
2021         /* Are we looking at the work tree? */
2022         if (s->dirty_submodule)
2023                 dirty = "-dirty";
2024
2025         len = snprintf(data, 100,
2026                        "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2027         s->data = data;
2028         s->size = len;
2029         s->should_free = 1;
2030         if (size_only) {
2031                 s->data = NULL;
2032                 free(data);
2033         }
2034         return 0;
2035 }
2036
2037 /*
2038  * While doing rename detection and pickaxe operation, we may need to
2039  * grab the data for the blob (or file) for our own in-core comparison.
2040  * diff_filespec has data and size fields for this purpose.
2041  */
2042 int diff_populate_filespec(struct diff_filespec *s, int size_only)
2043 {
2044         int err = 0;
2045         if (!DIFF_FILE_VALID(s))
2046                 die("internal error: asking to populate invalid file.");
2047         if (S_ISDIR(s->mode))
2048                 return -1;
2049
2050         if (s->data)
2051                 return 0;
2052
2053         if (size_only && 0 < s->size)
2054                 return 0;
2055
2056         if (S_ISGITLINK(s->mode))
2057                 return diff_populate_gitlink(s, size_only);
2058
2059         if (!s->sha1_valid ||
2060             reuse_worktree_file(s->path, s->sha1, 0)) {
2061                 struct strbuf buf = STRBUF_INIT;
2062                 struct stat st;
2063                 int fd;
2064
2065                 if (!strcmp(s->path, "-"))
2066                         return populate_from_stdin(s);
2067
2068                 if (lstat(s->path, &st) < 0) {
2069                         if (errno == ENOENT) {
2070                         err_empty:
2071                                 err = -1;
2072                         empty:
2073                                 s->data = (char *)"";
2074                                 s->size = 0;
2075                                 return err;
2076                         }
2077                 }
2078                 s->size = xsize_t(st.st_size);
2079                 if (!s->size)
2080                         goto empty;
2081                 if (S_ISLNK(st.st_mode)) {
2082                         struct strbuf sb = STRBUF_INIT;
2083
2084                         if (strbuf_readlink(&sb, s->path, s->size))
2085                                 goto err_empty;
2086                         s->size = sb.len;
2087                         s->data = strbuf_detach(&sb, NULL);
2088                         s->should_free = 1;
2089                         return 0;
2090                 }
2091                 if (size_only)
2092                         return 0;
2093                 fd = open(s->path, O_RDONLY);
2094                 if (fd < 0)
2095                         goto err_empty;
2096                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2097                 close(fd);
2098                 s->should_munmap = 1;
2099
2100                 /*
2101                  * Convert from working tree format to canonical git format
2102                  */
2103                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2104                         size_t size = 0;
2105                         munmap(s->data, s->size);
2106                         s->should_munmap = 0;
2107                         s->data = strbuf_detach(&buf, &size);
2108                         s->size = size;
2109                         s->should_free = 1;
2110                 }
2111         }
2112         else {
2113                 enum object_type type;
2114                 if (size_only)
2115                         type = sha1_object_info(s->sha1, &s->size);
2116                 else {
2117                         s->data = read_sha1_file(s->sha1, &type, &s->size);
2118                         s->should_free = 1;
2119                 }
2120         }
2121         return 0;
2122 }
2123
2124 void diff_free_filespec_blob(struct diff_filespec *s)
2125 {
2126         if (s->should_free)
2127                 free(s->data);
2128         else if (s->should_munmap)
2129                 munmap(s->data, s->size);
2130
2131         if (s->should_free || s->should_munmap) {
2132                 s->should_free = s->should_munmap = 0;
2133                 s->data = NULL;
2134         }
2135 }
2136
2137 void diff_free_filespec_data(struct diff_filespec *s)
2138 {
2139         diff_free_filespec_blob(s);
2140         free(s->cnt_data);
2141         s->cnt_data = NULL;
2142 }
2143
2144 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2145                            void *blob,
2146                            unsigned long size,
2147                            const unsigned char *sha1,
2148                            int mode)
2149 {
2150         int fd;
2151         struct strbuf buf = STRBUF_INIT;
2152         struct strbuf template = STRBUF_INIT;
2153         char *path_dup = xstrdup(path);
2154         const char *base = basename(path_dup);
2155
2156         /* Generate "XXXXXX_basename.ext" */
2157         strbuf_addstr(&template, "XXXXXX_");
2158         strbuf_addstr(&template, base);
2159
2160         fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2161                         strlen(base) + 1);
2162         if (fd < 0)
2163                 die_errno("unable to create temp-file");
2164         if (convert_to_working_tree(path,
2165                         (const char *)blob, (size_t)size, &buf)) {
2166                 blob = buf.buf;
2167                 size = buf.len;
2168         }
2169         if (write_in_full(fd, blob, size) != size)
2170                 die_errno("unable to write temp-file");
2171         close(fd);
2172         temp->name = temp->tmp_path;
2173         strcpy(temp->hex, sha1_to_hex(sha1));
2174         temp->hex[40] = 0;
2175         sprintf(temp->mode, "%06o", mode);
2176         strbuf_release(&buf);
2177         strbuf_release(&template);
2178         free(path_dup);
2179 }
2180
2181 static struct diff_tempfile *prepare_temp_file(const char *name,
2182                 struct diff_filespec *one)
2183 {
2184         struct diff_tempfile *temp = claim_diff_tempfile();
2185
2186         if (!DIFF_FILE_VALID(one)) {
2187         not_a_valid_file:
2188                 /* A '-' entry produces this for file-2, and
2189                  * a '+' entry produces this for file-1.
2190                  */
2191                 temp->name = "/dev/null";
2192                 strcpy(temp->hex, ".");
2193                 strcpy(temp->mode, ".");
2194                 return temp;
2195         }
2196
2197         if (!remove_tempfile_installed) {
2198                 atexit(remove_tempfile);
2199                 sigchain_push_common(remove_tempfile_on_signal);
2200                 remove_tempfile_installed = 1;
2201         }
2202
2203         if (!one->sha1_valid ||
2204             reuse_worktree_file(name, one->sha1, 1)) {
2205                 struct stat st;
2206                 if (lstat(name, &st) < 0) {
2207                         if (errno == ENOENT)
2208                                 goto not_a_valid_file;
2209                         die_errno("stat(%s)", name);
2210                 }
2211                 if (S_ISLNK(st.st_mode)) {
2212                         struct strbuf sb = STRBUF_INIT;
2213                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2214                                 die_errno("readlink(%s)", name);
2215                         prep_temp_blob(name, temp, sb.buf, sb.len,
2216                                        (one->sha1_valid ?
2217                                         one->sha1 : null_sha1),
2218                                        (one->sha1_valid ?
2219                                         one->mode : S_IFLNK));
2220                         strbuf_release(&sb);
2221                 }
2222                 else {
2223                         /* we can borrow from the file in the work tree */
2224                         temp->name = name;
2225                         if (!one->sha1_valid)
2226                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2227                         else
2228                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2229                         /* Even though we may sometimes borrow the
2230                          * contents from the work tree, we always want
2231                          * one->mode.  mode is trustworthy even when
2232                          * !(one->sha1_valid), as long as
2233                          * DIFF_FILE_VALID(one).
2234                          */
2235                         sprintf(temp->mode, "%06o", one->mode);
2236                 }
2237                 return temp;
2238         }
2239         else {
2240                 if (diff_populate_filespec(one, 0))
2241                         die("cannot read data blob for %s", one->path);
2242                 prep_temp_blob(name, temp, one->data, one->size,
2243                                one->sha1, one->mode);
2244         }
2245         return temp;
2246 }
2247
2248 /* An external diff command takes:
2249  *
2250  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2251  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2252  *
2253  */
2254 static void run_external_diff(const char *pgm,
2255                               const char *name,
2256                               const char *other,
2257                               struct diff_filespec *one,
2258                               struct diff_filespec *two,
2259                               const char *xfrm_msg,
2260                               int complete_rewrite)
2261 {
2262         const char *spawn_arg[10];
2263         int retval;
2264         const char **arg = &spawn_arg[0];
2265
2266         if (one && two) {
2267                 struct diff_tempfile *temp_one, *temp_two;
2268                 const char *othername = (other ? other : name);
2269                 temp_one = prepare_temp_file(name, one);
2270                 temp_two = prepare_temp_file(othername, two);
2271                 *arg++ = pgm;
2272                 *arg++ = name;
2273                 *arg++ = temp_one->name;
2274                 *arg++ = temp_one->hex;
2275                 *arg++ = temp_one->mode;
2276                 *arg++ = temp_two->name;
2277                 *arg++ = temp_two->hex;
2278                 *arg++ = temp_two->mode;
2279                 if (other) {
2280                         *arg++ = other;
2281                         *arg++ = xfrm_msg;
2282                 }
2283         } else {
2284                 *arg++ = pgm;
2285                 *arg++ = name;
2286         }
2287         *arg = NULL;
2288         fflush(NULL);
2289         retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2290         remove_tempfile();
2291         if (retval) {
2292                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2293                 exit(1);
2294         }
2295 }
2296
2297 static int similarity_index(struct diff_filepair *p)
2298 {
2299         return p->score * 100 / MAX_SCORE;
2300 }
2301
2302 static void fill_metainfo(struct strbuf *msg,
2303                           const char *name,
2304                           const char *other,
2305                           struct diff_filespec *one,
2306                           struct diff_filespec *two,
2307                           struct diff_options *o,
2308                           struct diff_filepair *p)
2309 {
2310         strbuf_init(msg, PATH_MAX * 2 + 300);
2311         switch (p->status) {
2312         case DIFF_STATUS_COPIED:
2313                 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2314                 strbuf_addstr(msg, "\ncopy from ");
2315                 quote_c_style(name, msg, NULL, 0);
2316                 strbuf_addstr(msg, "\ncopy to ");
2317                 quote_c_style(other, msg, NULL, 0);
2318                 strbuf_addch(msg, '\n');
2319                 break;
2320         case DIFF_STATUS_RENAMED:
2321                 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2322                 strbuf_addstr(msg, "\nrename from ");
2323                 quote_c_style(name, msg, NULL, 0);
2324                 strbuf_addstr(msg, "\nrename to ");
2325                 quote_c_style(other, msg, NULL, 0);
2326                 strbuf_addch(msg, '\n');
2327                 break;
2328         case DIFF_STATUS_MODIFIED:
2329                 if (p->score) {
2330                         strbuf_addf(msg, "dissimilarity index %d%%\n",
2331                                     similarity_index(p));
2332                         break;
2333                 }
2334                 /* fallthru */
2335         default:
2336                 /* nothing */
2337                 ;
2338         }
2339         if (one && two && hashcmp(one->sha1, two->sha1)) {
2340                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2341
2342                 if (DIFF_OPT_TST(o, BINARY)) {
2343                         mmfile_t mf;
2344                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2345                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2346                                 abbrev = 40;
2347                 }
2348                 strbuf_addf(msg, "index %.*s..%.*s",
2349                             abbrev, sha1_to_hex(one->sha1),
2350                             abbrev, sha1_to_hex(two->sha1));
2351                 if (one->mode == two->mode)
2352                         strbuf_addf(msg, " %06o", one->mode);
2353                 strbuf_addch(msg, '\n');
2354         }
2355         if (msg->len)
2356                 strbuf_setlen(msg, msg->len - 1);
2357 }
2358
2359 static void run_diff_cmd(const char *pgm,
2360                          const char *name,
2361                          const char *other,
2362                          const char *attr_path,
2363                          struct diff_filespec *one,
2364                          struct diff_filespec *two,
2365                          struct strbuf *msg,
2366                          struct diff_options *o,
2367                          struct diff_filepair *p)
2368 {
2369         const char *xfrm_msg = NULL;
2370         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2371
2372         if (msg) {
2373                 fill_metainfo(msg, name, other, one, two, o, p);
2374                 xfrm_msg = msg->len ? msg->buf : NULL;
2375         }
2376
2377         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2378                 pgm = NULL;
2379         else {
2380                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2381                 if (drv && drv->external)
2382                         pgm = drv->external;
2383         }
2384
2385         if (pgm) {
2386                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2387                                   complete_rewrite);
2388                 return;
2389         }
2390         if (one && two)
2391                 builtin_diff(name, other ? other : name,
2392                              one, two, xfrm_msg, o, complete_rewrite);
2393         else
2394                 fprintf(o->file, "* Unmerged path %s\n", name);
2395 }
2396
2397 static void diff_fill_sha1_info(struct diff_filespec *one)
2398 {
2399         if (DIFF_FILE_VALID(one)) {
2400                 if (!one->sha1_valid) {
2401                         struct stat st;
2402                         if (!strcmp(one->path, "-")) {
2403                                 hashcpy(one->sha1, null_sha1);
2404                                 return;
2405                         }
2406                         if (lstat(one->path, &st) < 0)
2407                                 die_errno("stat '%s'", one->path);
2408                         if (index_path(one->sha1, one->path, &st, 0))
2409                                 die("cannot hash %s", one->path);
2410                 }
2411         }
2412         else
2413                 hashclr(one->sha1);
2414 }
2415
2416 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2417 {
2418         /* Strip the prefix but do not molest /dev/null and absolute paths */
2419         if (*namep && **namep != '/')
2420                 *namep += prefix_length;
2421         if (*otherp && **otherp != '/')
2422                 *otherp += prefix_length;
2423 }
2424
2425 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2426 {
2427         const char *pgm = external_diff();
2428         struct strbuf msg;
2429         struct diff_filespec *one = p->one;
2430         struct diff_filespec *two = p->two;
2431         const char *name;
2432         const char *other;
2433         const char *attr_path;
2434
2435         name  = p->one->path;
2436         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2437         attr_path = name;
2438         if (o->prefix_length)
2439                 strip_prefix(o->prefix_length, &name, &other);
2440
2441         if (DIFF_PAIR_UNMERGED(p)) {
2442                 run_diff_cmd(pgm, name, NULL, attr_path,
2443                              NULL, NULL, NULL, o, p);
2444                 return;
2445         }
2446
2447         diff_fill_sha1_info(one);
2448         diff_fill_sha1_info(two);
2449
2450         if (!pgm &&
2451             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2452             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2453                 /*
2454                  * a filepair that changes between file and symlink
2455                  * needs to be split into deletion and creation.
2456                  */
2457                 struct diff_filespec *null = alloc_filespec(two->path);
2458                 run_diff_cmd(NULL, name, other, attr_path,
2459                              one, null, &msg, o, p);
2460                 free(null);
2461                 strbuf_release(&msg);
2462
2463                 null = alloc_filespec(one->path);
2464                 run_diff_cmd(NULL, name, other, attr_path,
2465                              null, two, &msg, o, p);
2466                 free(null);
2467         }
2468         else
2469                 run_diff_cmd(pgm, name, other, attr_path,
2470                              one, two, &msg, o, p);
2471
2472         strbuf_release(&msg);
2473 }
2474
2475 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2476                          struct diffstat_t *diffstat)
2477 {
2478         const char *name;
2479         const char *other;
2480         int complete_rewrite = 0;
2481
2482         if (DIFF_PAIR_UNMERGED(p)) {
2483                 /* unmerged */
2484                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2485                 return;
2486         }
2487
2488         name = p->one->path;
2489         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2490
2491         if (o->prefix_length)
2492                 strip_prefix(o->prefix_length, &name, &other);
2493
2494         diff_fill_sha1_info(p->one);
2495         diff_fill_sha1_info(p->two);
2496
2497         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2498                 complete_rewrite = 1;
2499         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2500 }
2501
2502 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2503 {
2504         const char *name;
2505         const char *other;
2506         const char *attr_path;
2507
2508         if (DIFF_PAIR_UNMERGED(p)) {
2509                 /* unmerged */
2510                 return;
2511         }
2512
2513         name = p->one->path;
2514         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2515         attr_path = other ? other : name;
2516
2517         if (o->prefix_length)
2518                 strip_prefix(o->prefix_length, &name, &other);
2519
2520         diff_fill_sha1_info(p->one);
2521         diff_fill_sha1_info(p->two);
2522
2523         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2524 }
2525
2526 void diff_setup(struct diff_options *options)
2527 {
2528         memset(options, 0, sizeof(*options));
2529
2530         options->file = stdout;
2531
2532         options->line_termination = '\n';
2533         options->break_opt = -1;
2534         options->rename_limit = -1;
2535         options->dirstat_percent = 3;
2536         options->context = 3;
2537
2538         options->change = diff_change;
2539         options->add_remove = diff_addremove;
2540         if (diff_use_color_default > 0)
2541                 DIFF_OPT_SET(options, COLOR_DIFF);
2542         options->detect_rename = diff_detect_rename_default;
2543
2544         if (!diff_mnemonic_prefix) {
2545                 options->a_prefix = "a/";
2546                 options->b_prefix = "b/";
2547         }
2548 }
2549
2550 int diff_setup_done(struct diff_options *options)
2551 {
2552         int count = 0;
2553
2554         if (options->output_format & DIFF_FORMAT_NAME)
2555                 count++;
2556         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2557                 count++;
2558         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2559                 count++;
2560         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2561                 count++;
2562         if (count > 1)
2563                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2564
2565         /*
2566          * Most of the time we can say "there are changes"
2567          * only by checking if there are changed paths, but
2568          * --ignore-whitespace* options force us to look
2569          * inside contents.
2570          */
2571
2572         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
2573             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
2574             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
2575                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
2576         else
2577                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
2578
2579         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2580                 options->detect_rename = DIFF_DETECT_COPY;
2581
2582         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2583                 options->prefix = NULL;
2584         if (options->prefix)
2585                 options->prefix_length = strlen(options->prefix);
2586         else
2587                 options->prefix_length = 0;
2588
2589         if (options->output_format & (DIFF_FORMAT_NAME |
2590                                       DIFF_FORMAT_NAME_STATUS |
2591                                       DIFF_FORMAT_CHECKDIFF |
2592                                       DIFF_FORMAT_NO_OUTPUT))
2593                 options->output_format &= ~(DIFF_FORMAT_RAW |
2594                                             DIFF_FORMAT_NUMSTAT |
2595                                             DIFF_FORMAT_DIFFSTAT |
2596                                             DIFF_FORMAT_SHORTSTAT |
2597                                             DIFF_FORMAT_DIRSTAT |
2598                                             DIFF_FORMAT_SUMMARY |
2599                                             DIFF_FORMAT_PATCH);
2600
2601         /*
2602          * These cases always need recursive; we do not drop caller-supplied
2603          * recursive bits for other formats here.
2604          */
2605         if (options->output_format & (DIFF_FORMAT_PATCH |
2606                                       DIFF_FORMAT_NUMSTAT |
2607                                       DIFF_FORMAT_DIFFSTAT |
2608                                       DIFF_FORMAT_SHORTSTAT |
2609                                       DIFF_FORMAT_DIRSTAT |
2610                                       DIFF_FORMAT_SUMMARY |
2611                                       DIFF_FORMAT_CHECKDIFF))
2612                 DIFF_OPT_SET(options, RECURSIVE);
2613         /*
2614          * Also pickaxe would not work very well if you do not say recursive
2615          */
2616         if (options->pickaxe)
2617                 DIFF_OPT_SET(options, RECURSIVE);
2618         /*
2619          * When patches are generated, submodules diffed against the work tree
2620          * must be checked for dirtiness too so it can be shown in the output
2621          */
2622         if (options->output_format & DIFF_FORMAT_PATCH)
2623                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
2624
2625         if (options->detect_rename && options->rename_limit < 0)
2626                 options->rename_limit = diff_rename_limit_default;
2627         if (options->setup & DIFF_SETUP_USE_CACHE) {
2628                 if (!active_cache)
2629                         /* read-cache does not die even when it fails
2630                          * so it is safe for us to do this here.  Also
2631                          * it does not smudge active_cache or active_nr
2632                          * when it fails, so we do not have to worry about
2633                          * cleaning it up ourselves either.
2634                          */
2635                         read_cache();
2636         }
2637         if (options->abbrev <= 0 || 40 < options->abbrev)
2638                 options->abbrev = 40; /* full */
2639
2640         /*
2641          * It does not make sense to show the first hit we happened
2642          * to have found.  It does not make sense not to return with
2643          * exit code in such a case either.
2644          */
2645         if (DIFF_OPT_TST(options, QUICK)) {
2646                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2647                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2648         }
2649
2650         return 0;
2651 }
2652
2653 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2654 {
2655         char c, *eq;
2656         int len;
2657
2658         if (*arg != '-')
2659                 return 0;
2660         c = *++arg;
2661         if (!c)
2662                 return 0;
2663         if (c == arg_short) {
2664                 c = *++arg;
2665                 if (!c)
2666                         return 1;
2667                 if (val && isdigit(c)) {
2668                         char *end;
2669                         int n = strtoul(arg, &end, 10);
2670                         if (*end)
2671                                 return 0;
2672                         *val = n;
2673                         return 1;
2674                 }
2675                 return 0;
2676         }
2677         if (c != '-')
2678                 return 0;
2679         arg++;
2680         eq = strchr(arg, '=');
2681         if (eq)
2682                 len = eq - arg;
2683         else
2684                 len = strlen(arg);
2685         if (!len || strncmp(arg, arg_long, len))
2686                 return 0;
2687         if (eq) {
2688                 int n;
2689                 char *end;
2690                 if (!isdigit(*++eq))
2691                         return 0;
2692                 n = strtoul(eq, &end, 10);
2693                 if (*end)
2694                         return 0;
2695                 *val = n;
2696         }
2697         return 1;
2698 }
2699
2700 static int diff_scoreopt_parse(const char *opt);
2701
2702 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2703 {
2704         const char *arg = av[0];
2705
2706         /* Output format options */
2707         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2708                 options->output_format |= DIFF_FORMAT_PATCH;
2709         else if (opt_arg(arg, 'U', "unified", &options->context))
2710                 options->output_format |= DIFF_FORMAT_PATCH;
2711         else if (!strcmp(arg, "--raw"))
2712                 options->output_format |= DIFF_FORMAT_RAW;
2713         else if (!strcmp(arg, "--patch-with-raw"))
2714                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2715         else if (!strcmp(arg, "--numstat"))
2716                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2717         else if (!strcmp(arg, "--shortstat"))
2718                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2719         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2720                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2721         else if (!strcmp(arg, "--cumulative")) {
2722                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2723                 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2724         } else if (opt_arg(arg, 0, "dirstat-by-file",
2725                            &options->dirstat_percent)) {
2726                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2727                 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
2728         }
2729         else if (!strcmp(arg, "--check"))
2730                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2731         else if (!strcmp(arg, "--summary"))
2732                 options->output_format |= DIFF_FORMAT_SUMMARY;
2733         else if (!strcmp(arg, "--patch-with-stat"))
2734                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2735         else if (!strcmp(arg, "--name-only"))
2736                 options->output_format |= DIFF_FORMAT_NAME;
2737         else if (!strcmp(arg, "--name-status"))
2738                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2739         else if (!strcmp(arg, "-s"))
2740                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2741         else if (!prefixcmp(arg, "--stat")) {
2742                 char *end;
2743                 int width = options->stat_width;
2744                 int name_width = options->stat_name_width;
2745                 arg += 6;
2746                 end = (char *)arg;
2747
2748                 switch (*arg) {
2749                 case '-':
2750                         if (!prefixcmp(arg, "-width="))
2751                                 width = strtoul(arg + 7, &end, 10);
2752                         else if (!prefixcmp(arg, "-name-width="))
2753                                 name_width = strtoul(arg + 12, &end, 10);
2754                         break;
2755                 case '=':
2756                         width = strtoul(arg+1, &end, 10);
2757                         if (*end == ',')
2758                                 name_width = strtoul(end+1, &end, 10);
2759                 }
2760
2761                 /* Important! This checks all the error cases! */
2762                 if (*end)
2763                         return 0;
2764                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2765                 options->stat_name_width = name_width;
2766                 options->stat_width = width;
2767         }
2768
2769         /* renames options */
2770         else if (!prefixcmp(arg, "-B")) {
2771                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2772                         return -1;
2773         }
2774         else if (!prefixcmp(arg, "-M")) {
2775                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2776                         return -1;
2777                 options->detect_rename = DIFF_DETECT_RENAME;
2778         }
2779         else if (!prefixcmp(arg, "-C")) {
2780                 if (options->detect_rename == DIFF_DETECT_COPY)
2781                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2782                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2783                         return -1;
2784                 options->detect_rename = DIFF_DETECT_COPY;
2785         }
2786         else if (!strcmp(arg, "--no-renames"))
2787                 options->detect_rename = 0;
2788         else if (!strcmp(arg, "--relative"))
2789                 DIFF_OPT_SET(options, RELATIVE_NAME);
2790         else if (!prefixcmp(arg, "--relative=")) {
2791                 DIFF_OPT_SET(options, RELATIVE_NAME);
2792                 options->prefix = arg + 11;
2793         }
2794
2795         /* xdiff options */
2796         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2797                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
2798         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2799                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
2800         else if (!strcmp(arg, "--ignore-space-at-eol"))
2801                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
2802         else if (!strcmp(arg, "--patience"))
2803                 DIFF_XDL_SET(options, PATIENCE_DIFF);
2804
2805         /* flags options */
2806         else if (!strcmp(arg, "--binary")) {
2807                 options->output_format |= DIFF_FORMAT_PATCH;
2808                 DIFF_OPT_SET(options, BINARY);
2809         }
2810         else if (!strcmp(arg, "--full-index"))
2811                 DIFF_OPT_SET(options, FULL_INDEX);
2812         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2813                 DIFF_OPT_SET(options, TEXT);
2814         else if (!strcmp(arg, "-R"))
2815                 DIFF_OPT_SET(options, REVERSE_DIFF);
2816         else if (!strcmp(arg, "--find-copies-harder"))
2817                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2818         else if (!strcmp(arg, "--follow"))
2819                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2820         else if (!strcmp(arg, "--color"))
2821                 DIFF_OPT_SET(options, COLOR_DIFF);
2822         else if (!prefixcmp(arg, "--color=")) {
2823                 int value = git_config_colorbool(NULL, arg+8, -1);
2824                 if (value == 0)
2825                         DIFF_OPT_CLR(options, COLOR_DIFF);
2826                 else if (value > 0)
2827                         DIFF_OPT_SET(options, COLOR_DIFF);
2828                 else
2829                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
2830         }
2831         else if (!strcmp(arg, "--no-color"))
2832                 DIFF_OPT_CLR(options, COLOR_DIFF);
2833         else if (!strcmp(arg, "--color-words")) {
2834                 DIFF_OPT_SET(options, COLOR_DIFF);
2835                 DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2836         }
2837         else if (!prefixcmp(arg, "--color-words=")) {
2838                 DIFF_OPT_SET(options, COLOR_DIFF);
2839                 DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2840                 options->word_regex = arg + 14;
2841         }
2842         else if (!strcmp(arg, "--exit-code"))
2843                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2844         else if (!strcmp(arg, "--quiet"))
2845                 DIFF_OPT_SET(options, QUICK);
2846         else if (!strcmp(arg, "--ext-diff"))
2847                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2848         else if (!strcmp(arg, "--no-ext-diff"))
2849                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2850         else if (!strcmp(arg, "--textconv"))
2851                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
2852         else if (!strcmp(arg, "--no-textconv"))
2853                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
2854         else if (!strcmp(arg, "--ignore-submodules"))
2855                 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2856         else if (!strcmp(arg, "--submodule"))
2857                 DIFF_OPT_SET(options, SUBMODULE_LOG);
2858         else if (!prefixcmp(arg, "--submodule=")) {
2859                 if (!strcmp(arg + 12, "log"))
2860                         DIFF_OPT_SET(options, SUBMODULE_LOG);
2861         }
2862
2863         /* misc options */
2864         else if (!strcmp(arg, "-z"))
2865                 options->line_termination = 0;
2866         else if (!prefixcmp(arg, "-l"))
2867                 options->rename_limit = strtoul(arg+2, NULL, 10);
2868         else if (!prefixcmp(arg, "-S"))
2869                 options->pickaxe = arg + 2;
2870         else if (!strcmp(arg, "--pickaxe-all"))
2871                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2872         else if (!strcmp(arg, "--pickaxe-regex"))
2873                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2874         else if (!prefixcmp(arg, "-O"))
2875                 options->orderfile = arg + 2;
2876         else if (!prefixcmp(arg, "--diff-filter="))
2877                 options->filter = arg + 14;
2878         else if (!strcmp(arg, "--abbrev"))
2879                 options->abbrev = DEFAULT_ABBREV;
2880         else if (!prefixcmp(arg, "--abbrev=")) {
2881                 options->abbrev = strtoul(arg + 9, NULL, 10);
2882                 if (options->abbrev < MINIMUM_ABBREV)
2883                         options->abbrev = MINIMUM_ABBREV;
2884                 else if (40 < options->abbrev)
2885                         options->abbrev = 40;
2886         }
2887         else if (!prefixcmp(arg, "--src-prefix="))
2888                 options->a_prefix = arg + 13;
2889         else if (!prefixcmp(arg, "--dst-prefix="))
2890                 options->b_prefix = arg + 13;
2891         else if (!strcmp(arg, "--no-prefix"))
2892                 options->a_prefix = options->b_prefix = "";
2893         else if (opt_arg(arg, '\0', "inter-hunk-context",
2894                          &options->interhunkcontext))
2895                 ;
2896         else if (!prefixcmp(arg, "--output=")) {
2897                 options->file = fopen(arg + strlen("--output="), "w");
2898                 if (!options->file)
2899                         die_errno("Could not open '%s'", arg + strlen("--output="));
2900                 options->close_file = 1;
2901         } else
2902                 return 0;
2903         return 1;
2904 }
2905
2906 static int parse_num(const char **cp_p)
2907 {
2908         unsigned long num, scale;
2909         int ch, dot;
2910         const char *cp = *cp_p;
2911
2912         num = 0;
2913         scale = 1;
2914         dot = 0;
2915         for (;;) {
2916                 ch = *cp;
2917                 if ( !dot && ch == '.' ) {
2918                         scale = 1;
2919                         dot = 1;
2920                 } else if ( ch == '%' ) {
2921                         scale = dot ? scale*100 : 100;
2922                         cp++;   /* % is always at the end */
2923                         break;
2924                 } else if ( ch >= '0' && ch <= '9' ) {
2925                         if ( scale < 100000 ) {
2926                                 scale *= 10;
2927                                 num = (num*10) + (ch-'0');
2928                         }
2929                 } else {
2930                         break;
2931                 }
2932                 cp++;
2933         }
2934         *cp_p = cp;
2935
2936         /* user says num divided by scale and we say internally that
2937          * is MAX_SCORE * num / scale.
2938          */
2939         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2940 }
2941
2942 static int diff_scoreopt_parse(const char *opt)
2943 {
2944         int opt1, opt2, cmd;
2945
2946         if (*opt++ != '-')
2947                 return -1;
2948         cmd = *opt++;
2949         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2950                 return -1; /* that is not a -M, -C nor -B option */
2951
2952         opt1 = parse_num(&opt);
2953         if (cmd != 'B')
2954                 opt2 = 0;
2955         else {
2956                 if (*opt == 0)
2957                         opt2 = 0;
2958                 else if (*opt != '/')
2959                         return -1; /* we expect -B80/99 or -B80 */
2960                 else {
2961                         opt++;
2962                         opt2 = parse_num(&opt);
2963                 }
2964         }
2965         if (*opt != 0)
2966                 return -1;
2967         return opt1 | (opt2 << 16);
2968 }
2969
2970 struct diff_queue_struct diff_queued_diff;
2971
2972 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2973 {
2974         if (queue->alloc <= queue->nr) {
2975                 queue->alloc = alloc_nr(queue->alloc);
2976                 queue->queue = xrealloc(queue->queue,
2977                                         sizeof(dp) * queue->alloc);
2978         }
2979         queue->queue[queue->nr++] = dp;
2980 }
2981
2982 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2983                                  struct diff_filespec *one,
2984                                  struct diff_filespec *two)
2985 {
2986         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2987         dp->one = one;
2988         dp->two = two;
2989         if (queue)
2990                 diff_q(queue, dp);
2991         return dp;
2992 }
2993
2994 void diff_free_filepair(struct diff_filepair *p)
2995 {
2996         free_filespec(p->one);
2997         free_filespec(p->two);
2998         free(p);
2999 }
3000
3001 /* This is different from find_unique_abbrev() in that
3002  * it stuffs the result with dots for alignment.
3003  */
3004 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3005 {
3006         int abblen;
3007         const char *abbrev;
3008         if (len == 40)
3009                 return sha1_to_hex(sha1);
3010
3011         abbrev = find_unique_abbrev(sha1, len);
3012         abblen = strlen(abbrev);
3013         if (abblen < 37) {
3014                 static char hex[41];
3015                 if (len < abblen && abblen <= len + 2)
3016                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3017                 else
3018                         sprintf(hex, "%s...", abbrev);
3019                 return hex;
3020         }
3021         return sha1_to_hex(sha1);
3022 }
3023
3024 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3025 {
3026         int line_termination = opt->line_termination;
3027         int inter_name_termination = line_termination ? '\t' : '\0';
3028
3029         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3030                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3031                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
3032                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3033         }
3034         if (p->score) {
3035                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3036                         inter_name_termination);
3037         } else {
3038                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
3039         }
3040
3041         if (p->status == DIFF_STATUS_COPIED ||
3042             p->status == DIFF_STATUS_RENAMED) {
3043                 const char *name_a, *name_b;
3044                 name_a = p->one->path;
3045                 name_b = p->two->path;
3046                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3047                 write_name_quoted(name_a, opt->file, inter_name_termination);
3048                 write_name_quoted(name_b, opt->file, line_termination);
3049         } else {
3050                 const char *name_a, *name_b;
3051                 name_a = p->one->mode ? p->one->path : p->two->path;
3052                 name_b = NULL;
3053                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3054                 write_name_quoted(name_a, opt->file, line_termination);
3055         }
3056 }
3057
3058 int diff_unmodified_pair(struct diff_filepair *p)
3059 {
3060         /* This function is written stricter than necessary to support
3061          * the currently implemented transformers, but the idea is to
3062          * let transformers to produce diff_filepairs any way they want,
3063          * and filter and clean them up here before producing the output.
3064          */
3065         struct diff_filespec *one = p->one, *two = p->two;
3066
3067         if (DIFF_PAIR_UNMERGED(p))
3068                 return 0; /* unmerged is interesting */
3069
3070         /* deletion, addition, mode or type change
3071          * and rename are all interesting.
3072          */
3073         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
3074             DIFF_PAIR_MODE_CHANGED(p) ||
3075             strcmp(one->path, two->path))
3076                 return 0;
3077
3078         /* both are valid and point at the same path.  that is, we are
3079          * dealing with a change.
3080          */
3081         if (one->sha1_valid && two->sha1_valid &&
3082             !hashcmp(one->sha1, two->sha1) &&
3083             !one->dirty_submodule && !two->dirty_submodule)
3084                 return 1; /* no change */
3085         if (!one->sha1_valid && !two->sha1_valid)
3086                 return 1; /* both look at the same file on the filesystem. */
3087         return 0;
3088 }
3089
3090 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
3091 {
3092         if (diff_unmodified_pair(p))
3093                 return;
3094
3095         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3096             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3097                 return; /* no tree diffs in patch format */
3098
3099         run_diff(p, o);
3100 }
3101
3102 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
3103                             struct diffstat_t *diffstat)
3104 {
3105         if (diff_unmodified_pair(p))
3106                 return;
3107
3108         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3109             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3110                 return; /* no tree diffs in patch format */
3111
3112         run_diffstat(p, o, diffstat);
3113 }
3114
3115 static void diff_flush_checkdiff(struct diff_filepair *p,
3116                 struct diff_options *o)
3117 {
3118         if (diff_unmodified_pair(p))
3119                 return;
3120
3121         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3122             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3123                 return; /* no tree diffs in patch format */
3124
3125         run_checkdiff(p, o);
3126 }
3127
3128 int diff_queue_is_empty(void)
3129 {
3130         struct diff_queue_struct *q = &diff_queued_diff;
3131         int i;
3132         for (i = 0; i < q->nr; i++)
3133                 if (!diff_unmodified_pair(q->queue[i]))
3134                         return 0;
3135         return 1;
3136 }
3137
3138 #if DIFF_DEBUG
3139 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
3140 {
3141         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
3142                 x, one ? one : "",
3143                 s->path,
3144                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
3145                 s->mode,
3146                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
3147         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
3148                 x, one ? one : "",
3149                 s->size, s->xfrm_flags);
3150 }
3151
3152 void diff_debug_filepair(const struct diff_filepair *p, int i)
3153 {
3154         diff_debug_filespec(p->one, i, "one");
3155         diff_debug_filespec(p->two, i, "two");
3156         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
3157                 p->score, p->status ? p->status : '?',
3158                 p->one->rename_used, p->broken_pair);
3159 }
3160
3161 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
3162 {
3163         int i;
3164         if (msg)
3165                 fprintf(stderr, "%s\n", msg);
3166         fprintf(stderr, "q->nr = %d\n", q->nr);
3167         for (i = 0; i < q->nr; i++) {
3168                 struct diff_filepair *p = q->queue[i];
3169                 diff_debug_filepair(p, i);
3170         }
3171 }
3172 #endif
3173
3174 static void diff_resolve_rename_copy(void)
3175 {
3176         int i;
3177         struct diff_filepair *p;
3178         struct diff_queue_struct *q = &diff_queued_diff;
3179
3180         diff_debug_queue("resolve-rename-copy", q);
3181
3182         for (i = 0; i < q->nr; i++) {
3183                 p = q->queue[i];
3184                 p->status = 0; /* undecided */
3185                 if (DIFF_PAIR_UNMERGED(p))
3186                         p->status = DIFF_STATUS_UNMERGED;
3187                 else if (!DIFF_FILE_VALID(p->one))
3188                         p->status = DIFF_STATUS_ADDED;
3189                 else if (!DIFF_FILE_VALID(p->two))
3190                         p->status = DIFF_STATUS_DELETED;
3191                 else if (DIFF_PAIR_TYPE_CHANGED(p))
3192                         p->status = DIFF_STATUS_TYPE_CHANGED;
3193
3194                 /* from this point on, we are dealing with a pair
3195                  * whose both sides are valid and of the same type, i.e.
3196                  * either in-place edit or rename/copy edit.
3197                  */
3198                 else if (DIFF_PAIR_RENAME(p)) {
3199                         /*
3200                          * A rename might have re-connected a broken
3201                          * pair up, causing the pathnames to be the
3202                          * same again. If so, that's not a rename at
3203                          * all, just a modification..
3204                          *
3205                          * Otherwise, see if this source was used for
3206                          * multiple renames, in which case we decrement
3207                          * the count, and call it a copy.
3208                          */
3209                         if (!strcmp(p->one->path, p->two->path))
3210                                 p->status = DIFF_STATUS_MODIFIED;
3211                         else if (--p->one->rename_used > 0)
3212                                 p->status = DIFF_STATUS_COPIED;
3213                         else
3214                                 p->status = DIFF_STATUS_RENAMED;
3215                 }
3216                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
3217                          p->one->mode != p->two->mode ||
3218                          p->one->dirty_submodule ||
3219                          p->two->dirty_submodule ||
3220                          is_null_sha1(p->one->sha1))
3221                         p->status = DIFF_STATUS_MODIFIED;
3222                 else {
3223                         /* This is a "no-change" entry and should not
3224                          * happen anymore, but prepare for broken callers.
3225                          */
3226                         error("feeding unmodified %s to diffcore",
3227                               p->one->path);
3228                         p->status = DIFF_STATUS_UNKNOWN;
3229                 }
3230         }
3231         diff_debug_queue("resolve-rename-copy done", q);
3232 }
3233
3234 static int check_pair_status(struct diff_filepair *p)
3235 {
3236         switch (p->status) {
3237         case DIFF_STATUS_UNKNOWN:
3238                 return 0;
3239         case 0:
3240                 die("internal error in diff-resolve-rename-copy");
3241         default:
3242                 return 1;
3243         }
3244 }
3245
3246 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3247 {
3248         int fmt = opt->output_format;
3249
3250         if (fmt & DIFF_FORMAT_CHECKDIFF)
3251                 diff_flush_checkdiff(p, opt);
3252         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3253                 diff_flush_raw(p, opt);
3254         else if (fmt & DIFF_FORMAT_NAME) {
3255                 const char *name_a, *name_b;
3256                 name_a = p->two->path;
3257                 name_b = NULL;
3258                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3259                 write_name_quoted(name_a, opt->file, opt->line_termination);
3260         }
3261 }
3262
3263 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3264 {
3265         if (fs->mode)
3266                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3267         else
3268                 fprintf(file, " %s ", newdelete);
3269         write_name_quoted(fs->path, file, '\n');
3270 }
3271
3272
3273 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
3274 {
3275         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3276                 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
3277                         show_name ? ' ' : '\n');
3278                 if (show_name) {
3279                         write_name_quoted(p->two->path, file, '\n');
3280                 }
3281         }
3282 }
3283
3284 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3285 {
3286         char *names = pprint_rename(p->one->path, p->two->path);
3287
3288         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3289         free(names);
3290         show_mode_change(file, p, 0);
3291 }
3292
3293 static void diff_summary(FILE *file, struct diff_filepair *p)
3294 {
3295         switch(p->status) {
3296         case DIFF_STATUS_DELETED:
3297                 show_file_mode_name(file, "delete", p->one);
3298                 break;
3299         case DIFF_STATUS_ADDED:
3300                 show_file_mode_name(file, "create", p->two);
3301                 break;
3302         case DIFF_STATUS_COPIED:
3303                 show_rename_copy(file, "copy", p);
3304                 break;
3305         case DIFF_STATUS_RENAMED:
3306                 show_rename_copy(file, "rename", p);
3307                 break;
3308         default:
3309                 if (p->score) {
3310                         fputs(" rewrite ", file);
3311                         write_name_quoted(p->two->path, file, ' ');
3312                         fprintf(file, "(%d%%)\n", similarity_index(p));
3313                 }
3314                 show_mode_change(file, p, !p->score);
3315                 break;
3316         }
3317 }
3318
3319 struct patch_id_t {
3320         git_SHA_CTX *ctx;
3321         int patchlen;
3322 };
3323
3324 static int remove_space(char *line, int len)
3325 {
3326         int i;
3327         char *dst = line;
3328         unsigned char c;
3329
3330         for (i = 0; i < len; i++)
3331                 if (!isspace((c = line[i])))
3332                         *dst++ = c;
3333
3334         return dst - line;
3335 }
3336
3337 static void patch_id_consume(void *priv, char *line, unsigned long len)
3338 {
3339         struct patch_id_t *data = priv;
3340         int new_len;
3341
3342         /* Ignore line numbers when computing the SHA1 of the patch */
3343         if (!prefixcmp(line, "@@ -"))
3344                 return;
3345
3346         new_len = remove_space(line, len);
3347
3348         git_SHA1_Update(data->ctx, line, new_len);
3349         data->patchlen += new_len;
3350 }
3351
3352 /* returns 0 upon success, and writes result into sha1 */
3353 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3354 {
3355         struct diff_queue_struct *q = &diff_queued_diff;
3356         int i;
3357         git_SHA_CTX ctx;
3358         struct patch_id_t data;
3359         char buffer[PATH_MAX * 4 + 20];
3360
3361         git_SHA1_Init(&ctx);
3362         memset(&data, 0, sizeof(struct patch_id_t));
3363         data.ctx = &ctx;
3364
3365         for (i = 0; i < q->nr; i++) {
3366                 xpparam_t xpp;
3367                 xdemitconf_t xecfg;
3368                 xdemitcb_t ecb;
3369                 mmfile_t mf1, mf2;
3370                 struct diff_filepair *p = q->queue[i];
3371                 int len1, len2;
3372
3373                 memset(&xpp, 0, sizeof(xpp));
3374                 memset(&xecfg, 0, sizeof(xecfg));
3375                 if (p->status == 0)
3376                         return error("internal diff status error");
3377                 if (p->status == DIFF_STATUS_UNKNOWN)
3378                         continue;
3379                 if (diff_unmodified_pair(p))
3380                         continue;
3381                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3382                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3383                         continue;
3384                 if (DIFF_PAIR_UNMERGED(p))
3385                         continue;
3386
3387                 diff_fill_sha1_info(p->one);
3388                 diff_fill_sha1_info(p->two);
3389                 if (fill_mmfile(&mf1, p->one) < 0 ||
3390                                 fill_mmfile(&mf2, p->two) < 0)
3391                         return error("unable to read files to diff");
3392
3393                 len1 = remove_space(p->one->path, strlen(p->one->path));
3394                 len2 = remove_space(p->two->path, strlen(p->two->path));
3395                 if (p->one->mode == 0)
3396                         len1 = snprintf(buffer, sizeof(buffer),
3397                                         "diff--gita/%.*sb/%.*s"
3398                                         "newfilemode%06o"
3399                                         "---/dev/null"
3400                                         "+++b/%.*s",
3401                                         len1, p->one->path,
3402                                         len2, p->two->path,
3403                                         p->two->mode,
3404                                         len2, p->two->path);
3405                 else if (p->two->mode == 0)
3406                         len1 = snprintf(buffer, sizeof(buffer),
3407                                         "diff--gita/%.*sb/%.*s"
3408                                         "deletedfilemode%06o"
3409                                         "---a/%.*s"
3410                                         "+++/dev/null",
3411                                         len1, p->one->path,
3412                                         len2, p->two->path,
3413                                         p->one->mode,
3414                                         len1, p->one->path);
3415                 else
3416                         len1 = snprintf(buffer, sizeof(buffer),
3417                                         "diff--gita/%.*sb/%.*s"
3418                                         "---a/%.*s"
3419                                         "+++b/%.*s",
3420                                         len1, p->one->path,
3421                                         len2, p->two->path,
3422                                         len1, p->one->path,
3423                                         len2, p->two->path);
3424                 git_SHA1_Update(&ctx, buffer, len1);
3425
3426                 xpp.flags = XDF_NEED_MINIMAL;
3427                 xecfg.ctxlen = 3;
3428                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3429                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3430                               &xpp, &xecfg, &ecb);
3431         }
3432
3433         git_SHA1_Final(sha1, &ctx);
3434         return 0;
3435 }
3436
3437 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3438 {
3439         struct diff_queue_struct *q = &diff_queued_diff;
3440         int i;
3441         int result = diff_get_patch_id(options, sha1);
3442
3443         for (i = 0; i < q->nr; i++)
3444                 diff_free_filepair(q->queue[i]);
3445
3446         free(q->queue);
3447         q->queue = NULL;
3448         q->nr = q->alloc = 0;
3449
3450         return result;
3451 }
3452
3453 static int is_summary_empty(const struct diff_queue_struct *q)
3454 {
3455         int i;
3456
3457         for (i = 0; i < q->nr; i++) {
3458                 const struct diff_filepair *p = q->queue[i];
3459
3460                 switch (p->status) {
3461                 case DIFF_STATUS_DELETED:
3462                 case DIFF_STATUS_ADDED:
3463                 case DIFF_STATUS_COPIED:
3464                 case DIFF_STATUS_RENAMED:
3465                         return 0;
3466                 default:
3467                         if (p->score)
3468                                 return 0;
3469                         if (p->one->mode && p->two->mode &&
3470                             p->one->mode != p->two->mode)
3471                                 return 0;
3472                         break;
3473                 }
3474         }
3475         return 1;
3476 }
3477
3478 void diff_flush(struct diff_options *options)
3479 {
3480         struct diff_queue_struct *q = &diff_queued_diff;
3481         int i, output_format = options->output_format;
3482         int separator = 0;
3483
3484         /*
3485          * Order: raw, stat, summary, patch
3486          * or:    name/name-status/checkdiff (other bits clear)
3487          */
3488         if (!q->nr)
3489                 goto free_queue;
3490
3491         if (output_format & (DIFF_FORMAT_RAW |
3492                              DIFF_FORMAT_NAME |
3493                              DIFF_FORMAT_NAME_STATUS |
3494                              DIFF_FORMAT_CHECKDIFF)) {
3495                 for (i = 0; i < q->nr; i++) {
3496                         struct diff_filepair *p = q->queue[i];
3497                         if (check_pair_status(p))
3498                                 flush_one_pair(p, options);
3499                 }
3500                 separator++;
3501         }
3502
3503         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3504                 struct diffstat_t diffstat;
3505
3506                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3507                 for (i = 0; i < q->nr; i++) {
3508                         struct diff_filepair *p = q->queue[i];
3509                         if (check_pair_status(p))
3510                                 diff_flush_stat(p, options, &diffstat);
3511                 }
3512                 if (output_format & DIFF_FORMAT_NUMSTAT)
3513                         show_numstat(&diffstat, options);
3514                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3515                         show_stats(&diffstat, options);
3516                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3517                         show_shortstats(&diffstat, options);
3518                 free_diffstat_info(&diffstat);
3519                 separator++;
3520         }
3521         if (output_format & DIFF_FORMAT_DIRSTAT)
3522                 show_dirstat(options);
3523
3524         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3525                 for (i = 0; i < q->nr; i++)
3526                         diff_summary(options->file, q->queue[i]);
3527                 separator++;
3528         }
3529
3530         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
3531             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
3532             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
3533                 /*
3534                  * run diff_flush_patch for the exit status. setting
3535                  * options->file to /dev/null should be safe, becaue we
3536                  * aren't supposed to produce any output anyway.
3537                  */
3538                 if (options->close_file)
3539                         fclose(options->file);
3540                 options->file = fopen("/dev/null", "w");
3541                 if (!options->file)
3542                         die_errno("Could not open /dev/null");
3543                 options->close_file = 1;
3544                 for (i = 0; i < q->nr; i++) {
3545                         struct diff_filepair *p = q->queue[i];
3546                         if (check_pair_status(p))
3547                                 diff_flush_patch(p, options);
3548                         if (options->found_changes)
3549                                 break;
3550                 }
3551         }
3552
3553         if (output_format & DIFF_FORMAT_PATCH) {
3554                 if (separator) {
3555                         putc(options->line_termination, options->file);
3556                         if (options->stat_sep) {
3557                                 /* attach patch instead of inline */
3558                                 fputs(options->stat_sep, options->file);
3559                         }
3560                 }
3561
3562                 for (i = 0; i < q->nr; i++) {
3563                         struct diff_filepair *p = q->queue[i];
3564                         if (check_pair_status(p))
3565                                 diff_flush_patch(p, options);
3566                 }
3567         }
3568
3569         if (output_format & DIFF_FORMAT_CALLBACK)
3570                 options->format_callback(q, options, options->format_callback_data);
3571
3572         for (i = 0; i < q->nr; i++)
3573                 diff_free_filepair(q->queue[i]);
3574 free_queue:
3575         free(q->queue);
3576         q->queue = NULL;
3577         q->nr = q->alloc = 0;
3578         if (options->close_file)
3579                 fclose(options->file);
3580
3581         /*
3582          * Report the content-level differences with HAS_CHANGES;
3583          * diff_addremove/diff_change does not set the bit when
3584          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
3585          */
3586         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
3587                 if (options->found_changes)
3588                         DIFF_OPT_SET(options, HAS_CHANGES);
3589                 else
3590                         DIFF_OPT_CLR(options, HAS_CHANGES);
3591         }
3592 }
3593
3594 static void diffcore_apply_filter(const char *filter)
3595 {
3596         int i;
3597         struct diff_queue_struct *q = &diff_queued_diff;
3598         struct diff_queue_struct outq;
3599         outq.queue = NULL;
3600         outq.nr = outq.alloc = 0;
3601
3602         if (!filter)
3603                 return;
3604
3605         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3606                 int found;
3607                 for (i = found = 0; !found && i < q->nr; i++) {
3608                         struct diff_filepair *p = q->queue[i];
3609                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3610                              ((p->score &&
3611                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3612                               (!p->score &&
3613                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3614                             ((p->status != DIFF_STATUS_MODIFIED) &&
3615                              strchr(filter, p->status)))
3616                                 found++;
3617                 }
3618                 if (found)
3619                         return;
3620
3621                 /* otherwise we will clear the whole queue
3622                  * by copying the empty outq at the end of this
3623                  * function, but first clear the current entries
3624                  * in the queue.
3625                  */
3626                 for (i = 0; i < q->nr; i++)
3627                         diff_free_filepair(q->queue[i]);
3628         }
3629         else {
3630                 /* Only the matching ones */
3631                 for (i = 0; i < q->nr; i++) {
3632                         struct diff_filepair *p = q->queue[i];
3633
3634                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3635                              ((p->score &&
3636                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3637                               (!p->score &&
3638                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3639                             ((p->status != DIFF_STATUS_MODIFIED) &&
3640                              strchr(filter, p->status)))
3641                                 diff_q(&outq, p);
3642                         else
3643                                 diff_free_filepair(p);
3644                 }
3645         }
3646         free(q->queue);
3647         *q = outq;
3648 }
3649
3650 /* Check whether two filespecs with the same mode and size are identical */
3651 static int diff_filespec_is_identical(struct diff_filespec *one,
3652                                       struct diff_filespec *two)
3653 {
3654         if (S_ISGITLINK(one->mode))
3655                 return 0;
3656         if (diff_populate_filespec(one, 0))
3657                 return 0;
3658         if (diff_populate_filespec(two, 0))
3659                 return 0;
3660         return !memcmp(one->data, two->data, one->size);
3661 }
3662
3663 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3664 {
3665         int i;
3666         struct diff_queue_struct *q = &diff_queued_diff;
3667         struct diff_queue_struct outq;
3668         outq.queue = NULL;
3669         outq.nr = outq.alloc = 0;
3670
3671         for (i = 0; i < q->nr; i++) {
3672                 struct diff_filepair *p = q->queue[i];
3673
3674                 /*
3675                  * 1. Entries that come from stat info dirtiness
3676                  *    always have both sides (iow, not create/delete),
3677                  *    one side of the object name is unknown, with
3678                  *    the same mode and size.  Keep the ones that
3679                  *    do not match these criteria.  They have real
3680                  *    differences.
3681                  *
3682                  * 2. At this point, the file is known to be modified,
3683                  *    with the same mode and size, and the object
3684                  *    name of one side is unknown.  Need to inspect
3685                  *    the identical contents.
3686                  */
3687                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3688                     !DIFF_FILE_VALID(p->two) ||
3689                     (p->one->sha1_valid && p->two->sha1_valid) ||
3690                     (p->one->mode != p->two->mode) ||
3691                     diff_populate_filespec(p->one, 1) ||
3692                     diff_populate_filespec(p->two, 1) ||
3693                     (p->one->size != p->two->size) ||
3694                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3695                         diff_q(&outq, p);
3696                 else {
3697                         /*
3698                          * The caller can subtract 1 from skip_stat_unmatch
3699                          * to determine how many paths were dirty only
3700                          * due to stat info mismatch.
3701                          */
3702                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3703                                 diffopt->skip_stat_unmatch++;
3704                         diff_free_filepair(p);
3705                 }
3706         }
3707         free(q->queue);
3708         *q = outq;
3709 }
3710
3711 static int diffnamecmp(const void *a_, const void *b_)
3712 {
3713         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
3714         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
3715         const char *name_a, *name_b;
3716
3717         name_a = a->one ? a->one->path : a->two->path;
3718         name_b = b->one ? b->one->path : b->two->path;
3719         return strcmp(name_a, name_b);
3720 }
3721
3722 void diffcore_fix_diff_index(struct diff_options *options)
3723 {
3724         struct diff_queue_struct *q = &diff_queued_diff;
3725         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
3726 }
3727
3728 void diffcore_std(struct diff_options *options)
3729 {
3730         if (options->skip_stat_unmatch)
3731                 diffcore_skip_stat_unmatch(options);
3732         if (options->break_opt != -1)
3733                 diffcore_break(options->break_opt);
3734         if (options->detect_rename)
3735                 diffcore_rename(options);
3736         if (options->break_opt != -1)
3737                 diffcore_merge_broken();
3738         if (options->pickaxe)
3739                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3740         if (options->orderfile)
3741                 diffcore_order(options->orderfile);
3742         diff_resolve_rename_copy();
3743         diffcore_apply_filter(options->filter);
3744
3745         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3746                 DIFF_OPT_SET(options, HAS_CHANGES);
3747         else
3748                 DIFF_OPT_CLR(options, HAS_CHANGES);
3749 }
3750
3751 int diff_result_code(struct diff_options *opt, int status)
3752 {
3753         int result = 0;
3754         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3755             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3756                 return status;
3757         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3758             DIFF_OPT_TST(opt, HAS_CHANGES))
3759                 result |= 01;
3760         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3761             DIFF_OPT_TST(opt, CHECK_FAILED))
3762                 result |= 02;
3763         return result;
3764 }
3765
3766 void diff_addremove(struct diff_options *options,
3767                     int addremove, unsigned mode,
3768                     const unsigned char *sha1,
3769                     const char *concatpath, unsigned dirty_submodule)
3770 {
3771         struct diff_filespec *one, *two;
3772
3773         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3774                 return;
3775
3776         /* This may look odd, but it is a preparation for
3777          * feeding "there are unchanged files which should
3778          * not produce diffs, but when you are doing copy
3779          * detection you would need them, so here they are"
3780          * entries to the diff-core.  They will be prefixed
3781          * with something like '=' or '*' (I haven't decided
3782          * which but should not make any difference).
3783          * Feeding the same new and old to diff_change()
3784          * also has the same effect.
3785          * Before the final output happens, they are pruned after
3786          * merged into rename/copy pairs as appropriate.
3787          */
3788         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3789                 addremove = (addremove == '+' ? '-' :
3790                              addremove == '-' ? '+' : addremove);
3791
3792         if (options->prefix &&
3793             strncmp(concatpath, options->prefix, options->prefix_length))
3794                 return;
3795
3796         one = alloc_filespec(concatpath);
3797         two = alloc_filespec(concatpath);
3798
3799         if (addremove != '+')
3800                 fill_filespec(one, sha1, mode);
3801         if (addremove != '-') {
3802                 fill_filespec(two, sha1, mode);
3803                 two->dirty_submodule = dirty_submodule;
3804         }
3805
3806         diff_queue(&diff_queued_diff, one, two);
3807         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3808                 DIFF_OPT_SET(options, HAS_CHANGES);
3809 }
3810
3811 void diff_change(struct diff_options *options,
3812                  unsigned old_mode, unsigned new_mode,
3813                  const unsigned char *old_sha1,
3814                  const unsigned char *new_sha1,
3815                  const char *concatpath,
3816                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
3817 {
3818         struct diff_filespec *one, *two;
3819
3820         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3821                         && S_ISGITLINK(new_mode))
3822                 return;
3823
3824         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3825                 unsigned tmp;
3826                 const unsigned char *tmp_c;
3827                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3828                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3829                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
3830                         new_dirty_submodule = tmp;
3831         }
3832
3833         if (options->prefix &&
3834             strncmp(concatpath, options->prefix, options->prefix_length))
3835                 return;
3836
3837         one = alloc_filespec(concatpath);
3838         two = alloc_filespec(concatpath);
3839         fill_filespec(one, old_sha1, old_mode);
3840         fill_filespec(two, new_sha1, new_mode);
3841         one->dirty_submodule = old_dirty_submodule;
3842         two->dirty_submodule = new_dirty_submodule;
3843
3844         diff_queue(&diff_queued_diff, one, two);
3845         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3846                 DIFF_OPT_SET(options, HAS_CHANGES);
3847 }
3848
3849 void diff_unmerge(struct diff_options *options,
3850                   const char *path,
3851                   unsigned mode, const unsigned char *sha1)
3852 {
3853         struct diff_filespec *one, *two;
3854
3855         if (options->prefix &&
3856             strncmp(path, options->prefix, options->prefix_length))
3857                 return;
3858
3859         one = alloc_filespec(path);
3860         two = alloc_filespec(path);
3861         fill_filespec(one, sha1, mode);
3862         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3863 }
3864
3865 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
3866                 size_t *outsize)
3867 {
3868         struct diff_tempfile *temp;
3869         const char *argv[3];
3870         const char **arg = argv;
3871         struct child_process child;
3872         struct strbuf buf = STRBUF_INIT;
3873         int err = 0;
3874
3875         temp = prepare_temp_file(spec->path, spec);
3876         *arg++ = pgm;
3877         *arg++ = temp->name;
3878         *arg = NULL;
3879
3880         memset(&child, 0, sizeof(child));
3881         child.use_shell = 1;
3882         child.argv = argv;
3883         child.out = -1;
3884         if (start_command(&child)) {
3885                 remove_tempfile();
3886                 return NULL;
3887         }
3888
3889         if (strbuf_read(&buf, child.out, 0) < 0)
3890                 err = error("error reading from textconv command '%s'", pgm);
3891         close(child.out);
3892
3893         if (finish_command(&child) || err) {
3894                 strbuf_release(&buf);
3895                 remove_tempfile();
3896                 return NULL;
3897         }
3898         remove_tempfile();
3899
3900         return strbuf_detach(&buf, outsize);
3901 }
3902
3903 static size_t fill_textconv(struct userdiff_driver *driver,
3904                             struct diff_filespec *df,
3905                             char **outbuf)
3906 {
3907         size_t size;
3908
3909         if (!driver || !driver->textconv) {
3910                 if (!DIFF_FILE_VALID(df)) {
3911                         *outbuf = "";
3912                         return 0;
3913                 }
3914                 if (diff_populate_filespec(df, 0))
3915                         die("unable to read files to diff");
3916                 *outbuf = df->data;
3917                 return df->size;
3918         }
3919
3920         if (driver->textconv_cache) {
3921                 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
3922                                           &size);
3923                 if (*outbuf)
3924                         return size;
3925         }
3926
3927         *outbuf = run_textconv(driver->textconv, df, &size);
3928         if (!*outbuf)
3929                 die("unable to read files to diff");
3930
3931         if (driver->textconv_cache) {
3932                 /* ignore errors, as we might be in a readonly repository */
3933                 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
3934                                 size);
3935                 /*
3936                  * we could save up changes and flush them all at the end,
3937                  * but we would need an extra call after all diffing is done.
3938                  * Since generating a cache entry is the slow path anyway,
3939                  * this extra overhead probably isn't a big deal.
3940                  */
3941                 notes_cache_write(driver->textconv_cache);
3942         }
3943
3944         return size;
3945 }