diff.c

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "quote.h"
   6 #include "diff.h"
   7 #include "diffcore.h"
   8 #include "delta.h"
   9 #include "xdiff-interface.h"
  10 #include "color.h"
  11 #include "attr.h"
  12 #include "run-command.h"
  13 #include "utf8.h"
  14 #include "userdiff.h"
  15 #include "sigchain.h"
  16 #include "submodule.h"
  17 #include "ll-merge.h"
  18
  19 #ifdef NO_FAST_WORKING_DIRECTORY
  20 #define FAST_WORKING_DIRECTORY 0
  21 #else
  22 #define FAST_WORKING_DIRECTORY 1
  23 #endif
  24
  25 static int diff_detect_rename_default;
  26 static int diff_rename_limit_default = 200;
  27 static int diff_suppress_blank_empty;
  28 int diff_use_color_default = -1;
  29 static const char *diff_word_regex_cfg;
  30 static const char *external_diff_cmd_cfg;
  31 int diff_auto_refresh_index = 1;
  32 static int diff_mnemonic_prefix;
  33
  34 static char diff_colors[][COLOR_MAXLEN] = {
  35         GIT_COLOR_RESET,
  36         GIT_COLOR_NORMAL,       /* PLAIN */
  37         GIT_COLOR_BOLD,         /* METAINFO */
  38         GIT_COLOR_CYAN,         /* FRAGINFO */
  39         GIT_COLOR_RED,          /* OLD */
  40         GIT_COLOR_GREEN,        /* NEW */
  41         GIT_COLOR_YELLOW,       /* COMMIT */
  42         GIT_COLOR_BG_RED,       /* WHITESPACE */
  43         GIT_COLOR_NORMAL,       /* FUNCINFO */
  44 };
  45
  46 static void diff_filespec_load_driver(struct diff_filespec *one);
  47 static char *run_textconv(const char *, struct diff_filespec *, size_t *);
  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                               const char *textconv_one,
 470                               const char *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         const 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         diff_populate_filespec(one, 0);
 504         diff_populate_filespec(two, 0);
 505         if (textconv_one) {
 506                 data_one = run_textconv(textconv_one, one, &size_one);
 507                 if (!data_one)
 508                         die("unable to read files to diff");
 509         }
 510         else {
 511                 data_one = one->data;
 512                 size_one = one->size;
 513         }
 514         if (textconv_two) {
 515                 data_two = run_textconv(textconv_two, two, &size_two);
 516                 if (!data_two)
 517                         die("unable to read files to diff");
 518         }
 519         else {
 520                 data_two = two->data;
 521                 size_two = two->size;
 522         }
 523
 524         memset(&ecbdata, 0, sizeof(ecbdata));
 525         ecbdata.color_diff = color_diff;
 526         ecbdata.found_changesp = &o->found_changes;
 527         ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
 528         ecbdata.file = o->file;
 529         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
 530                 mmfile_t mf1, mf2;
 531                 mf1.ptr = (char *)data_one;
 532                 mf2.ptr = (char *)data_two;
 533                 mf1.size = size_one;
 534                 mf2.size = size_two;
 535                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
 536         }
 537         ecbdata.lno_in_preimage = 1;
 538         ecbdata.lno_in_postimage = 1;
 539
 540         lc_a = count_lines(data_one, size_one);
 541         lc_b = count_lines(data_two, size_two);
 542         fprintf(o->file,
 543                 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
 544                 metainfo, a_name.buf, name_a_tab, reset,
 545                 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
 546         print_line_count(o->file, lc_a);
 547         fprintf(o->file, " +");
 548         print_line_count(o->file, lc_b);
 549         fprintf(o->file, " @@%s\n", reset);
 550         if (lc_a)
 551                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
 552         if (lc_b)
 553                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
 554         if (textconv_one)
 555                 free((char *)data_one);
 556         if (textconv_two)
 557                 free((char *)data_two);
 558 }
 559
 560 struct diff_words_buffer {
 561         mmfile_t text;
 562         long alloc;
 563         struct diff_words_orig {
 564                 const char *begin, *end;
 565         } *orig;
 566         int orig_nr, orig_alloc;
 567 };
 568
 569 static void diff_words_append(char *line, unsigned long len,
 570                 struct diff_words_buffer *buffer)
 571 {
 572         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
 573         line++;
 574         len--;
 575         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 576         buffer->text.size += len;
 577         buffer->text.ptr[buffer->text.size] = '\0';
 578 }
 579
 580 struct diff_words_data {
 581         struct diff_words_buffer minus, plus;
 582         const char *current_plus;
 583         FILE *file;
 584         regex_t *word_regex;
 585 };
 586
 587 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 588 {
 589         struct diff_words_data *diff_words = priv;
 590         int minus_first, minus_len, plus_first, plus_len;
 591         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 592
 593         if (line[0] != '@' || parse_hunk_header(line, len,
 594                         &minus_first, &minus_len, &plus_first, &plus_len))
 595                 return;
 596
 597         /* POSIX requires that first be decremented by one if len == 0... */
 598         if (minus_len) {
 599                 minus_begin = diff_words->minus.orig[minus_first].begin;
 600                 minus_end =
 601                         diff_words->minus.orig[minus_first + minus_len - 1].end;
 602         } else
 603                 minus_begin = minus_end =
 604                         diff_words->minus.orig[minus_first].end;
 605
 606         if (plus_len) {
 607                 plus_begin = diff_words->plus.orig[plus_first].begin;
 608                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
 609         } else
 610                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
 611
 612         if (diff_words->current_plus != plus_begin)
 613                 fwrite(diff_words->current_plus,
 614                                 plus_begin - diff_words->current_plus, 1,
 615                                 diff_words->file);
 616         if (minus_begin != minus_end)
 617                 color_fwrite_lines(diff_words->file,
 618                                 diff_get_color(1, DIFF_FILE_OLD),
 619                                 minus_end - minus_begin, minus_begin);
 620         if (plus_begin != plus_end)
 621                 color_fwrite_lines(diff_words->file,
 622                                 diff_get_color(1, DIFF_FILE_NEW),
 623                                 plus_end - plus_begin, plus_begin);
 624
 625         diff_words->current_plus = plus_end;
 626 }
 627
 628 /* This function starts looking at *begin, and returns 0 iff a word was found. */
 629 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
 630                 int *begin, int *end)
 631 {
 632         if (word_regex && *begin < buffer->size) {
 633                 regmatch_t match[1];
 634                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
 635                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
 636                                         '\n', match[0].rm_eo - match[0].rm_so);
 637                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
 638                         *begin += match[0].rm_so;
 639                         return *begin >= *end;
 640                 }
 641                 return -1;
 642         }
 643
 644         /* find the next word */
 645         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
 646                 (*begin)++;
 647         if (*begin >= buffer->size)
 648                 return -1;
 649
 650         /* find the end of the word */
 651         *end = *begin + 1;
 652         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
 653                 (*end)++;
 654
 655         return 0;
 656 }
 657
 658 /*
 659  * This function splits the words in buffer->text, stores the list with
 660  * newline separator into out, and saves the offsets of the original words
 661  * in buffer->orig.
 662  */
 663 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
 664                 regex_t *word_regex)
 665 {
 666         int i, j;
 667         long alloc = 0;
 668
 669         out->size = 0;
 670         out->ptr = NULL;
 671
 672         /* fake an empty "0th" word */
 673         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
 674         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
 675         buffer->orig_nr = 1;
 676
 677         for (i = 0; i < buffer->text.size; i++) {
 678                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
 679                         return;
 680
 681                 /* store original boundaries */
 682                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
 683                                 buffer->orig_alloc);
 684                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
 685                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
 686                 buffer->orig_nr++;
 687
 688                 /* store one word */
 689                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
 690                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
 691                 out->ptr[out->size + j - i] = '\n';
 692                 out->size += j - i + 1;
 693
 694                 i = j - 1;
 695         }
 696 }
 697
 698 /* this executes the word diff on the accumulated buffers */
 699 static void diff_words_show(struct diff_words_data *diff_words)
 700 {
 701         xpparam_t xpp;
 702         xdemitconf_t xecfg;
 703         mmfile_t minus, plus;
 704
 705         /* special case: only removal */
 706         if (!diff_words->plus.text.size) {
 707                 color_fwrite_lines(diff_words->file,
 708                         diff_get_color(1, DIFF_FILE_OLD),
 709                         diff_words->minus.text.size, diff_words->minus.text.ptr);
 710                 diff_words->minus.text.size = 0;
 711                 return;
 712         }
 713
 714         diff_words->current_plus = diff_words->plus.text.ptr;
 715
 716         memset(&xpp, 0, sizeof(xpp));
 717         memset(&xecfg, 0, sizeof(xecfg));
 718         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
 719         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
 720         xpp.flags = 0;
 721         /* as only the hunk header will be parsed, we need a 0-context */
 722         xecfg.ctxlen = 0;
 723         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
 724                       &xpp, &xecfg);
 725         free(minus.ptr);
 726         free(plus.ptr);
 727         if (diff_words->current_plus != diff_words->plus.text.ptr +
 728                         diff_words->plus.text.size)
 729                 fwrite(diff_words->current_plus,
 730                         diff_words->plus.text.ptr + diff_words->plus.text.size
 731                         - diff_words->current_plus, 1,
 732                         diff_words->file);
 733         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 734 }
 735
 736 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
 737 static void diff_words_flush(struct emit_callback *ecbdata)
 738 {
 739         if (ecbdata->diff_words->minus.text.size ||
 740             ecbdata->diff_words->plus.text.size)
 741                 diff_words_show(ecbdata->diff_words);
 742 }
 743
 744 static void free_diff_words_data(struct emit_callback *ecbdata)
 745 {
 746         if (ecbdata->diff_words) {
 747                 diff_words_flush(ecbdata);
 748                 free (ecbdata->diff_words->minus.text.ptr);
 749                 free (ecbdata->diff_words->minus.orig);
 750                 free (ecbdata->diff_words->plus.text.ptr);
 751                 free (ecbdata->diff_words->plus.orig);
 752                 free(ecbdata->diff_words->word_regex);
 753                 free(ecbdata->diff_words);
 754                 ecbdata->diff_words = NULL;
 755         }
 756 }
 757
 758 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 759 {
 760         if (diff_use_color)
 761                 return diff_colors[ix];
 762         return "";
 763 }
 764
 765 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
 766 {
 767         const char *cp;
 768         unsigned long allot;
 769         size_t l = len;
 770
 771         if (ecb->truncate)
 772                 return ecb->truncate(line, len);
 773         cp = line;
 774         allot = l;
 775         while (0 < l) {
 776                 (void) utf8_width(&cp, &l);
 777                 if (!cp)
 778                         break; /* truncated in the middle? */
 779         }
 780         return allot - l;
 781 }
 782
 783 static void find_lno(const char *line, struct emit_callback *ecbdata)
 784 {
 785         const char *p;
 786         ecbdata->lno_in_preimage = 0;
 787         ecbdata->lno_in_postimage = 0;
 788         p = strchr(line, '-');
 789         if (!p)
 790                 return; /* cannot happen */
 791         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
 792         p = strchr(p, '+');
 793         if (!p)
 794                 return; /* cannot happen */
 795         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
 796 }
 797
 798 static void fn_out_consume(void *priv, char *line, unsigned long len)
 799 {
 800         struct emit_callback *ecbdata = priv;
 801         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 802         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 803         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 804
 805         if (ecbdata->header) {
 806                 fprintf(ecbdata->file, "%s", ecbdata->header->buf);
 807                 strbuf_reset(ecbdata->header);
 808                 ecbdata->header = NULL;
 809         }
 810         *(ecbdata->found_changesp) = 1;
 811
 812         if (ecbdata->label_path[0]) {
 813                 const char *name_a_tab, *name_b_tab;
 814
 815                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 816                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 817
 818                 fprintf(ecbdata->file, "%s--- %s%s%s\n",
 819                         meta, ecbdata->label_path[0], reset, name_a_tab);
 820                 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
 821                         meta, ecbdata->label_path[1], reset, name_b_tab);
 822                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 823         }
 824
 825         if (diff_suppress_blank_empty
 826             && len == 2 && line[0] == ' ' && line[1] == '\n') {
 827                 line[0] = '\n';
 828                 len = 1;
 829         }
 830
 831         if (line[0] == '@') {
 832                 if (ecbdata->diff_words)
 833                         diff_words_flush(ecbdata);
 834                 len = sane_truncate_line(ecbdata, line, len);
 835                 find_lno(line, ecbdata);
 836                 emit_hunk_header(ecbdata, line, len);
 837                 if (line[len-1] != '\n')
 838                         putc('\n', ecbdata->file);
 839                 return;
 840         }
 841
 842         if (len < 1) {
 843                 emit_line(ecbdata->file, reset, reset, line, len);
 844                 return;
 845         }
 846
 847         if (ecbdata->diff_words) {
 848                 if (line[0] == '-') {
 849                         diff_words_append(line, len,
 850                                           &ecbdata->diff_words->minus);
 851                         return;
 852                 } else if (line[0] == '+') {
 853                         diff_words_append(line, len,
 854                                           &ecbdata->diff_words->plus);
 855                         return;
 856                 }
 857                 diff_words_flush(ecbdata);
 858                 line++;
 859                 len--;
 860                 emit_line(ecbdata->file, plain, reset, line, len);
 861                 return;
 862         }
 863
 864         if (line[0] != '+') {
 865                 const char *color =
 866                         diff_get_color(ecbdata->color_diff,
 867                                        line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
 868                 ecbdata->lno_in_preimage++;
 869                 if (line[0] == ' ')
 870                         ecbdata->lno_in_postimage++;
 871                 emit_line(ecbdata->file, color, reset, line, len);
 872         } else {
 873                 ecbdata->lno_in_postimage++;
 874                 emit_add_line(reset, ecbdata, line + 1, len - 1);
 875         }
 876 }
 877
 878 static char *pprint_rename(const char *a, const char *b)
 879 {
 880         const char *old = a;
 881         const char *new = b;
 882         struct strbuf name = STRBUF_INIT;
 883         int pfx_length, sfx_length;
 884         int len_a = strlen(a);
 885         int len_b = strlen(b);
 886         int a_midlen, b_midlen;
 887         int qlen_a = quote_c_style(a, NULL, NULL, 0);
 888         int qlen_b = quote_c_style(b, NULL, NULL, 0);
 889
 890         if (qlen_a || qlen_b) {
 891                 quote_c_style(a, &name, NULL, 0);
 892                 strbuf_addstr(&name, " => ");
 893                 quote_c_style(b, &name, NULL, 0);
 894                 return strbuf_detach(&name, NULL);
 895         }
 896
 897         /* Find common prefix */
 898         pfx_length = 0;
 899         while (*old && *new && *old == *new) {
 900                 if (*old == '/')
 901                         pfx_length = old - a + 1;
 902                 old++;
 903                 new++;
 904         }
 905
 906         /* Find common suffix */
 907         old = a + len_a;
 908         new = b + len_b;
 909         sfx_length = 0;
 910         while (a <= old && b <= new && *old == *new) {
 911                 if (*old == '/')
 912                         sfx_length = len_a - (old - a);
 913                 old--;
 914                 new--;
 915         }
 916
 917         /*
 918          * pfx{mid-a => mid-b}sfx
 919          * {pfx-a => pfx-b}sfx
 920          * pfx{sfx-a => sfx-b}
 921          * name-a => name-b
 922          */
 923         a_midlen = len_a - pfx_length - sfx_length;
 924         b_midlen = len_b - pfx_length - sfx_length;
 925         if (a_midlen < 0)
 926                 a_midlen = 0;
 927         if (b_midlen < 0)
 928                 b_midlen = 0;
 929
 930         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 931         if (pfx_length + sfx_length) {
 932                 strbuf_add(&name, a, pfx_length);
 933                 strbuf_addch(&name, '{');
 934         }
 935         strbuf_add(&name, a + pfx_length, a_midlen);
 936         strbuf_addstr(&name, " => ");
 937         strbuf_add(&name, b + pfx_length, b_midlen);
 938         if (pfx_length + sfx_length) {
 939                 strbuf_addch(&name, '}');
 940                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 941         }
 942         return strbuf_detach(&name, NULL);
 943 }
 944
 945 struct diffstat_t {
 946         int nr;
 947         int alloc;
 948         struct diffstat_file {
 949                 char *from_name;
 950                 char *name;
 951                 char *print_name;
 952                 unsigned is_unmerged:1;
 953                 unsigned is_binary:1;
 954                 unsigned is_renamed:1;
 955                 uintmax_t added, deleted;
 956         } **files;
 957 };
 958
 959 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 960                                           const char *name_a,
 961                                           const char *name_b)
 962 {
 963         struct diffstat_file *x;
 964         x = xcalloc(sizeof (*x), 1);
 965         if (diffstat->nr == diffstat->alloc) {
 966                 diffstat->alloc = alloc_nr(diffstat->alloc);
 967                 diffstat->files = xrealloc(diffstat->files,
 968                                 diffstat->alloc * sizeof(x));
 969         }
 970         diffstat->files[diffstat->nr++] = x;
 971         if (name_b) {
 972                 x->from_name = xstrdup(name_a);
 973                 x->name = xstrdup(name_b);
 974                 x->is_renamed = 1;
 975         }
 976         else {
 977                 x->from_name = NULL;
 978                 x->name = xstrdup(name_a);
 979         }
 980         return x;
 981 }
 982
 983 static void diffstat_consume(void *priv, char *line, unsigned long len)
 984 {
 985         struct diffstat_t *diffstat = priv;
 986         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 987
 988         if (line[0] == '+')
 989                 x->added++;
 990         else if (line[0] == '-')
 991                 x->deleted++;
 992 }
 993
 994 const char mime_boundary_leader[] = "------------";
 995
 996 static int scale_linear(int it, int width, int max_change)
 997 {
 998         /*
 999          * make sure that at least one '-' is printed if there were deletions,
1000          * and likewise for '+'.
1001          */
1002         if (max_change < 2)
1003                 return it;
1004         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
1005 }
1006
1007 static void show_name(FILE *file,
1008                       const char *prefix, const char *name, int len)
1009 {
1010         fprintf(file, " %s%-*s |", prefix, len, name);
1011 }
1012
1013 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1014 {
1015         if (cnt <= 0)
1016                 return;
1017         fprintf(file, "%s", set);
1018         while (cnt--)
1019                 putc(ch, file);
1020         fprintf(file, "%s", reset);
1021 }
1022
1023 static void fill_print_name(struct diffstat_file *file)
1024 {
1025         char *pname;
1026
1027         if (file->print_name)
1028                 return;
1029
1030         if (!file->is_renamed) {
1031                 struct strbuf buf = STRBUF_INIT;
1032                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1033                         pname = strbuf_detach(&buf, NULL);
1034                 } else {
1035                         pname = file->name;
1036                         strbuf_release(&buf);
1037                 }
1038         } else {
1039                 pname = pprint_rename(file->from_name, file->name);
1040         }
1041         file->print_name = pname;
1042 }
1043
1044 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1045 {
1046         int i, len, add, del, adds = 0, dels = 0;
1047         uintmax_t max_change = 0, max_len = 0;
1048         int total_files = data->nr;
1049         int width, name_width;
1050         const char *reset, *set, *add_c, *del_c;
1051
1052         if (data->nr == 0)
1053                 return;
1054
1055         width = options->stat_width ? options->stat_width : 80;
1056         name_width = options->stat_name_width ? options->stat_name_width : 50;
1057
1058         /* Sanity: give at least 5 columns to the graph,
1059          * but leave at least 10 columns for the name.
1060          */
1061         if (width < 25)
1062                 width = 25;
1063         if (name_width < 10)
1064                 name_width = 10;
1065         else if (width < name_width + 15)
1066                 name_width = width - 15;
1067
1068         /* Find the longest filename and max number of changes */
1069         reset = diff_get_color_opt(options, DIFF_RESET);
1070         set   = diff_get_color_opt(options, DIFF_PLAIN);
1071         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1072         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1073
1074         for (i = 0; i < data->nr; i++) {
1075                 struct diffstat_file *file = data->files[i];
1076                 uintmax_t change = file->added + file->deleted;
1077                 fill_print_name(file);
1078                 len = strlen(file->print_name);
1079                 if (max_len < len)
1080                         max_len = len;
1081
1082                 if (file->is_binary || file->is_unmerged)
1083                         continue;
1084                 if (max_change < change)
1085                         max_change = change;
1086         }
1087
1088         /* Compute the width of the graph part;
1089          * 10 is for one blank at the beginning of the line plus
1090          * " | count " between the name and the graph.
1091          *
1092          * From here on, name_width is the width of the name area,
1093          * and width is the width of the graph area.
1094          */
1095         name_width = (name_width < max_len) ? name_width : max_len;
1096         if (width < (name_width + 10) + max_change)
1097                 width = width - (name_width + 10);
1098         else
1099                 width = max_change;
1100
1101         for (i = 0; i < data->nr; i++) {
1102                 const char *prefix = "";
1103                 char *name = data->files[i]->print_name;
1104                 uintmax_t added = data->files[i]->added;
1105                 uintmax_t deleted = data->files[i]->deleted;
1106                 int name_len;
1107
1108                 /*
1109                  * "scale" the filename
1110                  */
1111                 len = name_width;
1112                 name_len = strlen(name);
1113                 if (name_width < name_len) {
1114                         char *slash;
1115                         prefix = "...";
1116                         len -= 3;
1117                         name += name_len - len;
1118                         slash = strchr(name, '/');
1119                         if (slash)
1120                                 name = slash;
1121                 }
1122
1123                 if (data->files[i]->is_binary) {
1124                         show_name(options->file, prefix, name, len);
1125                         fprintf(options->file, "  Bin ");
1126                         fprintf(options->file, "%s%"PRIuMAX"%s",
1127                                 del_c, deleted, reset);
1128                         fprintf(options->file, " -> ");
1129                         fprintf(options->file, "%s%"PRIuMAX"%s",
1130                                 add_c, added, reset);
1131                         fprintf(options->file, " bytes");
1132                         fprintf(options->file, "\n");
1133                         continue;
1134                 }
1135                 else if (data->files[i]->is_unmerged) {
1136                         show_name(options->file, prefix, name, len);
1137                         fprintf(options->file, "  Unmerged\n");
1138                         continue;
1139                 }
1140                 else if (!data->files[i]->is_renamed &&
1141                          (added + deleted == 0)) {
1142                         total_files--;
1143                         continue;
1144                 }
1145
1146                 /*
1147                  * scale the add/delete
1148                  */
1149                 add = added;
1150                 del = deleted;
1151                 adds += add;
1152                 dels += del;
1153
1154                 if (width <= max_change) {
1155                         add = scale_linear(add, width, max_change);
1156                         del = scale_linear(del, width, max_change);
1157                 }
1158                 show_name(options->file, prefix, name, len);
1159                 fprintf(options->file, "%5"PRIuMAX"%s", added + deleted,
1160                                 added + deleted ? " " : "");
1161                 show_graph(options->file, '+', add, add_c, reset);
1162                 show_graph(options->file, '-', del, del_c, reset);
1163                 fprintf(options->file, "\n");
1164         }
1165         fprintf(options->file,
1166                " %d files changed, %d insertions(+), %d deletions(-)\n",
1167                total_files, adds, dels);
1168 }
1169
1170 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1171 {
1172         int i, adds = 0, dels = 0, total_files = data->nr;
1173
1174         if (data->nr == 0)
1175                 return;
1176
1177         for (i = 0; i < data->nr; i++) {
1178                 if (!data->files[i]->is_binary &&
1179                     !data->files[i]->is_unmerged) {
1180                         int added = data->files[i]->added;
1181                         int deleted= data->files[i]->deleted;
1182                         if (!data->files[i]->is_renamed &&
1183                             (added + deleted == 0)) {
1184                                 total_files--;
1185                         } else {
1186                                 adds += added;
1187                                 dels += deleted;
1188                         }
1189                 }
1190         }
1191         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
1192                total_files, adds, dels);
1193 }
1194
1195 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1196 {
1197         int i;
1198
1199         if (data->nr == 0)
1200                 return;
1201
1202         for (i = 0; i < data->nr; i++) {
1203                 struct diffstat_file *file = data->files[i];
1204
1205                 if (file->is_binary)
1206                         fprintf(options->file, "-\t-\t");
1207                 else
1208                         fprintf(options->file,
1209                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1210                                 file->added, file->deleted);
1211                 if (options->line_termination) {
1212                         fill_print_name(file);
1213                         if (!file->is_renamed)
1214                                 write_name_quoted(file->name, options->file,
1215                                                   options->line_termination);
1216                         else {
1217                                 fputs(file->print_name, options->file);
1218                                 putc(options->line_termination, options->file);
1219                         }
1220                 } else {
1221                         if (file->is_renamed) {
1222                                 putc('\0', options->file);
1223                                 write_name_quoted(file->from_name, options->file, '\0');
1224                         }
1225                         write_name_quoted(file->name, options->file, '\0');
1226                 }
1227         }
1228 }
1229
1230 struct dirstat_file {
1231         const char *name;
1232         unsigned long changed;
1233 };
1234
1235 struct dirstat_dir {
1236         struct dirstat_file *files;
1237         int alloc, nr, percent, cumulative;
1238 };
1239
1240 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1241 {
1242         unsigned long this_dir = 0;
1243         unsigned int sources = 0;
1244
1245         while (dir->nr) {
1246                 struct dirstat_file *f = dir->files;
1247                 int namelen = strlen(f->name);
1248                 unsigned long this;
1249                 char *slash;
1250
1251                 if (namelen < baselen)
1252                         break;
1253                 if (memcmp(f->name, base, baselen))
1254                         break;
1255                 slash = strchr(f->name + baselen, '/');
1256                 if (slash) {
1257                         int newbaselen = slash + 1 - f->name;
1258                         this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1259                         sources++;
1260                 } else {
1261                         this = f->changed;
1262                         dir->files++;
1263                         dir->nr--;
1264                         sources += 2;
1265                 }
1266                 this_dir += this;
1267         }
1268
1269         /*
1270          * We don't report dirstat's for
1271          *  - the top level
1272          *  - or cases where everything came from a single directory
1273          *    under this directory (sources == 1).
1274          */
1275         if (baselen && sources != 1) {
1276                 int permille = this_dir * 1000 / changed;
1277                 if (permille) {
1278                         int percent = permille / 10;
1279                         if (percent >= dir->percent) {
1280                                 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1281                                 if (!dir->cumulative)
1282                                         return 0;
1283                         }
1284                 }
1285         }
1286         return this_dir;
1287 }
1288
1289 static int dirstat_compare(const void *_a, const void *_b)
1290 {
1291         const struct dirstat_file *a = _a;
1292         const struct dirstat_file *b = _b;
1293         return strcmp(a->name, b->name);
1294 }
1295
1296 static void show_dirstat(struct diff_options *options)
1297 {
1298         int i;
1299         unsigned long changed;
1300         struct dirstat_dir dir;
1301         struct diff_queue_struct *q = &diff_queued_diff;
1302
1303         dir.files = NULL;
1304         dir.alloc = 0;
1305         dir.nr = 0;
1306         dir.percent = options->dirstat_percent;
1307         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1308
1309         changed = 0;
1310         for (i = 0; i < q->nr; i++) {
1311                 struct diff_filepair *p = q->queue[i];
1312                 const char *name;
1313                 unsigned long copied, added, damage;
1314
1315                 name = p->one->path ? p->one->path : p->two->path;
1316
1317                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1318                         diff_populate_filespec(p->one, 0);
1319                         diff_populate_filespec(p->two, 0);
1320                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1321                                                &copied, &added);
1322                         diff_free_filespec_data(p->one);
1323                         diff_free_filespec_data(p->two);
1324                 } else if (DIFF_FILE_VALID(p->one)) {
1325                         diff_populate_filespec(p->one, 1);
1326                         copied = added = 0;
1327                         diff_free_filespec_data(p->one);
1328                 } else if (DIFF_FILE_VALID(p->two)) {
1329                         diff_populate_filespec(p->two, 1);
1330                         copied = 0;
1331                         added = p->two->size;
1332                         diff_free_filespec_data(p->two);
1333                 } else
1334                         continue;
1335
1336                 /*
1337                  * Original minus copied is the removed material,
1338                  * added is the new material.  They are both damages
1339                  * made to the preimage. In --dirstat-by-file mode, count
1340                  * damaged files, not damaged lines. This is done by
1341                  * counting only a single damaged line per file.
1342                  */
1343                 damage = (p->one->size - copied) + added;
1344                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1345                         damage = 1;
1346
1347                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1348                 dir.files[dir.nr].name = name;
1349                 dir.files[dir.nr].changed = damage;
1350                 changed += damage;
1351                 dir.nr++;
1352         }
1353
1354         /* This can happen even with many files, if everything was renames */
1355         if (!changed)
1356                 return;
1357
1358         /* Show all directories with more than x% of the changes */
1359         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1360         gather_dirstat(options->file, &dir, changed, "", 0);
1361 }
1362
1363 static void free_diffstat_info(struct diffstat_t *diffstat)
1364 {
1365         int i;
1366         for (i = 0; i < diffstat->nr; i++) {
1367                 struct diffstat_file *f = diffstat->files[i];
1368                 if (f->name != f->print_name)
1369                         free(f->print_name);
1370                 free(f->name);
1371                 free(f->from_name);
1372                 free(f);
1373         }
1374         free(diffstat->files);
1375 }
1376
1377 struct checkdiff_t {
1378         const char *filename;
1379         int lineno;
1380         int conflict_marker_size;
1381         struct diff_options *o;
1382         unsigned ws_rule;
1383         unsigned status;
1384 };
1385
1386 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
1387 {
1388         char firstchar;
1389         int cnt;
1390
1391         if (len < marker_size + 1)
1392                 return 0;
1393         firstchar = line[0];
1394         switch (firstchar) {
1395         case '=': case '>': case '<': case '|':
1396                 break;
1397         default:
1398                 return 0;
1399         }
1400         for (cnt = 1; cnt < marker_size; cnt++)
1401                 if (line[cnt] != firstchar)
1402                         return 0;
1403         /* line[1] thru line[marker_size-1] are same as firstchar */
1404         if (len < marker_size + 1 || !isspace(line[marker_size]))
1405                 return 0;
1406         return 1;
1407 }
1408
1409 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1410 {
1411         struct checkdiff_t *data = priv;
1412         int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1413         int marker_size = data->conflict_marker_size;
1414         const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1415         const char *reset = diff_get_color(color_diff, DIFF_RESET);
1416         const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1417         char *err;
1418
1419         if (line[0] == '+') {
1420                 unsigned bad;
1421                 data->lineno++;
1422                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
1423                         data->status |= 1;
1424                         fprintf(data->o->file,
1425                                 "%s:%d: leftover conflict marker\n",
1426                                 data->filename, data->lineno);
1427                 }
1428                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1429                 if (!bad)
1430                         return;
1431                 data->status |= bad;
1432                 err = whitespace_error_string(bad);
1433                 fprintf(data->o->file, "%s:%d: %s.\n",
1434                         data->filename, data->lineno, err);
1435                 free(err);
1436                 emit_line(data->o->file, set, reset, line, 1);
1437                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1438                               data->o->file, set, reset, ws);
1439         } else if (line[0] == ' ') {
1440                 data->lineno++;
1441         } else if (line[0] == '@') {
1442                 char *plus = strchr(line, '+');
1443                 if (plus)
1444                         data->lineno = strtol(plus, NULL, 10) - 1;
1445                 else
1446                         die("invalid diff");
1447         }
1448 }
1449
1450 static unsigned char *deflate_it(char *data,
1451                                  unsigned long size,
1452                                  unsigned long *result_size)
1453 {
1454         int bound;
1455         unsigned char *deflated;
1456         z_stream stream;
1457
1458         memset(&stream, 0, sizeof(stream));
1459         deflateInit(&stream, zlib_compression_level);
1460         bound = deflateBound(&stream, size);
1461         deflated = xmalloc(bound);
1462         stream.next_out = deflated;
1463         stream.avail_out = bound;
1464
1465         stream.next_in = (unsigned char *)data;
1466         stream.avail_in = size;
1467         while (deflate(&stream, Z_FINISH) == Z_OK)
1468                 ; /* nothing */
1469         deflateEnd(&stream);
1470         *result_size = stream.total_out;
1471         return deflated;
1472 }
1473
1474 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1475 {
1476         void *cp;
1477         void *delta;
1478         void *deflated;
1479         void *data;
1480         unsigned long orig_size;
1481         unsigned long delta_size;
1482         unsigned long deflate_size;
1483         unsigned long data_size;
1484
1485         /* We could do deflated delta, or we could do just deflated two,
1486          * whichever is smaller.
1487          */
1488         delta = NULL;
1489         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1490         if (one->size && two->size) {
1491                 delta = diff_delta(one->ptr, one->size,
1492                                    two->ptr, two->size,
1493                                    &delta_size, deflate_size);
1494                 if (delta) {
1495                         void *to_free = delta;
1496                         orig_size = delta_size;
1497                         delta = deflate_it(delta, delta_size, &delta_size);
1498                         free(to_free);
1499                 }
1500         }
1501
1502         if (delta && delta_size < deflate_size) {
1503                 fprintf(file, "delta %lu\n", orig_size);
1504                 free(deflated);
1505                 data = delta;
1506                 data_size = delta_size;
1507         }
1508         else {
1509                 fprintf(file, "literal %lu\n", two->size);
1510                 free(delta);
1511                 data = deflated;
1512                 data_size = deflate_size;
1513         }
1514
1515         /* emit data encoded in base85 */
1516         cp = data;
1517         while (data_size) {
1518                 int bytes = (52 < data_size) ? 52 : data_size;
1519                 char line[70];
1520                 data_size -= bytes;
1521                 if (bytes <= 26)
1522                         line[0] = bytes + 'A' - 1;
1523                 else
1524                         line[0] = bytes - 26 + 'a' - 1;
1525                 encode_85(line + 1, cp, bytes);
1526                 cp = (char *) cp + bytes;
1527                 fputs(line, file);
1528                 fputc('\n', file);
1529         }
1530         fprintf(file, "\n");
1531         free(data);
1532 }
1533
1534 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1535 {
1536         fprintf(file, "GIT binary patch\n");
1537         emit_binary_diff_body(file, one, two);
1538         emit_binary_diff_body(file, two, one);
1539 }
1540
1541 static void diff_filespec_load_driver(struct diff_filespec *one)
1542 {
1543         if (!one->driver)
1544                 one->driver = userdiff_find_by_path(one->path);
1545         if (!one->driver)
1546                 one->driver = userdiff_find_by_name("default");
1547 }
1548
1549 int diff_filespec_is_binary(struct diff_filespec *one)
1550 {
1551         if (one->is_binary == -1) {
1552                 diff_filespec_load_driver(one);
1553                 if (one->driver->binary != -1)
1554                         one->is_binary = one->driver->binary;
1555                 else {
1556                         if (!one->data && DIFF_FILE_VALID(one))
1557                                 diff_populate_filespec(one, 0);
1558                         if (one->data)
1559                                 one->is_binary = buffer_is_binary(one->data,
1560                                                 one->size);
1561                         if (one->is_binary == -1)
1562                                 one->is_binary = 0;
1563                 }
1564         }
1565         return one->is_binary;
1566 }
1567
1568 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1569 {
1570         diff_filespec_load_driver(one);
1571         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1572 }
1573
1574 static const char *userdiff_word_regex(struct diff_filespec *one)
1575 {
1576         diff_filespec_load_driver(one);
1577         return one->driver->word_regex;
1578 }
1579
1580 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1581 {
1582         if (!options->a_prefix)
1583                 options->a_prefix = a;
1584         if (!options->b_prefix)
1585                 options->b_prefix = b;
1586 }
1587
1588 static const char *get_textconv(struct diff_filespec *one)
1589 {
1590         if (!DIFF_FILE_VALID(one))
1591                 return NULL;
1592         if (!S_ISREG(one->mode))
1593                 return NULL;
1594         diff_filespec_load_driver(one);
1595         return one->driver->textconv;
1596 }
1597
1598 static void builtin_diff(const char *name_a,
1599                          const char *name_b,
1600                          struct diff_filespec *one,
1601                          struct diff_filespec *two,
1602                          const char *xfrm_msg,
1603                          int must_show_header,
1604                          struct diff_options *o,
1605                          int complete_rewrite)
1606 {
1607         mmfile_t mf1, mf2;
1608         const char *lbl[2];
1609         char *a_one, *b_two;
1610         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1611         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1612         const char *a_prefix, *b_prefix;
1613         const char *textconv_one = NULL, *textconv_two = NULL;
1614         struct strbuf header = STRBUF_INIT;
1615
1616         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
1617                         (!one->mode || S_ISGITLINK(one->mode)) &&
1618                         (!two->mode || S_ISGITLINK(two->mode))) {
1619                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
1620                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
1621                 show_submodule_summary(o->file, one ? one->path : two->path,
1622                                 one->sha1, two->sha1, two->dirty_submodule,
1623                                 del, add, reset);
1624                 return;
1625         }
1626
1627         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
1628                 textconv_one = get_textconv(one);
1629                 textconv_two = get_textconv(two);
1630         }
1631
1632         diff_set_mnemonic_prefix(o, "a/", "b/");
1633         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1634                 a_prefix = o->b_prefix;
1635                 b_prefix = o->a_prefix;
1636         } else {
1637                 a_prefix = o->a_prefix;
1638                 b_prefix = o->b_prefix;
1639         }
1640
1641         /* Never use a non-valid filename anywhere if at all possible */
1642         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1643         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1644
1645         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1646         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1647         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1648         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1649         strbuf_addf(&header, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1650         if (lbl[0][0] == '/') {
1651                 /* /dev/null */
1652                 strbuf_addf(&header, "%snew file mode %06o%s\n", set, two->mode, reset);
1653                 if (xfrm_msg)
1654                         strbuf_addstr(&header, xfrm_msg);
1655                 must_show_header = 1;
1656         }
1657         else if (lbl[1][0] == '/') {
1658                 strbuf_addf(&header, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1659                 if (xfrm_msg)
1660                         strbuf_addstr(&header, xfrm_msg);
1661                 must_show_header = 1;
1662         }
1663         else {
1664                 if (one->mode != two->mode) {
1665                         strbuf_addf(&header, "%sold mode %06o%s\n", set, one->mode, reset);
1666                         strbuf_addf(&header, "%snew mode %06o%s\n", set, two->mode, reset);
1667                         must_show_header = 1;
1668                 }
1669                 if (xfrm_msg)
1670                         strbuf_addstr(&header, xfrm_msg);
1671
1672                 /*
1673                  * we do not run diff between different kind
1674                  * of objects.
1675                  */
1676                 if ((one->mode ^ two->mode) & S_IFMT)
1677                         goto free_ab_and_return;
1678                 if (complete_rewrite &&
1679                     (textconv_one || !diff_filespec_is_binary(one)) &&
1680                     (textconv_two || !diff_filespec_is_binary(two))) {
1681                         fprintf(o->file, "%s", header.buf);
1682                         strbuf_reset(&header);
1683                         emit_rewrite_diff(name_a, name_b, one, two,
1684                                                 textconv_one, textconv_two, o);
1685                         o->found_changes = 1;
1686                         goto free_ab_and_return;
1687                 }
1688         }
1689
1690         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1691                 die("unable to read files to diff");
1692
1693         if (!DIFF_OPT_TST(o, TEXT) &&
1694             ( (diff_filespec_is_binary(one) && !textconv_one) ||
1695               (diff_filespec_is_binary(two) && !textconv_two) )) {
1696                 /* Quite common confusing case */
1697                 if (mf1.size == mf2.size &&
1698                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
1699                         if (must_show_header)
1700                                 fprintf(o->file, "%s", header.buf);
1701                         goto free_ab_and_return;
1702                 }
1703                 fprintf(o->file, "%s", header.buf);
1704                 strbuf_reset(&header);
1705                 if (DIFF_OPT_TST(o, BINARY))
1706                         emit_binary_diff(o->file, &mf1, &mf2);
1707                 else
1708                         fprintf(o->file, "Binary files %s and %s differ\n",
1709                                 lbl[0], lbl[1]);
1710                 o->found_changes = 1;
1711         }
1712         else {
1713                 /* Crazy xdl interfaces.. */
1714                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1715                 xpparam_t xpp;
1716                 xdemitconf_t xecfg;
1717                 struct emit_callback ecbdata;
1718                 const struct userdiff_funcname *pe;
1719
1720                 if (!DIFF_XDL_TST(o, WHITESPACE_FLAGS) || must_show_header) {
1721                         fprintf(o->file, "%s", header.buf);
1722                         strbuf_reset(&header);
1723                 }
1724
1725                 if (textconv_one) {
1726                         size_t size;
1727                         mf1.ptr = run_textconv(textconv_one, one, &size);
1728                         if (!mf1.ptr)
1729                                 die("unable to read files to diff");
1730                         mf1.size = size;
1731                 }
1732                 if (textconv_two) {
1733                         size_t size;
1734                         mf2.ptr = run_textconv(textconv_two, two, &size);
1735                         if (!mf2.ptr)
1736                                 die("unable to read files to diff");
1737                         mf2.size = size;
1738                 }
1739
1740                 pe = diff_funcname_pattern(one);
1741                 if (!pe)
1742                         pe = diff_funcname_pattern(two);
1743
1744                 memset(&xpp, 0, sizeof(xpp));
1745                 memset(&xecfg, 0, sizeof(xecfg));
1746                 memset(&ecbdata, 0, sizeof(ecbdata));
1747                 ecbdata.label_path = lbl;
1748                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1749                 ecbdata.found_changesp = &o->found_changes;
1750                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1751                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
1752                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
1753                 ecbdata.file = o->file;
1754                 ecbdata.header = header.len ? &header : NULL;
1755                 xpp.flags = o->xdl_opts;
1756                 xecfg.ctxlen = o->context;
1757                 xecfg.interhunkctxlen = o->interhunkcontext;
1758                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1759                 if (pe)
1760                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1761                 if (!diffopts)
1762                         ;
1763                 else if (!prefixcmp(diffopts, "--unified="))
1764                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1765                 else if (!prefixcmp(diffopts, "-u"))
1766                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1767                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1768                         ecbdata.diff_words =
1769                                 xcalloc(1, sizeof(struct diff_words_data));
1770                         ecbdata.diff_words->file = o->file;
1771                         if (!o->word_regex)
1772                                 o->word_regex = userdiff_word_regex(one);
1773                         if (!o->word_regex)
1774                                 o->word_regex = userdiff_word_regex(two);
1775                         if (!o->word_regex)
1776                                 o->word_regex = diff_word_regex_cfg;
1777                         if (o->word_regex) {
1778                                 ecbdata.diff_words->word_regex = (regex_t *)
1779                                         xmalloc(sizeof(regex_t));
1780                                 if (regcomp(ecbdata.diff_words->word_regex,
1781                                                 o->word_regex,
1782                                                 REG_EXTENDED | REG_NEWLINE))
1783                                         die ("Invalid regular expression: %s",
1784                                                         o->word_regex);
1785                         }
1786                 }
1787                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1788                               &xpp, &xecfg);
1789                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1790                         free_diff_words_data(&ecbdata);
1791                 if (textconv_one)
1792                         free(mf1.ptr);
1793                 if (textconv_two)
1794                         free(mf2.ptr);
1795                 xdiff_clear_find_func(&xecfg);
1796         }
1797
1798  free_ab_and_return:
1799         strbuf_release(&header);
1800         diff_free_filespec_data(one);
1801         diff_free_filespec_data(two);
1802         free(a_one);
1803         free(b_two);
1804         return;
1805 }
1806
1807 static void builtin_diffstat(const char *name_a, const char *name_b,
1808                              struct diff_filespec *one,
1809                              struct diff_filespec *two,
1810                              struct diffstat_t *diffstat,
1811                              struct diff_options *o,
1812                              int complete_rewrite)
1813 {
1814         mmfile_t mf1, mf2;
1815         struct diffstat_file *data;
1816
1817         data = diffstat_add(diffstat, name_a, name_b);
1818
1819         if (!one || !two) {
1820                 data->is_unmerged = 1;
1821                 return;
1822         }
1823         if (complete_rewrite) {
1824                 diff_populate_filespec(one, 0);
1825                 diff_populate_filespec(two, 0);
1826                 data->deleted = count_lines(one->data, one->size);
1827                 data->added = count_lines(two->data, two->size);
1828                 goto free_and_return;
1829         }
1830         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1831                 die("unable to read files to diff");
1832
1833         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1834                 data->is_binary = 1;
1835                 data->added = mf2.size;
1836                 data->deleted = mf1.size;
1837         } else {
1838                 /* Crazy xdl interfaces.. */
1839                 xpparam_t xpp;
1840                 xdemitconf_t xecfg;
1841
1842                 memset(&xpp, 0, sizeof(xpp));
1843                 memset(&xecfg, 0, sizeof(xecfg));
1844                 xpp.flags = o->xdl_opts;
1845                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1846                               &xpp, &xecfg);
1847         }
1848
1849  free_and_return:
1850         diff_free_filespec_data(one);
1851         diff_free_filespec_data(two);
1852 }
1853
1854 static void builtin_checkdiff(const char *name_a, const char *name_b,
1855                               const char *attr_path,
1856                               struct diff_filespec *one,
1857                               struct diff_filespec *two,
1858                               struct diff_options *o)
1859 {
1860         mmfile_t mf1, mf2;
1861         struct checkdiff_t data;
1862
1863         if (!two)
1864                 return;
1865
1866         memset(&data, 0, sizeof(data));
1867         data.filename = name_b ? name_b : name_a;
1868         data.lineno = 0;
1869         data.o = o;
1870         data.ws_rule = whitespace_rule(attr_path);
1871         data.conflict_marker_size = ll_merge_marker_size(attr_path);
1872
1873         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1874                 die("unable to read files to diff");
1875
1876         /*
1877          * All the other codepaths check both sides, but not checking
1878          * the "old" side here is deliberate.  We are checking the newly
1879          * introduced changes, and as long as the "new" side is text, we
1880          * can and should check what it introduces.
1881          */
1882         if (diff_filespec_is_binary(two))
1883                 goto free_and_return;
1884         else {
1885                 /* Crazy xdl interfaces.. */
1886                 xpparam_t xpp;
1887                 xdemitconf_t xecfg;
1888
1889                 memset(&xpp, 0, sizeof(xpp));
1890                 memset(&xecfg, 0, sizeof(xecfg));
1891                 xecfg.ctxlen = 1; /* at least one context line */
1892                 xpp.flags = 0;
1893                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1894                               &xpp, &xecfg);
1895
1896                 if (data.ws_rule & WS_BLANK_AT_EOF) {
1897                         struct emit_callback ecbdata;
1898                         int blank_at_eof;
1899
1900                         ecbdata.ws_rule = data.ws_rule;
1901                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
1902                         blank_at_eof = ecbdata.blank_at_eof_in_preimage;
1903
1904                         if (blank_at_eof) {
1905                                 static char *err;
1906                                 if (!err)
1907                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
1908                                 fprintf(o->file, "%s:%d: %s.\n",
1909                                         data.filename, blank_at_eof, err);
1910                                 data.status = 1; /* report errors */
1911                         }
1912                 }
1913         }
1914  free_and_return:
1915         diff_free_filespec_data(one);
1916         diff_free_filespec_data(two);
1917         if (data.status)
1918                 DIFF_OPT_SET(o, CHECK_FAILED);
1919 }
1920
1921 struct diff_filespec *alloc_filespec(const char *path)
1922 {
1923         int namelen = strlen(path);
1924         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1925
1926         memset(spec, 0, sizeof(*spec));
1927         spec->path = (char *)(spec + 1);
1928         memcpy(spec->path, path, namelen+1);
1929         spec->count = 1;
1930         spec->is_binary = -1;
1931         return spec;
1932 }
1933
1934 void free_filespec(struct diff_filespec *spec)
1935 {
1936         if (!--spec->count) {
1937                 diff_free_filespec_data(spec);
1938                 free(spec);
1939         }
1940 }
1941
1942 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1943                    unsigned short mode)
1944 {
1945         if (mode) {
1946                 spec->mode = canon_mode(mode);
1947                 hashcpy(spec->sha1, sha1);
1948                 spec->sha1_valid = !is_null_sha1(sha1);
1949         }
1950 }
1951
1952 /*
1953  * Given a name and sha1 pair, if the index tells us the file in
1954  * the work tree has that object contents, return true, so that
1955  * prepare_temp_file() does not have to inflate and extract.
1956  */
1957 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1958 {
1959         struct cache_entry *ce;
1960         struct stat st;
1961         int pos, len;
1962
1963         /*
1964          * We do not read the cache ourselves here, because the
1965          * benchmark with my previous version that always reads cache
1966          * shows that it makes things worse for diff-tree comparing
1967          * two linux-2.6 kernel trees in an already checked out work
1968          * tree.  This is because most diff-tree comparisons deal with
1969          * only a small number of files, while reading the cache is
1970          * expensive for a large project, and its cost outweighs the
1971          * savings we get by not inflating the object to a temporary
1972          * file.  Practically, this code only helps when we are used
1973          * by diff-cache --cached, which does read the cache before
1974          * calling us.
1975          */
1976         if (!active_cache)
1977                 return 0;
1978
1979         /* We want to avoid the working directory if our caller
1980          * doesn't need the data in a normal file, this system
1981          * is rather slow with its stat/open/mmap/close syscalls,
1982          * and the object is contained in a pack file.  The pack
1983          * is probably already open and will be faster to obtain
1984          * the data through than the working directory.  Loose
1985          * objects however would tend to be slower as they need
1986          * to be individually opened and inflated.
1987          */
1988         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
1989                 return 0;
1990
1991         len = strlen(name);
1992         pos = cache_name_pos(name, len);
1993         if (pos < 0)
1994                 return 0;
1995         ce = active_cache[pos];
1996
1997         /*
1998          * This is not the sha1 we are looking for, or
1999          * unreusable because it is not a regular file.
2000          */
2001         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2002                 return 0;
2003
2004         /*
2005          * If ce is marked as "assume unchanged", there is no
2006          * guarantee that work tree matches what we are looking for.
2007          */
2008         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2009                 return 0;
2010
2011         /*
2012          * If ce matches the file in the work tree, we can reuse it.
2013          */
2014         if (ce_uptodate(ce) ||
2015             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2016                 return 1;
2017
2018         return 0;
2019 }
2020
2021 static int populate_from_stdin(struct diff_filespec *s)
2022 {
2023         struct strbuf buf = STRBUF_INIT;
2024         size_t size = 0;
2025
2026         if (strbuf_read(&buf, 0, 0) < 0)
2027                 return error("error while reading from stdin %s",
2028                                      strerror(errno));
2029
2030         s->should_munmap = 0;
2031         s->data = strbuf_detach(&buf, &size);
2032         s->size = size;
2033         s->should_free = 1;
2034         return 0;
2035 }
2036
2037 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2038 {
2039         int len;
2040         char *data = xmalloc(100), *dirty = "";
2041
2042         /* Are we looking at the work tree? */
2043         if (s->dirty_submodule)
2044                 dirty = "-dirty";
2045
2046         len = snprintf(data, 100,
2047                        "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2048         s->data = data;
2049         s->size = len;
2050         s->should_free = 1;
2051         if (size_only) {
2052                 s->data = NULL;
2053                 free(data);
2054         }
2055         return 0;
2056 }
2057
2058 /*
2059  * While doing rename detection and pickaxe operation, we may need to
2060  * grab the data for the blob (or file) for our own in-core comparison.
2061  * diff_filespec has data and size fields for this purpose.
2062  */
2063 int diff_populate_filespec(struct diff_filespec *s, int size_only)
2064 {
2065         int err = 0;
2066         if (!DIFF_FILE_VALID(s))
2067                 die("internal error: asking to populate invalid file.");
2068         if (S_ISDIR(s->mode))
2069                 return -1;
2070
2071         if (s->data)
2072                 return 0;
2073
2074         if (size_only && 0 < s->size)
2075                 return 0;
2076
2077         if (S_ISGITLINK(s->mode))
2078                 return diff_populate_gitlink(s, size_only);
2079
2080         if (!s->sha1_valid ||
2081             reuse_worktree_file(s->path, s->sha1, 0)) {
2082                 struct strbuf buf = STRBUF_INIT;
2083                 struct stat st;
2084                 int fd;
2085
2086                 if (!strcmp(s->path, "-"))
2087                         return populate_from_stdin(s);
2088
2089                 if (lstat(s->path, &st) < 0) {
2090                         if (errno == ENOENT) {
2091                         err_empty:
2092                                 err = -1;
2093                         empty:
2094                                 s->data = (char *)"";
2095                                 s->size = 0;
2096                                 return err;
2097                         }
2098                 }
2099                 s->size = xsize_t(st.st_size);
2100                 if (!s->size)
2101                         goto empty;
2102                 if (S_ISLNK(st.st_mode)) {
2103                         struct strbuf sb = STRBUF_INIT;
2104
2105                         if (strbuf_readlink(&sb, s->path, s->size))
2106                                 goto err_empty;
2107                         s->size = sb.len;
2108                         s->data = strbuf_detach(&sb, NULL);
2109                         s->should_free = 1;
2110                         return 0;
2111                 }
2112                 if (size_only)
2113                         return 0;
2114                 fd = open(s->path, O_RDONLY);
2115                 if (fd < 0)
2116                         goto err_empty;
2117                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2118                 close(fd);
2119                 s->should_munmap = 1;
2120
2121                 /*
2122                  * Convert from working tree format to canonical git format
2123                  */
2124                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2125                         size_t size = 0;
2126                         munmap(s->data, s->size);
2127                         s->should_munmap = 0;
2128                         s->data = strbuf_detach(&buf, &size);
2129                         s->size = size;
2130                         s->should_free = 1;
2131                 }
2132         }
2133         else {
2134                 enum object_type type;
2135                 if (size_only)
2136                         type = sha1_object_info(s->sha1, &s->size);
2137                 else {
2138                         s->data = read_sha1_file(s->sha1, &type, &s->size);
2139                         s->should_free = 1;
2140                 }
2141         }
2142         return 0;
2143 }
2144
2145 void diff_free_filespec_blob(struct diff_filespec *s)
2146 {
2147         if (s->should_free)
2148                 free(s->data);
2149         else if (s->should_munmap)
2150                 munmap(s->data, s->size);
2151
2152         if (s->should_free || s->should_munmap) {
2153                 s->should_free = s->should_munmap = 0;
2154                 s->data = NULL;
2155         }
2156 }
2157
2158 void diff_free_filespec_data(struct diff_filespec *s)
2159 {
2160         diff_free_filespec_blob(s);
2161         free(s->cnt_data);
2162         s->cnt_data = NULL;
2163 }
2164
2165 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2166                            void *blob,
2167                            unsigned long size,
2168                            const unsigned char *sha1,
2169                            int mode)
2170 {
2171         int fd;
2172         struct strbuf buf = STRBUF_INIT;
2173         struct strbuf template = STRBUF_INIT;
2174         char *path_dup = xstrdup(path);
2175         const char *base = basename(path_dup);
2176
2177         /* Generate "XXXXXX_basename.ext" */
2178         strbuf_addstr(&template, "XXXXXX_");
2179         strbuf_addstr(&template, base);
2180
2181         fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2182                         strlen(base) + 1);
2183         if (fd < 0)
2184                 die_errno("unable to create temp-file");
2185         if (convert_to_working_tree(path,
2186                         (const char *)blob, (size_t)size, &buf)) {
2187                 blob = buf.buf;
2188                 size = buf.len;
2189         }
2190         if (write_in_full(fd, blob, size) != size)
2191                 die_errno("unable to write temp-file");
2192         close(fd);
2193         temp->name = temp->tmp_path;
2194         strcpy(temp->hex, sha1_to_hex(sha1));
2195         temp->hex[40] = 0;
2196         sprintf(temp->mode, "%06o", mode);
2197         strbuf_release(&buf);
2198         strbuf_release(&template);
2199         free(path_dup);
2200 }
2201
2202 static struct diff_tempfile *prepare_temp_file(const char *name,
2203                 struct diff_filespec *one)
2204 {
2205         struct diff_tempfile *temp = claim_diff_tempfile();
2206
2207         if (!DIFF_FILE_VALID(one)) {
2208         not_a_valid_file:
2209                 /* A '-' entry produces this for file-2, and
2210                  * a '+' entry produces this for file-1.
2211                  */
2212                 temp->name = "/dev/null";
2213                 strcpy(temp->hex, ".");
2214                 strcpy(temp->mode, ".");
2215                 return temp;
2216         }
2217
2218         if (!remove_tempfile_installed) {
2219                 atexit(remove_tempfile);
2220                 sigchain_push_common(remove_tempfile_on_signal);
2221                 remove_tempfile_installed = 1;
2222         }
2223
2224         if (!one->sha1_valid ||
2225             reuse_worktree_file(name, one->sha1, 1)) {
2226                 struct stat st;
2227                 if (lstat(name, &st) < 0) {
2228                         if (errno == ENOENT)
2229                                 goto not_a_valid_file;
2230                         die_errno("stat(%s)", name);
2231                 }
2232                 if (S_ISLNK(st.st_mode)) {
2233                         struct strbuf sb = STRBUF_INIT;
2234                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2235                                 die_errno("readlink(%s)", name);
2236                         prep_temp_blob(name, temp, sb.buf, sb.len,
2237                                        (one->sha1_valid ?
2238                                         one->sha1 : null_sha1),
2239                                        (one->sha1_valid ?
2240                                         one->mode : S_IFLNK));
2241                         strbuf_release(&sb);
2242                 }
2243                 else {
2244                         /* we can borrow from the file in the work tree */
2245                         temp->name = name;
2246                         if (!one->sha1_valid)
2247                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2248                         else
2249                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2250                         /* Even though we may sometimes borrow the
2251                          * contents from the work tree, we always want
2252                          * one->mode.  mode is trustworthy even when
2253                          * !(one->sha1_valid), as long as
2254                          * DIFF_FILE_VALID(one).
2255                          */
2256                         sprintf(temp->mode, "%06o", one->mode);
2257                 }
2258                 return temp;
2259         }
2260         else {
2261                 if (diff_populate_filespec(one, 0))
2262                         die("cannot read data blob for %s", one->path);
2263                 prep_temp_blob(name, temp, one->data, one->size,
2264                                one->sha1, one->mode);
2265         }
2266         return temp;
2267 }
2268
2269 /* An external diff command takes:
2270  *
2271  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2272  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2273  *
2274  */
2275 static void run_external_diff(const char *pgm,
2276                               const char *name,
2277                               const char *other,
2278                               struct diff_filespec *one,
2279                               struct diff_filespec *two,
2280                               const char *xfrm_msg,
2281                               int complete_rewrite)
2282 {
2283         const char *spawn_arg[10];
2284         int retval;
2285         const char **arg = &spawn_arg[0];
2286
2287         if (one && two) {
2288                 struct diff_tempfile *temp_one, *temp_two;
2289                 const char *othername = (other ? other : name);
2290                 temp_one = prepare_temp_file(name, one);
2291                 temp_two = prepare_temp_file(othername, two);
2292                 *arg++ = pgm;
2293                 *arg++ = name;
2294                 *arg++ = temp_one->name;
2295                 *arg++ = temp_one->hex;
2296                 *arg++ = temp_one->mode;
2297                 *arg++ = temp_two->name;
2298                 *arg++ = temp_two->hex;
2299                 *arg++ = temp_two->mode;
2300                 if (other) {
2301                         *arg++ = other;
2302                         *arg++ = xfrm_msg;
2303                 }
2304         } else {
2305                 *arg++ = pgm;
2306                 *arg++ = name;
2307         }
2308         *arg = NULL;
2309         fflush(NULL);
2310         retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2311         remove_tempfile();
2312         if (retval) {
2313                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2314                 exit(1);
2315         }
2316 }
2317
2318 static int similarity_index(struct diff_filepair *p)
2319 {
2320         return p->score * 100 / MAX_SCORE;
2321 }
2322
2323 static void fill_metainfo(struct strbuf *msg,
2324                           const char *name,
2325                           const char *other,
2326                           struct diff_filespec *one,
2327                           struct diff_filespec *two,
2328                           struct diff_options *o,
2329                           struct diff_filepair *p,
2330                           int *must_show_header,
2331                           int use_color)
2332 {
2333         const char *set = diff_get_color(use_color, DIFF_METAINFO);
2334         const char *reset = diff_get_color(use_color, DIFF_RESET);
2335
2336         *must_show_header = 1;
2337         strbuf_init(msg, PATH_MAX * 2 + 300);
2338         switch (p->status) {
2339         case DIFF_STATUS_COPIED:
2340                 strbuf_addf(msg, "%ssimilarity index %d%%",
2341                             set, similarity_index(p));
2342                 strbuf_addf(msg, "%s\n%scopy from ", reset, set);
2343                 quote_c_style(name, msg, NULL, 0);
2344                 strbuf_addf(msg, "%s\n%scopy to ", reset, set);
2345                 quote_c_style(other, msg, NULL, 0);
2346                 strbuf_addf(msg, "%s\n", reset);
2347                 break;
2348         case DIFF_STATUS_RENAMED:
2349                 strbuf_addf(msg, "%ssimilarity index %d%%",
2350                             set, similarity_index(p));
2351                 strbuf_addf(msg, "%s\n%srename from ", reset, set);
2352                 quote_c_style(name, msg, NULL, 0);
2353                 strbuf_addf(msg, "%s\n%srename to ", reset, set);
2354                 quote_c_style(other, msg, NULL, 0);
2355                 strbuf_addf(msg, "%s\n", reset);
2356                 break;
2357         case DIFF_STATUS_MODIFIED:
2358                 if (p->score) {
2359                         strbuf_addf(msg, "%sdissimilarity index %d%%%s\n",
2360                                     set, similarity_index(p), reset);
2361                         break;
2362                 }
2363                 /* fallthru */
2364         default:
2365                 *must_show_header = 0;
2366         }
2367         if (one && two && hashcmp(one->sha1, two->sha1)) {
2368                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2369
2370                 if (DIFF_OPT_TST(o, BINARY)) {
2371                         mmfile_t mf;
2372                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2373                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2374                                 abbrev = 40;
2375                 }
2376                 strbuf_addf(msg, "%sindex %.*s..%.*s", set,
2377                             abbrev, sha1_to_hex(one->sha1),
2378                             abbrev, sha1_to_hex(two->sha1));
2379                 if (one->mode == two->mode)
2380                         strbuf_addf(msg, " %06o", one->mode);
2381                 strbuf_addf(msg, "%s\n", reset);
2382         }
2383 }
2384
2385 static void run_diff_cmd(const char *pgm,
2386                          const char *name,
2387                          const char *other,
2388                          const char *attr_path,
2389                          struct diff_filespec *one,
2390                          struct diff_filespec *two,
2391                          struct strbuf *msg,
2392                          struct diff_options *o,
2393                          struct diff_filepair *p)
2394 {
2395         const char *xfrm_msg = NULL;
2396         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2397         int must_show_header = 0;
2398
2399         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2400                 pgm = NULL;
2401         else {
2402                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2403                 if (drv && drv->external)
2404                         pgm = drv->external;
2405         }
2406
2407         if (msg) {
2408                 /*
2409                  * don't use colors when the header is intended for an
2410                  * external diff driver
2411                  */
2412                 fill_metainfo(msg, name, other, one, two, o, p,
2413                               &must_show_header,
2414                               DIFF_OPT_TST(o, COLOR_DIFF) && !pgm);
2415                 xfrm_msg = msg->len ? msg->buf : NULL;
2416         }
2417
2418         if (pgm) {
2419                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2420                                   complete_rewrite);
2421                 return;
2422         }
2423         if (one && two)
2424                 builtin_diff(name, other ? other : name,
2425                              one, two, xfrm_msg, must_show_header,
2426                              o, complete_rewrite);
2427         else
2428                 fprintf(o->file, "* Unmerged path %s\n", name);
2429 }
2430
2431 static void diff_fill_sha1_info(struct diff_filespec *one)
2432 {
2433         if (DIFF_FILE_VALID(one)) {
2434                 if (!one->sha1_valid) {
2435                         struct stat st;
2436                         if (!strcmp(one->path, "-")) {
2437                                 hashcpy(one->sha1, null_sha1);
2438                                 return;
2439                         }
2440                         if (lstat(one->path, &st) < 0)
2441                                 die_errno("stat '%s'", one->path);
2442                         if (index_path(one->sha1, one->path, &st, 0))
2443                                 die("cannot hash %s", one->path);
2444                 }
2445         }
2446         else
2447                 hashclr(one->sha1);
2448 }
2449
2450 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2451 {
2452         /* Strip the prefix but do not molest /dev/null and absolute paths */
2453         if (*namep && **namep != '/')
2454                 *namep += prefix_length;
2455         if (*otherp && **otherp != '/')
2456                 *otherp += prefix_length;
2457 }
2458
2459 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2460 {
2461         const char *pgm = external_diff();
2462         struct strbuf msg;
2463         struct diff_filespec *one = p->one;
2464         struct diff_filespec *two = p->two;
2465         const char *name;
2466         const char *other;
2467         const char *attr_path;
2468
2469         name  = p->one->path;
2470         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2471         attr_path = name;
2472         if (o->prefix_length)
2473                 strip_prefix(o->prefix_length, &name, &other);
2474
2475         if (DIFF_PAIR_UNMERGED(p)) {
2476                 run_diff_cmd(pgm, name, NULL, attr_path,
2477                              NULL, NULL, NULL, o, p);
2478                 return;
2479         }
2480
2481         diff_fill_sha1_info(one);
2482         diff_fill_sha1_info(two);
2483
2484         if (!pgm &&
2485             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2486             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2487                 /*
2488                  * a filepair that changes between file and symlink
2489                  * needs to be split into deletion and creation.
2490                  */
2491                 struct diff_filespec *null = alloc_filespec(two->path);
2492                 run_diff_cmd(NULL, name, other, attr_path,
2493                              one, null, &msg, o, p);
2494                 free(null);
2495                 strbuf_release(&msg);
2496
2497                 null = alloc_filespec(one->path);
2498                 run_diff_cmd(NULL, name, other, attr_path,
2499                              null, two, &msg, o, p);
2500                 free(null);
2501         }
2502         else
2503                 run_diff_cmd(pgm, name, other, attr_path,
2504                              one, two, &msg, o, p);
2505
2506         strbuf_release(&msg);
2507 }
2508
2509 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2510                          struct diffstat_t *diffstat)
2511 {
2512         const char *name;
2513         const char *other;
2514         int complete_rewrite = 0;
2515
2516         if (DIFF_PAIR_UNMERGED(p)) {
2517                 /* unmerged */
2518                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2519                 return;
2520         }
2521
2522         name = p->one->path;
2523         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2524
2525         if (o->prefix_length)
2526                 strip_prefix(o->prefix_length, &name, &other);
2527
2528         diff_fill_sha1_info(p->one);
2529         diff_fill_sha1_info(p->two);
2530
2531         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2532                 complete_rewrite = 1;
2533         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2534 }
2535
2536 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2537 {
2538         const char *name;
2539         const char *other;
2540         const char *attr_path;
2541
2542         if (DIFF_PAIR_UNMERGED(p)) {
2543                 /* unmerged */
2544                 return;
2545         }
2546
2547         name = p->one->path;
2548         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2549         attr_path = other ? other : name;
2550
2551         if (o->prefix_length)
2552                 strip_prefix(o->prefix_length, &name, &other);
2553
2554         diff_fill_sha1_info(p->one);
2555         diff_fill_sha1_info(p->two);
2556
2557         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2558 }
2559
2560 void diff_setup(struct diff_options *options)
2561 {
2562         memset(options, 0, sizeof(*options));
2563
2564         options->file = stdout;
2565
2566         options->line_termination = '\n';
2567         options->break_opt = -1;
2568         options->rename_limit = -1;
2569         options->dirstat_percent = 3;
2570         options->context = 3;
2571
2572         options->change = diff_change;
2573         options->add_remove = diff_addremove;
2574         if (diff_use_color_default > 0)
2575                 DIFF_OPT_SET(options, COLOR_DIFF);
2576         options->detect_rename = diff_detect_rename_default;
2577
2578         if (!diff_mnemonic_prefix) {
2579                 options->a_prefix = "a/";
2580                 options->b_prefix = "b/";
2581         }
2582 }
2583
2584 int diff_setup_done(struct diff_options *options)
2585 {
2586         int count = 0;
2587
2588         if (options->output_format & DIFF_FORMAT_NAME)
2589                 count++;
2590         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2591                 count++;
2592         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2593                 count++;
2594         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2595                 count++;
2596         if (count > 1)
2597                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2598
2599         /*
2600          * Most of the time we can say "there are changes"
2601          * only by checking if there are changed paths, but
2602          * --ignore-whitespace* options force us to look
2603          * inside contents.
2604          */
2605
2606         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
2607             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
2608             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
2609                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
2610         else
2611                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
2612
2613         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2614                 options->detect_rename = DIFF_DETECT_COPY;
2615
2616         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2617                 options->prefix = NULL;
2618         if (options->prefix)
2619                 options->prefix_length = strlen(options->prefix);
2620         else
2621                 options->prefix_length = 0;
2622
2623         if (options->output_format & (DIFF_FORMAT_NAME |
2624                                       DIFF_FORMAT_NAME_STATUS |
2625                                       DIFF_FORMAT_CHECKDIFF |
2626                                       DIFF_FORMAT_NO_OUTPUT))
2627                 options->output_format &= ~(DIFF_FORMAT_RAW |
2628                                             DIFF_FORMAT_NUMSTAT |
2629                                             DIFF_FORMAT_DIFFSTAT |
2630                                             DIFF_FORMAT_SHORTSTAT |
2631                                             DIFF_FORMAT_DIRSTAT |
2632                                             DIFF_FORMAT_SUMMARY |
2633                                             DIFF_FORMAT_PATCH);
2634
2635         /*
2636          * These cases always need recursive; we do not drop caller-supplied
2637          * recursive bits for other formats here.
2638          */
2639         if (options->output_format & (DIFF_FORMAT_PATCH |
2640                                       DIFF_FORMAT_NUMSTAT |
2641                                       DIFF_FORMAT_DIFFSTAT |
2642                                       DIFF_FORMAT_SHORTSTAT |
2643                                       DIFF_FORMAT_DIRSTAT |
2644                                       DIFF_FORMAT_SUMMARY |
2645                                       DIFF_FORMAT_CHECKDIFF))
2646                 DIFF_OPT_SET(options, RECURSIVE);
2647         /*
2648          * Also pickaxe would not work very well if you do not say recursive
2649          */
2650         if (options->pickaxe)
2651                 DIFF_OPT_SET(options, RECURSIVE);
2652         /*
2653          * When patches are generated, submodules diffed against the work tree
2654          * must be checked for dirtiness too so it can be shown in the output
2655          */
2656         if (options->output_format & DIFF_FORMAT_PATCH)
2657                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
2658
2659         if (options->detect_rename && options->rename_limit < 0)
2660                 options->rename_limit = diff_rename_limit_default;
2661         if (options->setup & DIFF_SETUP_USE_CACHE) {
2662                 if (!active_cache)
2663                         /* read-cache does not die even when it fails
2664                          * so it is safe for us to do this here.  Also
2665                          * it does not smudge active_cache or active_nr
2666                          * when it fails, so we do not have to worry about
2667                          * cleaning it up ourselves either.
2668                          */
2669                         read_cache();
2670         }
2671         if (options->abbrev <= 0 || 40 < options->abbrev)
2672                 options->abbrev = 40; /* full */
2673
2674         /*
2675          * It does not make sense to show the first hit we happened
2676          * to have found.  It does not make sense not to return with
2677          * exit code in such a case either.
2678          */
2679         if (DIFF_OPT_TST(options, QUICK)) {
2680                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2681                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2682         }
2683
2684         return 0;
2685 }
2686
2687 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2688 {
2689         char c, *eq;
2690         int len;
2691
2692         if (*arg != '-')
2693                 return 0;
2694         c = *++arg;
2695         if (!c)
2696                 return 0;
2697         if (c == arg_short) {
2698                 c = *++arg;
2699                 if (!c)
2700                         return 1;
2701                 if (val && isdigit(c)) {
2702                         char *end;
2703                         int n = strtoul(arg, &end, 10);
2704                         if (*end)
2705                                 return 0;
2706                         *val = n;
2707                         return 1;
2708                 }
2709                 return 0;
2710         }
2711         if (c != '-')
2712                 return 0;
2713         arg++;
2714         eq = strchr(arg, '=');
2715         if (eq)
2716                 len = eq - arg;
2717         else
2718                 len = strlen(arg);
2719         if (!len || strncmp(arg, arg_long, len))
2720                 return 0;
2721         if (eq) {
2722                 int n;
2723                 char *end;
2724                 if (!isdigit(*++eq))
2725                         return 0;
2726                 n = strtoul(eq, &end, 10);
2727                 if (*end)
2728                         return 0;
2729                 *val = n;
2730         }
2731         return 1;
2732 }
2733
2734 static int diff_scoreopt_parse(const char *opt);
2735
2736 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2737 {
2738         const char *arg = av[0];
2739
2740         /* Output format options */
2741         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2742                 options->output_format |= DIFF_FORMAT_PATCH;
2743         else if (opt_arg(arg, 'U', "unified", &options->context))
2744                 options->output_format |= DIFF_FORMAT_PATCH;
2745         else if (!strcmp(arg, "--raw"))
2746                 options->output_format |= DIFF_FORMAT_RAW;
2747         else if (!strcmp(arg, "--patch-with-raw"))
2748                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2749         else if (!strcmp(arg, "--numstat"))
2750                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2751         else if (!strcmp(arg, "--shortstat"))
2752                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2753         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2754                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2755         else if (!strcmp(arg, "--cumulative")) {
2756                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2757                 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2758         } else if (opt_arg(arg, 0, "dirstat-by-file",
2759                            &options->dirstat_percent)) {
2760                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2761                 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
2762         }
2763         else if (!strcmp(arg, "--check"))
2764                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2765         else if (!strcmp(arg, "--summary"))
2766                 options->output_format |= DIFF_FORMAT_SUMMARY;
2767         else if (!strcmp(arg, "--patch-with-stat"))
2768                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2769         else if (!strcmp(arg, "--name-only"))
2770                 options->output_format |= DIFF_FORMAT_NAME;
2771         else if (!strcmp(arg, "--name-status"))
2772                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2773         else if (!strcmp(arg, "-s"))
2774                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2775         else if (!prefixcmp(arg, "--stat")) {
2776                 char *end;
2777                 int width = options->stat_width;
2778                 int name_width = options->stat_name_width;
2779                 arg += 6;
2780                 end = (char *)arg;
2781
2782                 switch (*arg) {
2783                 case '-':
2784                         if (!prefixcmp(arg, "-width="))
2785                                 width = strtoul(arg + 7, &end, 10);
2786                         else if (!prefixcmp(arg, "-name-width="))
2787                                 name_width = strtoul(arg + 12, &end, 10);
2788                         break;
2789                 case '=':
2790                         width = strtoul(arg+1, &end, 10);
2791                         if (*end == ',')
2792                                 name_width = strtoul(end+1, &end, 10);
2793                 }
2794
2795                 /* Important! This checks all the error cases! */
2796                 if (*end)
2797                         return 0;
2798                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2799                 options->stat_name_width = name_width;
2800                 options->stat_width = width;
2801         }
2802
2803         /* renames options */
2804         else if (!prefixcmp(arg, "-B")) {
2805                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2806                         return -1;
2807         }
2808         else if (!prefixcmp(arg, "-M")) {
2809                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2810                         return -1;
2811                 options->detect_rename = DIFF_DETECT_RENAME;
2812         }
2813         else if (!prefixcmp(arg, "-C")) {
2814                 if (options->detect_rename == DIFF_DETECT_COPY)
2815                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2816                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2817                         return -1;
2818                 options->detect_rename = DIFF_DETECT_COPY;
2819         }
2820         else if (!strcmp(arg, "--no-renames"))
2821                 options->detect_rename = 0;
2822         else if (!strcmp(arg, "--relative"))
2823                 DIFF_OPT_SET(options, RELATIVE_NAME);
2824         else if (!prefixcmp(arg, "--relative=")) {
2825                 DIFF_OPT_SET(options, RELATIVE_NAME);
2826                 options->prefix = arg + 11;
2827         }
2828
2829         /* xdiff options */
2830         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2831                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
2832         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2833                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
2834         else if (!strcmp(arg, "--ignore-space-at-eol"))
2835                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
2836         else if (!strcmp(arg, "--patience"))
2837                 DIFF_XDL_SET(options, PATIENCE_DIFF);
2838
2839         /* flags options */
2840         else if (!strcmp(arg, "--binary")) {
2841                 options->output_format |= DIFF_FORMAT_PATCH;
2842                 DIFF_OPT_SET(options, BINARY);
2843         }
2844         else if (!strcmp(arg, "--full-index"))
2845                 DIFF_OPT_SET(options, FULL_INDEX);
2846         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2847                 DIFF_OPT_SET(options, TEXT);
2848         else if (!strcmp(arg, "-R"))
2849                 DIFF_OPT_SET(options, REVERSE_DIFF);
2850         else if (!strcmp(arg, "--find-copies-harder"))
2851                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2852         else if (!strcmp(arg, "--follow"))
2853                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2854         else if (!strcmp(arg, "--color"))
2855                 DIFF_OPT_SET(options, COLOR_DIFF);
2856         else if (!prefixcmp(arg, "--color=")) {
2857                 int value = git_config_colorbool(NULL, arg+8, -1);
2858                 if (value == 0)
2859                         DIFF_OPT_CLR(options, COLOR_DIFF);
2860                 else if (value > 0)
2861                         DIFF_OPT_SET(options, COLOR_DIFF);
2862                 else
2863                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
2864         }
2865         else if (!strcmp(arg, "--no-color"))
2866                 DIFF_OPT_CLR(options, COLOR_DIFF);
2867         else if (!strcmp(arg, "--color-words")) {
2868                 DIFF_OPT_SET(options, COLOR_DIFF);
2869                 DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2870         }
2871         else if (!prefixcmp(arg, "--color-words=")) {
2872                 DIFF_OPT_SET(options, COLOR_DIFF);
2873                 DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2874                 options->word_regex = arg + 14;
2875         }
2876         else if (!strcmp(arg, "--exit-code"))
2877                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2878         else if (!strcmp(arg, "--quiet"))
2879                 DIFF_OPT_SET(options, QUICK);
2880         else if (!strcmp(arg, "--ext-diff"))
2881                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2882         else if (!strcmp(arg, "--no-ext-diff"))
2883                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2884         else if (!strcmp(arg, "--textconv"))
2885                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
2886         else if (!strcmp(arg, "--no-textconv"))
2887                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
2888         else if (!strcmp(arg, "--ignore-submodules"))
2889                 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2890         else if (!strcmp(arg, "--submodule"))
2891                 DIFF_OPT_SET(options, SUBMODULE_LOG);
2892         else if (!prefixcmp(arg, "--submodule=")) {
2893                 if (!strcmp(arg + 12, "log"))
2894                         DIFF_OPT_SET(options, SUBMODULE_LOG);
2895         }
2896
2897         /* misc options */
2898         else if (!strcmp(arg, "-z"))
2899                 options->line_termination = 0;
2900         else if (!prefixcmp(arg, "-l"))
2901                 options->rename_limit = strtoul(arg+2, NULL, 10);
2902         else if (!prefixcmp(arg, "-S"))
2903                 options->pickaxe = arg + 2;
2904         else if (!strcmp(arg, "--pickaxe-all"))
2905                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2906         else if (!strcmp(arg, "--pickaxe-regex"))
2907                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2908         else if (!prefixcmp(arg, "-O"))
2909                 options->orderfile = arg + 2;
2910         else if (!prefixcmp(arg, "--diff-filter="))
2911                 options->filter = arg + 14;
2912         else if (!strcmp(arg, "--abbrev"))
2913                 options->abbrev = DEFAULT_ABBREV;
2914         else if (!prefixcmp(arg, "--abbrev=")) {
2915                 options->abbrev = strtoul(arg + 9, NULL, 10);
2916                 if (options->abbrev < MINIMUM_ABBREV)
2917                         options->abbrev = MINIMUM_ABBREV;
2918                 else if (40 < options->abbrev)
2919                         options->abbrev = 40;
2920         }
2921         else if (!prefixcmp(arg, "--src-prefix="))
2922                 options->a_prefix = arg + 13;
2923         else if (!prefixcmp(arg, "--dst-prefix="))
2924                 options->b_prefix = arg + 13;
2925         else if (!strcmp(arg, "--no-prefix"))
2926                 options->a_prefix = options->b_prefix = "";
2927         else if (opt_arg(arg, '\0', "inter-hunk-context",
2928                          &options->interhunkcontext))
2929                 ;
2930         else if (!prefixcmp(arg, "--output=")) {
2931                 options->file = fopen(arg + strlen("--output="), "w");
2932                 if (!options->file)
2933                         die_errno("Could not open '%s'", arg + strlen("--output="));
2934                 options->close_file = 1;
2935         } else
2936                 return 0;
2937         return 1;
2938 }
2939
2940 static int parse_num(const char **cp_p)
2941 {
2942         unsigned long num, scale;
2943         int ch, dot;
2944         const char *cp = *cp_p;
2945
2946         num = 0;
2947         scale = 1;
2948         dot = 0;
2949         for (;;) {
2950                 ch = *cp;
2951                 if ( !dot && ch == '.' ) {
2952                         scale = 1;
2953                         dot = 1;
2954                 } else if ( ch == '%' ) {
2955                         scale = dot ? scale*100 : 100;
2956                         cp++;   /* % is always at the end */
2957                         break;
2958                 } else if ( ch >= '0' && ch <= '9' ) {
2959                         if ( scale < 100000 ) {
2960                                 scale *= 10;
2961                                 num = (num*10) + (ch-'0');
2962                         }
2963                 } else {
2964                         break;
2965                 }
2966                 cp++;
2967         }
2968         *cp_p = cp;
2969
2970         /* user says num divided by scale and we say internally that
2971          * is MAX_SCORE * num / scale.
2972          */
2973         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2974 }
2975
2976 static int diff_scoreopt_parse(const char *opt)
2977 {
2978         int opt1, opt2, cmd;
2979
2980         if (*opt++ != '-')
2981                 return -1;
2982         cmd = *opt++;
2983         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2984                 return -1; /* that is not a -M, -C nor -B option */
2985
2986         opt1 = parse_num(&opt);
2987         if (cmd != 'B')
2988                 opt2 = 0;
2989         else {
2990                 if (*opt == 0)
2991                         opt2 = 0;
2992                 else if (*opt != '/')
2993                         return -1; /* we expect -B80/99 or -B80 */
2994                 else {
2995                         opt++;
2996                         opt2 = parse_num(&opt);
2997                 }
2998         }
2999         if (*opt != 0)
3000                 return -1;
3001         return opt1 | (opt2 << 16);
3002 }
3003
3004 struct diff_queue_struct diff_queued_diff;
3005
3006 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
3007 {
3008         if (queue->alloc <= queue->nr) {
3009                 queue->alloc = alloc_nr(queue->alloc);
3010                 queue->queue = xrealloc(queue->queue,
3011                                         sizeof(dp) * queue->alloc);
3012         }
3013         queue->queue[queue->nr++] = dp;
3014 }
3015
3016 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
3017                                  struct diff_filespec *one,
3018                                  struct diff_filespec *two)
3019 {
3020         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3021         dp->one = one;
3022         dp->two = two;
3023         if (queue)
3024                 diff_q(queue, dp);
3025         return dp;
3026 }
3027
3028 void diff_free_filepair(struct diff_filepair *p)
3029 {
3030         free_filespec(p->one);
3031         free_filespec(p->two);
3032         free(p);
3033 }
3034
3035 /* This is different from find_unique_abbrev() in that
3036  * it stuffs the result with dots for alignment.
3037  */
3038 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3039 {
3040         int abblen;
3041         const char *abbrev;
3042         if (len == 40)
3043                 return sha1_to_hex(sha1);
3044
3045         abbrev = find_unique_abbrev(sha1, len);
3046         abblen = strlen(abbrev);
3047         if (abblen < 37) {
3048                 static char hex[41];
3049                 if (len < abblen && abblen <= len + 2)
3050                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3051                 else
3052                         sprintf(hex, "%s...", abbrev);
3053                 return hex;
3054         }
3055         return sha1_to_hex(sha1);
3056 }
3057
3058 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3059 {
3060         int line_termination = opt->line_termination;
3061         int inter_name_termination = line_termination ? '\t' : '\0';
3062
3063         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3064                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3065                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
3066                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3067         }
3068         if (p->score) {
3069                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3070                         inter_name_termination);
3071         } else {
3072                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
3073         }
3074
3075         if (p->status == DIFF_STATUS_COPIED ||
3076             p->status == DIFF_STATUS_RENAMED) {
3077                 const char *name_a, *name_b;
3078                 name_a = p->one->path;
3079                 name_b = p->two->path;
3080                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3081                 write_name_quoted(name_a, opt->file, inter_name_termination);
3082                 write_name_quoted(name_b, opt->file, line_termination);
3083         } else {
3084                 const char *name_a, *name_b;
3085                 name_a = p->one->mode ? p->one->path : p->two->path;
3086                 name_b = NULL;
3087                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3088                 write_name_quoted(name_a, opt->file, line_termination);
3089         }
3090 }
3091
3092 int diff_unmodified_pair(struct diff_filepair *p)
3093 {
3094         /* This function is written stricter than necessary to support
3095          * the currently implemented transformers, but the idea is to
3096          * let transformers to produce diff_filepairs any way they want,
3097          * and filter and clean them up here before producing the output.
3098          */
3099         struct diff_filespec *one = p->one, *two = p->two;
3100
3101         if (DIFF_PAIR_UNMERGED(p))
3102                 return 0; /* unmerged is interesting */
3103
3104         /* deletion, addition, mode or type change
3105          * and rename are all interesting.
3106          */
3107         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
3108             DIFF_PAIR_MODE_CHANGED(p) ||
3109             strcmp(one->path, two->path))
3110                 return 0;
3111
3112         /* both are valid and point at the same path.  that is, we are
3113          * dealing with a change.
3114          */
3115         if (one->sha1_valid && two->sha1_valid &&
3116             !hashcmp(one->sha1, two->sha1) &&
3117             !one->dirty_submodule && !two->dirty_submodule)
3118                 return 1; /* no change */
3119         if (!one->sha1_valid && !two->sha1_valid)
3120                 return 1; /* both look at the same file on the filesystem. */
3121         return 0;
3122 }
3123
3124 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
3125 {
3126         if (diff_unmodified_pair(p))
3127                 return;
3128
3129         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3130             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3131                 return; /* no tree diffs in patch format */
3132
3133         run_diff(p, o);
3134 }
3135
3136 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
3137                             struct diffstat_t *diffstat)
3138 {
3139         if (diff_unmodified_pair(p))
3140                 return;
3141
3142         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3143             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3144                 return; /* no tree diffs in patch format */
3145
3146         run_diffstat(p, o, diffstat);
3147 }
3148
3149 static void diff_flush_checkdiff(struct diff_filepair *p,
3150                 struct diff_options *o)
3151 {
3152         if (diff_unmodified_pair(p))
3153                 return;
3154
3155         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3156             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3157                 return; /* no tree diffs in patch format */
3158
3159         run_checkdiff(p, o);
3160 }
3161
3162 int diff_queue_is_empty(void)
3163 {
3164         struct diff_queue_struct *q = &diff_queued_diff;
3165         int i;
3166         for (i = 0; i < q->nr; i++)
3167                 if (!diff_unmodified_pair(q->queue[i]))
3168                         return 0;
3169         return 1;
3170 }
3171
3172 #if DIFF_DEBUG
3173 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
3174 {
3175         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
3176                 x, one ? one : "",
3177                 s->path,
3178                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
3179                 s->mode,
3180                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
3181         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
3182                 x, one ? one : "",
3183                 s->size, s->xfrm_flags);
3184 }
3185
3186 void diff_debug_filepair(const struct diff_filepair *p, int i)
3187 {
3188         diff_debug_filespec(p->one, i, "one");
3189         diff_debug_filespec(p->two, i, "two");
3190         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
3191                 p->score, p->status ? p->status : '?',
3192                 p->one->rename_used, p->broken_pair);
3193 }
3194
3195 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
3196 {
3197         int i;
3198         if (msg)
3199                 fprintf(stderr, "%s\n", msg);
3200         fprintf(stderr, "q->nr = %d\n", q->nr);
3201         for (i = 0; i < q->nr; i++) {
3202                 struct diff_filepair *p = q->queue[i];
3203                 diff_debug_filepair(p, i);
3204         }
3205 }
3206 #endif
3207
3208 static void diff_resolve_rename_copy(void)
3209 {
3210         int i;
3211         struct diff_filepair *p;
3212         struct diff_queue_struct *q = &diff_queued_diff;
3213
3214         diff_debug_queue("resolve-rename-copy", q);
3215
3216         for (i = 0; i < q->nr; i++) {
3217                 p = q->queue[i];
3218                 p->status = 0; /* undecided */
3219                 if (DIFF_PAIR_UNMERGED(p))
3220                         p->status = DIFF_STATUS_UNMERGED;
3221                 else if (!DIFF_FILE_VALID(p->one))
3222                         p->status = DIFF_STATUS_ADDED;
3223                 else if (!DIFF_FILE_VALID(p->two))
3224                         p->status = DIFF_STATUS_DELETED;
3225                 else if (DIFF_PAIR_TYPE_CHANGED(p))
3226                         p->status = DIFF_STATUS_TYPE_CHANGED;
3227
3228                 /* from this point on, we are dealing with a pair
3229                  * whose both sides are valid and of the same type, i.e.
3230                  * either in-place edit or rename/copy edit.
3231                  */
3232                 else if (DIFF_PAIR_RENAME(p)) {
3233                         /*
3234                          * A rename might have re-connected a broken
3235                          * pair up, causing the pathnames to be the
3236                          * same again. If so, that's not a rename at
3237                          * all, just a modification..
3238                          *
3239                          * Otherwise, see if this source was used for
3240                          * multiple renames, in which case we decrement
3241                          * the count, and call it a copy.
3242                          */
3243                         if (!strcmp(p->one->path, p->two->path))
3244                                 p->status = DIFF_STATUS_MODIFIED;
3245                         else if (--p->one->rename_used > 0)
3246                                 p->status = DIFF_STATUS_COPIED;
3247                         else
3248                                 p->status = DIFF_STATUS_RENAMED;
3249                 }
3250                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
3251                          p->one->mode != p->two->mode ||
3252                          p->one->dirty_submodule ||
3253                          p->two->dirty_submodule ||
3254                          is_null_sha1(p->one->sha1))
3255                         p->status = DIFF_STATUS_MODIFIED;
3256                 else {
3257                         /* This is a "no-change" entry and should not
3258                          * happen anymore, but prepare for broken callers.
3259                          */
3260                         error("feeding unmodified %s to diffcore",
3261                               p->one->path);
3262                         p->status = DIFF_STATUS_UNKNOWN;
3263                 }
3264         }
3265         diff_debug_queue("resolve-rename-copy done", q);
3266 }
3267
3268 static int check_pair_status(struct diff_filepair *p)
3269 {
3270         switch (p->status) {
3271         case DIFF_STATUS_UNKNOWN:
3272                 return 0;
3273         case 0:
3274                 die("internal error in diff-resolve-rename-copy");
3275         default:
3276                 return 1;
3277         }
3278 }
3279
3280 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3281 {
3282         int fmt = opt->output_format;
3283
3284         if (fmt & DIFF_FORMAT_CHECKDIFF)
3285                 diff_flush_checkdiff(p, opt);
3286         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3287                 diff_flush_raw(p, opt);
3288         else if (fmt & DIFF_FORMAT_NAME) {
3289                 const char *name_a, *name_b;
3290                 name_a = p->two->path;
3291                 name_b = NULL;
3292                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3293                 write_name_quoted(name_a, opt->file, opt->line_termination);
3294         }
3295 }
3296
3297 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3298 {
3299         if (fs->mode)
3300                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3301         else
3302                 fprintf(file, " %s ", newdelete);
3303         write_name_quoted(fs->path, file, '\n');
3304 }
3305
3306
3307 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
3308 {
3309         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3310                 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
3311                         show_name ? ' ' : '\n');
3312                 if (show_name) {
3313                         write_name_quoted(p->two->path, file, '\n');
3314                 }
3315         }
3316 }
3317
3318 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3319 {
3320         char *names = pprint_rename(p->one->path, p->two->path);
3321
3322         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3323         free(names);
3324         show_mode_change(file, p, 0);
3325 }
3326
3327 static void diff_summary(FILE *file, struct diff_filepair *p)
3328 {
3329         switch(p->status) {
3330         case DIFF_STATUS_DELETED:
3331                 show_file_mode_name(file, "delete", p->one);
3332                 break;
3333         case DIFF_STATUS_ADDED:
3334                 show_file_mode_name(file, "create", p->two);
3335                 break;
3336         case DIFF_STATUS_COPIED:
3337                 show_rename_copy(file, "copy", p);
3338                 break;
3339         case DIFF_STATUS_RENAMED:
3340                 show_rename_copy(file, "rename", p);
3341                 break;
3342         default:
3343                 if (p->score) {
3344                         fputs(" rewrite ", file);
3345                         write_name_quoted(p->two->path, file, ' ');
3346                         fprintf(file, "(%d%%)\n", similarity_index(p));
3347                 }
3348                 show_mode_change(file, p, !p->score);
3349                 break;
3350         }
3351 }
3352
3353 struct patch_id_t {
3354         git_SHA_CTX *ctx;
3355         int patchlen;
3356 };
3357
3358 static int remove_space(char *line, int len)
3359 {
3360         int i;
3361         char *dst = line;
3362         unsigned char c;
3363
3364         for (i = 0; i < len; i++)
3365                 if (!isspace((c = line[i])))
3366                         *dst++ = c;
3367
3368         return dst - line;
3369 }
3370
3371 static void patch_id_consume(void *priv, char *line, unsigned long len)
3372 {
3373         struct patch_id_t *data = priv;
3374         int new_len;
3375
3376         /* Ignore line numbers when computing the SHA1 of the patch */
3377         if (!prefixcmp(line, "@@ -"))
3378                 return;
3379
3380         new_len = remove_space(line, len);
3381
3382         git_SHA1_Update(data->ctx, line, new_len);
3383         data->patchlen += new_len;
3384 }
3385
3386 /* returns 0 upon success, and writes result into sha1 */
3387 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3388 {
3389         struct diff_queue_struct *q = &diff_queued_diff;
3390         int i;
3391         git_SHA_CTX ctx;
3392         struct patch_id_t data;
3393         char buffer[PATH_MAX * 4 + 20];
3394
3395         git_SHA1_Init(&ctx);
3396         memset(&data, 0, sizeof(struct patch_id_t));
3397         data.ctx = &ctx;
3398
3399         for (i = 0; i < q->nr; i++) {
3400                 xpparam_t xpp;
3401                 xdemitconf_t xecfg;
3402                 mmfile_t mf1, mf2;
3403                 struct diff_filepair *p = q->queue[i];
3404                 int len1, len2;
3405
3406                 memset(&xpp, 0, sizeof(xpp));
3407                 memset(&xecfg, 0, sizeof(xecfg));
3408                 if (p->status == 0)
3409                         return error("internal diff status error");
3410                 if (p->status == DIFF_STATUS_UNKNOWN)
3411                         continue;
3412                 if (diff_unmodified_pair(p))
3413                         continue;
3414                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3415                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3416                         continue;
3417                 if (DIFF_PAIR_UNMERGED(p))
3418                         continue;
3419
3420                 diff_fill_sha1_info(p->one);
3421                 diff_fill_sha1_info(p->two);
3422                 if (fill_mmfile(&mf1, p->one) < 0 ||
3423                                 fill_mmfile(&mf2, p->two) < 0)
3424                         return error("unable to read files to diff");
3425
3426                 len1 = remove_space(p->one->path, strlen(p->one->path));
3427                 len2 = remove_space(p->two->path, strlen(p->two->path));
3428                 if (p->one->mode == 0)
3429                         len1 = snprintf(buffer, sizeof(buffer),
3430                                         "diff--gita/%.*sb/%.*s"
3431                                         "newfilemode%06o"
3432                                         "---/dev/null"
3433                                         "+++b/%.*s",
3434                                         len1, p->one->path,
3435                                         len2, p->two->path,
3436                                         p->two->mode,
3437                                         len2, p->two->path);
3438                 else if (p->two->mode == 0)
3439                         len1 = snprintf(buffer, sizeof(buffer),
3440                                         "diff--gita/%.*sb/%.*s"
3441                                         "deletedfilemode%06o"
3442                                         "---a/%.*s"
3443                                         "+++/dev/null",
3444                                         len1, p->one->path,
3445                                         len2, p->two->path,
3446                                         p->one->mode,
3447                                         len1, p->one->path);
3448                 else
3449                         len1 = snprintf(buffer, sizeof(buffer),
3450                                         "diff--gita/%.*sb/%.*s"
3451                                         "---a/%.*s"
3452                                         "+++b/%.*s",
3453                                         len1, p->one->path,
3454                                         len2, p->two->path,
3455                                         len1, p->one->path,
3456                                         len2, p->two->path);
3457                 git_SHA1_Update(&ctx, buffer, len1);
3458
3459                 xpp.flags = 0;
3460                 xecfg.ctxlen = 3;
3461                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3462                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3463                               &xpp, &xecfg);
3464         }
3465
3466         git_SHA1_Final(sha1, &ctx);
3467         return 0;
3468 }
3469
3470 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3471 {
3472         struct diff_queue_struct *q = &diff_queued_diff;
3473         int i;
3474         int result = diff_get_patch_id(options, sha1);
3475
3476         for (i = 0; i < q->nr; i++)
3477                 diff_free_filepair(q->queue[i]);
3478
3479         free(q->queue);
3480         q->queue = NULL;
3481         q->nr = q->alloc = 0;
3482
3483         return result;
3484 }
3485
3486 static int is_summary_empty(const struct diff_queue_struct *q)
3487 {
3488         int i;
3489
3490         for (i = 0; i < q->nr; i++) {
3491                 const struct diff_filepair *p = q->queue[i];
3492
3493                 switch (p->status) {
3494                 case DIFF_STATUS_DELETED:
3495                 case DIFF_STATUS_ADDED:
3496                 case DIFF_STATUS_COPIED:
3497                 case DIFF_STATUS_RENAMED:
3498                         return 0;
3499                 default:
3500                         if (p->score)
3501                                 return 0;
3502                         if (p->one->mode && p->two->mode &&
3503                             p->one->mode != p->two->mode)
3504                                 return 0;
3505                         break;
3506                 }
3507         }
3508         return 1;
3509 }
3510
3511 void diff_flush(struct diff_options *options)
3512 {
3513         struct diff_queue_struct *q = &diff_queued_diff;
3514         int i, output_format = options->output_format;
3515         int separator = 0;
3516
3517         /*
3518          * Order: raw, stat, summary, patch
3519          * or:    name/name-status/checkdiff (other bits clear)
3520          */
3521         if (!q->nr)
3522                 goto free_queue;
3523
3524         if (output_format & (DIFF_FORMAT_RAW |
3525                              DIFF_FORMAT_NAME |
3526                              DIFF_FORMAT_NAME_STATUS |
3527                              DIFF_FORMAT_CHECKDIFF)) {
3528                 for (i = 0; i < q->nr; i++) {
3529                         struct diff_filepair *p = q->queue[i];
3530                         if (check_pair_status(p))
3531                                 flush_one_pair(p, options);
3532                 }
3533                 separator++;
3534         }
3535
3536         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3537                 struct diffstat_t diffstat;
3538
3539                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3540                 for (i = 0; i < q->nr; i++) {
3541                         struct diff_filepair *p = q->queue[i];
3542                         if (check_pair_status(p))
3543                                 diff_flush_stat(p, options, &diffstat);
3544                 }
3545                 if (output_format & DIFF_FORMAT_NUMSTAT)
3546                         show_numstat(&diffstat, options);
3547                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3548                         show_stats(&diffstat, options);
3549                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3550                         show_shortstats(&diffstat, options);
3551                 free_diffstat_info(&diffstat);
3552                 separator++;
3553         }
3554         if (output_format & DIFF_FORMAT_DIRSTAT)
3555                 show_dirstat(options);
3556
3557         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3558                 for (i = 0; i < q->nr; i++)
3559                         diff_summary(options->file, q->queue[i]);
3560                 separator++;
3561         }
3562
3563         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
3564             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
3565             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
3566                 /*
3567                  * run diff_flush_patch for the exit status. setting
3568                  * options->file to /dev/null should be safe, becaue we
3569                  * aren't supposed to produce any output anyway.
3570                  */
3571                 if (options->close_file)
3572                         fclose(options->file);
3573                 options->file = fopen("/dev/null", "w");
3574                 if (!options->file)
3575                         die_errno("Could not open /dev/null");
3576                 options->close_file = 1;
3577                 for (i = 0; i < q->nr; i++) {
3578                         struct diff_filepair *p = q->queue[i];
3579                         if (check_pair_status(p))
3580                                 diff_flush_patch(p, options);
3581                         if (options->found_changes)
3582                                 break;
3583                 }
3584         }
3585
3586         if (output_format & DIFF_FORMAT_PATCH) {
3587                 if (separator) {
3588                         putc(options->line_termination, options->file);
3589                         if (options->stat_sep) {
3590                                 /* attach patch instead of inline */
3591                                 fputs(options->stat_sep, options->file);
3592                         }
3593                 }
3594
3595                 for (i = 0; i < q->nr; i++) {
3596                         struct diff_filepair *p = q->queue[i];
3597                         if (check_pair_status(p))
3598                                 diff_flush_patch(p, options);
3599                 }
3600         }
3601
3602         if (output_format & DIFF_FORMAT_CALLBACK)
3603                 options->format_callback(q, options, options->format_callback_data);
3604
3605         for (i = 0; i < q->nr; i++)
3606                 diff_free_filepair(q->queue[i]);
3607 free_queue:
3608         free(q->queue);
3609         q->queue = NULL;
3610         q->nr = q->alloc = 0;
3611         if (options->close_file)
3612                 fclose(options->file);
3613
3614         /*
3615          * Report the content-level differences with HAS_CHANGES;
3616          * diff_addremove/diff_change does not set the bit when
3617          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
3618          */
3619         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
3620                 if (options->found_changes)
3621                         DIFF_OPT_SET(options, HAS_CHANGES);
3622                 else
3623                         DIFF_OPT_CLR(options, HAS_CHANGES);
3624         }
3625 }
3626
3627 static void diffcore_apply_filter(const char *filter)
3628 {
3629         int i;
3630         struct diff_queue_struct *q = &diff_queued_diff;
3631         struct diff_queue_struct outq;
3632         outq.queue = NULL;
3633         outq.nr = outq.alloc = 0;
3634
3635         if (!filter)
3636                 return;
3637
3638         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3639                 int found;
3640                 for (i = found = 0; !found && i < q->nr; i++) {
3641                         struct diff_filepair *p = q->queue[i];
3642                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3643                              ((p->score &&
3644                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3645                               (!p->score &&
3646                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3647                             ((p->status != DIFF_STATUS_MODIFIED) &&
3648                              strchr(filter, p->status)))
3649                                 found++;
3650                 }
3651                 if (found)
3652                         return;
3653
3654                 /* otherwise we will clear the whole queue
3655                  * by copying the empty outq at the end of this
3656                  * function, but first clear the current entries
3657                  * in the queue.
3658                  */
3659                 for (i = 0; i < q->nr; i++)
3660                         diff_free_filepair(q->queue[i]);
3661         }
3662         else {
3663                 /* Only the matching ones */
3664                 for (i = 0; i < q->nr; i++) {
3665                         struct diff_filepair *p = q->queue[i];
3666
3667                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3668                              ((p->score &&
3669                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3670                               (!p->score &&
3671                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3672                             ((p->status != DIFF_STATUS_MODIFIED) &&
3673                              strchr(filter, p->status)))
3674                                 diff_q(&outq, p);
3675                         else
3676                                 diff_free_filepair(p);
3677                 }
3678         }
3679         free(q->queue);
3680         *q = outq;
3681 }
3682
3683 /* Check whether two filespecs with the same mode and size are identical */
3684 static int diff_filespec_is_identical(struct diff_filespec *one,
3685                                       struct diff_filespec *two)
3686 {
3687         if (S_ISGITLINK(one->mode))
3688                 return 0;
3689         if (diff_populate_filespec(one, 0))
3690                 return 0;
3691         if (diff_populate_filespec(two, 0))
3692                 return 0;
3693         return !memcmp(one->data, two->data, one->size);
3694 }
3695
3696 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3697 {
3698         int i;
3699         struct diff_queue_struct *q = &diff_queued_diff;
3700         struct diff_queue_struct outq;
3701         outq.queue = NULL;
3702         outq.nr = outq.alloc = 0;
3703
3704         for (i = 0; i < q->nr; i++) {
3705                 struct diff_filepair *p = q->queue[i];
3706
3707                 /*
3708                  * 1. Entries that come from stat info dirtiness
3709                  *    always have both sides (iow, not create/delete),
3710                  *    one side of the object name is unknown, with
3711                  *    the same mode and size.  Keep the ones that
3712                  *    do not match these criteria.  They have real
3713                  *    differences.
3714                  *
3715                  * 2. At this point, the file is known to be modified,
3716                  *    with the same mode and size, and the object
3717                  *    name of one side is unknown.  Need to inspect
3718                  *    the identical contents.
3719                  */
3720                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3721                     !DIFF_FILE_VALID(p->two) ||
3722                     (p->one->sha1_valid && p->two->sha1_valid) ||
3723                     (p->one->mode != p->two->mode) ||
3724                     diff_populate_filespec(p->one, 1) ||
3725                     diff_populate_filespec(p->two, 1) ||
3726                     (p->one->size != p->two->size) ||
3727                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3728                         diff_q(&outq, p);
3729                 else {
3730                         /*
3731                          * The caller can subtract 1 from skip_stat_unmatch
3732                          * to determine how many paths were dirty only
3733                          * due to stat info mismatch.
3734                          */
3735                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3736                                 diffopt->skip_stat_unmatch++;
3737                         diff_free_filepair(p);
3738                 }
3739         }
3740         free(q->queue);
3741         *q = outq;
3742 }
3743
3744 static int diffnamecmp(const void *a_, const void *b_)
3745 {
3746         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
3747         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
3748         const char *name_a, *name_b;
3749
3750         name_a = a->one ? a->one->path : a->two->path;
3751         name_b = b->one ? b->one->path : b->two->path;
3752         return strcmp(name_a, name_b);
3753 }
3754
3755 void diffcore_fix_diff_index(struct diff_options *options)
3756 {
3757         struct diff_queue_struct *q = &diff_queued_diff;
3758         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
3759 }
3760
3761 void diffcore_std(struct diff_options *options)
3762 {
3763         if (options->skip_stat_unmatch)
3764                 diffcore_skip_stat_unmatch(options);
3765         if (options->break_opt != -1)
3766                 diffcore_break(options->break_opt);
3767         if (options->detect_rename)
3768                 diffcore_rename(options);
3769         if (options->break_opt != -1)
3770                 diffcore_merge_broken();
3771         if (options->pickaxe)
3772                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3773         if (options->orderfile)
3774                 diffcore_order(options->orderfile);
3775         diff_resolve_rename_copy();
3776         diffcore_apply_filter(options->filter);
3777
3778         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3779                 DIFF_OPT_SET(options, HAS_CHANGES);
3780         else
3781                 DIFF_OPT_CLR(options, HAS_CHANGES);
3782 }
3783
3784 int diff_result_code(struct diff_options *opt, int status)
3785 {
3786         int result = 0;
3787         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3788             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3789                 return status;
3790         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3791             DIFF_OPT_TST(opt, HAS_CHANGES))
3792                 result |= 01;
3793         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3794             DIFF_OPT_TST(opt, CHECK_FAILED))
3795                 result |= 02;
3796         return result;
3797 }
3798
3799 void diff_addremove(struct diff_options *options,
3800                     int addremove, unsigned mode,
3801                     const unsigned char *sha1,
3802                     const char *concatpath, unsigned dirty_submodule)
3803 {
3804         struct diff_filespec *one, *two;
3805
3806         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3807                 return;
3808
3809         /* This may look odd, but it is a preparation for
3810          * feeding "there are unchanged files which should
3811          * not produce diffs, but when you are doing copy
3812          * detection you would need them, so here they are"
3813          * entries to the diff-core.  They will be prefixed
3814          * with something like '=' or '*' (I haven't decided
3815          * which but should not make any difference).
3816          * Feeding the same new and old to diff_change()
3817          * also has the same effect.
3818          * Before the final output happens, they are pruned after
3819          * merged into rename/copy pairs as appropriate.
3820          */
3821         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3822                 addremove = (addremove == '+' ? '-' :
3823                              addremove == '-' ? '+' : addremove);
3824
3825         if (options->prefix &&
3826             strncmp(concatpath, options->prefix, options->prefix_length))
3827                 return;
3828
3829         one = alloc_filespec(concatpath);
3830         two = alloc_filespec(concatpath);
3831
3832         if (addremove != '+')
3833                 fill_filespec(one, sha1, mode);
3834         if (addremove != '-') {
3835                 fill_filespec(two, sha1, mode);
3836                 two->dirty_submodule = dirty_submodule;
3837         }
3838
3839         diff_queue(&diff_queued_diff, one, two);
3840         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3841                 DIFF_OPT_SET(options, HAS_CHANGES);
3842 }
3843
3844 void diff_change(struct diff_options *options,
3845                  unsigned old_mode, unsigned new_mode,
3846                  const unsigned char *old_sha1,
3847                  const unsigned char *new_sha1,
3848                  const char *concatpath,
3849                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
3850 {
3851         struct diff_filespec *one, *two;
3852
3853         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3854                         && S_ISGITLINK(new_mode))
3855                 return;
3856
3857         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3858                 unsigned tmp;
3859                 const unsigned char *tmp_c;
3860                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3861                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3862                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
3863                         new_dirty_submodule = tmp;
3864         }
3865
3866         if (options->prefix &&
3867             strncmp(concatpath, options->prefix, options->prefix_length))
3868                 return;
3869
3870         one = alloc_filespec(concatpath);
3871         two = alloc_filespec(concatpath);
3872         fill_filespec(one, old_sha1, old_mode);
3873         fill_filespec(two, new_sha1, new_mode);
3874         one->dirty_submodule = old_dirty_submodule;
3875         two->dirty_submodule = new_dirty_submodule;
3876
3877         diff_queue(&diff_queued_diff, one, two);
3878         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3879                 DIFF_OPT_SET(options, HAS_CHANGES);
3880 }
3881
3882 void diff_unmerge(struct diff_options *options,
3883                   const char *path,
3884                   unsigned mode, const unsigned char *sha1)
3885 {
3886         struct diff_filespec *one, *two;
3887
3888         if (options->prefix &&
3889             strncmp(path, options->prefix, options->prefix_length))
3890                 return;
3891
3892         one = alloc_filespec(path);
3893         two = alloc_filespec(path);
3894         fill_filespec(one, sha1, mode);
3895         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3896 }
3897
3898 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
3899                 size_t *outsize)
3900 {
3901         struct diff_tempfile *temp;
3902         const char *argv[3];
3903         const char **arg = argv;
3904         struct child_process child;
3905         struct strbuf buf = STRBUF_INIT;
3906         int err = 0;
3907
3908         temp = prepare_temp_file(spec->path, spec);
3909         *arg++ = pgm;
3910         *arg++ = temp->name;
3911         *arg = NULL;
3912
3913         memset(&child, 0, sizeof(child));
3914         child.use_shell = 1;
3915         child.argv = argv;
3916         child.out = -1;
3917         if (start_command(&child)) {
3918                 remove_tempfile();
3919                 return NULL;
3920         }
3921
3922         if (strbuf_read(&buf, child.out, 0) < 0)
3923                 err = error("error reading from textconv command '%s'", pgm);
3924         close(child.out);
3925
3926         if (finish_command(&child) || err) {
3927                 strbuf_release(&buf);
3928                 remove_tempfile();
3929                 return NULL;
3930         }
3931         remove_tempfile();
3932
3933         return strbuf_detach(&buf, outsize);
3934 }