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 static int diff_no_prefix;
  34
  35 static char diff_colors[][COLOR_MAXLEN] = {
  36         GIT_COLOR_RESET,
  37         GIT_COLOR_NORMAL,       /* PLAIN */
  38         GIT_COLOR_BOLD,         /* METAINFO */
  39         GIT_COLOR_CYAN,         /* FRAGINFO */
  40         GIT_COLOR_RED,          /* OLD */
  41         GIT_COLOR_GREEN,        /* NEW */
  42         GIT_COLOR_YELLOW,       /* COMMIT */
  43         GIT_COLOR_BG_RED,       /* WHITESPACE */
  44         GIT_COLOR_NORMAL,       /* FUNCINFO */
  45 };
  46
  47 static void diff_filespec_load_driver(struct diff_filespec *one);
  48 static size_t fill_textconv(struct userdiff_driver *driver,
  49                             struct diff_filespec *df, char **outbuf);
  50
  51 static int parse_diff_color_slot(const char *var, int ofs)
  52 {
  53         if (!strcasecmp(var+ofs, "plain"))
  54                 return DIFF_PLAIN;
  55         if (!strcasecmp(var+ofs, "meta"))
  56                 return DIFF_METAINFO;
  57         if (!strcasecmp(var+ofs, "frag"))
  58                 return DIFF_FRAGINFO;
  59         if (!strcasecmp(var+ofs, "old"))
  60                 return DIFF_FILE_OLD;
  61         if (!strcasecmp(var+ofs, "new"))
  62                 return DIFF_FILE_NEW;
  63         if (!strcasecmp(var+ofs, "commit"))
  64                 return DIFF_COMMIT;
  65         if (!strcasecmp(var+ofs, "whitespace"))
  66                 return DIFF_WHITESPACE;
  67         if (!strcasecmp(var+ofs, "func"))
  68                 return DIFF_FUNCINFO;
  69         return -1;
  70 }
  71
  72 static int git_config_rename(const char *var, const char *value)
  73 {
  74         if (!value)
  75                 return DIFF_DETECT_RENAME;
  76         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
  77                 return  DIFF_DETECT_COPY;
  78         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
  79 }
  80
  81 /*
  82  * These are to give UI layer defaults.
  83  * The core-level commands such as git-diff-files should
  84  * never be affected by the setting of diff.renames
  85  * the user happens to have in the configuration file.
  86  */
  87 int git_diff_ui_config(const char *var, const char *value, void *cb)
  88 {
  89         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
  90                 diff_use_color_default = git_config_colorbool(var, value, -1);
  91                 return 0;
  92         }
  93         if (!strcmp(var, "diff.renames")) {
  94                 diff_detect_rename_default = git_config_rename(var, value);
  95                 return 0;
  96         }
  97         if (!strcmp(var, "diff.autorefreshindex")) {
  98                 diff_auto_refresh_index = git_config_bool(var, value);
  99                 return 0;
 100         }
 101         if (!strcmp(var, "diff.mnemonicprefix")) {
 102                 diff_mnemonic_prefix = git_config_bool(var, value);
 103                 return 0;
 104         }
 105         if (!strcmp(var, "diff.noprefix")) {
 106                 diff_no_prefix = git_config_bool(var, value);
 107                 return 0;
 108         }
 109         if (!strcmp(var, "diff.external"))
 110                 return git_config_string(&external_diff_cmd_cfg, var, value);
 111         if (!strcmp(var, "diff.wordregex"))
 112                 return git_config_string(&diff_word_regex_cfg, var, value);
 113
 114         return git_diff_basic_config(var, value, cb);
 115 }
 116
 117 int git_diff_basic_config(const char *var, const char *value, void *cb)
 118 {
 119         if (!strcmp(var, "diff.renamelimit")) {
 120                 diff_rename_limit_default = git_config_int(var, value);
 121                 return 0;
 122         }
 123
 124         switch (userdiff_config(var, value)) {
 125                 case 0: break;
 126                 case -1: return -1;
 127                 default: return 0;
 128         }
 129
 130         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 131                 int slot = parse_diff_color_slot(var, 11);
 132                 if (slot < 0)
 133                         return 0;
 134                 if (!value)
 135                         return config_error_nonbool(var);
 136                 color_parse(value, var, diff_colors[slot]);
 137                 return 0;
 138         }
 139
 140         /* like GNU diff's --suppress-blank-empty option  */
 141         if (!strcmp(var, "diff.suppressblankempty") ||
 142                         /* for backwards compatibility */
 143                         !strcmp(var, "diff.suppress-blank-empty")) {
 144                 diff_suppress_blank_empty = git_config_bool(var, value);
 145                 return 0;
 146         }
 147
 148         return git_color_default_config(var, value, cb);
 149 }
 150
 151 static char *quote_two(const char *one, const char *two)
 152 {
 153         int need_one = quote_c_style(one, NULL, NULL, 1);
 154         int need_two = quote_c_style(two, NULL, NULL, 1);
 155         struct strbuf res = STRBUF_INIT;
 156
 157         if (need_one + need_two) {
 158                 strbuf_addch(&res, '"');
 159                 quote_c_style(one, &res, NULL, 1);
 160                 quote_c_style(two, &res, NULL, 1);
 161                 strbuf_addch(&res, '"');
 162         } else {
 163                 strbuf_addstr(&res, one);
 164                 strbuf_addstr(&res, two);
 165         }
 166         return strbuf_detach(&res, NULL);
 167 }
 168
 169 static const char *external_diff(void)
 170 {
 171         static const char *external_diff_cmd = NULL;
 172         static int done_preparing = 0;
 173
 174         if (done_preparing)
 175                 return external_diff_cmd;
 176         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 177         if (!external_diff_cmd)
 178                 external_diff_cmd = external_diff_cmd_cfg;
 179         done_preparing = 1;
 180         return external_diff_cmd;
 181 }
 182
 183 static struct diff_tempfile {
 184         const char *name; /* filename external diff should read from */
 185         char hex[41];
 186         char mode[10];
 187         char tmp_path[PATH_MAX];
 188 } diff_temp[2];
 189
 190 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 191
 192 struct emit_callback {
 193         int color_diff;
 194         unsigned ws_rule;
 195         int blank_at_eof_in_preimage;
 196         int blank_at_eof_in_postimage;
 197         int lno_in_preimage;
 198         int lno_in_postimage;
 199         sane_truncate_fn truncate;
 200         const char **label_path;
 201         struct diff_words_data *diff_words;
 202         int *found_changesp;
 203         FILE *file;
 204         struct strbuf *header;
 205 };
 206
 207 static int count_lines(const char *data, int size)
 208 {
 209         int count, ch, completely_empty = 1, nl_just_seen = 0;
 210         count = 0;
 211         while (0 < size--) {
 212                 ch = *data++;
 213                 if (ch == '\n') {
 214                         count++;
 215                         nl_just_seen = 1;
 216                         completely_empty = 0;
 217                 }
 218                 else {
 219                         nl_just_seen = 0;
 220                         completely_empty = 0;
 221                 }
 222         }
 223         if (completely_empty)
 224                 return 0;
 225         if (!nl_just_seen)
 226                 count++; /* no trailing newline */
 227         return count;
 228 }
 229
 230 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 231 {
 232         if (!DIFF_FILE_VALID(one)) {
 233                 mf->ptr = (char *)""; /* does not matter */
 234                 mf->size = 0;
 235                 return 0;
 236         }
 237         else if (diff_populate_filespec(one, 0))
 238                 return -1;
 239
 240         mf->ptr = one->data;
 241         mf->size = one->size;
 242         return 0;
 243 }
 244
 245 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 246 {
 247         char *ptr = mf->ptr;
 248         long size = mf->size;
 249         int cnt = 0;
 250
 251         if (!size)
 252                 return cnt;
 253         ptr += size - 1; /* pointing at the very end */
 254         if (*ptr != '\n')
 255                 ; /* incomplete line */
 256         else
 257                 ptr--; /* skip the last LF */
 258         while (mf->ptr < ptr) {
 259                 char *prev_eol;
 260                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 261                         if (*prev_eol == '\n')
 262                                 break;
 263                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 264                         break;
 265                 cnt++;
 266                 ptr = prev_eol - 1;
 267         }
 268         return cnt;
 269 }
 270
 271 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 272                                struct emit_callback *ecbdata)
 273 {
 274         int l1, l2, at;
 275         unsigned ws_rule = ecbdata->ws_rule;
 276         l1 = count_trailing_blank(mf1, ws_rule);
 277         l2 = count_trailing_blank(mf2, ws_rule);
 278         if (l2 <= l1) {
 279                 ecbdata->blank_at_eof_in_preimage = 0;
 280                 ecbdata->blank_at_eof_in_postimage = 0;
 281                 return;
 282         }
 283         at = count_lines(mf1->ptr, mf1->size);
 284         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 285
 286         at = count_lines(mf2->ptr, mf2->size);
 287         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 288 }
 289
 290 static void emit_line_0(FILE *file, const char *set, const char *reset,
 291                         int first, const char *line, int len)
 292 {
 293         int has_trailing_newline, has_trailing_carriage_return;
 294         int nofirst;
 295
 296         if (len == 0) {
 297                 has_trailing_newline = (first == '\n');
 298                 has_trailing_carriage_return = (!has_trailing_newline &&
 299                                                 (first == '\r'));
 300                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 301         } else {
 302                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 303                 if (has_trailing_newline)
 304                         len--;
 305                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 306                 if (has_trailing_carriage_return)
 307                         len--;
 308                 nofirst = 0;
 309         }
 310
 311         if (len || !nofirst) {
 312                 fputs(set, file);
 313                 if (!nofirst)
 314                         fputc(first, file);
 315                 fwrite(line, len, 1, file);
 316                 fputs(reset, file);
 317         }
 318         if (has_trailing_carriage_return)
 319                 fputc('\r', file);
 320         if (has_trailing_newline)
 321                 fputc('\n', file);
 322 }
 323
 324 static void emit_line(FILE *file, const char *set, const char *reset,
 325                       const char *line, int len)
 326 {
 327         emit_line_0(file, set, reset, line[0], line+1, len-1);
 328 }
 329
 330 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
 331 {
 332         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
 333               ecbdata->blank_at_eof_in_preimage &&
 334               ecbdata->blank_at_eof_in_postimage &&
 335               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
 336               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
 337                 return 0;
 338         return ws_blank_line(line, len, ecbdata->ws_rule);
 339 }
 340
 341 static void emit_add_line(const char *reset,
 342                           struct emit_callback *ecbdata,
 343                           const char *line, int len)
 344 {
 345         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 346         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 347
 348         if (!*ws)
 349                 emit_line_0(ecbdata->file, set, reset, '+', line, len);
 350         else if (new_blank_line_at_eof(ecbdata, line, len))
 351                 /* Blank line at EOF - paint '+' as well */
 352                 emit_line_0(ecbdata->file, ws, reset, '+', line, len);
 353         else {
 354                 /* Emit just the prefix, then the rest. */
 355                 emit_line_0(ecbdata->file, set, reset, '+', "", 0);
 356                 ws_check_emit(line, len, ecbdata->ws_rule,
 357                               ecbdata->file, set, reset, ws);
 358         }
 359 }
 360
 361 static void emit_hunk_header(struct emit_callback *ecbdata,
 362                              const char *line, int len)
 363 {
 364         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 365         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
 366         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
 367         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 368         static const char atat[2] = { '@', '@' };
 369         const char *cp, *ep;
 370
 371         /*
 372          * As a hunk header must begin with "@@ -<old>, +<new> @@",
 373          * it always is at least 10 bytes long.
 374          */
 375         if (len < 10 ||
 376             memcmp(line, atat, 2) ||
 377             !(ep = memmem(line + 2, len - 2, atat, 2))) {
 378                 emit_line(ecbdata->file, plain, reset, line, len);
 379                 return;
 380         }
 381         ep += 2; /* skip over @@ */
 382
 383         /* The hunk header in fraginfo color */
 384         emit_line(ecbdata->file, frag, reset, line, ep - line);
 385
 386         /* blank before the func header */
 387         for (cp = ep; ep - line < len; ep++)
 388                 if (*ep != ' ' && *ep != '\t')
 389                         break;
 390         if (ep != cp)
 391                 emit_line(ecbdata->file, plain, reset, cp, ep - cp);
 392
 393         if (ep < line + len)
 394                 emit_line(ecbdata->file, func, reset, ep, line + len - ep);
 395 }
 396
 397 static struct diff_tempfile *claim_diff_tempfile(void) {
 398         int i;
 399         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
 400                 if (!diff_temp[i].name)
 401                         return diff_temp + i;
 402         die("BUG: diff is failing to clean up its tempfiles");
 403 }
 404
 405 static int remove_tempfile_installed;
 406
 407 static void remove_tempfile(void)
 408 {
 409         int i;
 410         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
 411                 if (diff_temp[i].name == diff_temp[i].tmp_path)
 412                         unlink_or_warn(diff_temp[i].name);
 413                 diff_temp[i].name = NULL;
 414         }
 415 }
 416
 417 static void remove_tempfile_on_signal(int signo)
 418 {
 419         remove_tempfile();
 420         sigchain_pop(signo);
 421         raise(signo);
 422 }
 423
 424 static void print_line_count(FILE *file, int count)
 425 {
 426         switch (count) {
 427         case 0:
 428                 fprintf(file, "0,0");
 429                 break;
 430         case 1:
 431                 fprintf(file, "1");
 432                 break;
 433         default:
 434                 fprintf(file, "1,%d", count);
 435                 break;
 436         }
 437 }
 438
 439 static void emit_rewrite_lines(struct emit_callback *ecb,
 440                                int prefix, const char *data, int size)
 441 {
 442         const char *endp = NULL;
 443         static const char *nneof = " No newline at end of file\n";
 444         const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
 445         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
 446
 447         while (0 < size) {
 448                 int len;
 449
 450                 endp = memchr(data, '\n', size);
 451                 len = endp ? (endp - data + 1) : size;
 452                 if (prefix != '+') {
 453                         ecb->lno_in_preimage++;
 454                         emit_line_0(ecb->file, old, reset, '-',
 455                                     data, len);
 456                 } else {
 457                         ecb->lno_in_postimage++;
 458                         emit_add_line(reset, ecb, data, len);
 459                 }
 460                 size -= len;
 461                 data += len;
 462         }
 463         if (!endp) {
 464                 const char *plain = diff_get_color(ecb->color_diff,
 465                                                    DIFF_PLAIN);
 466                 emit_line_0(ecb->file, plain, reset, '\\',
 467                             nneof, strlen(nneof));
 468         }
 469 }
 470
 471 static void emit_rewrite_diff(const char *name_a,
 472                               const char *name_b,
 473                               struct diff_filespec *one,
 474                               struct diff_filespec *two,
 475                               struct userdiff_driver *textconv_one,
 476                               struct userdiff_driver *textconv_two,
 477                               struct diff_options *o)
 478 {
 479         int lc_a, lc_b;
 480         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
 481         const char *name_a_tab, *name_b_tab;
 482         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 483         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 484         const char *reset = diff_get_color(color_diff, DIFF_RESET);
 485         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 486         const char *a_prefix, *b_prefix;
 487         char *data_one, *data_two;
 488         size_t size_one, size_two;
 489         struct emit_callback ecbdata;
 490
 491         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 492                 a_prefix = o->b_prefix;
 493                 b_prefix = o->a_prefix;
 494         } else {
 495                 a_prefix = o->a_prefix;
 496                 b_prefix = o->b_prefix;
 497         }
 498
 499         name_a += (*name_a == '/');
 500         name_b += (*name_b == '/');
 501         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 502         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 503
 504         strbuf_reset(&a_name);
 505         strbuf_reset(&b_name);
 506         quote_two_c_style(&a_name, a_prefix, name_a, 0);
 507         quote_two_c_style(&b_name, b_prefix, name_b, 0);
 508
 509         size_one = fill_textconv(textconv_one, one, &data_one);
 510         size_two = fill_textconv(textconv_two, two, &data_two);
 511
 512         memset(&ecbdata, 0, sizeof(ecbdata));
 513         ecbdata.color_diff = color_diff;
 514         ecbdata.found_changesp = &o->found_changes;
 515         ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
 516         ecbdata.file = o->file;
 517         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
 518                 mmfile_t mf1, mf2;
 519                 mf1.ptr = (char *)data_one;
 520                 mf2.ptr = (char *)data_two;
 521                 mf1.size = size_one;
 522                 mf2.size = size_two;
 523                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
 524         }
 525         ecbdata.lno_in_preimage = 1;
 526         ecbdata.lno_in_postimage = 1;
 527
 528         lc_a = count_lines(data_one, size_one);
 529         lc_b = count_lines(data_two, size_two);
 530         fprintf(o->file,
 531                 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
 532                 metainfo, a_name.buf, name_a_tab, reset,
 533                 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
 534         print_line_count(o->file, lc_a);
 535         fprintf(o->file, " +");
 536         print_line_count(o->file, lc_b);
 537         fprintf(o->file, " @@%s\n", reset);
 538         if (lc_a)
 539                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
 540         if (lc_b)
 541                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
 542         if (textconv_one)
 543                 free((char *)data_one);
 544         if (textconv_two)
 545                 free((char *)data_two);
 546 }
 547
 548 struct diff_words_buffer {
 549         mmfile_t text;
 550         long alloc;
 551         struct diff_words_orig {
 552                 const char *begin, *end;
 553         } *orig;
 554         int orig_nr, orig_alloc;
 555 };
 556
 557 static void diff_words_append(char *line, unsigned long len,
 558                 struct diff_words_buffer *buffer)
 559 {
 560         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
 561         line++;
 562         len--;
 563         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 564         buffer->text.size += len;
 565         buffer->text.ptr[buffer->text.size] = '\0';
 566 }
 567
 568 struct diff_words_data {
 569         struct diff_words_buffer minus, plus;
 570         const char *current_plus;
 571         FILE *file;
 572         regex_t *word_regex;
 573 };
 574
 575 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 576 {
 577         struct diff_words_data *diff_words = priv;
 578         int minus_first, minus_len, plus_first, plus_len;
 579         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 580
 581         if (line[0] != '@' || parse_hunk_header(line, len,
 582                         &minus_first, &minus_len, &plus_first, &plus_len))
 583                 return;
 584
 585         /* POSIX requires that first be decremented by one if len == 0... */
 586         if (minus_len) {
 587                 minus_begin = diff_words->minus.orig[minus_first].begin;
 588                 minus_end =
 589                         diff_words->minus.orig[minus_first + minus_len - 1].end;
 590         } else
 591                 minus_begin = minus_end =
 592                         diff_words->minus.orig[minus_first].end;
 593
 594         if (plus_len) {
 595                 plus_begin = diff_words->plus.orig[plus_first].begin;
 596                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
 597         } else
 598                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
 599
 600         if (diff_words->current_plus != plus_begin)
 601                 fwrite(diff_words->current_plus,
 602                                 plus_begin - diff_words->current_plus, 1,
 603                                 diff_words->file);
 604         if (minus_begin != minus_end)
 605                 color_fwrite_lines(diff_words->file,
 606                                 diff_get_color(1, DIFF_FILE_OLD),
 607                                 minus_end - minus_begin, minus_begin);
 608         if (plus_begin != plus_end)
 609                 color_fwrite_lines(diff_words->file,
 610                                 diff_get_color(1, DIFF_FILE_NEW),
 611                                 plus_end - plus_begin, plus_begin);
 612
 613         diff_words->current_plus = plus_end;
 614 }
 615
 616 /* This function starts looking at *begin, and returns 0 iff a word was found. */
 617 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
 618                 int *begin, int *end)
 619 {
 620         if (word_regex && *begin < buffer->size) {
 621                 regmatch_t match[1];
 622                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
 623                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
 624                                         '\n', match[0].rm_eo - match[0].rm_so);
 625                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
 626                         *begin += match[0].rm_so;
 627                         return *begin >= *end;
 628                 }
 629                 return -1;
 630         }
 631
 632         /* find the next word */
 633         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
 634                 (*begin)++;
 635         if (*begin >= buffer->size)
 636                 return -1;
 637
 638         /* find the end of the word */
 639         *end = *begin + 1;
 640         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
 641                 (*end)++;
 642
 643         return 0;
 644 }
 645
 646 /*
 647  * This function splits the words in buffer->text, stores the list with
 648  * newline separator into out, and saves the offsets of the original words
 649  * in buffer->orig.
 650  */
 651 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
 652                 regex_t *word_regex)
 653 {
 654         int i, j;
 655         long alloc = 0;
 656
 657         out->size = 0;
 658         out->ptr = NULL;
 659
 660         /* fake an empty "0th" word */
 661         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
 662         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
 663         buffer->orig_nr = 1;
 664
 665         for (i = 0; i < buffer->text.size; i++) {
 666                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
 667                         return;
 668
 669                 /* store original boundaries */
 670                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
 671                                 buffer->orig_alloc);
 672                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
 673                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
 674                 buffer->orig_nr++;
 675
 676                 /* store one word */
 677                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
 678                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
 679                 out->ptr[out->size + j - i] = '\n';
 680                 out->size += j - i + 1;
 681
 682                 i = j - 1;
 683         }
 684 }
 685
 686 /* this executes the word diff on the accumulated buffers */
 687 static void diff_words_show(struct diff_words_data *diff_words)
 688 {
 689         xpparam_t xpp;
 690         xdemitconf_t xecfg;
 691         mmfile_t minus, plus;
 692
 693         /* special case: only removal */
 694         if (!diff_words->plus.text.size) {
 695                 color_fwrite_lines(diff_words->file,
 696                         diff_get_color(1, DIFF_FILE_OLD),
 697                         diff_words->minus.text.size, diff_words->minus.text.ptr);
 698                 diff_words->minus.text.size = 0;
 699                 return;
 700         }
 701
 702         diff_words->current_plus = diff_words->plus.text.ptr;
 703
 704         memset(&xpp, 0, sizeof(xpp));
 705         memset(&xecfg, 0, sizeof(xecfg));
 706         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
 707         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
 708         xpp.flags = XDF_NEED_MINIMAL;
 709         /* as only the hunk header will be parsed, we need a 0-context */
 710         xecfg.ctxlen = 0;
 711         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
 712                       &xpp, &xecfg);
 713         free(minus.ptr);
 714         free(plus.ptr);
 715         if (diff_words->current_plus != diff_words->plus.text.ptr +
 716                         diff_words->plus.text.size)
 717                 fwrite(diff_words->current_plus,
 718                         diff_words->plus.text.ptr + diff_words->plus.text.size
 719                         - diff_words->current_plus, 1,
 720                         diff_words->file);
 721         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 722 }
 723
 724 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
 725 static void diff_words_flush(struct emit_callback *ecbdata)
 726 {
 727         if (ecbdata->diff_words->minus.text.size ||
 728             ecbdata->diff_words->plus.text.size)
 729                 diff_words_show(ecbdata->diff_words);
 730 }
 731
 732 static void free_diff_words_data(struct emit_callback *ecbdata)
 733 {
 734         if (ecbdata->diff_words) {
 735                 diff_words_flush(ecbdata);
 736                 free (ecbdata->diff_words->minus.text.ptr);
 737                 free (ecbdata->diff_words->minus.orig);
 738                 free (ecbdata->diff_words->plus.text.ptr);
 739                 free (ecbdata->diff_words->plus.orig);
 740                 free(ecbdata->diff_words->word_regex);
 741                 free(ecbdata->diff_words);
 742                 ecbdata->diff_words = NULL;
 743         }
 744 }
 745
 746 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 747 {
 748         if (diff_use_color)
 749                 return diff_colors[ix];
 750         return "";
 751 }
 752
 753 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
 754 {
 755         const char *cp;
 756         unsigned long allot;
 757         size_t l = len;
 758
 759         if (ecb->truncate)
 760                 return ecb->truncate(line, len);
 761         cp = line;
 762         allot = l;
 763         while (0 < l) {
 764                 (void) utf8_width(&cp, &l);
 765                 if (!cp)
 766                         break; /* truncated in the middle? */
 767         }
 768         return allot - l;
 769 }
 770
 771 static void find_lno(const char *line, struct emit_callback *ecbdata)
 772 {
 773         const char *p;
 774         ecbdata->lno_in_preimage = 0;
 775         ecbdata->lno_in_postimage = 0;
 776         p = strchr(line, '-');
 777         if (!p)
 778                 return; /* cannot happen */
 779         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
 780         p = strchr(p, '+');
 781         if (!p)
 782                 return; /* cannot happen */
 783         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
 784 }
 785
 786 static void fn_out_consume(void *priv, char *line, unsigned long len)
 787 {
 788         struct emit_callback *ecbdata = priv;
 789         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 790         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 791         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 792
 793         if (ecbdata->header) {
 794                 fprintf(ecbdata->file, "%s", ecbdata->header->buf);
 795                 strbuf_reset(ecbdata->header);
 796                 ecbdata->header = NULL;
 797         }
 798         *(ecbdata->found_changesp) = 1;
 799
 800         if (ecbdata->label_path[0]) {
 801                 const char *name_a_tab, *name_b_tab;
 802
 803                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 804                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 805
 806                 fprintf(ecbdata->file, "%s--- %s%s%s\n",
 807                         meta, ecbdata->label_path[0], reset, name_a_tab);
 808                 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
 809                         meta, ecbdata->label_path[1], reset, name_b_tab);
 810                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 811         }
 812
 813         if (diff_suppress_blank_empty
 814             && len == 2 && line[0] == ' ' && line[1] == '\n') {
 815                 line[0] = '\n';
 816                 len = 1;
 817         }
 818
 819         if (line[0] == '@') {
 820                 if (ecbdata->diff_words)
 821                         diff_words_flush(ecbdata);
 822                 len = sane_truncate_line(ecbdata, line, len);
 823                 find_lno(line, ecbdata);
 824                 emit_hunk_header(ecbdata, line, len);
 825                 if (line[len-1] != '\n')
 826                         putc('\n', ecbdata->file);
 827                 return;
 828         }
 829
 830         if (len < 1) {
 831                 emit_line(ecbdata->file, reset, reset, line, len);
 832                 return;
 833         }
 834
 835         if (ecbdata->diff_words) {
 836                 if (line[0] == '-') {
 837                         diff_words_append(line, len,
 838                                           &ecbdata->diff_words->minus);
 839                         return;
 840                 } else if (line[0] == '+') {
 841                         diff_words_append(line, len,
 842                                           &ecbdata->diff_words->plus);
 843                         return;
 844                 }
 845                 diff_words_flush(ecbdata);
 846                 line++;
 847                 len--;
 848                 emit_line(ecbdata->file, plain, reset, line, len);
 849                 return;
 850         }
 851
 852         if (line[0] != '+') {
 853                 const char *color =
 854                         diff_get_color(ecbdata->color_diff,
 855                                        line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
 856                 ecbdata->lno_in_preimage++;
 857                 if (line[0] == ' ')
 858                         ecbdata->lno_in_postimage++;
 859                 emit_line(ecbdata->file, color, reset, line, len);
 860         } else {
 861                 ecbdata->lno_in_postimage++;
 862                 emit_add_line(reset, ecbdata, line + 1, len - 1);
 863         }
 864 }
 865
 866 static char *pprint_rename(const char *a, const char *b)
 867 {
 868         const char *old = a;
 869         const char *new = b;
 870         struct strbuf name = STRBUF_INIT;
 871         int pfx_length, sfx_length;
 872         int len_a = strlen(a);
 873         int len_b = strlen(b);
 874         int a_midlen, b_midlen;
 875         int qlen_a = quote_c_style(a, NULL, NULL, 0);
 876         int qlen_b = quote_c_style(b, NULL, NULL, 0);
 877
 878         if (qlen_a || qlen_b) {
 879                 quote_c_style(a, &name, NULL, 0);
 880                 strbuf_addstr(&name, " => ");
 881                 quote_c_style(b, &name, NULL, 0);
 882                 return strbuf_detach(&name, NULL);
 883         }
 884
 885         /* Find common prefix */
 886         pfx_length = 0;
 887         while (*old && *new && *old == *new) {
 888                 if (*old == '/')
 889                         pfx_length = old - a + 1;
 890                 old++;
 891                 new++;
 892         }
 893
 894         /* Find common suffix */
 895         old = a + len_a;
 896         new = b + len_b;
 897         sfx_length = 0;
 898         while (a <= old && b <= new && *old == *new) {
 899                 if (*old == '/')
 900                         sfx_length = len_a - (old - a);
 901                 old--;
 902                 new--;
 903         }
 904
 905         /*
 906          * pfx{mid-a => mid-b}sfx
 907          * {pfx-a => pfx-b}sfx
 908          * pfx{sfx-a => sfx-b}
 909          * name-a => name-b
 910          */
 911         a_midlen = len_a - pfx_length - sfx_length;
 912         b_midlen = len_b - pfx_length - sfx_length;
 913         if (a_midlen < 0)
 914                 a_midlen = 0;
 915         if (b_midlen < 0)
 916                 b_midlen = 0;
 917
 918         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 919         if (pfx_length + sfx_length) {
 920                 strbuf_add(&name, a, pfx_length);
 921                 strbuf_addch(&name, '{');
 922         }
 923         strbuf_add(&name, a + pfx_length, a_midlen);
 924         strbuf_addstr(&name, " => ");
 925         strbuf_add(&name, b + pfx_length, b_midlen);
 926         if (pfx_length + sfx_length) {
 927                 strbuf_addch(&name, '}');
 928                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 929         }
 930         return strbuf_detach(&name, NULL);
 931 }
 932
 933 struct diffstat_t {
 934         int nr;
 935         int alloc;
 936         struct diffstat_file {
 937                 char *from_name;
 938                 char *name;
 939                 char *print_name;
 940                 unsigned is_unmerged:1;
 941                 unsigned is_binary:1;
 942                 unsigned is_renamed:1;
 943                 uintmax_t added, deleted;
 944         } **files;
 945 };
 946
 947 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 948                                           const char *name_a,
 949                                           const char *name_b)
 950 {
 951         struct diffstat_file *x;
 952         x = xcalloc(sizeof (*x), 1);
 953         if (diffstat->nr == diffstat->alloc) {
 954                 diffstat->alloc = alloc_nr(diffstat->alloc);
 955                 diffstat->files = xrealloc(diffstat->files,
 956                                 diffstat->alloc * sizeof(x));
 957         }
 958         diffstat->files[diffstat->nr++] = x;
 959         if (name_b) {
 960                 x->from_name = xstrdup(name_a);
 961                 x->name = xstrdup(name_b);
 962                 x->is_renamed = 1;
 963         }
 964         else {
 965                 x->from_name = NULL;
 966                 x->name = xstrdup(name_a);
 967         }
 968         return x;
 969 }
 970
 971 static void diffstat_consume(void *priv, char *line, unsigned long len)
 972 {
 973         struct diffstat_t *diffstat = priv;
 974         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 975
 976         if (line[0] == '+')
 977                 x->added++;
 978         else if (line[0] == '-')
 979                 x->deleted++;
 980 }
 981
 982 const char mime_boundary_leader[] = "------------";
 983
 984 static int scale_linear(int it, int width, int max_change)
 985 {
 986         /*
 987          * make sure that at least one '-' is printed if there were deletions,
 988          * and likewise for '+'.
 989          */
 990         if (max_change < 2)
 991                 return it;
 992         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
 993 }
 994
 995 static void show_name(FILE *file,
 996                       const char *prefix, const char *name, int len)
 997 {
 998         fprintf(file, " %s%-*s |", prefix, len, name);
 999 }
1000
1001 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1002 {
1003         if (cnt <= 0)
1004                 return;
1005         fprintf(file, "%s", set);
1006         while (cnt--)
1007                 putc(ch, file);
1008         fprintf(file, "%s", reset);
1009 }
1010
1011 static void fill_print_name(struct diffstat_file *file)
1012 {
1013         char *pname;
1014
1015         if (file->print_name)
1016                 return;
1017
1018         if (!file->is_renamed) {
1019                 struct strbuf buf = STRBUF_INIT;
1020                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1021                         pname = strbuf_detach(&buf, NULL);
1022                 } else {
1023                         pname = file->name;
1024                         strbuf_release(&buf);
1025                 }
1026         } else {
1027                 pname = pprint_rename(file->from_name, file->name);
1028         }
1029         file->print_name = pname;
1030 }
1031
1032 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1033 {
1034         int i, len, add, del, adds = 0, dels = 0;
1035         uintmax_t max_change = 0, max_len = 0;
1036         int total_files = data->nr;
1037         int width, name_width;
1038         const char *reset, *set, *add_c, *del_c;
1039
1040         if (data->nr == 0)
1041                 return;
1042
1043         width = options->stat_width ? options->stat_width : 80;
1044         name_width = options->stat_name_width ? options->stat_name_width : 50;
1045
1046         /* Sanity: give at least 5 columns to the graph,
1047          * but leave at least 10 columns for the name.
1048          */
1049         if (width < 25)
1050                 width = 25;
1051         if (name_width < 10)
1052                 name_width = 10;
1053         else if (width < name_width + 15)
1054                 name_width = width - 15;
1055
1056         /* Find the longest filename and max number of changes */
1057         reset = diff_get_color_opt(options, DIFF_RESET);
1058         set   = diff_get_color_opt(options, DIFF_PLAIN);
1059         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1060         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1061
1062         for (i = 0; i < data->nr; i++) {
1063                 struct diffstat_file *file = data->files[i];
1064                 uintmax_t change = file->added + file->deleted;
1065                 fill_print_name(file);
1066                 len = strlen(file->print_name);
1067                 if (max_len < len)
1068                         max_len = len;
1069
1070                 if (file->is_binary || file->is_unmerged)
1071                         continue;
1072                 if (max_change < change)
1073                         max_change = change;
1074         }
1075
1076         /* Compute the width of the graph part;
1077          * 10 is for one blank at the beginning of the line plus
1078          * " | count " between the name and the graph.
1079          *
1080          * From here on, name_width is the width of the name area,
1081          * and width is the width of the graph area.
1082          */
1083         name_width = (name_width < max_len) ? name_width : max_len;
1084         if (width < (name_width + 10) + max_change)
1085                 width = width - (name_width + 10);
1086         else
1087                 width = max_change;
1088
1089         for (i = 0; i < data->nr; i++) {
1090                 const char *prefix = "";
1091                 char *name = data->files[i]->print_name;
1092                 uintmax_t added = data->files[i]->added;
1093                 uintmax_t deleted = data->files[i]->deleted;
1094                 int name_len;
1095
1096                 /*
1097                  * "scale" the filename
1098                  */
1099                 len = name_width;
1100                 name_len = strlen(name);
1101                 if (name_width < name_len) {
1102                         char *slash;
1103                         prefix = "...";
1104                         len -= 3;
1105                         name += name_len - len;
1106                         slash = strchr(name, '/');
1107                         if (slash)
1108                                 name = slash;
1109                 }
1110
1111                 if (data->files[i]->is_binary) {
1112                         show_name(options->file, prefix, name, len);
1113                         fprintf(options->file, "  Bin ");
1114                         fprintf(options->file, "%s%"PRIuMAX"%s",
1115                                 del_c, deleted, reset);
1116                         fprintf(options->file, " -> ");
1117                         fprintf(options->file, "%s%"PRIuMAX"%s",
1118                                 add_c, added, reset);
1119                         fprintf(options->file, " bytes");
1120                         fprintf(options->file, "\n");
1121                         continue;
1122                 }
1123                 else if (data->files[i]->is_unmerged) {
1124                         show_name(options->file, prefix, name, len);
1125                         fprintf(options->file, "  Unmerged\n");
1126                         continue;
1127                 }
1128                 else if (!data->files[i]->is_renamed &&
1129                          (added + deleted == 0)) {
1130                         total_files--;
1131                         continue;
1132                 }
1133
1134                 /*
1135                  * scale the add/delete
1136                  */
1137                 add = added;
1138                 del = deleted;
1139                 adds += add;
1140                 dels += del;
1141
1142                 if (width <= max_change) {
1143                         add = scale_linear(add, width, max_change);
1144                         del = scale_linear(del, width, max_change);
1145                 }
1146                 show_name(options->file, prefix, name, len);
1147                 fprintf(options->file, "%5"PRIuMAX"%s", added + deleted,
1148                                 added + deleted ? " " : "");
1149                 show_graph(options->file, '+', add, add_c, reset);
1150                 show_graph(options->file, '-', del, del_c, reset);
1151                 fprintf(options->file, "\n");
1152         }
1153         fprintf(options->file,
1154                " %d files changed, %d insertions(+), %d deletions(-)\n",
1155                total_files, adds, dels);
1156 }
1157
1158 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1159 {
1160         int i, adds = 0, dels = 0, total_files = data->nr;
1161
1162         if (data->nr == 0)
1163                 return;
1164
1165         for (i = 0; i < data->nr; i++) {
1166                 if (!data->files[i]->is_binary &&
1167                     !data->files[i]->is_unmerged) {
1168                         int added = data->files[i]->added;
1169                         int deleted= data->files[i]->deleted;
1170                         if (!data->files[i]->is_renamed &&
1171                             (added + deleted == 0)) {
1172                                 total_files--;
1173                         } else {
1174                                 adds += added;
1175                                 dels += deleted;
1176                         }
1177                 }
1178         }
1179         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
1180                total_files, adds, dels);
1181 }
1182
1183 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1184 {
1185         int i;
1186
1187         if (data->nr == 0)
1188                 return;
1189
1190         for (i = 0; i < data->nr; i++) {
1191                 struct diffstat_file *file = data->files[i];
1192
1193                 if (file->is_binary)
1194                         fprintf(options->file, "-\t-\t");
1195                 else
1196                         fprintf(options->file,
1197                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1198                                 file->added, file->deleted);
1199                 if (options->line_termination) {
1200                         fill_print_name(file);
1201                         if (!file->is_renamed)
1202                                 write_name_quoted(file->name, options->file,
1203                                                   options->line_termination);
1204                         else {
1205                                 fputs(file->print_name, options->file);
1206                                 putc(options->line_termination, options->file);
1207                         }
1208                 } else {
1209                         if (file->is_renamed) {
1210                                 putc('\0', options->file);
1211                                 write_name_quoted(file->from_name, options->file, '\0');
1212                         }
1213                         write_name_quoted(file->name, options->file, '\0');
1214                 }
1215         }
1216 }
1217
1218 struct dirstat_file {
1219         const char *name;
1220         unsigned long changed;
1221 };
1222
1223 struct dirstat_dir {
1224         struct dirstat_file *files;
1225         int alloc, nr, percent, cumulative;
1226 };
1227
1228 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1229 {
1230         unsigned long this_dir = 0;
1231         unsigned int sources = 0;
1232
1233         while (dir->nr) {
1234                 struct dirstat_file *f = dir->files;
1235                 int namelen = strlen(f->name);
1236                 unsigned long this;
1237                 char *slash;
1238
1239                 if (namelen < baselen)
1240                         break;
1241                 if (memcmp(f->name, base, baselen))
1242                         break;
1243                 slash = strchr(f->name + baselen, '/');
1244                 if (slash) {
1245                         int newbaselen = slash + 1 - f->name;
1246                         this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1247                         sources++;
1248                 } else {
1249                         this = f->changed;
1250                         dir->files++;
1251                         dir->nr--;
1252                         sources += 2;
1253                 }
1254                 this_dir += this;
1255         }
1256
1257         /*
1258          * We don't report dirstat's for
1259          *  - the top level
1260          *  - or cases where everything came from a single directory
1261          *    under this directory (sources == 1).
1262          */
1263         if (baselen && sources != 1) {
1264                 int permille = this_dir * 1000 / changed;
1265                 if (permille) {
1266                         int percent = permille / 10;
1267                         if (percent >= dir->percent) {
1268                                 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1269                                 if (!dir->cumulative)
1270                                         return 0;
1271                         }
1272                 }
1273         }
1274         return this_dir;
1275 }
1276
1277 static int dirstat_compare(const void *_a, const void *_b)
1278 {
1279         const struct dirstat_file *a = _a;
1280         const struct dirstat_file *b = _b;
1281         return strcmp(a->name, b->name);
1282 }
1283
1284 static void show_dirstat(struct diff_options *options)
1285 {
1286         int i;
1287         unsigned long changed;
1288         struct dirstat_dir dir;
1289         struct diff_queue_struct *q = &diff_queued_diff;
1290
1291         dir.files = NULL;
1292         dir.alloc = 0;
1293         dir.nr = 0;
1294         dir.percent = options->dirstat_percent;
1295         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1296
1297         changed = 0;
1298         for (i = 0; i < q->nr; i++) {
1299                 struct diff_filepair *p = q->queue[i];
1300                 const char *name;
1301                 unsigned long copied, added, damage;
1302
1303                 name = p->one->path ? p->one->path : p->two->path;
1304
1305                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1306                         diff_populate_filespec(p->one, 0);
1307                         diff_populate_filespec(p->two, 0);
1308                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1309                                                &copied, &added);
1310                         diff_free_filespec_data(p->one);
1311                         diff_free_filespec_data(p->two);
1312                 } else if (DIFF_FILE_VALID(p->one)) {
1313                         diff_populate_filespec(p->one, 1);
1314                         copied = added = 0;
1315                         diff_free_filespec_data(p->one);
1316                 } else if (DIFF_FILE_VALID(p->two)) {
1317                         diff_populate_filespec(p->two, 1);
1318                         copied = 0;
1319                         added = p->two->size;
1320                         diff_free_filespec_data(p->two);
1321                 } else
1322                         continue;
1323
1324                 /*
1325                  * Original minus copied is the removed material,
1326                  * added is the new material.  They are both damages
1327                  * made to the preimage. In --dirstat-by-file mode, count
1328                  * damaged files, not damaged lines. This is done by
1329                  * counting only a single damaged line per file.
1330                  */
1331                 damage = (p->one->size - copied) + added;
1332                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1333                         damage = 1;
1334
1335                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1336                 dir.files[dir.nr].name = name;
1337                 dir.files[dir.nr].changed = damage;
1338                 changed += damage;
1339                 dir.nr++;
1340         }
1341
1342         /* This can happen even with many files, if everything was renames */
1343         if (!changed)
1344                 return;
1345
1346         /* Show all directories with more than x% of the changes */
1347         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1348         gather_dirstat(options->file, &dir, changed, "", 0);
1349 }
1350
1351 static void free_diffstat_info(struct diffstat_t *diffstat)
1352 {
1353         int i;
1354         for (i = 0; i < diffstat->nr; i++) {
1355                 struct diffstat_file *f = diffstat->files[i];
1356                 if (f->name != f->print_name)
1357                         free(f->print_name);
1358                 free(f->name);
1359                 free(f->from_name);
1360                 free(f);
1361         }
1362         free(diffstat->files);
1363 }
1364
1365 struct checkdiff_t {
1366         const char *filename;
1367         int lineno;
1368         int conflict_marker_size;
1369         struct diff_options *o;
1370         unsigned ws_rule;
1371         unsigned status;
1372 };
1373
1374 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
1375 {
1376         char firstchar;
1377         int cnt;
1378
1379         if (len < marker_size + 1)
1380                 return 0;
1381         firstchar = line[0];
1382         switch (firstchar) {
1383         case '=': case '>': case '<': case '|':
1384                 break;
1385         default:
1386                 return 0;
1387         }
1388         for (cnt = 1; cnt < marker_size; cnt++)
1389                 if (line[cnt] != firstchar)
1390                         return 0;
1391         /* line[1] thru line[marker_size-1] are same as firstchar */
1392         if (len < marker_size + 1 || !isspace(line[marker_size]))
1393                 return 0;
1394         return 1;
1395 }
1396
1397 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1398 {
1399         struct checkdiff_t *data = priv;
1400         int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1401         int marker_size = data->conflict_marker_size;
1402         const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1403         const char *reset = diff_get_color(color_diff, DIFF_RESET);
1404         const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1405         char *err;
1406
1407         if (line[0] == '+') {
1408                 unsigned bad;
1409                 data->lineno++;
1410                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
1411                         data->status |= 1;
1412                         fprintf(data->o->file,
1413                                 "%s:%d: leftover conflict marker\n",
1414                                 data->filename, data->lineno);
1415                 }
1416                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1417                 if (!bad)
1418                         return;
1419                 data->status |= bad;
1420                 err = whitespace_error_string(bad);
1421                 fprintf(data->o->file, "%s:%d: %s.\n",
1422                         data->filename, data->lineno, err);
1423                 free(err);
1424                 emit_line(data->o->file, set, reset, line, 1);
1425                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1426                               data->o->file, set, reset, ws);
1427         } else if (line[0] == ' ') {
1428                 data->lineno++;
1429         } else if (line[0] == '@') {
1430                 char *plus = strchr(line, '+');
1431                 if (plus)
1432                         data->lineno = strtol(plus, NULL, 10) - 1;
1433                 else
1434                         die("invalid diff");
1435         }
1436 }
1437
1438 static unsigned char *deflate_it(char *data,
1439                                  unsigned long size,
1440                                  unsigned long *result_size)
1441 {
1442         int bound;
1443         unsigned char *deflated;
1444         z_stream stream;
1445
1446         memset(&stream, 0, sizeof(stream));
1447         deflateInit(&stream, zlib_compression_level);
1448         bound = deflateBound(&stream, size);
1449         deflated = xmalloc(bound);
1450         stream.next_out = deflated;
1451         stream.avail_out = bound;
1452
1453         stream.next_in = (unsigned char *)data;
1454         stream.avail_in = size;
1455         while (deflate(&stream, Z_FINISH) == Z_OK)
1456                 ; /* nothing */
1457         deflateEnd(&stream);
1458         *result_size = stream.total_out;
1459         return deflated;
1460 }
1461
1462 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1463 {
1464         void *cp;
1465         void *delta;
1466         void *deflated;
1467         void *data;
1468         unsigned long orig_size;
1469         unsigned long delta_size;
1470         unsigned long deflate_size;
1471         unsigned long data_size;
1472
1473         /* We could do deflated delta, or we could do just deflated two,
1474          * whichever is smaller.
1475          */
1476         delta = NULL;
1477         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1478         if (one->size && two->size) {
1479                 delta = diff_delta(one->ptr, one->size,
1480                                    two->ptr, two->size,
1481                                    &delta_size, deflate_size);
1482                 if (delta) {
1483                         void *to_free = delta;
1484                         orig_size = delta_size;
1485                         delta = deflate_it(delta, delta_size, &delta_size);
1486                         free(to_free);
1487                 }
1488         }
1489
1490         if (delta && delta_size < deflate_size) {
1491                 fprintf(file, "delta %lu\n", orig_size);
1492                 free(deflated);
1493                 data = delta;
1494                 data_size = delta_size;
1495         }
1496         else {
1497                 fprintf(file, "literal %lu\n", two->size);
1498                 free(delta);
1499                 data = deflated;
1500                 data_size = deflate_size;
1501         }
1502
1503         /* emit data encoded in base85 */
1504         cp = data;
1505         while (data_size) {
1506                 int bytes = (52 < data_size) ? 52 : data_size;
1507                 char line[70];
1508                 data_size -= bytes;
1509                 if (bytes <= 26)
1510                         line[0] = bytes + 'A' - 1;
1511                 else
1512                         line[0] = bytes - 26 + 'a' - 1;
1513                 encode_85(line + 1, cp, bytes);
1514                 cp = (char *) cp + bytes;
1515                 fputs(line, file);
1516                 fputc('\n', file);
1517         }
1518         fprintf(file, "\n");
1519         free(data);
1520 }
1521
1522 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1523 {
1524         fprintf(file, "GIT binary patch\n");
1525         emit_binary_diff_body(file, one, two);
1526         emit_binary_diff_body(file, two, one);
1527 }
1528
1529 static void diff_filespec_load_driver(struct diff_filespec *one)
1530 {
1531         if (!one->driver)
1532                 one->driver = userdiff_find_by_path(one->path);
1533         if (!one->driver)
1534                 one->driver = userdiff_find_by_name("default");
1535 }
1536
1537 int diff_filespec_is_binary(struct diff_filespec *one)
1538 {
1539         if (one->is_binary == -1) {
1540                 diff_filespec_load_driver(one);
1541                 if (one->driver->binary != -1)
1542                         one->is_binary = one->driver->binary;
1543                 else {
1544                         if (!one->data && DIFF_FILE_VALID(one))
1545                                 diff_populate_filespec(one, 0);
1546                         if (one->data)
1547                                 one->is_binary = buffer_is_binary(one->data,
1548                                                 one->size);
1549                         if (one->is_binary == -1)
1550                                 one->is_binary = 0;
1551                 }
1552         }
1553         return one->is_binary;
1554 }
1555
1556 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1557 {
1558         diff_filespec_load_driver(one);
1559         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1560 }
1561
1562 static const char *userdiff_word_regex(struct diff_filespec *one)
1563 {
1564         diff_filespec_load_driver(one);
1565         return one->driver->word_regex;
1566 }
1567
1568 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1569 {
1570         if (!options->a_prefix)
1571                 options->a_prefix = a;
1572         if (!options->b_prefix)
1573                 options->b_prefix = b;
1574 }
1575
1576 static struct userdiff_driver *get_textconv(struct diff_filespec *one)
1577 {
1578         if (!DIFF_FILE_VALID(one))
1579                 return NULL;
1580         if (!S_ISREG(one->mode))
1581                 return NULL;
1582         diff_filespec_load_driver(one);
1583         if (!one->driver->textconv)
1584                 return NULL;
1585
1586         if (one->driver->textconv_want_cache && !one->driver->textconv_cache) {
1587                 struct notes_cache *c = xmalloc(sizeof(*c));
1588                 struct strbuf name = STRBUF_INIT;
1589
1590                 strbuf_addf(&name, "textconv/%s", one->driver->name);
1591                 notes_cache_init(c, name.buf, one->driver->textconv);
1592                 one->driver->textconv_cache = c;
1593         }
1594
1595         return one->driver;
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                          struct diff_options *o,
1604                          int complete_rewrite)
1605 {
1606         mmfile_t mf1, mf2;
1607         const char *lbl[2];
1608         char *a_one, *b_two;
1609         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1610         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1611         const char *a_prefix, *b_prefix;
1612         struct userdiff_driver *textconv_one = NULL;
1613         struct userdiff_driver *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 && xfrm_msg[0])
1654                         strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1655         }
1656         else if (lbl[1][0] == '/') {
1657                 strbuf_addf(&header, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1658                 if (xfrm_msg && xfrm_msg[0])
1659                         strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1660         }
1661         else {
1662                 if (one->mode != two->mode) {
1663                         strbuf_addf(&header, "%sold mode %06o%s\n", set, one->mode, reset);
1664                         strbuf_addf(&header, "%snew mode %06o%s\n", set, two->mode, reset);
1665                 }
1666                 if (xfrm_msg && xfrm_msg[0])
1667                         strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1668
1669                 /*
1670                  * we do not run diff between different kind
1671                  * of objects.
1672                  */
1673                 if ((one->mode ^ two->mode) & S_IFMT)
1674                         goto free_ab_and_return;
1675                 if (complete_rewrite &&
1676                     (textconv_one || !diff_filespec_is_binary(one)) &&
1677                     (textconv_two || !diff_filespec_is_binary(two))) {
1678                         fprintf(o->file, "%s", header.buf);
1679                         strbuf_reset(&header);
1680                         emit_rewrite_diff(name_a, name_b, one, two,
1681                                                 textconv_one, textconv_two, o);
1682                         o->found_changes = 1;
1683                         goto free_ab_and_return;
1684                 }
1685         }
1686
1687         if (!DIFF_OPT_TST(o, TEXT) &&
1688             ( (!textconv_one && diff_filespec_is_binary(one)) ||
1689               (!textconv_two && diff_filespec_is_binary(two)) )) {
1690                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1691                         die("unable to read files to diff");
1692                 /* Quite common confusing case */
1693                 if (mf1.size == mf2.size &&
1694                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1695                         goto free_ab_and_return;
1696                 fprintf(o->file, "%s", header.buf);
1697                 strbuf_reset(&header);
1698                 if (DIFF_OPT_TST(o, BINARY))
1699                         emit_binary_diff(o->file, &mf1, &mf2);
1700                 else
1701                         fprintf(o->file, "Binary files %s and %s differ\n",
1702                                 lbl[0], lbl[1]);
1703                 o->found_changes = 1;
1704         }
1705         else {
1706                 /* Crazy xdl interfaces.. */
1707                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1708                 xpparam_t xpp;
1709                 xdemitconf_t xecfg;
1710                 struct emit_callback ecbdata;
1711                 const struct userdiff_funcname *pe;
1712
1713                 if (!DIFF_XDL_TST(o, WHITESPACE_FLAGS)) {
1714                         fprintf(o->file, "%s", header.buf);
1715                         strbuf_reset(&header);
1716                 }
1717
1718                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
1719                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
1720
1721                 pe = diff_funcname_pattern(one);
1722                 if (!pe)
1723                         pe = diff_funcname_pattern(two);
1724
1725                 memset(&xpp, 0, sizeof(xpp));
1726                 memset(&xecfg, 0, sizeof(xecfg));
1727                 memset(&ecbdata, 0, sizeof(ecbdata));
1728                 ecbdata.label_path = lbl;
1729                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1730                 ecbdata.found_changesp = &o->found_changes;
1731                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1732                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
1733                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
1734                 ecbdata.file = o->file;
1735                 ecbdata.header = header.len ? &header : NULL;
1736                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1737                 xecfg.ctxlen = o->context;
1738                 xecfg.interhunkctxlen = o->interhunkcontext;
1739                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1740                 if (pe)
1741                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1742                 if (!diffopts)
1743                         ;
1744                 else if (!prefixcmp(diffopts, "--unified="))
1745                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1746                 else if (!prefixcmp(diffopts, "-u"))
1747                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1748                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1749                         ecbdata.diff_words =
1750                                 xcalloc(1, sizeof(struct diff_words_data));
1751                         ecbdata.diff_words->file = o->file;
1752                         if (!o->word_regex)
1753                                 o->word_regex = userdiff_word_regex(one);
1754                         if (!o->word_regex)
1755                                 o->word_regex = userdiff_word_regex(two);
1756                         if (!o->word_regex)
1757                                 o->word_regex = diff_word_regex_cfg;
1758                         if (o->word_regex) {
1759                                 ecbdata.diff_words->word_regex = (regex_t *)
1760                                         xmalloc(sizeof(regex_t));
1761                                 if (regcomp(ecbdata.diff_words->word_regex,
1762                                                 o->word_regex,
1763                                                 REG_EXTENDED | REG_NEWLINE))
1764                                         die ("Invalid regular expression: %s",
1765                                                         o->word_regex);
1766                         }
1767                 }
1768                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1769                               &xpp, &xecfg);
1770                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1771                         free_diff_words_data(&ecbdata);
1772                 if (textconv_one)
1773                         free(mf1.ptr);
1774                 if (textconv_two)
1775                         free(mf2.ptr);
1776                 xdiff_clear_find_func(&xecfg);
1777         }
1778
1779  free_ab_and_return:
1780         strbuf_release(&header);
1781         diff_free_filespec_data(one);
1782         diff_free_filespec_data(two);
1783         free(a_one);
1784         free(b_two);
1785         return;
1786 }
1787
1788 static void builtin_diffstat(const char *name_a, const char *name_b,
1789                              struct diff_filespec *one,
1790                              struct diff_filespec *two,
1791                              struct diffstat_t *diffstat,
1792                              struct diff_options *o,
1793                              int complete_rewrite)
1794 {
1795         mmfile_t mf1, mf2;
1796         struct diffstat_file *data;
1797
1798         data = diffstat_add(diffstat, name_a, name_b);
1799
1800         if (!one || !two) {
1801                 data->is_unmerged = 1;
1802                 return;
1803         }
1804         if (complete_rewrite) {
1805                 diff_populate_filespec(one, 0);
1806                 diff_populate_filespec(two, 0);
1807                 data->deleted = count_lines(one->data, one->size);
1808                 data->added = count_lines(two->data, two->size);
1809                 goto free_and_return;
1810         }
1811         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1812                 die("unable to read files to diff");
1813
1814         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1815                 data->is_binary = 1;
1816                 data->added = mf2.size;
1817                 data->deleted = mf1.size;
1818         } else {
1819                 /* Crazy xdl interfaces.. */
1820                 xpparam_t xpp;
1821                 xdemitconf_t xecfg;
1822
1823                 memset(&xpp, 0, sizeof(xpp));
1824                 memset(&xecfg, 0, sizeof(xecfg));
1825                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1826                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1827                               &xpp, &xecfg);
1828         }
1829
1830  free_and_return:
1831         diff_free_filespec_data(one);
1832         diff_free_filespec_data(two);
1833 }
1834
1835 static void builtin_checkdiff(const char *name_a, const char *name_b,
1836                               const char *attr_path,
1837                               struct diff_filespec *one,
1838                               struct diff_filespec *two,
1839                               struct diff_options *o)
1840 {
1841         mmfile_t mf1, mf2;
1842         struct checkdiff_t data;
1843
1844         if (!two)
1845                 return;
1846
1847         memset(&data, 0, sizeof(data));
1848         data.filename = name_b ? name_b : name_a;
1849         data.lineno = 0;
1850         data.o = o;
1851         data.ws_rule = whitespace_rule(attr_path);
1852         data.conflict_marker_size = ll_merge_marker_size(attr_path);
1853
1854         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1855                 die("unable to read files to diff");
1856
1857         /*
1858          * All the other codepaths check both sides, but not checking
1859          * the "old" side here is deliberate.  We are checking the newly
1860          * introduced changes, and as long as the "new" side is text, we
1861          * can and should check what it introduces.
1862          */
1863         if (diff_filespec_is_binary(two))
1864                 goto free_and_return;
1865         else {
1866                 /* Crazy xdl interfaces.. */
1867                 xpparam_t xpp;
1868                 xdemitconf_t xecfg;
1869
1870                 memset(&xpp, 0, sizeof(xpp));
1871                 memset(&xecfg, 0, sizeof(xecfg));
1872                 xecfg.ctxlen = 1; /* at least one context line */
1873                 xpp.flags = XDF_NEED_MINIMAL;
1874                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1875                               &xpp, &xecfg);
1876
1877                 if (data.ws_rule & WS_BLANK_AT_EOF) {
1878                         struct emit_callback ecbdata;
1879                         int blank_at_eof;
1880
1881                         ecbdata.ws_rule = data.ws_rule;
1882                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
1883                         blank_at_eof = ecbdata.blank_at_eof_in_preimage;
1884
1885                         if (blank_at_eof) {
1886                                 static char *err;
1887                                 if (!err)
1888                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
1889                                 fprintf(o->file, "%s:%d: %s.\n",
1890                                         data.filename, blank_at_eof, err);
1891                                 data.status = 1; /* report errors */
1892                         }
1893                 }
1894         }
1895  free_and_return:
1896         diff_free_filespec_data(one);
1897         diff_free_filespec_data(two);
1898         if (data.status)
1899                 DIFF_OPT_SET(o, CHECK_FAILED);
1900 }
1901
1902 struct diff_filespec *alloc_filespec(const char *path)
1903 {
1904         int namelen = strlen(path);
1905         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1906
1907         memset(spec, 0, sizeof(*spec));
1908         spec->path = (char *)(spec + 1);
1909         memcpy(spec->path, path, namelen+1);
1910         spec->count = 1;
1911         spec->is_binary = -1;
1912         return spec;
1913 }
1914
1915 void free_filespec(struct diff_filespec *spec)
1916 {
1917         if (!--spec->count) {
1918                 diff_free_filespec_data(spec);
1919                 free(spec);
1920         }
1921 }
1922
1923 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1924                    unsigned short mode)
1925 {
1926         if (mode) {
1927                 spec->mode = canon_mode(mode);
1928                 hashcpy(spec->sha1, sha1);
1929                 spec->sha1_valid = !is_null_sha1(sha1);
1930         }
1931 }
1932
1933 /*
1934  * Given a name and sha1 pair, if the index tells us the file in
1935  * the work tree has that object contents, return true, so that
1936  * prepare_temp_file() does not have to inflate and extract.
1937  */
1938 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1939 {
1940         struct cache_entry *ce;
1941         struct stat st;
1942         int pos, len;
1943
1944         /*
1945          * We do not read the cache ourselves here, because the
1946          * benchmark with my previous version that always reads cache
1947          * shows that it makes things worse for diff-tree comparing
1948          * two linux-2.6 kernel trees in an already checked out work
1949          * tree.  This is because most diff-tree comparisons deal with
1950          * only a small number of files, while reading the cache is
1951          * expensive for a large project, and its cost outweighs the
1952          * savings we get by not inflating the object to a temporary
1953          * file.  Practically, this code only helps when we are used
1954          * by diff-cache --cached, which does read the cache before
1955          * calling us.
1956          */
1957         if (!active_cache)
1958                 return 0;
1959
1960         /* We want to avoid the working directory if our caller
1961          * doesn't need the data in a normal file, this system
1962          * is rather slow with its stat/open/mmap/close syscalls,
1963          * and the object is contained in a pack file.  The pack
1964          * is probably already open and will be faster to obtain
1965          * the data through than the working directory.  Loose
1966          * objects however would tend to be slower as they need
1967          * to be individually opened and inflated.
1968          */
1969         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
1970                 return 0;
1971
1972         len = strlen(name);
1973         pos = cache_name_pos(name, len);
1974         if (pos < 0)
1975                 return 0;
1976         ce = active_cache[pos];
1977
1978         /*
1979          * This is not the sha1 we are looking for, or
1980          * unreusable because it is not a regular file.
1981          */
1982         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1983                 return 0;
1984
1985         /*
1986          * If ce is marked as "assume unchanged", there is no
1987          * guarantee that work tree matches what we are looking for.
1988          */
1989         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
1990                 return 0;
1991
1992         /*
1993          * If ce matches the file in the work tree, we can reuse it.
1994          */
1995         if (ce_uptodate(ce) ||
1996             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1997                 return 1;
1998
1999         return 0;
2000 }
2001
2002 static int populate_from_stdin(struct diff_filespec *s)
2003 {
2004         struct strbuf buf = STRBUF_INIT;
2005         size_t size = 0;
2006
2007         if (strbuf_read(&buf, 0, 0) < 0)
2008                 return error("error while reading from stdin %s",
2009                                      strerror(errno));
2010
2011         s->should_munmap = 0;
2012         s->data = strbuf_detach(&buf, &size);
2013         s->size = size;
2014         s->should_free = 1;
2015         return 0;
2016 }
2017
2018 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2019 {
2020         int len;
2021         char *data = xmalloc(100), *dirty = "";
2022
2023         /* Are we looking at the work tree? */
2024         if (s->dirty_submodule)
2025                 dirty = "-dirty";
2026
2027         len = snprintf(data, 100,
2028                        "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2029         s->data = data;
2030         s->size = len;
2031         s->should_free = 1;
2032         if (size_only) {
2033                 s->data = NULL;
2034                 free(data);
2035         }
2036         return 0;
2037 }
2038
2039 /*
2040  * While doing rename detection and pickaxe operation, we may need to
2041  * grab the data for the blob (or file) for our own in-core comparison.
2042  * diff_filespec has data and size fields for this purpose.
2043  */
2044 int diff_populate_filespec(struct diff_filespec *s, int size_only)
2045 {
2046         int err = 0;
2047         if (!DIFF_FILE_VALID(s))
2048                 die("internal error: asking to populate invalid file.");
2049         if (S_ISDIR(s->mode))
2050                 return -1;
2051
2052         if (s->data)
2053                 return 0;
2054
2055         if (size_only && 0 < s->size)
2056                 return 0;
2057
2058         if (S_ISGITLINK(s->mode))
2059                 return diff_populate_gitlink(s, size_only);
2060
2061         if (!s->sha1_valid ||
2062             reuse_worktree_file(s->path, s->sha1, 0)) {
2063                 struct strbuf buf = STRBUF_INIT;
2064                 struct stat st;
2065                 int fd;
2066
2067                 if (!strcmp(s->path, "-"))
2068                         return populate_from_stdin(s);
2069
2070                 if (lstat(s->path, &st) < 0) {
2071                         if (errno == ENOENT) {
2072                         err_empty:
2073                                 err = -1;
2074                         empty:
2075                                 s->data = (char *)"";
2076                                 s->size = 0;
2077                                 return err;
2078                         }
2079                 }
2080                 s->size = xsize_t(st.st_size);
2081                 if (!s->size)
2082                         goto empty;
2083                 if (S_ISLNK(st.st_mode)) {
2084                         struct strbuf sb = STRBUF_INIT;
2085
2086                         if (strbuf_readlink(&sb, s->path, s->size))
2087                                 goto err_empty;
2088                         s->size = sb.len;
2089                         s->data = strbuf_detach(&sb, NULL);
2090                         s->should_free = 1;
2091                         return 0;
2092                 }
2093                 if (size_only)
2094                         return 0;
2095                 fd = open(s->path, O_RDONLY);
2096                 if (fd < 0)
2097                         goto err_empty;
2098                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2099                 close(fd);
2100                 s->should_munmap = 1;
2101
2102                 /*
2103                  * Convert from working tree format to canonical git format
2104                  */
2105                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2106                         size_t size = 0;
2107                         munmap(s->data, s->size);
2108                         s->should_munmap = 0;
2109                         s->data = strbuf_detach(&buf, &size);
2110                         s->size = size;
2111                         s->should_free = 1;
2112                 }
2113         }
2114         else {
2115                 enum object_type type;
2116                 if (size_only)
2117                         type = sha1_object_info(s->sha1, &s->size);
2118                 else {
2119                         s->data = read_sha1_file(s->sha1, &type, &s->size);
2120                         s->should_free = 1;
2121                 }
2122         }
2123         return 0;
2124 }
2125
2126 void diff_free_filespec_blob(struct diff_filespec *s)
2127 {
2128         if (s->should_free)
2129                 free(s->data);
2130         else if (s->should_munmap)
2131                 munmap(s->data, s->size);
2132
2133         if (s->should_free || s->should_munmap) {
2134                 s->should_free = s->should_munmap = 0;
2135                 s->data = NULL;
2136         }
2137 }
2138
2139 void diff_free_filespec_data(struct diff_filespec *s)
2140 {
2141         diff_free_filespec_blob(s);
2142         free(s->cnt_data);
2143         s->cnt_data = NULL;
2144 }
2145
2146 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2147                            void *blob,
2148                            unsigned long size,
2149                            const unsigned char *sha1,
2150                            int mode)
2151 {
2152         int fd;
2153         struct strbuf buf = STRBUF_INIT;
2154         struct strbuf template = STRBUF_INIT;
2155         char *path_dup = xstrdup(path);
2156         const char *base = basename(path_dup);
2157
2158         /* Generate "XXXXXX_basename.ext" */
2159         strbuf_addstr(&template, "XXXXXX_");
2160         strbuf_addstr(&template, base);
2161
2162         fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2163                         strlen(base) + 1);
2164         if (fd < 0)
2165                 die_errno("unable to create temp-file");
2166         if (convert_to_working_tree(path,
2167                         (const char *)blob, (size_t)size, &buf)) {
2168                 blob = buf.buf;
2169                 size = buf.len;
2170         }
2171         if (write_in_full(fd, blob, size) != size)
2172                 die_errno("unable to write temp-file");
2173         close(fd);
2174         temp->name = temp->tmp_path;
2175         strcpy(temp->hex, sha1_to_hex(sha1));
2176         temp->hex[40] = 0;
2177         sprintf(temp->mode, "%06o", mode);
2178         strbuf_release(&buf);
2179         strbuf_release(&template);
2180         free(path_dup);
2181 }
2182
2183 static struct diff_tempfile *prepare_temp_file(const char *name,
2184                 struct diff_filespec *one)
2185 {
2186         struct diff_tempfile *temp = claim_diff_tempfile();
2187
2188         if (!DIFF_FILE_VALID(one)) {
2189         not_a_valid_file:
2190                 /* A '-' entry produces this for file-2, and
2191                  * a '+' entry produces this for file-1.
2192                  */
2193                 temp->name = "/dev/null";
2194                 strcpy(temp->hex, ".");
2195                 strcpy(temp->mode, ".");
2196                 return temp;
2197         }
2198
2199         if (!remove_tempfile_installed) {
2200                 atexit(remove_tempfile);
2201                 sigchain_push_common(remove_tempfile_on_signal);
2202                 remove_tempfile_installed = 1;
2203         }
2204
2205         if (!one->sha1_valid ||
2206             reuse_worktree_file(name, one->sha1, 1)) {
2207                 struct stat st;
2208                 if (lstat(name, &st) < 0) {
2209                         if (errno == ENOENT)
2210                                 goto not_a_valid_file;
2211                         die_errno("stat(%s)", name);
2212                 }
2213                 if (S_ISLNK(st.st_mode)) {
2214                         struct strbuf sb = STRBUF_INIT;
2215                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2216                                 die_errno("readlink(%s)", name);
2217                         prep_temp_blob(name, temp, sb.buf, sb.len,
2218                                        (one->sha1_valid ?
2219                                         one->sha1 : null_sha1),
2220                                        (one->sha1_valid ?
2221                                         one->mode : S_IFLNK));
2222                         strbuf_release(&sb);
2223                 }
2224                 else {
2225                         /* we can borrow from the file in the work tree */
2226                         temp->name = name;
2227                         if (!one->sha1_valid)
2228                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2229                         else
2230                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2231                         /* Even though we may sometimes borrow the
2232                          * contents from the work tree, we always want
2233                          * one->mode.  mode is trustworthy even when
2234                          * !(one->sha1_valid), as long as
2235                          * DIFF_FILE_VALID(one).
2236                          */
2237                         sprintf(temp->mode, "%06o", one->mode);
2238                 }
2239                 return temp;
2240         }
2241         else {
2242                 if (diff_populate_filespec(one, 0))
2243                         die("cannot read data blob for %s", one->path);
2244                 prep_temp_blob(name, temp, one->data, one->size,
2245                                one->sha1, one->mode);
2246         }
2247         return temp;
2248 }
2249
2250 /* An external diff command takes:
2251  *
2252  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2253  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2254  *
2255  */
2256 static void run_external_diff(const char *pgm,
2257                               const char *name,
2258                               const char *other,
2259                               struct diff_filespec *one,
2260                               struct diff_filespec *two,
2261                               const char *xfrm_msg,
2262                               int complete_rewrite)
2263 {
2264         const char *spawn_arg[10];
2265         int retval;
2266         const char **arg = &spawn_arg[0];
2267
2268         if (one && two) {
2269                 struct diff_tempfile *temp_one, *temp_two;
2270                 const char *othername = (other ? other : name);
2271                 temp_one = prepare_temp_file(name, one);
2272                 temp_two = prepare_temp_file(othername, two);
2273                 *arg++ = pgm;
2274                 *arg++ = name;
2275                 *arg++ = temp_one->name;
2276                 *arg++ = temp_one->hex;
2277                 *arg++ = temp_one->mode;
2278                 *arg++ = temp_two->name;
2279                 *arg++ = temp_two->hex;
2280                 *arg++ = temp_two->mode;
2281                 if (other) {
2282                         *arg++ = other;
2283                         *arg++ = xfrm_msg;
2284                 }
2285         } else {
2286                 *arg++ = pgm;
2287                 *arg++ = name;
2288         }
2289         *arg = NULL;
2290         fflush(NULL);
2291         retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2292         remove_tempfile();
2293         if (retval) {
2294                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2295                 exit(1);
2296         }
2297 }
2298
2299 static int similarity_index(struct diff_filepair *p)
2300 {
2301         return p->score * 100 / MAX_SCORE;
2302 }
2303
2304 static void fill_metainfo(struct strbuf *msg,
2305                           const char *name,
2306                           const char *other,
2307                           struct diff_filespec *one,
2308                           struct diff_filespec *two,
2309                           struct diff_options *o,
2310                           struct diff_filepair *p)
2311 {
2312         strbuf_init(msg, PATH_MAX * 2 + 300);
2313         switch (p->status) {
2314         case DIFF_STATUS_COPIED:
2315                 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2316                 strbuf_addstr(msg, "\ncopy from ");
2317                 quote_c_style(name, msg, NULL, 0);
2318                 strbuf_addstr(msg, "\ncopy to ");
2319                 quote_c_style(other, msg, NULL, 0);
2320                 strbuf_addch(msg, '\n');
2321                 break;
2322         case DIFF_STATUS_RENAMED:
2323                 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2324                 strbuf_addstr(msg, "\nrename from ");
2325                 quote_c_style(name, msg, NULL, 0);
2326                 strbuf_addstr(msg, "\nrename to ");
2327                 quote_c_style(other, msg, NULL, 0);
2328                 strbuf_addch(msg, '\n');
2329                 break;
2330         case DIFF_STATUS_MODIFIED:
2331                 if (p->score) {
2332                         strbuf_addf(msg, "dissimilarity index %d%%\n",
2333                                     similarity_index(p));
2334                         break;
2335                 }
2336                 /* fallthru */
2337         default:
2338                 /* nothing */
2339                 ;
2340         }
2341         if (one && two && hashcmp(one->sha1, two->sha1)) {
2342                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2343
2344                 if (DIFF_OPT_TST(o, BINARY)) {
2345                         mmfile_t mf;
2346                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2347                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2348                                 abbrev = 40;
2349                 }
2350                 strbuf_addf(msg, "index %.*s..%.*s",
2351                             abbrev, sha1_to_hex(one->sha1),
2352                             abbrev, sha1_to_hex(two->sha1));
2353                 if (one->mode == two->mode)
2354                         strbuf_addf(msg, " %06o", one->mode);
2355                 strbuf_addch(msg, '\n');
2356         }
2357         if (msg->len)
2358                 strbuf_setlen(msg, msg->len - 1);
2359 }
2360
2361 static void run_diff_cmd(const char *pgm,
2362                          const char *name,
2363                          const char *other,
2364                          const char *attr_path,
2365                          struct diff_filespec *one,
2366                          struct diff_filespec *two,
2367                          struct strbuf *msg,
2368                          struct diff_options *o,
2369                          struct diff_filepair *p)
2370 {
2371         const char *xfrm_msg = NULL;
2372         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2373
2374         if (msg) {
2375                 fill_metainfo(msg, name, other, one, two, o, p);
2376                 xfrm_msg = msg->len ? msg->buf : NULL;
2377         }
2378
2379         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2380                 pgm = NULL;
2381         else {
2382                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2383                 if (drv && drv->external)
2384                         pgm = drv->external;
2385         }
2386
2387         if (pgm) {
2388                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2389                                   complete_rewrite);
2390                 return;
2391         }
2392         if (one && two)
2393                 builtin_diff(name, other ? other : name,
2394                              one, two, xfrm_msg, o, complete_rewrite);
2395         else
2396                 fprintf(o->file, "* Unmerged path %s\n", name);
2397 }
2398
2399 static void diff_fill_sha1_info(struct diff_filespec *one)
2400 {
2401         if (DIFF_FILE_VALID(one)) {
2402                 if (!one->sha1_valid) {
2403                         struct stat st;
2404                         if (!strcmp(one->path, "-")) {
2405                                 hashcpy(one->sha1, null_sha1);
2406                                 return;
2407                         }
2408                         if (lstat(one->path, &st) < 0)
2409                                 die_errno("stat '%s'", one->path);
2410                         if (index_path(one->sha1, one->path, &st, 0))
2411                                 die("cannot hash %s", one->path);
2412                 }
2413         }
2414         else
2415                 hashclr(one->sha1);
2416 }
2417
2418 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2419 {
2420         /* Strip the prefix but do not molest /dev/null and absolute paths */
2421         if (*namep && **namep != '/')
2422                 *namep += prefix_length;
2423         if (*otherp && **otherp != '/')
2424                 *otherp += prefix_length;
2425 }
2426
2427 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2428 {
2429         const char *pgm = external_diff();
2430         struct strbuf msg;
2431         struct diff_filespec *one = p->one;
2432         struct diff_filespec *two = p->two;
2433         const char *name;
2434         const char *other;
2435         const char *attr_path;
2436
2437         name  = p->one->path;
2438         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2439         attr_path = name;
2440         if (o->prefix_length)
2441                 strip_prefix(o->prefix_length, &name, &other);
2442
2443         if (DIFF_PAIR_UNMERGED(p)) {
2444                 run_diff_cmd(pgm, name, NULL, attr_path,
2445                              NULL, NULL, NULL, o, p);
2446                 return;
2447         }
2448
2449         diff_fill_sha1_info(one);
2450         diff_fill_sha1_info(two);
2451
2452         if (!pgm &&
2453             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2454             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2455                 /*
2456                  * a filepair that changes between file and symlink
2457                  * needs to be split into deletion and creation.
2458                  */
2459                 struct diff_filespec *null = alloc_filespec(two->path);
2460                 run_diff_cmd(NULL, name, other, attr_path,
2461                              one, null, &msg, o, p);
2462                 free(null);
2463                 strbuf_release(&msg);
2464
2465                 null = alloc_filespec(one->path);
2466                 run_diff_cmd(NULL, name, other, attr_path,
2467                              null, two, &msg, o, p);
2468                 free(null);
2469         }
2470         else
2471                 run_diff_cmd(pgm, name, other, attr_path,
2472                              one, two, &msg, o, p);
2473
2474         strbuf_release(&msg);
2475 }
2476
2477 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2478                          struct diffstat_t *diffstat)
2479 {
2480         const char *name;
2481         const char *other;
2482         int complete_rewrite = 0;
2483
2484         if (DIFF_PAIR_UNMERGED(p)) {
2485                 /* unmerged */
2486                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2487                 return;
2488         }
2489
2490         name = p->one->path;
2491         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2492
2493         if (o->prefix_length)
2494                 strip_prefix(o->prefix_length, &name, &other);
2495
2496         diff_fill_sha1_info(p->one);
2497         diff_fill_sha1_info(p->two);
2498
2499         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2500                 complete_rewrite = 1;
2501         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2502 }
2503
2504 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2505 {
2506         const char *name;
2507         const char *other;
2508         const char *attr_path;
2509
2510         if (DIFF_PAIR_UNMERGED(p)) {
2511                 /* unmerged */
2512                 return;
2513         }
2514
2515         name = p->one->path;
2516         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2517         attr_path = other ? other : name;
2518
2519         if (o->prefix_length)
2520                 strip_prefix(o->prefix_length, &name, &other);
2521
2522         diff_fill_sha1_info(p->one);
2523         diff_fill_sha1_info(p->two);
2524
2525         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2526 }
2527
2528 void diff_setup(struct diff_options *options)
2529 {
2530         memset(options, 0, sizeof(*options));
2531
2532         options->file = stdout;
2533
2534         options->line_termination = '\n';
2535         options->break_opt = -1;
2536         options->rename_limit = -1;
2537         options->dirstat_percent = 3;
2538         options->context = 3;
2539
2540         options->change = diff_change;
2541         options->add_remove = diff_addremove;
2542         if (diff_use_color_default > 0)
2543                 DIFF_OPT_SET(options, COLOR_DIFF);
2544         options->detect_rename = diff_detect_rename_default;
2545
2546         if (diff_no_prefix) {
2547                 options->a_prefix = options->b_prefix = "";
2548         } else if (!diff_mnemonic_prefix) {
2549                 options->a_prefix = "a/";
2550                 options->b_prefix = "b/";
2551         }
2552 }
2553
2554 int diff_setup_done(struct diff_options *options)
2555 {
2556         int count = 0;
2557
2558         if (options->output_format & DIFF_FORMAT_NAME)
2559                 count++;
2560         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2561                 count++;
2562         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2563                 count++;
2564         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2565                 count++;
2566         if (count > 1)
2567                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2568
2569         /*
2570          * Most of the time we can say "there are changes"
2571          * only by checking if there are changed paths, but
2572          * --ignore-whitespace* options force us to look
2573          * inside contents.
2574          */
2575
2576         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
2577             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
2578             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
2579                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
2580         else
2581                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
2582
2583         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2584                 options->detect_rename = DIFF_DETECT_COPY;
2585
2586         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2587                 options->prefix = NULL;
2588         if (options->prefix)
2589                 options->prefix_length = strlen(options->prefix);
2590         else
2591                 options->prefix_length = 0;
2592
2593         if (options->output_format & (DIFF_FORMAT_NAME |
2594                                       DIFF_FORMAT_NAME_STATUS |
2595                                       DIFF_FORMAT_CHECKDIFF |
2596                                       DIFF_FORMAT_NO_OUTPUT))
2597                 options->output_format &= ~(DIFF_FORMAT_RAW |
2598                                             DIFF_FORMAT_NUMSTAT |
2599                                             DIFF_FORMAT_DIFFSTAT |
2600                                             DIFF_FORMAT_SHORTSTAT |
2601                                             DIFF_FORMAT_DIRSTAT |
2602                                             DIFF_FORMAT_SUMMARY |
2603                                             DIFF_FORMAT_PATCH);
2604
2605         /*
2606          * These cases always need recursive; we do not drop caller-supplied
2607          * recursive bits for other formats here.
2608          */
2609         if (options->output_format & (DIFF_FORMAT_PATCH |
2610                                       DIFF_FORMAT_NUMSTAT |
2611                                       DIFF_FORMAT_DIFFSTAT |
2612                                       DIFF_FORMAT_SHORTSTAT |
2613                                       DIFF_FORMAT_DIRSTAT |
2614                                       DIFF_FORMAT_SUMMARY |
2615                                       DIFF_FORMAT_CHECKDIFF))
2616                 DIFF_OPT_SET(options, RECURSIVE);
2617         /*
2618          * Also pickaxe would not work very well if you do not say recursive
2619          */
2620         if (options->pickaxe)
2621                 DIFF_OPT_SET(options, RECURSIVE);
2622         /*
2623          * When patches are generated, submodules diffed against the work tree
2624          * must be checked for dirtiness too so it can be shown in the output
2625          */
2626         if (options->output_format & DIFF_FORMAT_PATCH)
2627                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
2628
2629         if (options->detect_rename && options->rename_limit < 0)
2630                 options->rename_limit = diff_rename_limit_default;
2631         if (options->setup & DIFF_SETUP_USE_CACHE) {
2632                 if (!active_cache)
2633                         /* read-cache does not die even when it fails
2634                          * so it is safe for us to do this here.  Also
2635                          * it does not smudge active_cache or active_nr
2636                          * when it fails, so we do not have to worry about
2637                          * cleaning it up ourselves either.
2638                          */
2639                         read_cache();
2640         }
2641         if (options->abbrev <= 0 || 40 < options->abbrev)
2642                 options->abbrev = 40; /* full */
2643
2644         /*
2645          * It does not make sense to show the first hit we happened
2646          * to have found.  It does not make sense not to return with
2647          * exit code in such a case either.
2648          */
2649         if (DIFF_OPT_TST(options, QUICK)) {
2650                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2651                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2652         }
2653
2654         return 0;
2655 }
2656
2657 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2658 {
2659         char c, *eq;
2660         int len;
2661
2662         if (*arg != '-')
2663                 return 0;
2664         c = *++arg;
2665         if (!c)
2666                 return 0;
2667         if (c == arg_short) {
2668                 c = *++arg;
2669                 if (!c)
2670                         return 1;
2671                 if (val && isdigit(c)) {
2672                         char *end;
2673                         int n = strtoul(arg, &end, 10);
2674                         if (*end)
2675                                 return 0;
2676                         *val = n;
2677                         return 1;
2678                 }
2679                 return 0;
2680         }
2681         if (c != '-')
2682                 return 0;
2683         arg++;
2684         eq = strchr(arg, '=');
2685         if (eq)
2686                 len = eq - arg;
2687         else
2688                 len = strlen(arg);
2689         if (!len || strncmp(arg, arg_long, len))
2690                 return 0;
2691         if (eq) {
2692                 int n;
2693                 char *end;
2694                 if (!isdigit(*++eq))
2695                         return 0;
2696                 n = strtoul(eq, &end, 10);
2697                 if (*end)
2698                         return 0;
2699                 *val = n;
2700         }
2701         return 1;
2702 }
2703
2704 static int diff_scoreopt_parse(const char *opt);
2705
2706 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2707 {
2708         const char *arg = av[0];
2709
2710         /* Output format options */
2711         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2712                 options->output_format |= DIFF_FORMAT_PATCH;
2713         else if (opt_arg(arg, 'U', "unified", &options->context))
2714                 options->output_format |= DIFF_FORMAT_PATCH;
2715         else if (!strcmp(arg, "--raw"))
2716                 options->output_format |= DIFF_FORMAT_RAW;
2717         else if (!strcmp(arg, "--patch-with-raw"))
2718                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2719         else if (!strcmp(arg, "--numstat"))
2720                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2721         else if (!strcmp(arg, "--shortstat"))
2722                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2723         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2724                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2725         else if (!strcmp(arg, "--cumulative")) {
2726                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2727                 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2728         } else if (opt_arg(arg, 0, "dirstat-by-file",
2729                            &options->dirstat_percent)) {
2730                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2731                 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
2732         }
2733         else if (!strcmp(arg, "--check"))
2734                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2735         else if (!strcmp(arg, "--summary"))
2736                 options->output_format |= DIFF_FORMAT_SUMMARY;
2737         else if (!strcmp(arg, "--patch-with-stat"))
2738                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2739         else if (!strcmp(arg, "--name-only"))
2740                 options->output_format |= DIFF_FORMAT_NAME;
2741         else if (!strcmp(arg, "--name-status"))
2742                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2743         else if (!strcmp(arg, "-s"))
2744                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2745         else if (!prefixcmp(arg, "--stat")) {
2746                 char *end;
2747                 int width = options->stat_width;
2748                 int name_width = options->stat_name_width;
2749                 arg += 6;
2750                 end = (char *)arg;
2751
2752                 switch (*arg) {
2753                 case '-':
2754                         if (!prefixcmp(arg, "-width="))
2755                                 width = strtoul(arg + 7, &end, 10);
2756                         else if (!prefixcmp(arg, "-name-width="))
2757                                 name_width = strtoul(arg + 12, &end, 10);
2758                         break;
2759                 case '=':
2760                         width = strtoul(arg+1, &end, 10);
2761                         if (*end == ',')
2762                                 name_width = strtoul(end+1, &end, 10);
2763                 }
2764
2765                 /* Important! This checks all the error cases! */
2766                 if (*end)
2767                         return 0;
2768                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2769                 options->stat_name_width = name_width;
2770                 options->stat_width = width;
2771         }
2772
2773         /* renames options */
2774         else if (!prefixcmp(arg, "-B")) {
2775                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2776                         return -1;
2777         }
2778         else if (!prefixcmp(arg, "-M")) {
2779                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2780                         return -1;
2781                 options->detect_rename = DIFF_DETECT_RENAME;
2782         }
2783         else if (!prefixcmp(arg, "-C")) {
2784                 if (options->detect_rename == DIFF_DETECT_COPY)
2785                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2786                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2787                         return -1;
2788                 options->detect_rename = DIFF_DETECT_COPY;
2789         }
2790         else if (!strcmp(arg, "--no-renames"))
2791                 options->detect_rename = 0;
2792         else if (!strcmp(arg, "--relative"))
2793                 DIFF_OPT_SET(options, RELATIVE_NAME);
2794         else if (!prefixcmp(arg, "--relative=")) {
2795                 DIFF_OPT_SET(options, RELATIVE_NAME);
2796                 options->prefix = arg + 11;
2797         }
2798
2799         /* xdiff options */
2800         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2801                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
2802         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2803                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
2804         else if (!strcmp(arg, "--ignore-space-at-eol"))
2805                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
2806         else if (!strcmp(arg, "--patience"))
2807                 DIFF_XDL_SET(options, PATIENCE_DIFF);
2808
2809         /* flags options */
2810         else if (!strcmp(arg, "--binary")) {
2811                 options->output_format |= DIFF_FORMAT_PATCH;
2812                 DIFF_OPT_SET(options, BINARY);
2813         }
2814         else if (!strcmp(arg, "--full-index"))
2815                 DIFF_OPT_SET(options, FULL_INDEX);
2816         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2817                 DIFF_OPT_SET(options, TEXT);
2818         else if (!strcmp(arg, "-R"))
2819                 DIFF_OPT_SET(options, REVERSE_DIFF);
2820         else if (!strcmp(arg, "--find-copies-harder"))
2821                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2822         else if (!strcmp(arg, "--follow"))
2823                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2824         else if (!strcmp(arg, "--color"))
2825                 DIFF_OPT_SET(options, COLOR_DIFF);
2826         else if (!prefixcmp(arg, "--color=")) {
2827                 int value = git_config_colorbool(NULL, arg+8, -1);
2828                 if (value == 0)
2829                         DIFF_OPT_CLR(options, COLOR_DIFF);
2830                 else if (value > 0)
2831                         DIFF_OPT_SET(options, COLOR_DIFF);
2832                 else
2833                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
2834         }
2835         else if (!strcmp(arg, "--no-color"))
2836                 DIFF_OPT_CLR(options, COLOR_DIFF);
2837         else if (!strcmp(arg, "--color-words")) {
2838                 DIFF_OPT_SET(options, COLOR_DIFF);
2839                 DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2840         }
2841         else if (!prefixcmp(arg, "--color-words=")) {
2842                 DIFF_OPT_SET(options, COLOR_DIFF);
2843                 DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2844                 options->word_regex = arg + 14;
2845         }
2846         else if (!strcmp(arg, "--exit-code"))
2847                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2848         else if (!strcmp(arg, "--quiet"))
2849                 DIFF_OPT_SET(options, QUICK);
2850         else if (!strcmp(arg, "--ext-diff"))
2851                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2852         else if (!strcmp(arg, "--no-ext-diff"))
2853                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2854         else if (!strcmp(arg, "--textconv"))
2855                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
2856         else if (!strcmp(arg, "--no-textconv"))
2857                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
2858         else if (!strcmp(arg, "--ignore-submodules"))
2859                 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2860         else if (!strcmp(arg, "--submodule"))
2861                 DIFF_OPT_SET(options, SUBMODULE_LOG);
2862         else if (!prefixcmp(arg, "--submodule=")) {
2863                 if (!strcmp(arg + 12, "log"))
2864                         DIFF_OPT_SET(options, SUBMODULE_LOG);
2865         }
2866
2867         /* misc options */
2868         else if (!strcmp(arg, "-z"))
2869                 options->line_termination = 0;
2870         else if (!prefixcmp(arg, "-l"))
2871                 options->rename_limit = strtoul(arg+2, NULL, 10);
2872         else if (!prefixcmp(arg, "-S"))
2873                 options->pickaxe = arg + 2;
2874         else if (!strcmp(arg, "--pickaxe-all"))
2875                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2876         else if (!strcmp(arg, "--pickaxe-regex"))
2877                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2878         else if (!prefixcmp(arg, "-O"))
2879                 options->orderfile = arg + 2;
2880         else if (!prefixcmp(arg, "--diff-filter="))
2881                 options->filter = arg + 14;
2882         else if (!strcmp(arg, "--abbrev"))
2883                 options->abbrev = DEFAULT_ABBREV;
2884         else if (!prefixcmp(arg, "--abbrev=")) {
2885                 options->abbrev = strtoul(arg + 9, NULL, 10);
2886                 if (options->abbrev < MINIMUM_ABBREV)
2887                         options->abbrev = MINIMUM_ABBREV;
2888                 else if (40 < options->abbrev)
2889                         options->abbrev = 40;
2890         }
2891         else if (!prefixcmp(arg, "--src-prefix="))
2892                 options->a_prefix = arg + 13;
2893         else if (!prefixcmp(arg, "--dst-prefix="))
2894                 options->b_prefix = arg + 13;
2895         else if (!strcmp(arg, "--no-prefix"))
2896                 options->a_prefix = options->b_prefix = "";
2897         else if (opt_arg(arg, '\0', "inter-hunk-context",
2898                          &options->interhunkcontext))
2899                 ;
2900         else if (!prefixcmp(arg, "--output=")) {
2901                 options->file = fopen(arg + strlen("--output="), "w");
2902                 if (!options->file)
2903                         die_errno("Could not open '%s'", arg + strlen("--output="));
2904                 options->close_file = 1;
2905         } else
2906                 return 0;
2907         return 1;
2908 }
2909
2910 static int parse_num(const char **cp_p)
2911 {
2912         unsigned long num, scale;
2913         int ch, dot;
2914         const char *cp = *cp_p;
2915
2916         num = 0;
2917         scale = 1;
2918         dot = 0;
2919         for (;;) {
2920                 ch = *cp;
2921                 if ( !dot && ch == '.' ) {
2922                         scale = 1;
2923                         dot = 1;
2924                 } else if ( ch == '%' ) {
2925                         scale = dot ? scale*100 : 100;
2926                         cp++;   /* % is always at the end */
2927                         break;
2928                 } else if ( ch >= '0' && ch <= '9' ) {
2929                         if ( scale < 100000 ) {
2930                                 scale *= 10;
2931                                 num = (num*10) + (ch-'0');
2932                         }
2933                 } else {
2934                         break;
2935                 }
2936                 cp++;
2937         }
2938         *cp_p = cp;
2939
2940         /* user says num divided by scale and we say internally that
2941          * is MAX_SCORE * num / scale.
2942          */
2943         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2944 }
2945
2946 static int diff_scoreopt_parse(const char *opt)
2947 {
2948         int opt1, opt2, cmd;
2949
2950         if (*opt++ != '-')
2951                 return -1;
2952         cmd = *opt++;
2953         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2954                 return -1; /* that is not a -M, -C nor -B option */
2955
2956         opt1 = parse_num(&opt);
2957         if (cmd != 'B')
2958                 opt2 = 0;
2959         else {
2960                 if (*opt == 0)
2961                         opt2 = 0;
2962                 else if (*opt != '/')
2963                         return -1; /* we expect -B80/99 or -B80 */
2964                 else {
2965                         opt++;
2966                         opt2 = parse_num(&opt);
2967                 }
2968         }
2969         if (*opt != 0)
2970                 return -1;
2971         return opt1 | (opt2 << 16);
2972 }
2973
2974 struct diff_queue_struct diff_queued_diff;
2975
2976 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2977 {
2978         if (queue->alloc <= queue->nr) {
2979                 queue->alloc = alloc_nr(queue->alloc);
2980                 queue->queue = xrealloc(queue->queue,
2981                                         sizeof(dp) * queue->alloc);
2982         }
2983         queue->queue[queue->nr++] = dp;
2984 }
2985
2986 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2987                                  struct diff_filespec *one,
2988                                  struct diff_filespec *two)
2989 {
2990         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2991         dp->one = one;
2992         dp->two = two;
2993         if (queue)
2994                 diff_q(queue, dp);
2995         return dp;
2996 }
2997
2998 void diff_free_filepair(struct diff_filepair *p)
2999 {
3000         free_filespec(p->one);
3001         free_filespec(p->two);
3002         free(p);
3003 }
3004
3005 /* This is different from find_unique_abbrev() in that
3006  * it stuffs the result with dots for alignment.
3007  */
3008 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3009 {
3010         int abblen;
3011         const char *abbrev;
3012         if (len == 40)
3013                 return sha1_to_hex(sha1);
3014
3015         abbrev = find_unique_abbrev(sha1, len);
3016         abblen = strlen(abbrev);
3017         if (abblen < 37) {
3018                 static char hex[41];
3019                 if (len < abblen && abblen <= len + 2)
3020                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3021                 else
3022                         sprintf(hex, "%s...", abbrev);
3023                 return hex;
3024         }
3025         return sha1_to_hex(sha1);
3026 }
3027
3028 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3029 {
3030         int line_termination = opt->line_termination;
3031         int inter_name_termination = line_termination ? '\t' : '\0';
3032
3033         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3034                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3035                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
3036                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3037         }
3038         if (p->score) {
3039                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3040                         inter_name_termination);
3041         } else {
3042                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
3043         }
3044
3045         if (p->status == DIFF_STATUS_COPIED ||
3046             p->status == DIFF_STATUS_RENAMED) {
3047                 const char *name_a, *name_b;
3048                 name_a = p->one->path;
3049                 name_b = p->two->path;
3050                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3051                 write_name_quoted(name_a, opt->file, inter_name_termination);
3052                 write_name_quoted(name_b, opt->file, line_termination);
3053         } else {
3054                 const char *name_a, *name_b;
3055                 name_a = p->one->mode ? p->one->path : p->two->path;
3056                 name_b = NULL;
3057                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3058                 write_name_quoted(name_a, opt->file, line_termination);
3059         }
3060 }
3061
3062 int diff_unmodified_pair(struct diff_filepair *p)
3063 {
3064         /* This function is written stricter than necessary to support
3065          * the currently implemented transformers, but the idea is to
3066          * let transformers to produce diff_filepairs any way they want,
3067          * and filter and clean them up here before producing the output.
3068          */
3069         struct diff_filespec *one = p->one, *two = p->two;
3070
3071         if (DIFF_PAIR_UNMERGED(p))
3072                 return 0; /* unmerged is interesting */
3073
3074         /* deletion, addition, mode or type change
3075          * and rename are all interesting.
3076          */
3077         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
3078             DIFF_PAIR_MODE_CHANGED(p) ||
3079             strcmp(one->path, two->path))
3080                 return 0;
3081
3082         /* both are valid and point at the same path.  that is, we are
3083          * dealing with a change.
3084          */
3085         if (one->sha1_valid && two->sha1_valid &&
3086             !hashcmp(one->sha1, two->sha1) &&
3087             !one->dirty_submodule && !two->dirty_submodule)
3088                 return 1; /* no change */
3089         if (!one->sha1_valid && !two->sha1_valid)
3090                 return 1; /* both look at the same file on the filesystem. */
3091         return 0;
3092 }
3093
3094 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
3095 {
3096         if (diff_unmodified_pair(p))
3097                 return;
3098
3099         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3100             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3101                 return; /* no tree diffs in patch format */
3102
3103         run_diff(p, o);
3104 }
3105
3106 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
3107                             struct diffstat_t *diffstat)
3108 {
3109         if (diff_unmodified_pair(p))
3110                 return;
3111
3112         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3113             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3114                 return; /* no tree diffs in patch format */
3115
3116         run_diffstat(p, o, diffstat);
3117 }
3118
3119 static void diff_flush_checkdiff(struct diff_filepair *p,
3120                 struct diff_options *o)
3121 {
3122         if (diff_unmodified_pair(p))
3123                 return;
3124
3125         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3126             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3127                 return; /* no tree diffs in patch format */
3128
3129         run_checkdiff(p, o);
3130 }
3131
3132 int diff_queue_is_empty(void)
3133 {
3134         struct diff_queue_struct *q = &diff_queued_diff;
3135         int i;
3136         for (i = 0; i < q->nr; i++)
3137                 if (!diff_unmodified_pair(q->queue[i]))
3138                         return 0;
3139         return 1;
3140 }
3141
3142 #if DIFF_DEBUG
3143 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
3144 {
3145         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
3146                 x, one ? one : "",
3147                 s->path,
3148                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
3149                 s->mode,
3150                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
3151         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
3152                 x, one ? one : "",
3153                 s->size, s->xfrm_flags);
3154 }
3155
3156 void diff_debug_filepair(const struct diff_filepair *p, int i)
3157 {
3158         diff_debug_filespec(p->one, i, "one");
3159         diff_debug_filespec(p->two, i, "two");
3160         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
3161                 p->score, p->status ? p->status : '?',
3162                 p->one->rename_used, p->broken_pair);
3163 }
3164
3165 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
3166 {
3167         int i;
3168         if (msg)
3169                 fprintf(stderr, "%s\n", msg);
3170         fprintf(stderr, "q->nr = %d\n", q->nr);
3171         for (i = 0; i < q->nr; i++) {
3172                 struct diff_filepair *p = q->queue[i];
3173                 diff_debug_filepair(p, i);
3174         }
3175 }
3176 #endif
3177
3178 static void diff_resolve_rename_copy(void)
3179 {
3180         int i;
3181         struct diff_filepair *p;
3182         struct diff_queue_struct *q = &diff_queued_diff;
3183
3184         diff_debug_queue("resolve-rename-copy", q);
3185
3186         for (i = 0; i < q->nr; i++) {
3187                 p = q->queue[i];
3188                 p->status = 0; /* undecided */
3189                 if (DIFF_PAIR_UNMERGED(p))
3190                         p->status = DIFF_STATUS_UNMERGED;
3191                 else if (!DIFF_FILE_VALID(p->one))
3192                         p->status = DIFF_STATUS_ADDED;
3193                 else if (!DIFF_FILE_VALID(p->two))
3194                         p->status = DIFF_STATUS_DELETED;
3195                 else if (DIFF_PAIR_TYPE_CHANGED(p))
3196                         p->status = DIFF_STATUS_TYPE_CHANGED;
3197
3198                 /* from this point on, we are dealing with a pair
3199                  * whose both sides are valid and of the same type, i.e.
3200                  * either in-place edit or rename/copy edit.
3201                  */
3202                 else if (DIFF_PAIR_RENAME(p)) {
3203                         /*
3204                          * A rename might have re-connected a broken
3205                          * pair up, causing the pathnames to be the
3206                          * same again. If so, that's not a rename at
3207                          * all, just a modification..
3208                          *
3209                          * Otherwise, see if this source was used for
3210                          * multiple renames, in which case we decrement
3211                          * the count, and call it a copy.
3212                          */
3213                         if (!strcmp(p->one->path, p->two->path))
3214                                 p->status = DIFF_STATUS_MODIFIED;
3215                         else if (--p->one->rename_used > 0)
3216                                 p->status = DIFF_STATUS_COPIED;
3217                         else
3218                                 p->status = DIFF_STATUS_RENAMED;
3219                 }
3220                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
3221                          p->one->mode != p->two->mode ||
3222                          p->one->dirty_submodule ||
3223                          p->two->dirty_submodule ||
3224                          is_null_sha1(p->one->sha1))
3225                         p->status = DIFF_STATUS_MODIFIED;
3226                 else {
3227                         /* This is a "no-change" entry and should not
3228                          * happen anymore, but prepare for broken callers.
3229                          */
3230                         error("feeding unmodified %s to diffcore",
3231                               p->one->path);
3232                         p->status = DIFF_STATUS_UNKNOWN;
3233                 }
3234         }
3235         diff_debug_queue("resolve-rename-copy done", q);
3236 }
3237
3238 static int check_pair_status(struct diff_filepair *p)
3239 {
3240         switch (p->status) {
3241         case DIFF_STATUS_UNKNOWN:
3242                 return 0;
3243         case 0:
3244                 die("internal error in diff-resolve-rename-copy");
3245         default:
3246                 return 1;
3247         }
3248 }
3249
3250 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3251 {
3252         int fmt = opt->output_format;
3253
3254         if (fmt & DIFF_FORMAT_CHECKDIFF)
3255                 diff_flush_checkdiff(p, opt);
3256         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3257                 diff_flush_raw(p, opt);
3258         else if (fmt & DIFF_FORMAT_NAME) {
3259                 const char *name_a, *name_b;
3260                 name_a = p->two->path;
3261                 name_b = NULL;
3262                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3263                 write_name_quoted(name_a, opt->file, opt->line_termination);
3264         }
3265 }
3266
3267 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3268 {
3269         if (fs->mode)
3270                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3271         else
3272                 fprintf(file, " %s ", newdelete);
3273         write_name_quoted(fs->path, file, '\n');
3274 }
3275
3276
3277 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
3278 {
3279         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3280                 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
3281                         show_name ? ' ' : '\n');
3282                 if (show_name) {
3283                         write_name_quoted(p->two->path, file, '\n');
3284                 }
3285         }
3286 }
3287
3288 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3289 {
3290         char *names = pprint_rename(p->one->path, p->two->path);
3291
3292         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3293         free(names);
3294         show_mode_change(file, p, 0);
3295 }
3296
3297 static void diff_summary(FILE *file, struct diff_filepair *p)
3298 {
3299         switch(p->status) {
3300         case DIFF_STATUS_DELETED:
3301                 show_file_mode_name(file, "delete", p->one);
3302                 break;
3303         case DIFF_STATUS_ADDED:
3304                 show_file_mode_name(file, "create", p->two);
3305                 break;
3306         case DIFF_STATUS_COPIED:
3307                 show_rename_copy(file, "copy", p);
3308                 break;
3309         case DIFF_STATUS_RENAMED:
3310                 show_rename_copy(file, "rename", p);
3311                 break;
3312         default:
3313                 if (p->score) {
3314                         fputs(" rewrite ", file);
3315                         write_name_quoted(p->two->path, file, ' ');
3316                         fprintf(file, "(%d%%)\n", similarity_index(p));
3317                 }
3318                 show_mode_change(file, p, !p->score);
3319                 break;
3320         }
3321 }
3322
3323 struct patch_id_t {
3324         git_SHA_CTX *ctx;
3325         int patchlen;
3326 };
3327
3328 static int remove_space(char *line, int len)
3329 {
3330         int i;
3331         char *dst = line;
3332         unsigned char c;
3333
3334         for (i = 0; i < len; i++)
3335                 if (!isspace((c = line[i])))
3336                         *dst++ = c;
3337
3338         return dst - line;
3339 }
3340
3341 static void patch_id_consume(void *priv, char *line, unsigned long len)
3342 {
3343         struct patch_id_t *data = priv;
3344         int new_len;
3345
3346         /* Ignore line numbers when computing the SHA1 of the patch */
3347         if (!prefixcmp(line, "@@ -"))
3348                 return;
3349
3350         new_len = remove_space(line, len);
3351
3352         git_SHA1_Update(data->ctx, line, new_len);
3353         data->patchlen += new_len;
3354 }
3355
3356 /* returns 0 upon success, and writes result into sha1 */
3357 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3358 {
3359         struct diff_queue_struct *q = &diff_queued_diff;
3360         int i;
3361         git_SHA_CTX ctx;
3362         struct patch_id_t data;
3363         char buffer[PATH_MAX * 4 + 20];
3364
3365         git_SHA1_Init(&ctx);
3366         memset(&data, 0, sizeof(struct patch_id_t));
3367         data.ctx = &ctx;
3368
3369         for (i = 0; i < q->nr; i++) {
3370                 xpparam_t xpp;
3371                 xdemitconf_t xecfg;
3372                 mmfile_t mf1, mf2;
3373                 struct diff_filepair *p = q->queue[i];
3374                 int len1, len2;
3375
3376                 memset(&xpp, 0, sizeof(xpp));
3377                 memset(&xecfg, 0, sizeof(xecfg));
3378                 if (p->status == 0)
3379                         return error("internal diff status error");
3380                 if (p->status == DIFF_STATUS_UNKNOWN)
3381                         continue;
3382                 if (diff_unmodified_pair(p))
3383                         continue;
3384                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3385                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3386                         continue;
3387                 if (DIFF_PAIR_UNMERGED(p))
3388                         continue;
3389
3390                 diff_fill_sha1_info(p->one);
3391                 diff_fill_sha1_info(p->two);
3392                 if (fill_mmfile(&mf1, p->one) < 0 ||
3393                                 fill_mmfile(&mf2, p->two) < 0)
3394                         return error("unable to read files to diff");
3395
3396                 len1 = remove_space(p->one->path, strlen(p->one->path));
3397                 len2 = remove_space(p->two->path, strlen(p->two->path));
3398                 if (p->one->mode == 0)
3399                         len1 = snprintf(buffer, sizeof(buffer),
3400                                         "diff--gita/%.*sb/%.*s"
3401                                         "newfilemode%06o"
3402                                         "---/dev/null"
3403                                         "+++b/%.*s",
3404                                         len1, p->one->path,
3405                                         len2, p->two->path,
3406                                         p->two->mode,
3407                                         len2, p->two->path);
3408                 else if (p->two->mode == 0)
3409                         len1 = snprintf(buffer, sizeof(buffer),
3410                                         "diff--gita/%.*sb/%.*s"
3411                                         "deletedfilemode%06o"
3412                                         "---a/%.*s"
3413                                         "+++/dev/null",
3414                                         len1, p->one->path,
3415                                         len2, p->two->path,
3416                                         p->one->mode,
3417                                         len1, p->one->path);
3418                 else
3419                         len1 = snprintf(buffer, sizeof(buffer),
3420                                         "diff--gita/%.*sb/%.*s"
3421                                         "---a/%.*s"
3422                                         "+++b/%.*s",
3423                                         len1, p->one->path,
3424                                         len2, p->two->path,
3425                                         len1, p->one->path,
3426                                         len2, p->two->path);
3427                 git_SHA1_Update(&ctx, buffer, len1);
3428
3429                 xpp.flags = XDF_NEED_MINIMAL;
3430                 xecfg.ctxlen = 3;
3431                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3432                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3433                               &xpp, &xecfg);
3434         }
3435
3436         git_SHA1_Final(sha1, &ctx);
3437         return 0;
3438 }
3439
3440 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3441 {
3442         struct diff_queue_struct *q = &diff_queued_diff;
3443         int i;
3444         int result = diff_get_patch_id(options, sha1);
3445
3446         for (i = 0; i < q->nr; i++)
3447                 diff_free_filepair(q->queue[i]);
3448
3449         free(q->queue);
3450         q->queue = NULL;
3451         q->nr = q->alloc = 0;
3452
3453         return result;
3454 }
3455
3456 static int is_summary_empty(const struct diff_queue_struct *q)
3457 {
3458         int i;
3459
3460         for (i = 0; i < q->nr; i++) {
3461                 const struct diff_filepair *p = q->queue[i];
3462
3463                 switch (p->status) {
3464                 case DIFF_STATUS_DELETED:
3465                 case DIFF_STATUS_ADDED:
3466                 case DIFF_STATUS_COPIED:
3467                 case DIFF_STATUS_RENAMED:
3468                         return 0;
3469                 default:
3470                         if (p->score)
3471                                 return 0;
3472                         if (p->one->mode && p->two->mode &&
3473                             p->one->mode != p->two->mode)
3474                                 return 0;
3475                         break;
3476                 }
3477         }
3478         return 1;
3479 }
3480
3481 void diff_flush(struct diff_options *options)
3482 {
3483         struct diff_queue_struct *q = &diff_queued_diff;
3484         int i, output_format = options->output_format;
3485         int separator = 0;
3486
3487         /*
3488          * Order: raw, stat, summary, patch
3489          * or:    name/name-status/checkdiff (other bits clear)
3490          */
3491         if (!q->nr)
3492                 goto free_queue;
3493
3494         if (output_format & (DIFF_FORMAT_RAW |
3495                              DIFF_FORMAT_NAME |
3496                              DIFF_FORMAT_NAME_STATUS |
3497                              DIFF_FORMAT_CHECKDIFF)) {
3498                 for (i = 0; i < q->nr; i++) {
3499                         struct diff_filepair *p = q->queue[i];
3500                         if (check_pair_status(p))
3501                                 flush_one_pair(p, options);
3502                 }
3503                 separator++;
3504         }
3505
3506         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3507                 struct diffstat_t diffstat;
3508
3509                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3510                 for (i = 0; i < q->nr; i++) {
3511                         struct diff_filepair *p = q->queue[i];
3512                         if (check_pair_status(p))
3513                                 diff_flush_stat(p, options, &diffstat);
3514                 }
3515                 if (output_format & DIFF_FORMAT_NUMSTAT)
3516                         show_numstat(&diffstat, options);
3517                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3518                         show_stats(&diffstat, options);
3519                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3520                         show_shortstats(&diffstat, options);
3521                 free_diffstat_info(&diffstat);
3522                 separator++;
3523         }
3524         if (output_format & DIFF_FORMAT_DIRSTAT)
3525                 show_dirstat(options);
3526
3527         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3528                 for (i = 0; i < q->nr; i++)
3529                         diff_summary(options->file, q->queue[i]);
3530                 separator++;
3531         }
3532
3533         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
3534             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
3535             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
3536                 /*
3537                  * run diff_flush_patch for the exit status. setting
3538                  * options->file to /dev/null should be safe, becaue we
3539                  * aren't supposed to produce any output anyway.
3540                  */
3541                 if (options->close_file)
3542                         fclose(options->file);
3543                 options->file = fopen("/dev/null", "w");
3544                 if (!options->file)
3545                         die_errno("Could not open /dev/null");
3546                 options->close_file = 1;
3547                 for (i = 0; i < q->nr; i++) {
3548                         struct diff_filepair *p = q->queue[i];
3549                         if (check_pair_status(p))
3550                                 diff_flush_patch(p, options);
3551                         if (options->found_changes)
3552                                 break;
3553                 }
3554         }
3555
3556         if (output_format & DIFF_FORMAT_PATCH) {
3557                 if (separator) {
3558                         putc(options->line_termination, options->file);
3559                         if (options->stat_sep) {
3560                                 /* attach patch instead of inline */
3561                                 fputs(options->stat_sep, options->file);
3562                         }
3563                 }
3564
3565                 for (i = 0; i < q->nr; i++) {
3566                         struct diff_filepair *p = q->queue[i];
3567                         if (check_pair_status(p))
3568                                 diff_flush_patch(p, options);
3569                 }
3570         }
3571
3572         if (output_format & DIFF_FORMAT_CALLBACK)
3573                 options->format_callback(q, options, options->format_callback_data);
3574
3575         for (i = 0; i < q->nr; i++)
3576                 diff_free_filepair(q->queue[i]);
3577 free_queue:
3578         free(q->queue);
3579         q->queue = NULL;
3580         q->nr = q->alloc = 0;
3581         if (options->close_file)
3582                 fclose(options->file);
3583
3584         /*
3585          * Report the content-level differences with HAS_CHANGES;
3586          * diff_addremove/diff_change does not set the bit when
3587          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
3588          */
3589         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
3590                 if (options->found_changes)
3591                         DIFF_OPT_SET(options, HAS_CHANGES);
3592                 else
3593                         DIFF_OPT_CLR(options, HAS_CHANGES);
3594         }
3595 }
3596
3597 static void diffcore_apply_filter(const char *filter)
3598 {
3599         int i;
3600         struct diff_queue_struct *q = &diff_queued_diff;
3601         struct diff_queue_struct outq;
3602         outq.queue = NULL;
3603         outq.nr = outq.alloc = 0;
3604
3605         if (!filter)
3606                 return;
3607
3608         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3609                 int found;
3610                 for (i = found = 0; !found && i < q->nr; i++) {
3611                         struct diff_filepair *p = q->queue[i];
3612                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3613                              ((p->score &&
3614                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3615                               (!p->score &&
3616                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3617                             ((p->status != DIFF_STATUS_MODIFIED) &&
3618                              strchr(filter, p->status)))
3619                                 found++;
3620                 }
3621                 if (found)
3622                         return;
3623
3624                 /* otherwise we will clear the whole queue
3625                  * by copying the empty outq at the end of this
3626                  * function, but first clear the current entries
3627                  * in the queue.
3628                  */
3629                 for (i = 0; i < q->nr; i++)
3630                         diff_free_filepair(q->queue[i]);
3631         }
3632         else {
3633                 /* Only the matching ones */
3634                 for (i = 0; i < q->nr; i++) {
3635                         struct diff_filepair *p = q->queue[i];
3636
3637                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3638                              ((p->score &&
3639                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3640                               (!p->score &&
3641                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3642                             ((p->status != DIFF_STATUS_MODIFIED) &&
3643                              strchr(filter, p->status)))
3644                                 diff_q(&outq, p);
3645                         else
3646                                 diff_free_filepair(p);
3647                 }
3648         }
3649         free(q->queue);
3650         *q = outq;
3651 }
3652
3653 /* Check whether two filespecs with the same mode and size are identical */
3654 static int diff_filespec_is_identical(struct diff_filespec *one,
3655                                       struct diff_filespec *two)
3656 {
3657         if (S_ISGITLINK(one->mode))
3658                 return 0;
3659         if (diff_populate_filespec(one, 0))
3660                 return 0;
3661         if (diff_populate_filespec(two, 0))
3662                 return 0;
3663         return !memcmp(one->data, two->data, one->size);
3664 }
3665
3666 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3667 {
3668         int i;
3669         struct diff_queue_struct *q = &diff_queued_diff;
3670         struct diff_queue_struct outq;
3671         outq.queue = NULL;
3672         outq.nr = outq.alloc = 0;
3673
3674         for (i = 0; i < q->nr; i++) {
3675                 struct diff_filepair *p = q->queue[i];
3676
3677                 /*
3678                  * 1. Entries that come from stat info dirtiness
3679                  *    always have both sides (iow, not create/delete),
3680                  *    one side of the object name is unknown, with
3681                  *    the same mode and size.  Keep the ones that
3682                  *    do not match these criteria.  They have real
3683                  *    differences.
3684                  *
3685                  * 2. At this point, the file is known to be modified,
3686                  *    with the same mode and size, and the object
3687                  *    name of one side is unknown.  Need to inspect
3688                  *    the identical contents.
3689                  */
3690                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3691                     !DIFF_FILE_VALID(p->two) ||
3692                     (p->one->sha1_valid && p->two->sha1_valid) ||
3693                     (p->one->mode != p->two->mode) ||
3694                     diff_populate_filespec(p->one, 1) ||
3695                     diff_populate_filespec(p->two, 1) ||
3696                     (p->one->size != p->two->size) ||
3697                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3698                         diff_q(&outq, p);
3699                 else {
3700                         /*
3701                          * The caller can subtract 1 from skip_stat_unmatch
3702                          * to determine how many paths were dirty only
3703                          * due to stat info mismatch.
3704                          */
3705                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3706                                 diffopt->skip_stat_unmatch++;
3707                         diff_free_filepair(p);
3708                 }
3709         }
3710         free(q->queue);
3711         *q = outq;
3712 }
3713
3714 static int diffnamecmp(const void *a_, const void *b_)
3715 {
3716         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
3717         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
3718         const char *name_a, *name_b;
3719
3720         name_a = a->one ? a->one->path : a->two->path;
3721         name_b = b->one ? b->one->path : b->two->path;
3722         return strcmp(name_a, name_b);
3723 }
3724
3725 void diffcore_fix_diff_index(struct diff_options *options)
3726 {
3727         struct diff_queue_struct *q = &diff_queued_diff;
3728         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
3729 }
3730
3731 void diffcore_std(struct diff_options *options)
3732 {
3733         if (options->skip_stat_unmatch)
3734                 diffcore_skip_stat_unmatch(options);
3735         if (options->break_opt != -1)
3736                 diffcore_break(options->break_opt);
3737         if (options->detect_rename)
3738                 diffcore_rename(options);
3739         if (options->break_opt != -1)
3740                 diffcore_merge_broken();
3741         if (options->pickaxe)
3742                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3743         if (options->orderfile)
3744                 diffcore_order(options->orderfile);
3745         diff_resolve_rename_copy();
3746         diffcore_apply_filter(options->filter);
3747
3748         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3749                 DIFF_OPT_SET(options, HAS_CHANGES);
3750         else
3751                 DIFF_OPT_CLR(options, HAS_CHANGES);
3752 }
3753
3754 int diff_result_code(struct diff_options *opt, int status)
3755 {
3756         int result = 0;
3757         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3758             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3759                 return status;
3760         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3761             DIFF_OPT_TST(opt, HAS_CHANGES))
3762                 result |= 01;
3763         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3764             DIFF_OPT_TST(opt, CHECK_FAILED))
3765                 result |= 02;
3766         return result;
3767 }
3768
3769 void diff_addremove(struct diff_options *options,
3770                     int addremove, unsigned mode,
3771                     const unsigned char *sha1,
3772                     const char *concatpath, unsigned dirty_submodule)
3773 {
3774         struct diff_filespec *one, *two;
3775
3776         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3777                 return;
3778
3779         /* This may look odd, but it is a preparation for
3780          * feeding "there are unchanged files which should
3781          * not produce diffs, but when you are doing copy
3782          * detection you would need them, so here they are"
3783          * entries to the diff-core.  They will be prefixed
3784          * with something like '=' or '*' (I haven't decided
3785          * which but should not make any difference).
3786          * Feeding the same new and old to diff_change()
3787          * also has the same effect.
3788          * Before the final output happens, they are pruned after
3789          * merged into rename/copy pairs as appropriate.
3790          */
3791         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3792                 addremove = (addremove == '+' ? '-' :
3793                              addremove == '-' ? '+' : addremove);
3794
3795         if (options->prefix &&
3796             strncmp(concatpath, options->prefix, options->prefix_length))
3797                 return;
3798
3799         one = alloc_filespec(concatpath);
3800         two = alloc_filespec(concatpath);
3801
3802         if (addremove != '+')
3803                 fill_filespec(one, sha1, mode);
3804         if (addremove != '-') {
3805                 fill_filespec(two, sha1, mode);
3806                 two->dirty_submodule = dirty_submodule;
3807         }
3808
3809         diff_queue(&diff_queued_diff, one, two);
3810         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3811                 DIFF_OPT_SET(options, HAS_CHANGES);
3812 }
3813
3814 void diff_change(struct diff_options *options,
3815                  unsigned old_mode, unsigned new_mode,
3816                  const unsigned char *old_sha1,
3817                  const unsigned char *new_sha1,
3818                  const char *concatpath,
3819                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
3820 {
3821         struct diff_filespec *one, *two;
3822
3823         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3824                         && S_ISGITLINK(new_mode))
3825                 return;
3826
3827         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3828                 unsigned tmp;
3829                 const unsigned char *tmp_c;
3830                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3831                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3832                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
3833                         new_dirty_submodule = tmp;
3834         }
3835
3836         if (options->prefix &&
3837             strncmp(concatpath, options->prefix, options->prefix_length))
3838                 return;
3839
3840         one = alloc_filespec(concatpath);
3841         two = alloc_filespec(concatpath);
3842         fill_filespec(one, old_sha1, old_mode);
3843         fill_filespec(two, new_sha1, new_mode);
3844         one->dirty_submodule = old_dirty_submodule;
3845         two->dirty_submodule = new_dirty_submodule;
3846
3847         diff_queue(&diff_queued_diff, one, two);
3848         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3849                 DIFF_OPT_SET(options, HAS_CHANGES);
3850 }
3851
3852 void diff_unmerge(struct diff_options *options,
3853                   const char *path,
3854                   unsigned mode, const unsigned char *sha1)
3855 {
3856         struct diff_filespec *one, *two;
3857
3858         if (options->prefix &&
3859             strncmp(path, options->prefix, options->prefix_length))
3860                 return;
3861
3862         one = alloc_filespec(path);
3863         two = alloc_filespec(path);
3864         fill_filespec(one, sha1, mode);
3865         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3866 }
3867
3868 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
3869                 size_t *outsize)
3870 {
3871         struct diff_tempfile *temp;
3872         const char *argv[3];
3873         const char **arg = argv;
3874         struct child_process child;
3875         struct strbuf buf = STRBUF_INIT;
3876         int err = 0;
3877
3878         temp = prepare_temp_file(spec->path, spec);
3879         *arg++ = pgm;
3880         *arg++ = temp->name;
3881         *arg = NULL;
3882
3883         memset(&child, 0, sizeof(child));
3884         child.use_shell = 1;
3885         child.argv = argv;
3886         child.out = -1;
3887         if (start_command(&child)) {
3888                 remove_tempfile();
3889                 return NULL;
3890         }
3891
3892         if (strbuf_read(&buf, child.out, 0) < 0)
3893                 err = error("error reading from textconv command '%s'", pgm);
3894         close(child.out);
3895
3896         if (finish_command(&child) || err) {
3897                 strbuf_release(&buf);
3898                 remove_tempfile();
3899                 return NULL;
3900         }
3901         remove_tempfile();
3902
3903         return strbuf_detach(&buf, outsize);
3904 }
3905
3906 static size_t fill_textconv(struct userdiff_driver *driver,
3907                             struct diff_filespec *df,
3908                             char **outbuf)
3909 {
3910         size_t size;
3911
3912         if (!driver || !driver->textconv) {
3913                 if (!DIFF_FILE_VALID(df)) {
3914                         *outbuf = "";
3915                         return 0;
3916                 }
3917                 if (diff_populate_filespec(df, 0))
3918                         die("unable to read files to diff");
3919                 *outbuf = df->data;
3920                 return df->size;
3921         }
3922
3923         if (driver->textconv_cache) {
3924                 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
3925                                           &size);
3926                 if (*outbuf)
3927                         return size;
3928         }
3929
3930         *outbuf = run_textconv(driver->textconv, df, &size);
3931         if (!*outbuf)
3932                 die("unable to read files to diff");
3933
3934         if (driver->textconv_cache) {
3935                 /* ignore errors, as we might be in a readonly repository */
3936                 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
3937                                 size);
3938                 /*
3939                  * we could save up changes and flush them all at the end,
3940                  * but we would need an extra call after all diffing is done.
3941                  * Since generating a cache entry is the slow path anyway,
3942                  * this extra overhead probably isn't a big deal.
3943                  */
3944                 notes_cache_write(driver->textconv_cache);
3945         }
3946
3947         return size;
3948 }