src/attr_file.c

   1 #include "common.h"
   2 #include "repository.h"
   3 #include "filebuf.h"
   4 #include "attr_file.h"
   5 #include "attrcache.h"
   6 #include "git2/blob.h"
   7 #include "git2/tree.h"
   8 #include "index.h"
   9 #include <ctype.h>
  10
  11 static void attr_file_free(git_attr_file *file)
  12 {
  13         bool unlock = !git_mutex_lock(&file->lock);
  14         git_attr_file__clear_rules(file, false);
  15         git_pool_clear(&file->pool);
  16         if (unlock)
  17                 git_mutex_unlock(&file->lock);
  18         git_mutex_free(&file->lock);
  19
  20         git__memzero(file, sizeof(*file));
  21         git__free(file);
  22 }
  23
  24 int git_attr_file__new(
  25         git_attr_file **out,
  26         git_attr_file_entry *entry,
  27         git_attr_file_source source)
  28 {
  29         git_attr_file *attrs = git__calloc(1, sizeof(git_attr_file));
  30         GITERR_CHECK_ALLOC(attrs);
  31
  32         if (git_mutex_init(&attrs->lock) < 0) {
  33                 giterr_set(GITERR_OS, "Failed to initialize lock");
  34                 git__free(attrs);
  35                 return -1;
  36         }
  37
  38         if (git_pool_init(&attrs->pool, 1, 0) < 0) {
  39                 attr_file_free(attrs);
  40                 return -1;
  41         }
  42
  43         GIT_REFCOUNT_INC(attrs);
  44         attrs->entry  = entry;
  45         attrs->source = source;
  46         *out = attrs;
  47         return 0;
  48 }
  49
  50 int git_attr_file__clear_rules(git_attr_file *file, bool need_lock)
  51 {
  52         unsigned int i;
  53         git_attr_rule *rule;
  54
  55         if (need_lock && git_mutex_lock(&file->lock) < 0) {
  56                 giterr_set(GITERR_OS, "Failed to lock attribute file");
  57                 return -1;
  58         }
  59
  60         git_vector_foreach(&file->rules, i, rule)
  61                 git_attr_rule__free(rule);
  62         git_vector_free(&file->rules);
  63
  64         if (need_lock)
  65                 git_mutex_unlock(&file->lock);
  66
  67         return 0;
  68 }
  69
  70 void git_attr_file__free(git_attr_file *file)
  71 {
  72         if (!file)
  73                 return;
  74         GIT_REFCOUNT_DEC(file, attr_file_free);
  75 }
  76
  77 static int attr_file_oid_from_index(
  78         git_oid *oid, git_repository *repo, const char *path)
  79 {
  80         int error;
  81         git_index *idx;
  82         size_t pos;
  83         const git_index_entry *entry;
  84
  85         if ((error = git_repository_index__weakptr(&idx, repo)) < 0 ||
  86                 (error = git_index__find_pos(&pos, idx, path, 0, 0)) < 0)
  87                 return error;
  88
  89         if (!(entry = git_index_get_byindex(idx, pos)))
  90                 return GIT_ENOTFOUND;
  91
  92         *oid = entry->id;
  93         return 0;
  94 }
  95
  96 int git_attr_file__load(
  97         git_attr_file **out,
  98         git_repository *repo,
  99         git_attr_file_entry *entry,
 100         git_attr_file_source source,
 101         git_attr_file_parser parser)
 102 {
 103         int error = 0;
 104         git_blob *blob = NULL;
 105         git_buf content = GIT_BUF_INIT;
 106         git_attr_file *file;
 107         struct stat st;
 108
 109         *out = NULL;
 110
 111         switch (source) {
 112         case GIT_ATTR_FILE__IN_MEMORY:
 113                 /* in-memory attribute file doesn't need data */
 114                 break;
 115         case GIT_ATTR_FILE__FROM_INDEX: {
 116                 git_oid id;
 117
 118                 if ((error = attr_file_oid_from_index(&id, repo, entry->path)) < 0 ||
 119                         (error = git_blob_lookup(&blob, repo, &id)) < 0)
 120                         return error;
 121
 122                 /* Do not assume that data straight from the ODB is NULL-terminated;
 123                  * copy the contents of a file to a buffer to work on */
 124                 git_buf_put(&content, git_blob_rawcontent(blob), git_blob_rawsize(blob));
 125                 break;
 126         }
 127         case GIT_ATTR_FILE__FROM_FILE: {
 128                 int fd;
 129
 130                 if (p_stat(entry->fullpath, &st) < 0)
 131                         return git_path_set_error(errno, entry->fullpath, "stat");
 132                 if (S_ISDIR(st.st_mode))
 133                         return GIT_ENOTFOUND;
 134
 135                 /* For open or read errors, return ENOTFOUND to skip item */
 136                 /* TODO: issue warning when warning API is available */
 137
 138                 if ((fd = git_futils_open_ro(entry->fullpath)) < 0)
 139                         return GIT_ENOTFOUND;
 140
 141                 error = git_futils_readbuffer_fd(&content, fd, (size_t)st.st_size);
 142                 p_close(fd);
 143
 144                 if (error < 0)
 145                         return GIT_ENOTFOUND;
 146
 147                 break;
 148         }
 149         default:
 150                 giterr_set(GITERR_INVALID, "Unknown file source %d", source);
 151                 return -1;
 152         }
 153
 154         if ((error = git_attr_file__new(&file, entry, source)) < 0)
 155                 goto cleanup;
 156
 157         if (parser && (error = parser(repo, file, git_buf_cstr(&content))) < 0) {
 158                 git_attr_file__free(file);
 159                 goto cleanup;
 160         }
 161
 162         /* write cache breaker */
 163         if (source == GIT_ATTR_FILE__FROM_INDEX)
 164                 git_oid_cpy(&file->cache_data.oid, git_blob_id(blob));
 165         else if (source == GIT_ATTR_FILE__FROM_FILE)
 166                 git_futils_filestamp_set_from_stat(&file->cache_data.stamp, &st);
 167         /* else always cacheable */
 168
 169         *out = file;
 170
 171 cleanup:
 172         git_blob_free(blob);
 173         git_buf_free(&content);
 174
 175         return error;
 176 }
 177
 178 int git_attr_file__out_of_date(git_repository *repo, git_attr_file *file)
 179 {
 180         if (!file)
 181                 return 1;
 182
 183         switch (file->source) {
 184         case GIT_ATTR_FILE__IN_MEMORY:
 185                 return 0;
 186
 187         case GIT_ATTR_FILE__FROM_FILE:
 188                 return git_futils_filestamp_check(
 189                         &file->cache_data.stamp, file->entry->fullpath);
 190
 191         case GIT_ATTR_FILE__FROM_INDEX: {
 192                 int error;
 193                 git_oid id;
 194
 195                 if ((error = attr_file_oid_from_index(
 196                                 &id, repo, file->entry->path)) < 0)
 197                         return error;
 198
 199                 return (git_oid__cmp(&file->cache_data.oid, &id) != 0);
 200         }
 201
 202         default:
 203                 giterr_set(GITERR_INVALID, "Invalid file type %d", file->source);
 204                 return -1;
 205         }
 206 }
 207
 208 static int sort_by_hash_and_name(const void *a_raw, const void *b_raw);
 209 static void git_attr_rule__clear(git_attr_rule *rule);
 210 static bool parse_optimized_patterns(
 211         git_attr_fnmatch *spec,
 212         git_pool *pool,
 213         const char *pattern);
 214
 215 int git_attr_file__parse_buffer(
 216         git_repository *repo, git_attr_file *attrs, const char *data)
 217 {
 218         int error = 0;
 219         const char *scan = data, *context = NULL;
 220         git_attr_rule *rule = NULL;
 221
 222         /* if subdir file path, convert context for file paths */
 223         if (attrs->entry &&
 224                 git_path_root(attrs->entry->path) < 0 &&
 225                 !git__suffixcmp(attrs->entry->path, "/" GIT_ATTR_FILE))
 226                 context = attrs->entry->path;
 227
 228         if (git_mutex_lock(&attrs->lock) < 0) {
 229                 giterr_set(GITERR_OS, "Failed to lock attribute file");
 230                 return -1;
 231         }
 232
 233         while (!error && *scan) {
 234                 /* allocate rule if needed */
 235                 if (!rule && !(rule = git__calloc(1, sizeof(*rule)))) {
 236                         error = -1;
 237                         break;
 238                 }
 239
 240                 rule->match.flags =
 241                         GIT_ATTR_FNMATCH_ALLOWNEG | GIT_ATTR_FNMATCH_ALLOWMACRO;
 242
 243                 /* parse the next "pattern attr attr attr" line */
 244                 if (!(error = git_attr_fnmatch__parse(
 245                                 &rule->match, &attrs->pool, context, &scan)) &&
 246                         !(error = git_attr_assignment__parse(
 247                                 repo, &attrs->pool, &rule->assigns, &scan)))
 248                 {
 249                         if (rule->match.flags & GIT_ATTR_FNMATCH_MACRO)
 250                                 /* TODO: warning if macro found in file below repo root */
 251                                 error = git_attr_cache__insert_macro(repo, rule);
 252                         else
 253                                 error = git_vector_insert(&attrs->rules, rule);
 254                 }
 255
 256                 /* if the rule wasn't a pattern, on to the next */
 257                 if (error < 0) {
 258                         git_attr_rule__clear(rule); /* reset rule contents */
 259                         if (error == GIT_ENOTFOUND)
 260                                 error = 0;
 261                 } else {
 262                         rule = NULL; /* vector now "owns" the rule */
 263                 }
 264         }
 265
 266         git_mutex_unlock(&attrs->lock);
 267         git_attr_rule__free(rule);
 268
 269         return error;
 270 }
 271
 272 uint32_t git_attr_file__name_hash(const char *name)
 273 {
 274         uint32_t h = 5381;
 275         int c;
 276         assert(name);
 277         while ((c = (int)*name++) != 0)
 278                 h = ((h << 5) + h) + c;
 279         return h;
 280 }
 281
 282 int git_attr_file__lookup_one(
 283         git_attr_file *file,
 284         git_attr_path *path,
 285         const char *attr,
 286         const char **value)
 287 {
 288         size_t i;
 289         git_attr_name name;
 290         git_attr_rule *rule;
 291
 292         *value = NULL;
 293
 294         name.name = attr;
 295         name.name_hash = git_attr_file__name_hash(attr);
 296
 297         git_attr_file__foreach_matching_rule(file, path, i, rule) {
 298                 size_t pos;
 299
 300                 if (!git_vector_bsearch(&pos, &rule->assigns, &name)) {
 301                         *value = ((git_attr_assignment *)
 302                                           git_vector_get(&rule->assigns, pos))->value;
 303                         break;
 304                 }
 305         }
 306
 307         return 0;
 308 }
 309
 310 int git_attr_file__load_standalone(git_attr_file **out, const char *path)
 311 {
 312         int error;
 313         git_attr_file *file;
 314         git_buf content = GIT_BUF_INIT;
 315
 316         error = git_attr_file__new(&file, NULL, GIT_ATTR_FILE__FROM_FILE);
 317         if (error < 0)
 318                 return error;
 319
 320         error = git_attr_cache__alloc_file_entry(
 321                 &file->entry, NULL, path, &file->pool);
 322         if (error < 0) {
 323                 git_attr_file__free(file);
 324                 return error;
 325         }
 326         /* because the cache entry is allocated from the file's own pool, we
 327          * don't have to free it - freeing file+pool will free cache entry, too.
 328          */
 329
 330         if (!(error = git_futils_readbuffer(&content, path))) {
 331                 error = git_attr_file__parse_buffer(NULL, file, content.ptr);
 332                 git_buf_free(&content);
 333         }
 334
 335         if (error < 0)
 336                 git_attr_file__free(file);
 337         else
 338                 *out = file;
 339
 340         return error;
 341 }
 342
 343 bool git_attr_fnmatch__match(
 344         git_attr_fnmatch *match,
 345         git_attr_path *path)
 346 {
 347         const char *filename;
 348         int flags = 0;
 349
 350         /*
 351          * If the rule was generated in a subdirectory, we must only
 352          * use it for paths inside that directory. We can thus return
 353          * a non-match if the prefixes don't match.
 354          */
 355         if (match->containing_dir) {
 356                 if (match->flags & GIT_ATTR_FNMATCH_ICASE) {
 357                         if (git__strncasecmp(path->path, match->containing_dir, match->containing_dir_length))
 358                                 return 0;
 359                 } else {
 360                         if (git__prefixcmp(path->path, match->containing_dir))
 361                                 return 0;
 362                 }
 363         }
 364
 365         if (match->flags & GIT_ATTR_FNMATCH_ICASE)
 366                 flags |= FNM_CASEFOLD;
 367         if (match->flags & GIT_ATTR_FNMATCH_LEADINGDIR)
 368                 flags |= FNM_LEADING_DIR;
 369
 370         if (match->flags & GIT_ATTR_FNMATCH_FULLPATH) {
 371                 filename = path->path;
 372                 flags |= FNM_PATHNAME;
 373         } else {
 374                 filename = path->basename;
 375
 376                 if (path->is_dir)
 377                         flags |= FNM_LEADING_DIR;
 378         }
 379
 380         if ((match->flags & GIT_ATTR_FNMATCH_DIRECTORY) && !path->is_dir) {
 381                 int matchval;
 382
 383                 /* for attribute checks or root ignore checks, fail match */
 384                 if (!(match->flags & GIT_ATTR_FNMATCH_IGNORE) ||
 385                         path->basename == path->path)
 386                         return false;
 387
 388                 /* for ignore checks, use container of current item for check */
 389                 path->basename[-1] = '\0';
 390                 flags |= FNM_LEADING_DIR;
 391                 matchval = p_fnmatch(match->pattern, path->path, flags);
 392                 path->basename[-1] = '/';
 393                 return (matchval != FNM_NOMATCH);
 394         }
 395
 396         /* if path is a directory prefix of a negated pattern, then match */
 397         if ((match->flags & GIT_ATTR_FNMATCH_NEGATIVE) && path->is_dir) {
 398                 size_t pathlen = strlen(path->path);
 399                 bool prefixed = (pathlen <= match->length) &&
 400                         ((match->flags & GIT_ATTR_FNMATCH_ICASE) ?
 401                          !strncasecmp(match->pattern, path->path, pathlen) :
 402                          !strncmp(match->pattern, path->path, pathlen));
 403
 404                 if (prefixed && git_path_at_end_of_segment(&match->pattern[pathlen]))
 405                         return true;
 406         }
 407
 408         return (p_fnmatch(match->pattern, filename, flags) != FNM_NOMATCH);
 409 }
 410
 411 bool git_attr_rule__match(
 412         git_attr_rule *rule,
 413         git_attr_path *path)
 414 {
 415         bool matched = git_attr_fnmatch__match(&rule->match, path);
 416
 417         if (rule->match.flags & GIT_ATTR_FNMATCH_NEGATIVE)
 418                 matched = !matched;
 419
 420         return matched;
 421 }
 422
 423 git_attr_assignment *git_attr_rule__lookup_assignment(
 424         git_attr_rule *rule, const char *name)
 425 {
 426         size_t pos;
 427         git_attr_name key;
 428         key.name = name;
 429         key.name_hash = git_attr_file__name_hash(name);
 430
 431         if (git_vector_bsearch(&pos, &rule->assigns, &key))
 432                 return NULL;
 433
 434         return git_vector_get(&rule->assigns, pos);
 435 }
 436
 437 int git_attr_path__init(
 438         git_attr_path *info, const char *path, const char *base)
 439 {
 440         ssize_t root;
 441
 442         /* build full path as best we can */
 443         git_buf_init(&info->full, 0);
 444
 445         if (git_path_join_unrooted(&info->full, path, base, &root) < 0)
 446                 return -1;
 447
 448         info->path = info->full.ptr + root;
 449
 450         /* remove trailing slashes */
 451         while (info->full.size > 0) {
 452                 if (info->full.ptr[info->full.size - 1] != '/')
 453                         break;
 454                 info->full.size--;
 455         }
 456         info->full.ptr[info->full.size] = '\0';
 457
 458         /* skip leading slashes in path */
 459         while (*info->path == '/')
 460                 info->path++;
 461
 462         /* find trailing basename component */
 463         info->basename = strrchr(info->path, '/');
 464         if (info->basename)
 465                 info->basename++;
 466         if (!info->basename || !*info->basename)
 467                 info->basename = info->path;
 468
 469         info->is_dir = (int)git_path_isdir(info->full.ptr);
 470
 471         return 0;
 472 }
 473
 474 void git_attr_path__free(git_attr_path *info)
 475 {
 476         git_buf_free(&info->full);
 477         info->path = NULL;
 478         info->basename = NULL;
 479 }
 480
 481 /*
 482  * From gitattributes(5):
 483  *
 484  * Patterns have the following format:
 485  *
 486  * - A blank line matches no files, so it can serve as a separator for
 487  *   readability.
 488  *
 489  * - A line starting with # serves as a comment.
 490  *
 491  * - An optional prefix ! which negates the pattern; any matching file
 492  *   excluded by a previous pattern will become included again. If a negated
 493  *   pattern matches, this will override lower precedence patterns sources.
 494  *
 495  * - If the pattern ends with a slash, it is removed for the purpose of the
 496  *   following description, but it would only find a match with a directory. In
 497  *   other words, foo/ will match a directory foo and paths underneath it, but
 498  *   will not match a regular file or a symbolic link foo (this is consistent
 499  *   with the way how pathspec works in general in git).
 500  *
 501  * - If the pattern does not contain a slash /, git treats it as a shell glob
 502  *   pattern and checks for a match against the pathname without leading
 503  *   directories.
 504  *
 505  * - Otherwise, git treats the pattern as a shell glob suitable for consumption
 506  *   by fnmatch(3) with the FNM_PATHNAME flag: wildcards in the pattern will
 507  *   not match a / in the pathname. For example, "Documentation/\*.html" matches
 508  *   "Documentation/git.html" but not "Documentation/ppc/ppc.html". A leading
 509  *   slash matches the beginning of the pathname; for example, "/\*.c" matches
 510  *   "cat-file.c" but not "mozilla-sha1/sha1.c".
 511  */
 512
 513 /*
 514  * This will return 0 if the spec was filled out,
 515  * GIT_ENOTFOUND if the fnmatch does not require matching, or
 516  * another error code there was an actual problem.
 517  */
 518 int git_attr_fnmatch__parse(
 519         git_attr_fnmatch *spec,
 520         git_pool *pool,
 521         const char *context,
 522         const char **base)
 523 {
 524         const char *pattern, *scan;
 525         int slash_count, allow_space;
 526
 527         assert(spec && base && *base);
 528
 529         if (parse_optimized_patterns(spec, pool, *base))
 530                 return 0;
 531
 532         spec->flags = (spec->flags & GIT_ATTR_FNMATCH__INCOMING);
 533         allow_space = ((spec->flags & GIT_ATTR_FNMATCH_ALLOWSPACE) != 0);
 534
 535         pattern = *base;
 536
 537         while (git__isspace(*pattern)) pattern++;
 538         if (!*pattern || *pattern == '#') {
 539                 *base = git__next_line(pattern);
 540                 return GIT_ENOTFOUND;
 541         }
 542
 543         if (*pattern == '[' && (spec->flags & GIT_ATTR_FNMATCH_ALLOWMACRO) != 0) {
 544                 if (strncmp(pattern, "[attr]", 6) == 0) {
 545                         spec->flags = spec->flags | GIT_ATTR_FNMATCH_MACRO;
 546                         pattern += 6;
 547                 }
 548                 /* else a character range like [a-e]* which is accepted */
 549         }
 550
 551         if (*pattern == '!' && (spec->flags & GIT_ATTR_FNMATCH_ALLOWNEG) != 0) {
 552                 spec->flags = spec->flags |
 553                         GIT_ATTR_FNMATCH_NEGATIVE | GIT_ATTR_FNMATCH_LEADINGDIR;
 554                 pattern++;
 555         }
 556
 557         slash_count = 0;
 558         for (scan = pattern; *scan != '\0'; ++scan) {
 559                 /* scan until (non-escaped) white space */
 560                 if (git__isspace(*scan) && *(scan - 1) != '\\') {
 561                         if (!allow_space || (*scan != ' ' && *scan != '\t' && *scan != '\r'))
 562                                 break;
 563                 }
 564
 565                 if (*scan == '/') {
 566                         spec->flags = spec->flags | GIT_ATTR_FNMATCH_FULLPATH;
 567                         slash_count++;
 568                         if (pattern == scan)
 569                                 pattern++;
 570                 }
 571                 /* remember if we see an unescaped wildcard in pattern */
 572                 else if (git__iswildcard(*scan) &&
 573                         (scan == pattern || (*(scan - 1) != '\\')))
 574                         spec->flags = spec->flags | GIT_ATTR_FNMATCH_HASWILD;
 575         }
 576
 577         *base = scan;
 578
 579         if ((spec->length = scan - pattern) == 0)
 580                 return GIT_ENOTFOUND;
 581
 582         /*
 583          * Remove one trailing \r in case this is a CRLF delimited
 584          * file, in the case of Icon\r\r\n, we still leave the first
 585          * \r there to match against.
 586          */
 587         if (pattern[spec->length - 1] == '\r')
 588                 if (--spec->length == 0)
 589                         return GIT_ENOTFOUND;
 590
 591         if (pattern[spec->length - 1] == '/') {
 592                 spec->length--;
 593                 spec->flags = spec->flags | GIT_ATTR_FNMATCH_DIRECTORY;
 594                 if (--slash_count <= 0)
 595                         spec->flags = spec->flags & ~GIT_ATTR_FNMATCH_FULLPATH;
 596         }
 597         if ((spec->flags & GIT_ATTR_FNMATCH_NOLEADINGDIR) == 0 &&
 598                 spec->length >= 2 &&
 599                 pattern[spec->length - 1] == '*' &&
 600                 pattern[spec->length - 2] == '/') {
 601                 spec->length -= 2;
 602                 spec->flags = spec->flags | GIT_ATTR_FNMATCH_LEADINGDIR;
 603                 /* leave FULLPATH match on, however */
 604         }
 605
 606         if (context) {
 607                 char *slash = strchr(context, '/');
 608                 size_t len;
 609                 if (slash) {
 610                         /* include the slash for easier matching */
 611                         len = slash - context + 1;
 612                         spec->containing_dir = git_pool_strndup(pool, context, len);
 613                         spec->containing_dir_length = len;
 614                 }
 615         }
 616
 617         if ((spec->flags & GIT_ATTR_FNMATCH_FULLPATH) != 0 &&
 618                 context != NULL && git_path_root(pattern) < 0)
 619         {
 620                 /* use context path minus the trailing filename */
 621                 char *slash = strrchr(context, '/');
 622                 size_t contextlen = slash ? slash - context + 1 : 0;
 623
 624                 /* given an unrooted fullpath match from a file inside a repo,
 625                  * prefix the pattern with the relative directory of the source file
 626                  */
 627                 spec->pattern = git_pool_malloc(
 628                         pool, (uint32_t)(contextlen + spec->length + 1));
 629                 if (spec->pattern) {
 630                         memcpy(spec->pattern, context, contextlen);
 631                         memcpy(spec->pattern + contextlen, pattern, spec->length);
 632                         spec->length += contextlen;
 633                         spec->pattern[spec->length] = '\0';
 634                 }
 635         } else {
 636                 spec->pattern = git_pool_strndup(pool, pattern, spec->length);
 637         }
 638
 639         if (!spec->pattern) {
 640                 *base = git__next_line(pattern);
 641                 return -1;
 642         } else {
 643                 /* strip '\' that might have be used for internal whitespace */
 644                 spec->length = git__unescape(spec->pattern);
 645                 /* TODO: convert remaining '\' into '/' for POSIX ??? */
 646         }
 647
 648         return 0;
 649 }
 650
 651 static bool parse_optimized_patterns(
 652         git_attr_fnmatch *spec,
 653         git_pool *pool,
 654         const char *pattern)
 655 {
 656         if (!pattern[1] && (pattern[0] == '*' || pattern[0] == '.')) {
 657                 spec->flags = GIT_ATTR_FNMATCH_MATCH_ALL;
 658                 spec->pattern = git_pool_strndup(pool, pattern, 1);
 659                 spec->length = 1;
 660
 661                 return true;
 662         }
 663
 664         return false;
 665 }
 666
 667 static int sort_by_hash_and_name(const void *a_raw, const void *b_raw)
 668 {
 669         const git_attr_name *a = a_raw;
 670         const git_attr_name *b = b_raw;
 671
 672         if (b->name_hash < a->name_hash)
 673                 return 1;
 674         else if (b->name_hash > a->name_hash)
 675                 return -1;
 676         else
 677                 return strcmp(b->name, a->name);
 678 }
 679
 680 static void git_attr_assignment__free(git_attr_assignment *assign)
 681 {
 682         /* name and value are stored in a git_pool associated with the
 683          * git_attr_file, so they do not need to be freed here
 684          */
 685         assign->name = NULL;
 686         assign->value = NULL;
 687         git__free(assign);
 688 }
 689
 690 static int merge_assignments(void **old_raw, void *new_raw)
 691 {
 692         git_attr_assignment **old = (git_attr_assignment **)old_raw;
 693         git_attr_assignment *new = (git_attr_assignment *)new_raw;
 694
 695         GIT_REFCOUNT_DEC(*old, git_attr_assignment__free);
 696         *old = new;
 697         return GIT_EEXISTS;
 698 }
 699
 700 int git_attr_assignment__parse(
 701         git_repository *repo,
 702         git_pool *pool,
 703         git_vector *assigns,
 704         const char **base)
 705 {
 706         int error;
 707         const char *scan = *base;
 708         git_attr_assignment *assign = NULL;
 709
 710         assert(assigns && !assigns->length);
 711
 712         git_vector_set_cmp(assigns, sort_by_hash_and_name);
 713
 714         while (*scan && *scan != '\n') {
 715                 const char *name_start, *value_start;
 716
 717                 /* skip leading blanks */
 718                 while (git__isspace(*scan) && *scan != '\n') scan++;
 719
 720                 /* allocate assign if needed */
 721                 if (!assign) {
 722                         assign = git__calloc(1, sizeof(git_attr_assignment));
 723                         GITERR_CHECK_ALLOC(assign);
 724                         GIT_REFCOUNT_INC(assign);
 725                 }
 726
 727                 assign->name_hash = 5381;
 728                 assign->value = git_attr__true;
 729
 730                 /* look for magic name prefixes */
 731                 if (*scan == '-') {
 732                         assign->value = git_attr__false;
 733                         scan++;
 734                 } else if (*scan == '!') {
 735                         assign->value = git_attr__unset; /* explicit unspecified state */
 736                         scan++;
 737                 } else if (*scan == '#') /* comment rest of line */
 738                         break;
 739
 740                 /* find the name */
 741                 name_start = scan;
 742                 while (*scan && !git__isspace(*scan) && *scan != '=') {
 743                         assign->name_hash =
 744                                 ((assign->name_hash << 5) + assign->name_hash) + *scan;
 745                         scan++;
 746                 }
 747                 if (scan == name_start) {
 748                         /* must have found lone prefix (" - ") or leading = ("=foo")
 749                          * or end of buffer -- advance until whitespace and continue
 750                          */
 751                         while (*scan && !git__isspace(*scan)) scan++;
 752                         continue;
 753                 }
 754
 755                 /* allocate permanent storage for name */
 756                 assign->name = git_pool_strndup(pool, name_start, scan - name_start);
 757                 GITERR_CHECK_ALLOC(assign->name);
 758
 759                 /* if there is an equals sign, find the value */
 760                 if (*scan == '=') {
 761                         for (value_start = ++scan; *scan && !git__isspace(*scan); ++scan);
 762
 763                         /* if we found a value, allocate permanent storage for it */
 764                         if (scan > value_start) {
 765                                 assign->value = git_pool_strndup(pool, value_start, scan - value_start);
 766                                 GITERR_CHECK_ALLOC(assign->value);
 767                         }
 768                 }
 769
 770                 /* expand macros (if given a repo with a macro cache) */
 771                 if (repo != NULL && assign->value == git_attr__true) {
 772                         git_attr_rule *macro =
 773                                 git_attr_cache__lookup_macro(repo, assign->name);
 774
 775                         if (macro != NULL) {
 776                                 unsigned int i;
 777                                 git_attr_assignment *massign;
 778
 779                                 git_vector_foreach(&macro->assigns, i, massign) {
 780                                         GIT_REFCOUNT_INC(massign);
 781
 782                                         error = git_vector_insert_sorted(
 783                                                 assigns, massign, &merge_assignments);
 784                                         if (error < 0 && error != GIT_EEXISTS)
 785                                                 return error;
 786                                 }
 787                         }
 788                 }
 789
 790                 /* insert allocated assign into vector */
 791                 error = git_vector_insert_sorted(assigns, assign, &merge_assignments);
 792                 if (error < 0 && error != GIT_EEXISTS)
 793                         return error;
 794
 795                 /* clear assign since it is now "owned" by the vector */
 796                 assign = NULL;
 797         }
 798
 799         if (assign != NULL)
 800                 git_attr_assignment__free(assign);
 801
 802         *base = git__next_line(scan);
 803
 804         return (assigns->length == 0) ? GIT_ENOTFOUND : 0;
 805 }
 806
 807 static void git_attr_rule__clear(git_attr_rule *rule)
 808 {
 809         unsigned int i;
 810         git_attr_assignment *assign;
 811
 812         if (!rule)
 813                 return;
 814
 815         if (!(rule->match.flags & GIT_ATTR_FNMATCH_IGNORE)) {
 816                 git_vector_foreach(&rule->assigns, i, assign)
 817                         GIT_REFCOUNT_DEC(assign, git_attr_assignment__free);
 818                 git_vector_free(&rule->assigns);
 819         }
 820
 821         /* match.pattern is stored in a git_pool, so no need to free */
 822         rule->match.pattern = NULL;
 823         rule->match.length = 0;
 824 }
 825
 826 void git_attr_rule__free(git_attr_rule *rule)
 827 {
 828         git_attr_rule__clear(rule);
 829         git__free(rule);
 830 }
 831