X-Git-Url: https://wimlib.net/git/?a=blobdiff_plain;f=src%2Fwildcard.c;h=408b0fe32e73b1cd3df618c9be8b950a174633d7;hb=8b74fa415edfefeee5869704d39ee8de1e82b422;hp=c0f0adbdb6486b35b00227a43253b6c0402f9b95;hpb=0065a73c25d704b139d5880d6e3af73a0b056a54;p=wimlib diff --git a/src/wildcard.c b/src/wildcard.c index c0f0adbd..408b0fe3 100644 --- a/src/wildcard.c +++ b/src/wildcard.c @@ -32,6 +32,7 @@ #include "wimlib/encoding.h" #include "wimlib/error.h" #include "wimlib/metadata.h" +#include "wimlib/paths.h" #include "wimlib/wildcard.h" struct match_dentry_ctx { @@ -44,16 +45,6 @@ struct match_dentry_ctx { bool case_insensitive; }; -#define PLATFORM_SUPPORTS_FNMATCH - -#ifdef __WIN32__ -/* PathMatchSpec() could provide a fnmatch() alternative, but it isn't - * documented properly, nor does it work properly. For example, it returns that - * any name matches *.* even if that name doesn't actually contain a period. */ -# undef PLATFORM_SUPPORTS_FNMATCH -#endif - -#ifndef PLATFORM_SUPPORTS_FNMATCH static bool do_match_wildcard(const tchar *string, size_t string_len, const tchar *wildcard, size_t wildcard_len, @@ -88,30 +79,87 @@ do_match_wildcard(const tchar *string, size_t string_len, } } } -#endif /* ! PLATFORM_SUPPORTS_FNMATCH */ static bool -match_wildcard(const tchar *string, tchar *wildcard, +match_wildcard(const tchar *string, const tchar *wildcard, size_t wildcard_len, bool ignore_case) { -#ifdef PLATFORM_SUPPORTS_FNMATCH - char orig; - int ret; - int flags = FNM_NOESCAPE; - if (ignore_case) - flags |= FNM_CASEFOLD; - - orig = wildcard[wildcard_len]; - wildcard[wildcard_len] = T('\0'); - - ret = fnmatch(wildcard, string, flags); - - wildcard[wildcard_len] = orig; - return (ret == 0); -#else return do_match_wildcard(string, tstrlen(string), wildcard, wildcard_len, ignore_case); -#endif +} + +/* + * Determines whether a path matches a wildcard pattern. + * + * @path + * The path to match. Assumptions: All path separators must be @path_sep, + * there cannot be consecutive path separators, there cannot be a trailing + * path separator, and there must be exactly one leading path separator. + * + * @path_nchars + * Number of characters in @path. + * + * @wildcard + * The wildcard pattern to match. It can contain the wildcard characters + * '*' and '?'. The former matches zero or more characters except + * @path_sep, and the latter matches any character except @path_sep. All + * path separators in the pattern must be @path_sep, and there cannot be + * consecutive path separators, and there cannot be a trailing path + * separator. If there is a leading path separator, the match is attempted + * with the filename only; otherwise, the match is attempted with the whole + * path. + * + * @path_sep + * Path separator character used in @path and @wildcard. + * + * @prefix_ok + * If %true, allow a prefix of @path, terminated by a path separator, to + * match the pattern, in addition to @path itself. In other words, return + * %true if the pattern actually matches one of the ancestor directories of + * @path. + * + * Returns %true if there was a match; %false if there was not. + */ +bool +match_path(const tchar *path, size_t path_nchars, + const tchar *wildcard, tchar path_sep, bool prefix_ok) +{ + if (*wildcard != path_sep) { + /* Pattern doesn't begin with path separator. Try to match the + * file name only. */ + return match_wildcard(path_basename_with_len(path, path_nchars), + wildcard, tstrlen(wildcard), + default_ignore_case); + } else { + /* Pattern begins with path separator. Try to match the whole + * path. */ + do { + if (!*wildcard) { + /* Path has more components than pattern */ + return prefix_ok; + } + + size_t path_component_len = 0; + size_t wildcard_component_len = 0; + + do { + path_component_len++; + } while (path[path_component_len] != path_sep && + path[path_component_len] != T('\0')); + do { + wildcard_component_len++; + } while (wildcard[wildcard_component_len] != path_sep && + wildcard[wildcard_component_len] != T('\0')); + if (!do_match_wildcard(path, path_component_len, + wildcard, wildcard_component_len, + default_ignore_case)) + return false; + path += path_component_len; + wildcard += wildcard_component_len; + } while (*path); + + return (*wildcard == '\0'); + } } static int @@ -129,7 +177,7 @@ wildcard_status(const tchar *wildcard) { if (*wildcard == T('\0')) return WILDCARD_STATUS_DONE_FULLY; - while (is_any_path_separator(*wildcard)) + while (*wildcard == WIM_PATH_SEPARATOR) wildcard++; if (*wildcard == T('\0')) return WILDCARD_STATUS_DONE_TRAILING_SLASHES; @@ -138,32 +186,27 @@ wildcard_status(const tchar *wildcard) } static int -match_dentry(struct wim_dentry *cur_dentry, void *_ctx) +match_dentry(struct wim_dentry *cur_dentry, struct match_dentry_ctx *ctx) { - struct match_dentry_ctx *ctx = _ctx; - tchar *name; - size_t name_len; + const tchar *name; + size_t name_nchars; int ret; if (cur_dentry->file_name_nbytes == 0) return 0; -#if TCHAR_IS_UTF16LE - name = cur_dentry->file_name; - name_len = cur_dentry->file_name_nbytes; -#else - ret = utf16le_to_tstr(cur_dentry->file_name, - cur_dentry->file_name_nbytes, - &name, &name_len); + ret = utf16le_get_tstr(cur_dentry->file_name, + cur_dentry->file_name_nbytes, + &name, &name_nchars); if (ret) return ret; -#endif - name_len /= sizeof(tchar); + name_nchars /= sizeof(tchar); - if (match_wildcard(name, - &ctx->wildcard_path[ctx->cur_component_offset], - ctx->cur_component_len, - ctx->case_insensitive)) + if (do_match_wildcard(name, + name_nchars, + &ctx->wildcard_path[ctx->cur_component_offset], + ctx->cur_component_len, + ctx->case_insensitive)) { switch (wildcard_status(&ctx->wildcard_path[ ctx->cur_component_offset + @@ -188,9 +231,8 @@ match_dentry(struct wim_dentry *cur_dentry, void *_ctx) ret = 0; } -#if !TCHAR_IS_UTF16LE - FREE(name); -#endif + utf16le_put_tstr(name); + return ret; } @@ -205,16 +247,17 @@ expand_wildcard_recursive(struct wim_dentry *cur_dentry, size_t offset_save; size_t len_save; int ret; + struct wim_dentry *child; w = ctx->wildcard_path; begin = ctx->cur_component_offset + ctx->cur_component_len; - while (is_any_path_separator(w[begin])) + while (w[begin] == WIM_PATH_SEPARATOR) begin++; end = begin; - while (w[end] != T('\0') && !is_any_path_separator(w[end])) + while (w[end] != T('\0') && w[end] != WIM_PATH_SEPARATOR) end++; len = end - begin; @@ -228,7 +271,12 @@ expand_wildcard_recursive(struct wim_dentry *cur_dentry, ctx->cur_component_offset = begin; ctx->cur_component_len = len; - ret = for_dentry_child(cur_dentry, match_dentry, ctx); + ret = 0; + for_dentry_child(child, cur_dentry) { + ret = match_dentry(child, ctx); + if (ret) + break; + } ctx->cur_component_len = len_save; ctx->cur_component_offset = offset_save; @@ -243,9 +291,9 @@ expand_wildcard_recursive(struct wim_dentry *cur_dentry, * wildcard. * @wildcard_path * Wildcard path to expand, which may contain the '?' and '*' characters. - * Path separators may be either forward slashes, and leading path - * separators are ignored. Trailing path separators indicate that the - * wildcard can only match directories. + * Path separators must be WIM_PATH_SEPARATOR. Leading path separators are + * ignored, whereas one or more trailing path separators indicate that the + * wildcard path can only match directories (and not reparse points). * @consume_dentry * Callback function which will receive each directory entry matched by the * wildcard. @@ -268,9 +316,6 @@ expand_wildcard_recursive(struct wim_dentry *cur_dentry, * * @return 0 on success; a positive error code on error; or the first nonzero * value returned by @consume_dentry. - * - * Note: this function uses the @tmp_list field of dentries it attempts to - * match. */ int expand_wildcard(WIMStruct *wim, @@ -282,7 +327,7 @@ expand_wildcard(WIMStruct *wim, struct wim_dentry *root; int ret; - root = wim_root_dentry(wim); + root = wim_get_current_root_dentry(wim); if (root == NULL) goto no_match;