| 1 | // |
| 2 | // Glob.cpp |
| 3 | // |
| 4 | // Library: Foundation |
| 5 | // Package: Filesystem |
| 6 | // Module: Glob |
| 7 | // |
| 8 | // Copyright (c) 2004-2006, Applied Informatics Software Engineering GmbH. |
| 9 | // and Contributors. |
| 10 | // |
| 11 | // SPDX-License-Identifier: BSL-1.0 |
| 12 | // |
| 13 | |
| 14 | |
| 15 | #include "Poco/Glob.h" |
| 16 | #include "Poco/Path.h" |
| 17 | #include "Poco/Exception.h" |
| 18 | #include "Poco/DirectoryIterator.h" |
| 19 | #include "Poco/File.h" |
| 20 | #include "Poco/UTF8Encoding.h" |
| 21 | #include "Poco/Unicode.h" |
| 22 | |
| 23 | |
| 24 | namespace Poco { |
| 25 | |
| 26 | |
| 27 | Glob::Glob(const std::string& pattern, int options) |
| 28 | : _pattern(pattern), _options(options) |
| 29 | { |
| 30 | } |
| 31 | |
| 32 | |
| 33 | Glob::~Glob() |
| 34 | { |
| 35 | } |
| 36 | |
| 37 | |
| 38 | bool Glob::match(const std::string& subject) |
| 39 | { |
| 40 | UTF8Encoding utf8; |
| 41 | TextIterator itp(_pattern, utf8); |
| 42 | TextIterator endp(_pattern); |
| 43 | TextIterator its(subject, utf8); |
| 44 | TextIterator ends(subject); |
| 45 | |
| 46 | if ((_options & GLOB_DOT_SPECIAL) && its != ends && *its == '.' && (*itp == '?' || *itp == '*')) |
| 47 | return false; |
| 48 | else |
| 49 | return match(itp, endp, its, ends); |
| 50 | } |
| 51 | |
| 52 | |
| 53 | void Glob::glob(const std::string& pathPattern, std::set<std::string>& files, int options) |
| 54 | { |
| 55 | glob(Path(Path::expand(pathPattern), Path::PATH_GUESS), files, options); |
| 56 | } |
| 57 | |
| 58 | |
| 59 | void Glob::glob(const char* pathPattern, std::set<std::string>& files, int options) |
| 60 | { |
| 61 | glob(Path(Path::expand(pathPattern), Path::PATH_GUESS), files, options); |
| 62 | } |
| 63 | |
| 64 | |
| 65 | void Glob::glob(const Path& pathPattern, std::set<std::string>& files, int options) |
| 66 | { |
| 67 | Path pattern(pathPattern); |
| 68 | pattern.makeDirectory(); // to simplify pattern handling later on |
| 69 | Path base(pattern); |
| 70 | Path absBase(base); |
| 71 | absBase.makeAbsolute(); |
| 72 | // In case of UNC paths we must not pop the topmost directory |
| 73 | // (which must not contain wildcards), otherwise collect() will fail |
| 74 | // as one cannot create a DirectoryIterator with only a node name ("\\srv\"). |
| 75 | int minDepth = base.getNode().empty() ? 0 : 1; |
| 76 | while (base.depth() > minDepth && base[base.depth() - 1] != ".." ) |
| 77 | { |
| 78 | base.popDirectory(); |
| 79 | absBase.popDirectory(); |
| 80 | } |
| 81 | if (pathPattern.isDirectory()) |
| 82 | options |= GLOB_DIRS_ONLY; |
| 83 | collect(pattern, absBase, base, pathPattern[base.depth()], files, options); |
| 84 | } |
| 85 | |
| 86 | |
| 87 | void Glob::glob(const Path& pathPattern, const Path& basePath, std::set<std::string>& files, int options) |
| 88 | { |
| 89 | Path pattern(pathPattern); |
| 90 | pattern.makeDirectory(); // to simplify pattern handling later on |
| 91 | Path absBase(basePath); |
| 92 | absBase.makeAbsolute(); |
| 93 | if (pathPattern.isDirectory()) |
| 94 | options |= GLOB_DIRS_ONLY; |
| 95 | collect(pattern, absBase, basePath, pathPattern[basePath.depth()], files, options); |
| 96 | } |
| 97 | |
| 98 | |
| 99 | bool Glob::match(TextIterator& itp, const TextIterator& endp, TextIterator& its, const TextIterator& ends) |
| 100 | { |
| 101 | while (itp != endp) |
| 102 | { |
| 103 | if (its == ends) |
| 104 | { |
| 105 | while (itp != endp && *itp == '*') ++itp; |
| 106 | break; |
| 107 | } |
| 108 | switch (*itp) |
| 109 | { |
| 110 | case '?': |
| 111 | ++itp; ++its; |
| 112 | break; |
| 113 | case '*': |
| 114 | if (++itp != endp) |
| 115 | { |
| 116 | while (its != ends && !matchAfterAsterisk(itp, endp, its, ends)) ++its; |
| 117 | return its != ends; |
| 118 | } |
| 119 | return true; |
| 120 | case '[': |
| 121 | if (++itp != endp) |
| 122 | { |
| 123 | bool invert = *itp == '!'; |
| 124 | if (invert) ++itp; |
| 125 | if (itp != endp) |
| 126 | { |
| 127 | bool mtch = matchSet(itp, endp, *its++); |
| 128 | if ((invert && mtch) || (!invert && !mtch)) return false; |
| 129 | break; |
| 130 | } |
| 131 | } |
| 132 | throw SyntaxException("bad range syntax in glob pattern" ); |
| 133 | case '\\': |
| 134 | if (++itp == endp) throw SyntaxException("backslash must be followed by character in glob pattern" ); |
| 135 | // fallthrough |
| 136 | default: |
| 137 | if (_options & GLOB_CASELESS) |
| 138 | { |
| 139 | if (Unicode::toLower(*itp) != Unicode::toLower(*its)) return false; |
| 140 | } |
| 141 | else |
| 142 | { |
| 143 | if (*itp != *its) return false; |
| 144 | } |
| 145 | ++itp; ++its; |
| 146 | } |
| 147 | } |
| 148 | return itp == endp && its == ends; |
| 149 | } |
| 150 | |
| 151 | |
| 152 | bool Glob::matchAfterAsterisk(TextIterator itp, const TextIterator& endp, TextIterator its, const TextIterator& ends) |
| 153 | { |
| 154 | return match(itp, endp, its, ends); |
| 155 | } |
| 156 | |
| 157 | |
| 158 | bool Glob::matchSet(TextIterator& itp, const TextIterator& endp, int c) |
| 159 | { |
| 160 | if (_options & GLOB_CASELESS) |
| 161 | c = Unicode::toLower(c); |
| 162 | |
| 163 | while (itp != endp) |
| 164 | { |
| 165 | switch (*itp) |
| 166 | { |
| 167 | case ']': |
| 168 | ++itp; |
| 169 | return false; |
| 170 | case '\\': |
| 171 | if (++itp == endp) throw SyntaxException("backslash must be followed by character in glob pattern" ); |
| 172 | } |
| 173 | int first = *itp; |
| 174 | int last = first; |
| 175 | if (++itp != endp && *itp == '-') |
| 176 | { |
| 177 | if (++itp != endp) |
| 178 | last = *itp++; |
| 179 | else |
| 180 | throw SyntaxException("bad range syntax in glob pattern" ); |
| 181 | } |
| 182 | if (_options & GLOB_CASELESS) |
| 183 | { |
| 184 | first = Unicode::toLower(first); |
| 185 | last = Unicode::toLower(last); |
| 186 | } |
| 187 | if (first <= c && c <= last) |
| 188 | { |
| 189 | while (itp != endp) |
| 190 | { |
| 191 | switch (*itp) |
| 192 | { |
| 193 | case ']': |
| 194 | ++itp; |
| 195 | return true; |
| 196 | case '\\': |
| 197 | if (++itp == endp) break; |
| 198 | default: |
| 199 | ++itp; |
| 200 | } |
| 201 | } |
| 202 | throw SyntaxException("range must be terminated by closing bracket in glob pattern" ); |
| 203 | } |
| 204 | } |
| 205 | return false; |
| 206 | } |
| 207 | |
| 208 | |
| 209 | void Glob::collect(const Path& pathPattern, const Path& base, const Path& current, const std::string& pattern, std::set<std::string>& files, int options) |
| 210 | { |
| 211 | try |
| 212 | { |
| 213 | std::string pp = pathPattern.toString(); |
| 214 | std::string basep = base.toString(); |
| 215 | std::string curp = current.toString(); |
| 216 | Glob g(pattern, options); |
| 217 | DirectoryIterator it(base); |
| 218 | DirectoryIterator end; |
| 219 | while (it != end) |
| 220 | { |
| 221 | const std::string& name = it.name(); |
| 222 | if (g.match(name)) |
| 223 | { |
| 224 | Path p(current); |
| 225 | if (p.depth() < pathPattern.depth() - 1) |
| 226 | { |
| 227 | p.pushDirectory(name); |
| 228 | collect(pathPattern, it.path(), p, pathPattern[p.depth()], files, options); |
| 229 | } |
| 230 | else |
| 231 | { |
| 232 | p.setFileName(name); |
| 233 | if (isDirectory(p, (options & GLOB_FOLLOW_SYMLINKS) != 0)) |
| 234 | { |
| 235 | p.makeDirectory(); |
| 236 | files.insert(p.toString()); |
| 237 | } |
| 238 | else if (!(options & GLOB_DIRS_ONLY)) |
| 239 | { |
| 240 | files.insert(p.toString()); |
| 241 | } |
| 242 | } |
| 243 | } |
| 244 | ++it; |
| 245 | } |
| 246 | } |
| 247 | catch (Exception&) |
| 248 | { |
| 249 | } |
| 250 | } |
| 251 | |
| 252 | |
| 253 | bool Glob::isDirectory(const Path& path, bool followSymlink) |
| 254 | { |
| 255 | File f(path); |
| 256 | bool isDir = false; |
| 257 | try |
| 258 | { |
| 259 | isDir = f.isDirectory(); |
| 260 | } |
| 261 | catch (Poco::Exception&) |
| 262 | { |
| 263 | return false; |
| 264 | } |
| 265 | if (isDir) |
| 266 | { |
| 267 | return true; |
| 268 | } |
| 269 | else if (followSymlink && f.isLink()) |
| 270 | { |
| 271 | try |
| 272 | { |
| 273 | // Test if link resolves to a directory. |
| 274 | DirectoryIterator it(f); |
| 275 | return true; |
| 276 | } |
| 277 | catch (Exception&) |
| 278 | { |
| 279 | } |
| 280 | } |
| 281 | return false; |
| 282 | } |
| 283 | |
| 284 | |
| 285 | } // namespace Poco |
| 286 | |