1 // Copyright (c) 2012 The Chromium Authors. All rights reserved. 2 // Use of this source code is governed by a BSD-style license that can be 3 // found in the LICENSE file. 4 5 // FilePath is a container for pathnames stored in a platform's native string 6 // type, providing containers for manipulation in according with the 7 // platform's conventions for pathnames. It supports the following path 8 // types: 9 // 10 // POSIX Windows 11 // --------------- ---------------------------------- 12 // Fundamental type char[] wchar_t[] 13 // Encoding unspecified* UTF-16 14 // Separator / \, tolerant of / 15 // Drive letters no case-insensitive A-Z followed by : 16 // Alternate root // (surprise!) \\, for UNC paths 17 // 18 // * The encoding need not be specified on POSIX systems, although some 19 // POSIX-compliant systems do specify an encoding. Mac OS X uses UTF-8. 20 // Chrome OS also uses UTF-8. 21 // Linux does not specify an encoding, but in practice, the locale's 22 // character set may be used. 23 // 24 // For more arcane bits of path trivia, see below. 25 // 26 // FilePath objects are intended to be used anywhere paths are. An 27 // application may pass FilePath objects around internally, masking the 28 // underlying differences between systems, only differing in implementation 29 // where interfacing directly with the system. For example, a single 30 // OpenFile(const FilePath &) function may be made available, allowing all 31 // callers to operate without regard to the underlying implementation. On 32 // POSIX-like platforms, OpenFile might wrap fopen, and on Windows, it might 33 // wrap _wfopen_s, perhaps both by calling file_path.value().c_str(). This 34 // allows each platform to pass pathnames around without requiring conversions 35 // between encodings, which has an impact on performance, but more imporantly, 36 // has an impact on correctness on platforms that do not have well-defined 37 // encodings for pathnames. 38 // 39 // Several methods are available to perform common operations on a FilePath 40 // object, such as determining the parent directory (DirName), isolating the 41 // final path component (BaseName), and appending a relative pathname string 42 // to an existing FilePath object (Append). These methods are highly 43 // recommended over attempting to split and concatenate strings directly. 44 // These methods are based purely on string manipulation and knowledge of 45 // platform-specific pathname conventions, and do not consult the filesystem 46 // at all, making them safe to use without fear of blocking on I/O operations. 47 // These methods do not function as mutators but instead return distinct 48 // instances of FilePath objects, and are therefore safe to use on const 49 // objects. The objects themselves are safe to share between threads. 50 // 51 // To aid in initialization of FilePath objects from string literals, a 52 // FILE_PATH_LITERAL macro is provided, which accounts for the difference 53 // between char[]-based pathnames on POSIX systems and wchar_t[]-based 54 // pathnames on Windows. 55 // 56 // As a precaution against premature truncation, paths can't contain NULs. 57 // 58 // Because a FilePath object should not be instantiated at the global scope, 59 // instead, use a FilePath::CharType[] and initialize it with 60 // FILE_PATH_LITERAL. At runtime, a FilePath object can be created from the 61 // character array. Example: 62 // 63 // | const FilePath::CharType kLogFileName[] = FILE_PATH_LITERAL("log.txt"); 64 // | 65 // | void Function() { 66 // | FilePath log_file_path(kLogFileName); 67 // | [...] 68 // | } 69 // 70 // WARNING: FilePaths should ALWAYS be displayed with LTR directionality, even 71 // when the UI language is RTL. This means you always need to pass filepaths 72 // through base::i18n::WrapPathWithLTRFormatting() before displaying it in the 73 // RTL UI. 74 // 75 // This is a very common source of bugs, please try to keep this in mind. 76 // 77 // ARCANE BITS OF PATH TRIVIA 78 // 79 // - A double leading slash is actually part of the POSIX standard. Systems 80 // are allowed to treat // as an alternate root, as Windows does for UNC 81 // (network share) paths. Most POSIX systems don't do anything special 82 // with two leading slashes, but FilePath handles this case properly 83 // in case it ever comes across such a system. FilePath needs this support 84 // for Windows UNC paths, anyway. 85 // References: 86 // The Open Group Base Specifications Issue 7, sections 3.267 ("Pathname") 87 // and 4.12 ("Pathname Resolution"), available at: 88 // http://www.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap03.html#tag_03_267 89 // http://www.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap04.html#tag_04_12 90 // 91 // - Windows treats c:\\ the same way it treats \\. This was intended to 92 // allow older applications that require drive letters to support UNC paths 93 // like \\server\share\path, by permitting c:\\server\share\path as an 94 // equivalent. Since the OS treats these paths specially, FilePath needs 95 // to do the same. Since Windows can use either / or \ as the separator, 96 // FilePath treats c://, c:\\, //, and \\ all equivalently. 97 // Reference: 98 // The Old New Thing, "Why is a drive letter permitted in front of UNC 99 // paths (sometimes)?", available at: 100 // http://blogs.msdn.com/oldnewthing/archive/2005/11/22/495740.aspx 101 102 #ifndef BASE_FILES_FILE_PATH_H_ 103 #define BASE_FILES_FILE_PATH_H_ 104 105 #include <stddef.h> 106 107 #include <iosfwd> 108 #include <string> 109 #include <vector> 110 111 #include "base/base_export.h" 112 #include "base/compiler_specific.h" 113 #include "base/macros.h" 114 #include "base/strings/string16.h" 115 #include "base/strings/string_piece.h" 116 #include "build/build_config.h" 117 118 // Windows-style drive letter support and pathname separator characters can be 119 // enabled and disabled independently, to aid testing. These #defines are 120 // here so that the same setting can be used in both the implementation and 121 // in the unit test. 122 #if defined(OS_WIN) 123 #define FILE_PATH_USES_DRIVE_LETTERS 124 #define FILE_PATH_USES_WIN_SEPARATORS 125 #endif // OS_WIN 126 127 // To print path names portably use PRIsFP (based on PRIuS and friends from 128 // C99 and format_macros.h) like this: 129 // base::StringPrintf("Path is %" PRIsFP ".\n", path.value().c_str()); 130 #if defined(OS_WIN) 131 #define PRIsFP "ls" 132 #elif defined(OS_POSIX) || defined(OS_FUCHSIA) 133 #define PRIsFP "s" 134 #endif // OS_WIN 135 136 namespace base { 137 138 class Pickle; 139 class PickleIterator; 140 141 // An abstraction to isolate users from the differences between native 142 // pathnames on different platforms. 143 class BASE_EXPORT FilePath { 144 public: 145 #if defined(OS_WIN) 146 // On Windows, for Unicode-aware applications, native pathnames are wchar_t 147 // arrays encoded in UTF-16. 148 typedef std::wstring StringType; 149 #elif defined(OS_POSIX) || defined(OS_FUCHSIA) 150 // On most platforms, native pathnames are char arrays, and the encoding 151 // may or may not be specified. On Mac OS X, native pathnames are encoded 152 // in UTF-8. 153 typedef std::string StringType; 154 #endif // OS_WIN 155 156 typedef BasicStringPiece<StringType> StringPieceType; 157 typedef StringType::value_type CharType; 158 159 // Null-terminated array of separators used to separate components in 160 // hierarchical paths. Each character in this array is a valid separator, 161 // but kSeparators[0] is treated as the canonical separator and will be used 162 // when composing pathnames. 163 static const CharType kSeparators[]; 164 165 // arraysize(kSeparators). 166 static const size_t kSeparatorsLength; 167 168 // A special path component meaning "this directory." 169 static const CharType kCurrentDirectory[]; 170 171 // A special path component meaning "the parent directory." 172 static const CharType kParentDirectory[]; 173 174 // The character used to identify a file extension. 175 static const CharType kExtensionSeparator; 176 177 FilePath(); 178 FilePath(const FilePath& that); 179 explicit FilePath(StringPieceType path); 180 ~FilePath(); 181 FilePath& operator=(const FilePath& that); 182 183 // Constructs FilePath with the contents of |that|, which is left in valid but 184 // unspecified state. 185 FilePath(FilePath&& that) noexcept; 186 // Replaces the contents with those of |that|, which is left in valid but 187 // unspecified state. 188 FilePath& operator=(FilePath&& that); 189 190 bool operator==(const FilePath& that) const; 191 192 bool operator!=(const FilePath& that) const; 193 194 // Required for some STL containers and operations 195 bool operator<(const FilePath& that) const { 196 return path_ < that.path_; 197 } 198 value()199 const StringType& value() const { return path_; } 200 empty()201 bool empty() const { return path_.empty(); } 202 clear()203 void clear() { path_.clear(); } 204 205 // Returns true if |character| is in kSeparators. 206 static bool IsSeparator(CharType character); 207 208 // Returns a vector of all of the components of the provided path. It is 209 // equivalent to calling DirName().value() on the path's root component, 210 // and BaseName().value() on each child component. 211 // 212 // To make sure this is lossless so we can differentiate absolute and 213 // relative paths, the root slash will be included even though no other 214 // slashes will be. The precise behavior is: 215 // 216 // Posix: "/foo/bar" -> [ "/", "foo", "bar" ] 217 // Windows: "C:\foo\bar" -> [ "C:", "\\", "foo", "bar" ] 218 void GetComponents(std::vector<FilePath::StringType>* components) const; 219 220 // Returns true if this FilePath is a strict parent of the |child|. Absolute 221 // and relative paths are accepted i.e. is /foo parent to /foo/bar and 222 // is foo parent to foo/bar. Does not convert paths to absolute, follow 223 // symlinks or directory navigation (e.g. ".."). A path is *NOT* its own 224 // parent. 225 bool IsParent(const FilePath& child) const; 226 227 // If IsParent(child) holds, appends to path (if non-NULL) the 228 // relative path to child and returns true. For example, if parent 229 // holds "/Users/johndoe/Library/Application Support", child holds 230 // "/Users/johndoe/Library/Application Support/Google/Chrome/Default", and 231 // *path holds "/Users/johndoe/Library/Caches", then after 232 // parent.AppendRelativePath(child, path) is called *path will hold 233 // "/Users/johndoe/Library/Caches/Google/Chrome/Default". Otherwise, 234 // returns false. 235 bool AppendRelativePath(const FilePath& child, FilePath* path) const; 236 237 // Returns a FilePath corresponding to the directory containing the path 238 // named by this object, stripping away the file component. If this object 239 // only contains one component, returns a FilePath identifying 240 // kCurrentDirectory. If this object already refers to the root directory, 241 // returns a FilePath identifying the root directory. Please note that this 242 // doesn't resolve directory navigation, e.g. the result for "../a" is "..". 243 FilePath DirName() const WARN_UNUSED_RESULT; 244 245 // Returns a FilePath corresponding to the last path component of this 246 // object, either a file or a directory. If this object already refers to 247 // the root directory, returns a FilePath identifying the root directory; 248 // this is the only situation in which BaseName will return an absolute path. 249 FilePath BaseName() const WARN_UNUSED_RESULT; 250 251 // Returns ".jpg" for path "C:\pics\jojo.jpg", or an empty string if 252 // the file has no extension. If non-empty, Extension() will always start 253 // with precisely one ".". The following code should always work regardless 254 // of the value of path. For common double-extensions like .tar.gz and 255 // .user.js, this method returns the combined extension. For a single 256 // component, use FinalExtension(). 257 // new_path = path.RemoveExtension().value().append(path.Extension()); 258 // ASSERT(new_path == path.value()); 259 // NOTE: this is different from the original file_util implementation which 260 // returned the extension without a leading "." ("jpg" instead of ".jpg") 261 StringType Extension() const WARN_UNUSED_RESULT; 262 263 // Returns the path's file extension, as in Extension(), but will 264 // never return a double extension. 265 // 266 // TODO(davidben): Check all our extension-sensitive code to see if 267 // we can rename this to Extension() and the other to something like 268 // LongExtension(), defaulting to short extensions and leaving the 269 // long "extensions" to logic like base::GetUniquePathNumber(). 270 StringType FinalExtension() const WARN_UNUSED_RESULT; 271 272 // Returns "C:\pics\jojo" for path "C:\pics\jojo.jpg" 273 // NOTE: this is slightly different from the similar file_util implementation 274 // which returned simply 'jojo'. 275 FilePath RemoveExtension() const WARN_UNUSED_RESULT; 276 277 // Removes the path's file extension, as in RemoveExtension(), but 278 // ignores double extensions. 279 FilePath RemoveFinalExtension() const WARN_UNUSED_RESULT; 280 281 // Inserts |suffix| after the file name portion of |path| but before the 282 // extension. Returns "" if BaseName() == "." or "..". 283 // Examples: 284 // path == "C:\pics\jojo.jpg" suffix == " (1)", returns "C:\pics\jojo (1).jpg" 285 // path == "jojo.jpg" suffix == " (1)", returns "jojo (1).jpg" 286 // path == "C:\pics\jojo" suffix == " (1)", returns "C:\pics\jojo (1)" 287 // path == "C:\pics.old\jojo" suffix == " (1)", returns "C:\pics.old\jojo (1)" 288 FilePath InsertBeforeExtension( 289 StringPieceType suffix) const WARN_UNUSED_RESULT; 290 FilePath InsertBeforeExtensionASCII( 291 StringPiece suffix) const WARN_UNUSED_RESULT; 292 293 // Adds |extension| to |file_name|. Returns the current FilePath if 294 // |extension| is empty. Returns "" if BaseName() == "." or "..". 295 FilePath AddExtension(StringPieceType extension) const WARN_UNUSED_RESULT; 296 297 // Replaces the extension of |file_name| with |extension|. If |file_name| 298 // does not have an extension, then |extension| is added. If |extension| is 299 // empty, then the extension is removed from |file_name|. 300 // Returns "" if BaseName() == "." or "..". 301 FilePath ReplaceExtension(StringPieceType extension) const WARN_UNUSED_RESULT; 302 303 // Returns true if the file path matches the specified extension. The test is 304 // case insensitive. Don't forget the leading period if appropriate. 305 bool MatchesExtension(StringPieceType extension) const; 306 307 // Returns a FilePath by appending a separator and the supplied path 308 // component to this object's path. Append takes care to avoid adding 309 // excessive separators if this object's path already ends with a separator. 310 // If this object's path is kCurrentDirectory, a new FilePath corresponding 311 // only to |component| is returned. |component| must be a relative path; 312 // it is an error to pass an absolute path. 313 FilePath Append(StringPieceType component) const WARN_UNUSED_RESULT; 314 FilePath Append(const FilePath& component) const WARN_UNUSED_RESULT; 315 316 // Although Windows StringType is std::wstring, since the encoding it uses for 317 // paths is well defined, it can handle ASCII path components as well. 318 // Mac uses UTF8, and since ASCII is a subset of that, it works there as well. 319 // On Linux, although it can use any 8-bit encoding for paths, we assume that 320 // ASCII is a valid subset, regardless of the encoding, since many operating 321 // system paths will always be ASCII. 322 FilePath AppendASCII(StringPiece component) const WARN_UNUSED_RESULT; 323 324 // Returns true if this FilePath contains an absolute path. On Windows, an 325 // absolute path begins with either a drive letter specification followed by 326 // a separator character, or with two separator characters. On POSIX 327 // platforms, an absolute path begins with a separator character. 328 bool IsAbsolute() const; 329 330 // Returns true if the patch ends with a path separator character. 331 bool EndsWithSeparator() const WARN_UNUSED_RESULT; 332 333 // Returns a copy of this FilePath that ends with a trailing separator. If 334 // the input path is empty, an empty FilePath will be returned. 335 FilePath AsEndingWithSeparator() const WARN_UNUSED_RESULT; 336 337 // Returns a copy of this FilePath that does not end with a trailing 338 // separator. 339 FilePath StripTrailingSeparators() const WARN_UNUSED_RESULT; 340 341 // Returns true if this FilePath contains an attempt to reference a parent 342 // directory (e.g. has a path component that is ".."). 343 bool ReferencesParent() const; 344 345 // Return a Unicode human-readable version of this path. 346 // Warning: you can *not*, in general, go from a display name back to a real 347 // path. Only use this when displaying paths to users, not just when you 348 // want to stuff a string16 into some other API. 349 string16 LossyDisplayName() const; 350 351 // Return the path as ASCII, or the empty string if the path is not ASCII. 352 // This should only be used for cases where the FilePath is representing a 353 // known-ASCII filename. 354 std::string MaybeAsASCII() const; 355 356 // Return the path as UTF-8. 357 // 358 // This function is *unsafe* as there is no way to tell what encoding is 359 // used in file names on POSIX systems other than Mac and Chrome OS, 360 // although UTF-8 is practically used everywhere these days. To mitigate 361 // the encoding issue, this function internally calls 362 // SysNativeMBToWide() on POSIX systems other than Mac and Chrome OS, 363 // per assumption that the current locale's encoding is used in file 364 // names, but this isn't a perfect solution. 365 // 366 // Once it becomes safe to to stop caring about non-UTF-8 file names, 367 // the SysNativeMBToWide() hack will be removed from the code, along 368 // with "Unsafe" in the function name. 369 std::string AsUTF8Unsafe() const; 370 371 // Similar to AsUTF8Unsafe, but returns UTF-16 instead. 372 string16 AsUTF16Unsafe() const; 373 374 // Returns a FilePath object from a path name in UTF-8. This function 375 // should only be used for cases where you are sure that the input 376 // string is UTF-8. 377 // 378 // Like AsUTF8Unsafe(), this function is unsafe. This function 379 // internally calls SysWideToNativeMB() on POSIX systems other than Mac 380 // and Chrome OS, to mitigate the encoding issue. See the comment at 381 // AsUTF8Unsafe() for details. 382 static FilePath FromUTF8Unsafe(StringPiece utf8); 383 384 // Similar to FromUTF8Unsafe, but accepts UTF-16 instead. 385 static FilePath FromUTF16Unsafe(StringPiece16 utf16); 386 387 void WriteToPickle(Pickle* pickle) const; 388 bool ReadFromPickle(PickleIterator* iter); 389 390 // Normalize all path separators to backslash on Windows 391 // (if FILE_PATH_USES_WIN_SEPARATORS is true), or do nothing on POSIX systems. 392 FilePath NormalizePathSeparators() const; 393 394 // Normalize all path separattors to given type on Windows 395 // (if FILE_PATH_USES_WIN_SEPARATORS is true), or do nothing on POSIX systems. 396 FilePath NormalizePathSeparatorsTo(CharType separator) const; 397 398 // Compare two strings in the same way the file system does. 399 // Note that these always ignore case, even on file systems that are case- 400 // sensitive. If case-sensitive comparison is ever needed, add corresponding 401 // methods here. 402 // The methods are written as a static method so that they can also be used 403 // on parts of a file path, e.g., just the extension. 404 // CompareIgnoreCase() returns -1, 0 or 1 for less-than, equal-to and 405 // greater-than respectively. 406 static int CompareIgnoreCase(StringPieceType string1, 407 StringPieceType string2); CompareEqualIgnoreCase(StringPieceType string1,StringPieceType string2)408 static bool CompareEqualIgnoreCase(StringPieceType string1, 409 StringPieceType string2) { 410 return CompareIgnoreCase(string1, string2) == 0; 411 } CompareLessIgnoreCase(StringPieceType string1,StringPieceType string2)412 static bool CompareLessIgnoreCase(StringPieceType string1, 413 StringPieceType string2) { 414 return CompareIgnoreCase(string1, string2) < 0; 415 } 416 417 #if defined(OS_MACOSX) 418 // Returns the string in the special canonical decomposed form as defined for 419 // HFS, which is close to, but not quite, decomposition form D. See 420 // http://developer.apple.com/mac/library/technotes/tn/tn1150.html#UnicodeSubtleties 421 // for further comments. 422 // Returns the epmty string if the conversion failed. 423 static StringType GetHFSDecomposedForm(StringPieceType string); 424 425 // Special UTF-8 version of FastUnicodeCompare. Cf: 426 // http://developer.apple.com/mac/library/technotes/tn/tn1150.html#StringComparisonAlgorithm 427 // IMPORTANT: The input strings must be in the special HFS decomposed form! 428 // (cf. above GetHFSDecomposedForm method) 429 static int HFSFastUnicodeCompare(StringPieceType string1, 430 StringPieceType string2); 431 #endif 432 433 #if defined(OS_ANDROID) 434 // On android, file selection dialog can return a file with content uri 435 // scheme(starting with content://). Content uri needs to be opened with 436 // ContentResolver to guarantee that the app has appropriate permissions 437 // to access it. 438 // Returns true if the path is a content uri, or false otherwise. 439 bool IsContentUri() const; 440 #endif 441 442 private: 443 // Remove trailing separators from this object. If the path is absolute, it 444 // will never be stripped any more than to refer to the absolute root 445 // directory, so "////" will become "/", not "". A leading pair of 446 // separators is never stripped, to support alternate roots. This is used to 447 // support UNC paths on Windows. 448 void StripTrailingSeparatorsInternal(); 449 450 StringType path_; 451 }; 452 453 BASE_EXPORT std::ostream& operator<<(std::ostream& out, 454 const FilePath& file_path); 455 456 } // namespace base 457 458 // Macros for string literal initialization of FilePath::CharType[], and for 459 // using a FilePath::CharType[] in a printf-style format string. 460 #if defined(OS_WIN) 461 #define FILE_PATH_LITERAL(x) L ## x 462 #define PRFilePath "ls" 463 #elif defined(OS_POSIX) || defined(OS_FUCHSIA) 464 #define FILE_PATH_LITERAL(x) x 465 #define PRFilePath "s" 466 #endif // OS_WIN 467 468 namespace std { 469 470 template <> 471 struct hash<base::FilePath> { 472 typedef base::FilePath argument_type; 473 typedef std::size_t result_type; 474 result_type operator()(argument_type const& f) const { 475 return hash<base::FilePath::StringType>()(f.value()); 476 } 477 }; 478 479 } // namespace std 480 481 #endif // BASE_FILES_FILE_PATH_H_ 482