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.
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
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
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.
24 // For more arcane bits of path trivia, see below.
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.
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.
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.
56 // Paths can't contain NULs as a precaution agaist premature truncation.
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:
63 // | const FilePath::CharType kLogFileName[] = FILE_PATH_LITERAL("log.txt");
65 // | void Function() {
66 // | FilePath log_file_path(kLogFileName);
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
75 // This is a very common source of bugs, please try to keep this in mind.
77 // ARCANE BITS OF PATH TRIVIA
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.
86 // The Open Group Base Specifications Issue 7, sections 3.266 ("Pathname")
87 // and 4.12 ("Pathname Resolution"), available at:
88 // http://www.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap03.html#tag_03_266
89 // http://www.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap04.html#tag_04_12
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.
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
102 #ifndef BASE_FILES_FILE_PATH_H_
103 #define BASE_FILES_FILE_PATH_H_
109 #include "base/base_export.h"
110 #include "base/containers/hash_tables.h"
111 #include "base/strings/string16.h"
112 #include "base/strings/string_piece.h" // For implicit conversions.
113 #include "build/build_config.h"
115 // Windows-style drive letter support and pathname separator characters can be
116 // enabled and disabled independently, to aid testing. These #defines are
117 // here so that the same setting can be used in both the implementation and
120 #define FILE_PATH_USES_DRIVE_LETTERS
121 #define FILE_PATH_USES_WIN_SEPARATORS
125 class PickleIterator;
129 // An abstraction to isolate users from the differences between native
130 // pathnames on different platforms.
131 class BASE_EXPORT FilePath {
133 #if defined(OS_POSIX)
134 // On most platforms, native pathnames are char arrays, and the encoding
135 // may or may not be specified. On Mac OS X, native pathnames are encoded
137 typedef std::string StringType;
138 #elif defined(OS_WIN)
139 // On Windows, for Unicode-aware applications, native pathnames are wchar_t
140 // arrays encoded in UTF-16.
141 typedef std::wstring StringType;
144 typedef StringType::value_type CharType;
146 // Null-terminated array of separators used to separate components in
147 // hierarchical paths. Each character in this array is a valid separator,
148 // but kSeparators[0] is treated as the canonical separator and will be used
149 // when composing pathnames.
150 static const CharType kSeparators[];
152 // arraysize(kSeparators).
153 static const size_t kSeparatorsLength;
155 // A special path component meaning "this directory."
156 static const CharType kCurrentDirectory[];
158 // A special path component meaning "the parent directory."
159 static const CharType kParentDirectory[];
161 // The character used to identify a file extension.
162 static const CharType kExtensionSeparator;
165 FilePath(const FilePath& that);
166 explicit FilePath(const StringType& path);
168 FilePath& operator=(const FilePath& that);
170 bool operator==(const FilePath& that) const;
172 bool operator!=(const FilePath& that) const;
174 // Required for some STL containers and operations
175 bool operator<(const FilePath& that) const {
176 return path_ < that.path_;
179 const StringType& value() const { return path_; }
181 bool empty() const { return path_.empty(); }
183 void clear() { path_.clear(); }
185 // Returns true if |character| is in kSeparators.
186 static bool IsSeparator(CharType character);
188 // Returns a vector of all of the components of the provided path. It is
189 // equivalent to calling DirName().value() on the path's root component,
190 // and BaseName().value() on each child component.
192 // To make sure this is lossless so we can differentiate absolute and
193 // relative paths, the root slash will be included even though no other
194 // slashes will be. The precise behavior is:
196 // Posix: "/foo/bar" -> [ "/", "foo", "bar" ]
197 // Windows: "C:\foo\bar" -> [ "C:", "\\", "foo", "bar" ]
198 void GetComponents(std::vector<FilePath::StringType>* components) const;
200 // Returns true if this FilePath is a strict parent of the |child|. Absolute
201 // and relative paths are accepted i.e. is /foo parent to /foo/bar and
202 // is foo parent to foo/bar. Does not convert paths to absolute, follow
203 // symlinks or directory navigation (e.g. ".."). A path is *NOT* its own
205 bool IsParent(const FilePath& child) const;
207 // If IsParent(child) holds, appends to path (if non-NULL) the
208 // relative path to child and returns true. For example, if parent
209 // holds "/Users/johndoe/Library/Application Support", child holds
210 // "/Users/johndoe/Library/Application Support/Google/Chrome/Default", and
211 // *path holds "/Users/johndoe/Library/Caches", then after
212 // parent.AppendRelativePath(child, path) is called *path will hold
213 // "/Users/johndoe/Library/Caches/Google/Chrome/Default". Otherwise,
215 bool AppendRelativePath(const FilePath& child, FilePath* path) const;
217 // Returns a FilePath corresponding to the directory containing the path
218 // named by this object, stripping away the file component. If this object
219 // only contains one component, returns a FilePath identifying
220 // kCurrentDirectory. If this object already refers to the root directory,
221 // returns a FilePath identifying the root directory.
222 FilePath DirName() const WARN_UNUSED_RESULT;
224 // Returns a FilePath corresponding to the last path component of this
225 // object, either a file or a directory. If this object already refers to
226 // the root directory, returns a FilePath identifying the root directory;
227 // this is the only situation in which BaseName will return an absolute path.
228 FilePath BaseName() const WARN_UNUSED_RESULT;
230 // Returns ".jpg" for path "C:\pics\jojo.jpg", or an empty string if
231 // the file has no extension. If non-empty, Extension() will always start
232 // with precisely one ".". The following code should always work regardless
233 // of the value of path. For common double-extensions like .tar.gz and
234 // .user.js, this method returns the combined extension. For a single
235 // component, use FinalExtension().
236 // new_path = path.RemoveExtension().value().append(path.Extension());
237 // ASSERT(new_path == path.value());
238 // NOTE: this is different from the original file_util implementation which
239 // returned the extension without a leading "." ("jpg" instead of ".jpg")
240 StringType Extension() const;
242 // Returns the path's file extension, as in Extension(), but will
243 // never return a double extension.
245 // TODO(davidben): Check all our extension-sensitive code to see if
246 // we can rename this to Extension() and the other to something like
247 // LongExtension(), defaulting to short extensions and leaving the
248 // long "extensions" to logic like base::GetUniquePathNumber().
249 StringType FinalExtension() const;
251 // Returns "C:\pics\jojo" for path "C:\pics\jojo.jpg"
252 // NOTE: this is slightly different from the similar file_util implementation
253 // which returned simply 'jojo'.
254 FilePath RemoveExtension() const WARN_UNUSED_RESULT;
256 // Removes the path's file extension, as in RemoveExtension(), but
257 // ignores double extensions.
258 FilePath RemoveFinalExtension() const WARN_UNUSED_RESULT;
260 // Inserts |suffix| after the file name portion of |path| but before the
261 // extension. Returns "" if BaseName() == "." or "..".
263 // path == "C:\pics\jojo.jpg" suffix == " (1)", returns "C:\pics\jojo (1).jpg"
264 // path == "jojo.jpg" suffix == " (1)", returns "jojo (1).jpg"
265 // path == "C:\pics\jojo" suffix == " (1)", returns "C:\pics\jojo (1)"
266 // path == "C:\pics.old\jojo" suffix == " (1)", returns "C:\pics.old\jojo (1)"
267 FilePath InsertBeforeExtension(
268 const StringType& suffix) const WARN_UNUSED_RESULT;
269 FilePath InsertBeforeExtensionASCII(
270 const base::StringPiece& suffix) const WARN_UNUSED_RESULT;
272 // Adds |extension| to |file_name|. Returns the current FilePath if
273 // |extension| is empty. Returns "" if BaseName() == "." or "..".
274 FilePath AddExtension(
275 const StringType& extension) const WARN_UNUSED_RESULT;
277 // Replaces the extension of |file_name| with |extension|. If |file_name|
278 // does not have an extension, then |extension| is added. If |extension| is
279 // empty, then the extension is removed from |file_name|.
280 // Returns "" if BaseName() == "." or "..".
281 FilePath ReplaceExtension(
282 const StringType& extension) const WARN_UNUSED_RESULT;
284 // Returns true if the file path matches the specified extension. The test is
285 // case insensitive. Don't forget the leading period if appropriate.
286 bool MatchesExtension(const StringType& extension) const;
288 // Returns a FilePath by appending a separator and the supplied path
289 // component to this object's path. Append takes care to avoid adding
290 // excessive separators if this object's path already ends with a separator.
291 // If this object's path is kCurrentDirectory, a new FilePath corresponding
292 // only to |component| is returned. |component| must be a relative path;
293 // it is an error to pass an absolute path.
294 FilePath Append(const StringType& component) const WARN_UNUSED_RESULT;
295 FilePath Append(const FilePath& component) const WARN_UNUSED_RESULT;
297 // Although Windows StringType is std::wstring, since the encoding it uses for
298 // paths is well defined, it can handle ASCII path components as well.
299 // Mac uses UTF8, and since ASCII is a subset of that, it works there as well.
300 // On Linux, although it can use any 8-bit encoding for paths, we assume that
301 // ASCII is a valid subset, regardless of the encoding, since many operating
302 // system paths will always be ASCII.
303 FilePath AppendASCII(const base::StringPiece& component)
304 const WARN_UNUSED_RESULT;
306 // Returns true if this FilePath contains an absolute path. On Windows, an
307 // absolute path begins with either a drive letter specification followed by
308 // a separator character, or with two separator characters. On POSIX
309 // platforms, an absolute path begins with a separator character.
310 bool IsAbsolute() const;
312 // Returns true if the patch ends with a path separator character.
313 bool EndsWithSeparator() const WARN_UNUSED_RESULT;
315 // Returns a copy of this FilePath that ends with a trailing separator. If
316 // the input path is empty, an empty FilePath will be returned.
317 FilePath AsEndingWithSeparator() const WARN_UNUSED_RESULT;
319 // Returns a copy of this FilePath that does not end with a trailing
321 FilePath StripTrailingSeparators() const WARN_UNUSED_RESULT;
323 // Returns true if this FilePath contains an attempt to reference a parent
324 // directory (e.g. has a path component that is "..").
325 bool ReferencesParent() const;
327 // Return a Unicode human-readable version of this path.
328 // Warning: you can *not*, in general, go from a display name back to a real
329 // path. Only use this when displaying paths to users, not just when you
330 // want to stuff a string16 into some other API.
331 string16 LossyDisplayName() const;
333 // Return the path as ASCII, or the empty string if the path is not ASCII.
334 // This should only be used for cases where the FilePath is representing a
335 // known-ASCII filename.
336 std::string MaybeAsASCII() const;
338 // Return the path as UTF-8.
340 // This function is *unsafe* as there is no way to tell what encoding is
341 // used in file names on POSIX systems other than Mac and Chrome OS,
342 // although UTF-8 is practically used everywhere these days. To mitigate
343 // the encoding issue, this function internally calls
344 // SysNativeMBToWide() on POSIX systems other than Mac and Chrome OS,
345 // per assumption that the current locale's encoding is used in file
346 // names, but this isn't a perfect solution.
348 // Once it becomes safe to to stop caring about non-UTF-8 file names,
349 // the SysNativeMBToWide() hack will be removed from the code, along
350 // with "Unsafe" in the function name.
351 std::string AsUTF8Unsafe() const;
353 // Similar to AsUTF8Unsafe, but returns UTF-16 instead.
354 string16 AsUTF16Unsafe() const;
356 // Returns a FilePath object from a path name in UTF-8. This function
357 // should only be used for cases where you are sure that the input
360 // Like AsUTF8Unsafe(), this function is unsafe. This function
361 // internally calls SysWideToNativeMB() on POSIX systems other than Mac
362 // and Chrome OS, to mitigate the encoding issue. See the comment at
363 // AsUTF8Unsafe() for details.
364 static FilePath FromUTF8Unsafe(const std::string& utf8);
366 // Similar to FromUTF8Unsafe, but accepts UTF-16 instead.
367 static FilePath FromUTF16Unsafe(const string16& utf16);
369 void WriteToPickle(Pickle* pickle) const;
370 bool ReadFromPickle(PickleIterator* iter);
372 // Normalize all path separators to backslash on Windows
373 // (if FILE_PATH_USES_WIN_SEPARATORS is true), or do nothing on POSIX systems.
374 FilePath NormalizePathSeparators() const;
376 // Normalize all path separattors to given type on Windows
377 // (if FILE_PATH_USES_WIN_SEPARATORS is true), or do nothing on POSIX systems.
378 FilePath NormalizePathSeparatorsTo(CharType separator) const;
380 // Compare two strings in the same way the file system does.
381 // Note that these always ignore case, even on file systems that are case-
382 // sensitive. If case-sensitive comparison is ever needed, add corresponding
384 // The methods are written as a static method so that they can also be used
385 // on parts of a file path, e.g., just the extension.
386 // CompareIgnoreCase() returns -1, 0 or 1 for less-than, equal-to and
387 // greater-than respectively.
388 static int CompareIgnoreCase(const StringType& string1,
389 const StringType& string2);
390 static bool CompareEqualIgnoreCase(const StringType& string1,
391 const StringType& string2) {
392 return CompareIgnoreCase(string1, string2) == 0;
394 static bool CompareLessIgnoreCase(const StringType& string1,
395 const StringType& string2) {
396 return CompareIgnoreCase(string1, string2) < 0;
399 #if defined(OS_MACOSX)
400 // Returns the string in the special canonical decomposed form as defined for
401 // HFS, which is close to, but not quite, decomposition form D. See
402 // http://developer.apple.com/mac/library/technotes/tn/tn1150.html#UnicodeSubtleties
403 // for further comments.
404 // Returns the epmty string if the conversion failed.
405 static StringType GetHFSDecomposedForm(const FilePath::StringType& string);
407 // Special UTF-8 version of FastUnicodeCompare. Cf:
408 // http://developer.apple.com/mac/library/technotes/tn/tn1150.html#StringComparisonAlgorithm
409 // IMPORTANT: The input strings must be in the special HFS decomposed form!
410 // (cf. above GetHFSDecomposedForm method)
411 static int HFSFastUnicodeCompare(const StringType& string1,
412 const StringType& string2);
415 #if defined(OS_ANDROID)
416 // On android, file selection dialog can return a file with content uri
417 // scheme(starting with content://). Content uri needs to be opened with
418 // ContentResolver to guarantee that the app has appropriate permissions
420 // Returns true if the path is a content uri, or false otherwise.
421 bool IsContentUri() const;
425 // Remove trailing separators from this object. If the path is absolute, it
426 // will never be stripped any more than to refer to the absolute root
427 // directory, so "////" will become "/", not "". A leading pair of
428 // separators is never stripped, to support alternate roots. This is used to
429 // support UNC paths on Windows.
430 void StripTrailingSeparatorsInternal();
437 // This is required by googletest to print a readable output on test failures.
438 BASE_EXPORT extern void PrintTo(const base::FilePath& path, std::ostream* out);
440 // Macros for string literal initialization of FilePath::CharType[], and for
441 // using a FilePath::CharType[] in a printf-style format string.
442 #if defined(OS_POSIX)
443 #define FILE_PATH_LITERAL(x) x
444 #define PRFilePath "s"
445 #define PRFilePathLiteral "%s"
446 #elif defined(OS_WIN)
447 #define FILE_PATH_LITERAL(x) L ## x
448 #define PRFilePath "ls"
449 #define PRFilePathLiteral L"%ls"
452 // Provide a hash function so that hash_sets and maps can contain FilePath
454 namespace BASE_HASH_NAMESPACE {
457 struct hash<base::FilePath> {
458 size_t operator()(const base::FilePath& f) const {
459 return hash<base::FilePath::StringType>()(f.value());
463 } // namespace BASE_HASH_NAMESPACE
465 #endif // BASE_FILES_FILE_PATH_H_