1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/security/sandbox/chromium/base/files/file_path.h Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,458 @@
1.4 +// Copyright (c) 2012 The Chromium Authors. All rights reserved.
1.5 +// Use of this source code is governed by a BSD-style license that can be
1.6 +// found in the LICENSE file.
1.7 +
1.8 +// FilePath is a container for pathnames stored in a platform's native string
1.9 +// type, providing containers for manipulation in according with the
1.10 +// platform's conventions for pathnames. It supports the following path
1.11 +// types:
1.12 +//
1.13 +// POSIX Windows
1.14 +// --------------- ----------------------------------
1.15 +// Fundamental type char[] wchar_t[]
1.16 +// Encoding unspecified* UTF-16
1.17 +// Separator / \, tolerant of /
1.18 +// Drive letters no case-insensitive A-Z followed by :
1.19 +// Alternate root // (surprise!) \\, for UNC paths
1.20 +//
1.21 +// * The encoding need not be specified on POSIX systems, although some
1.22 +// POSIX-compliant systems do specify an encoding. Mac OS X uses UTF-8.
1.23 +// Chrome OS also uses UTF-8.
1.24 +// Linux does not specify an encoding, but in practice, the locale's
1.25 +// character set may be used.
1.26 +//
1.27 +// For more arcane bits of path trivia, see below.
1.28 +//
1.29 +// FilePath objects are intended to be used anywhere paths are. An
1.30 +// application may pass FilePath objects around internally, masking the
1.31 +// underlying differences between systems, only differing in implementation
1.32 +// where interfacing directly with the system. For example, a single
1.33 +// OpenFile(const FilePath &) function may be made available, allowing all
1.34 +// callers to operate without regard to the underlying implementation. On
1.35 +// POSIX-like platforms, OpenFile might wrap fopen, and on Windows, it might
1.36 +// wrap _wfopen_s, perhaps both by calling file_path.value().c_str(). This
1.37 +// allows each platform to pass pathnames around without requiring conversions
1.38 +// between encodings, which has an impact on performance, but more imporantly,
1.39 +// has an impact on correctness on platforms that do not have well-defined
1.40 +// encodings for pathnames.
1.41 +//
1.42 +// Several methods are available to perform common operations on a FilePath
1.43 +// object, such as determining the parent directory (DirName), isolating the
1.44 +// final path component (BaseName), and appending a relative pathname string
1.45 +// to an existing FilePath object (Append). These methods are highly
1.46 +// recommended over attempting to split and concatenate strings directly.
1.47 +// These methods are based purely on string manipulation and knowledge of
1.48 +// platform-specific pathname conventions, and do not consult the filesystem
1.49 +// at all, making them safe to use without fear of blocking on I/O operations.
1.50 +// These methods do not function as mutators but instead return distinct
1.51 +// instances of FilePath objects, and are therefore safe to use on const
1.52 +// objects. The objects themselves are safe to share between threads.
1.53 +//
1.54 +// To aid in initialization of FilePath objects from string literals, a
1.55 +// FILE_PATH_LITERAL macro is provided, which accounts for the difference
1.56 +// between char[]-based pathnames on POSIX systems and wchar_t[]-based
1.57 +// pathnames on Windows.
1.58 +//
1.59 +// Paths can't contain NULs as a precaution agaist premature truncation.
1.60 +//
1.61 +// Because a FilePath object should not be instantiated at the global scope,
1.62 +// instead, use a FilePath::CharType[] and initialize it with
1.63 +// FILE_PATH_LITERAL. At runtime, a FilePath object can be created from the
1.64 +// character array. Example:
1.65 +//
1.66 +// | const FilePath::CharType kLogFileName[] = FILE_PATH_LITERAL("log.txt");
1.67 +// |
1.68 +// | void Function() {
1.69 +// | FilePath log_file_path(kLogFileName);
1.70 +// | [...]
1.71 +// | }
1.72 +//
1.73 +// WARNING: FilePaths should ALWAYS be displayed with LTR directionality, even
1.74 +// when the UI language is RTL. This means you always need to pass filepaths
1.75 +// through base::i18n::WrapPathWithLTRFormatting() before displaying it in the
1.76 +// RTL UI.
1.77 +//
1.78 +// This is a very common source of bugs, please try to keep this in mind.
1.79 +//
1.80 +// ARCANE BITS OF PATH TRIVIA
1.81 +//
1.82 +// - A double leading slash is actually part of the POSIX standard. Systems
1.83 +// are allowed to treat // as an alternate root, as Windows does for UNC
1.84 +// (network share) paths. Most POSIX systems don't do anything special
1.85 +// with two leading slashes, but FilePath handles this case properly
1.86 +// in case it ever comes across such a system. FilePath needs this support
1.87 +// for Windows UNC paths, anyway.
1.88 +// References:
1.89 +// The Open Group Base Specifications Issue 7, sections 3.266 ("Pathname")
1.90 +// and 4.12 ("Pathname Resolution"), available at:
1.91 +// http://www.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap03.html#tag_03_266
1.92 +// http://www.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap04.html#tag_04_12
1.93 +//
1.94 +// - Windows treats c:\\ the same way it treats \\. This was intended to
1.95 +// allow older applications that require drive letters to support UNC paths
1.96 +// like \\server\share\path, by permitting c:\\server\share\path as an
1.97 +// equivalent. Since the OS treats these paths specially, FilePath needs
1.98 +// to do the same. Since Windows can use either / or \ as the separator,
1.99 +// FilePath treats c://, c:\\, //, and \\ all equivalently.
1.100 +// Reference:
1.101 +// The Old New Thing, "Why is a drive letter permitted in front of UNC
1.102 +// paths (sometimes)?", available at:
1.103 +// http://blogs.msdn.com/oldnewthing/archive/2005/11/22/495740.aspx
1.104 +
1.105 +#ifndef BASE_FILES_FILE_PATH_H_
1.106 +#define BASE_FILES_FILE_PATH_H_
1.107 +
1.108 +#include <stddef.h>
1.109 +#include <string>
1.110 +#include <vector>
1.111 +
1.112 +#include "base/base_export.h"
1.113 +#include "base/compiler_specific.h"
1.114 +#include "base/containers/hash_tables.h"
1.115 +#include "base/strings/string16.h"
1.116 +#include "base/strings/string_piece.h" // For implicit conversions.
1.117 +#include "build/build_config.h"
1.118 +
1.119 +// Windows-style drive letter support and pathname separator characters can be
1.120 +// enabled and disabled independently, to aid testing. These #defines are
1.121 +// here so that the same setting can be used in both the implementation and
1.122 +// in the unit test.
1.123 +#if defined(OS_WIN)
1.124 +#define FILE_PATH_USES_DRIVE_LETTERS
1.125 +#define FILE_PATH_USES_WIN_SEPARATORS
1.126 +#endif // OS_WIN
1.127 +
1.128 +class Pickle;
1.129 +class PickleIterator;
1.130 +
1.131 +namespace base {
1.132 +
1.133 +// An abstraction to isolate users from the differences between native
1.134 +// pathnames on different platforms.
1.135 +class BASE_EXPORT FilePath {
1.136 + public:
1.137 +#if defined(OS_POSIX)
1.138 + // On most platforms, native pathnames are char arrays, and the encoding
1.139 + // may or may not be specified. On Mac OS X, native pathnames are encoded
1.140 + // in UTF-8.
1.141 + typedef std::string StringType;
1.142 +#elif defined(OS_WIN)
1.143 + // On Windows, for Unicode-aware applications, native pathnames are wchar_t
1.144 + // arrays encoded in UTF-16.
1.145 + typedef std::wstring StringType;
1.146 +#endif // OS_WIN
1.147 +
1.148 + typedef StringType::value_type CharType;
1.149 +
1.150 + // Null-terminated array of separators used to separate components in
1.151 + // hierarchical paths. Each character in this array is a valid separator,
1.152 + // but kSeparators[0] is treated as the canonical separator and will be used
1.153 + // when composing pathnames.
1.154 + static const CharType kSeparators[];
1.155 +
1.156 + // arraysize(kSeparators).
1.157 + static const size_t kSeparatorsLength;
1.158 +
1.159 + // A special path component meaning "this directory."
1.160 + static const CharType kCurrentDirectory[];
1.161 +
1.162 + // A special path component meaning "the parent directory."
1.163 + static const CharType kParentDirectory[];
1.164 +
1.165 + // The character used to identify a file extension.
1.166 + static const CharType kExtensionSeparator;
1.167 +
1.168 + FilePath();
1.169 + FilePath(const FilePath& that);
1.170 + explicit FilePath(const StringType& path);
1.171 + ~FilePath();
1.172 + FilePath& operator=(const FilePath& that);
1.173 +
1.174 + bool operator==(const FilePath& that) const;
1.175 +
1.176 + bool operator!=(const FilePath& that) const;
1.177 +
1.178 + // Required for some STL containers and operations
1.179 + bool operator<(const FilePath& that) const {
1.180 + return path_ < that.path_;
1.181 + }
1.182 +
1.183 + const StringType& value() const { return path_; }
1.184 +
1.185 + bool empty() const { return path_.empty(); }
1.186 +
1.187 + void clear() { path_.clear(); }
1.188 +
1.189 + // Returns true if |character| is in kSeparators.
1.190 + static bool IsSeparator(CharType character);
1.191 +
1.192 + // Returns a vector of all of the components of the provided path. It is
1.193 + // equivalent to calling DirName().value() on the path's root component,
1.194 + // and BaseName().value() on each child component.
1.195 + void GetComponents(std::vector<FilePath::StringType>* components) const;
1.196 +
1.197 + // Returns true if this FilePath is a strict parent of the |child|. Absolute
1.198 + // and relative paths are accepted i.e. is /foo parent to /foo/bar and
1.199 + // is foo parent to foo/bar. Does not convert paths to absolute, follow
1.200 + // symlinks or directory navigation (e.g. ".."). A path is *NOT* its own
1.201 + // parent.
1.202 + bool IsParent(const FilePath& child) const;
1.203 +
1.204 + // If IsParent(child) holds, appends to path (if non-NULL) the
1.205 + // relative path to child and returns true. For example, if parent
1.206 + // holds "/Users/johndoe/Library/Application Support", child holds
1.207 + // "/Users/johndoe/Library/Application Support/Google/Chrome/Default", and
1.208 + // *path holds "/Users/johndoe/Library/Caches", then after
1.209 + // parent.AppendRelativePath(child, path) is called *path will hold
1.210 + // "/Users/johndoe/Library/Caches/Google/Chrome/Default". Otherwise,
1.211 + // returns false.
1.212 + bool AppendRelativePath(const FilePath& child, FilePath* path) const;
1.213 +
1.214 + // Returns a FilePath corresponding to the directory containing the path
1.215 + // named by this object, stripping away the file component. If this object
1.216 + // only contains one component, returns a FilePath identifying
1.217 + // kCurrentDirectory. If this object already refers to the root directory,
1.218 + // returns a FilePath identifying the root directory.
1.219 + FilePath DirName() const WARN_UNUSED_RESULT;
1.220 +
1.221 + // Returns a FilePath corresponding to the last path component of this
1.222 + // object, either a file or a directory. If this object already refers to
1.223 + // the root directory, returns a FilePath identifying the root directory;
1.224 + // this is the only situation in which BaseName will return an absolute path.
1.225 + FilePath BaseName() const WARN_UNUSED_RESULT;
1.226 +
1.227 + // Returns ".jpg" for path "C:\pics\jojo.jpg", or an empty string if
1.228 + // the file has no extension. If non-empty, Extension() will always start
1.229 + // with precisely one ".". The following code should always work regardless
1.230 + // of the value of path.
1.231 + // new_path = path.RemoveExtension().value().append(path.Extension());
1.232 + // ASSERT(new_path == path.value());
1.233 + // NOTE: this is different from the original file_util implementation which
1.234 + // returned the extension without a leading "." ("jpg" instead of ".jpg")
1.235 + StringType Extension() const;
1.236 +
1.237 + // Returns "C:\pics\jojo" for path "C:\pics\jojo.jpg"
1.238 + // NOTE: this is slightly different from the similar file_util implementation
1.239 + // which returned simply 'jojo'.
1.240 + FilePath RemoveExtension() const WARN_UNUSED_RESULT;
1.241 +
1.242 + // Inserts |suffix| after the file name portion of |path| but before the
1.243 + // extension. Returns "" if BaseName() == "." or "..".
1.244 + // Examples:
1.245 + // path == "C:\pics\jojo.jpg" suffix == " (1)", returns "C:\pics\jojo (1).jpg"
1.246 + // path == "jojo.jpg" suffix == " (1)", returns "jojo (1).jpg"
1.247 + // path == "C:\pics\jojo" suffix == " (1)", returns "C:\pics\jojo (1)"
1.248 + // path == "C:\pics.old\jojo" suffix == " (1)", returns "C:\pics.old\jojo (1)"
1.249 + FilePath InsertBeforeExtension(
1.250 + const StringType& suffix) const WARN_UNUSED_RESULT;
1.251 + FilePath InsertBeforeExtensionASCII(
1.252 + const base::StringPiece& suffix) const WARN_UNUSED_RESULT;
1.253 +
1.254 + // Adds |extension| to |file_name|. Returns the current FilePath if
1.255 + // |extension| is empty. Returns "" if BaseName() == "." or "..".
1.256 + FilePath AddExtension(
1.257 + const StringType& extension) const WARN_UNUSED_RESULT;
1.258 +
1.259 + // Replaces the extension of |file_name| with |extension|. If |file_name|
1.260 + // does not have an extension, then |extension| is added. If |extension| is
1.261 + // empty, then the extension is removed from |file_name|.
1.262 + // Returns "" if BaseName() == "." or "..".
1.263 + FilePath ReplaceExtension(
1.264 + const StringType& extension) const WARN_UNUSED_RESULT;
1.265 +
1.266 + // Returns true if the file path matches the specified extension. The test is
1.267 + // case insensitive. Don't forget the leading period if appropriate.
1.268 + bool MatchesExtension(const StringType& extension) const;
1.269 +
1.270 + // Returns a FilePath by appending a separator and the supplied path
1.271 + // component to this object's path. Append takes care to avoid adding
1.272 + // excessive separators if this object's path already ends with a separator.
1.273 + // If this object's path is kCurrentDirectory, a new FilePath corresponding
1.274 + // only to |component| is returned. |component| must be a relative path;
1.275 + // it is an error to pass an absolute path.
1.276 + FilePath Append(const StringType& component) const WARN_UNUSED_RESULT;
1.277 + FilePath Append(const FilePath& component) const WARN_UNUSED_RESULT;
1.278 +
1.279 + // Although Windows StringType is std::wstring, since the encoding it uses for
1.280 + // paths is well defined, it can handle ASCII path components as well.
1.281 + // Mac uses UTF8, and since ASCII is a subset of that, it works there as well.
1.282 + // On Linux, although it can use any 8-bit encoding for paths, we assume that
1.283 + // ASCII is a valid subset, regardless of the encoding, since many operating
1.284 + // system paths will always be ASCII.
1.285 + FilePath AppendASCII(const base::StringPiece& component)
1.286 + const WARN_UNUSED_RESULT;
1.287 +
1.288 + // Returns true if this FilePath contains an absolute path. On Windows, an
1.289 + // absolute path begins with either a drive letter specification followed by
1.290 + // a separator character, or with two separator characters. On POSIX
1.291 + // platforms, an absolute path begins with a separator character.
1.292 + bool IsAbsolute() const;
1.293 +
1.294 + // Returns true if the patch ends with a path separator character.
1.295 + bool EndsWithSeparator() const WARN_UNUSED_RESULT;
1.296 +
1.297 + // Returns a copy of this FilePath that ends with a trailing separator. If
1.298 + // the input path is empty, an empty FilePath will be returned.
1.299 + FilePath AsEndingWithSeparator() const WARN_UNUSED_RESULT;
1.300 +
1.301 + // Returns a copy of this FilePath that does not end with a trailing
1.302 + // separator.
1.303 + FilePath StripTrailingSeparators() const WARN_UNUSED_RESULT;
1.304 +
1.305 + // Returns true if this FilePath contains any attempt to reference a parent
1.306 + // directory (i.e. has a path component that is ".."
1.307 + bool ReferencesParent() const;
1.308 +
1.309 + // Return a Unicode human-readable version of this path.
1.310 + // Warning: you can *not*, in general, go from a display name back to a real
1.311 + // path. Only use this when displaying paths to users, not just when you
1.312 + // want to stuff a string16 into some other API.
1.313 + string16 LossyDisplayName() const;
1.314 +
1.315 + // Return the path as ASCII, or the empty string if the path is not ASCII.
1.316 + // This should only be used for cases where the FilePath is representing a
1.317 + // known-ASCII filename.
1.318 + std::string MaybeAsASCII() const;
1.319 +
1.320 + // Return the path as UTF-8.
1.321 + //
1.322 + // This function is *unsafe* as there is no way to tell what encoding is
1.323 + // used in file names on POSIX systems other than Mac and Chrome OS,
1.324 + // although UTF-8 is practically used everywhere these days. To mitigate
1.325 + // the encoding issue, this function internally calls
1.326 + // SysNativeMBToWide() on POSIX systems other than Mac and Chrome OS,
1.327 + // per assumption that the current locale's encoding is used in file
1.328 + // names, but this isn't a perfect solution.
1.329 + //
1.330 + // Once it becomes safe to to stop caring about non-UTF-8 file names,
1.331 + // the SysNativeMBToWide() hack will be removed from the code, along
1.332 + // with "Unsafe" in the function name.
1.333 + std::string AsUTF8Unsafe() const;
1.334 +
1.335 + // Similar to AsUTF8Unsafe, but returns UTF-16 instead.
1.336 + string16 AsUTF16Unsafe() const;
1.337 +
1.338 + // Older Chromium code assumes that paths are always wstrings.
1.339 + // This function converts wstrings to FilePaths, and is
1.340 + // useful to smooth porting that old code to the FilePath API.
1.341 + // It has "Hack" its name so people feel bad about using it.
1.342 + // http://code.google.com/p/chromium/issues/detail?id=24672
1.343 + //
1.344 + // If you are trying to be a good citizen and remove these, ask yourself:
1.345 + // - Am I interacting with other Chrome code that deals with files? Then
1.346 + // try to convert the API into using FilePath.
1.347 + // - Am I interacting with OS-native calls? Then use value() to get at an
1.348 + // OS-native string format.
1.349 + // - Am I using well-known file names, like "config.ini"? Then use the
1.350 + // ASCII functions (we require paths to always be supersets of ASCII).
1.351 + // - Am I displaying a string to the user in some UI? Then use the
1.352 + // LossyDisplayName() function, but keep in mind that you can't
1.353 + // ever use the result of that again as a path.
1.354 + static FilePath FromWStringHack(const std::wstring& wstring);
1.355 +
1.356 + // Returns a FilePath object from a path name in UTF-8. This function
1.357 + // should only be used for cases where you are sure that the input
1.358 + // string is UTF-8.
1.359 + //
1.360 + // Like AsUTF8Unsafe(), this function is unsafe. This function
1.361 + // internally calls SysWideToNativeMB() on POSIX systems other than Mac
1.362 + // and Chrome OS, to mitigate the encoding issue. See the comment at
1.363 + // AsUTF8Unsafe() for details.
1.364 + static FilePath FromUTF8Unsafe(const std::string& utf8);
1.365 +
1.366 + // Similar to FromUTF8Unsafe, but accepts UTF-16 instead.
1.367 + static FilePath FromUTF16Unsafe(const string16& utf16);
1.368 +
1.369 + void WriteToPickle(Pickle* pickle) const;
1.370 + bool ReadFromPickle(PickleIterator* iter);
1.371 +
1.372 + // Normalize all path separators to backslash on Windows
1.373 + // (if FILE_PATH_USES_WIN_SEPARATORS is true), or do nothing on POSIX systems.
1.374 + FilePath NormalizePathSeparators() const;
1.375 +
1.376 + // Compare two strings in the same way the file system does.
1.377 + // Note that these always ignore case, even on file systems that are case-
1.378 + // sensitive. If case-sensitive comparison is ever needed, add corresponding
1.379 + // methods here.
1.380 + // The methods are written as a static method so that they can also be used
1.381 + // on parts of a file path, e.g., just the extension.
1.382 + // CompareIgnoreCase() returns -1, 0 or 1 for less-than, equal-to and
1.383 + // greater-than respectively.
1.384 + static int CompareIgnoreCase(const StringType& string1,
1.385 + const StringType& string2);
1.386 + static bool CompareEqualIgnoreCase(const StringType& string1,
1.387 + const StringType& string2) {
1.388 + return CompareIgnoreCase(string1, string2) == 0;
1.389 + }
1.390 + static bool CompareLessIgnoreCase(const StringType& string1,
1.391 + const StringType& string2) {
1.392 + return CompareIgnoreCase(string1, string2) < 0;
1.393 + }
1.394 +
1.395 +#if defined(OS_MACOSX)
1.396 + // Returns the string in the special canonical decomposed form as defined for
1.397 + // HFS, which is close to, but not quite, decomposition form D. See
1.398 + // http://developer.apple.com/mac/library/technotes/tn/tn1150.html#UnicodeSubtleties
1.399 + // for further comments.
1.400 + // Returns the epmty string if the conversion failed.
1.401 + static StringType GetHFSDecomposedForm(const FilePath::StringType& string);
1.402 +
1.403 + // Special UTF-8 version of FastUnicodeCompare. Cf:
1.404 + // http://developer.apple.com/mac/library/technotes/tn/tn1150.html#StringComparisonAlgorithm
1.405 + // IMPORTANT: The input strings must be in the special HFS decomposed form!
1.406 + // (cf. above GetHFSDecomposedForm method)
1.407 + static int HFSFastUnicodeCompare(const StringType& string1,
1.408 + const StringType& string2);
1.409 +#endif
1.410 +
1.411 + private:
1.412 + // Remove trailing separators from this object. If the path is absolute, it
1.413 + // will never be stripped any more than to refer to the absolute root
1.414 + // directory, so "////" will become "/", not "". A leading pair of
1.415 + // separators is never stripped, to support alternate roots. This is used to
1.416 + // support UNC paths on Windows.
1.417 + void StripTrailingSeparatorsInternal();
1.418 +
1.419 + StringType path_;
1.420 +};
1.421 +
1.422 +} // namespace base
1.423 +
1.424 +// This is required by googletest to print a readable output on test failures.
1.425 +BASE_EXPORT extern void PrintTo(const base::FilePath& path, std::ostream* out);
1.426 +
1.427 +// Macros for string literal initialization of FilePath::CharType[], and for
1.428 +// using a FilePath::CharType[] in a printf-style format string.
1.429 +#if defined(OS_POSIX)
1.430 +#define FILE_PATH_LITERAL(x) x
1.431 +#define PRFilePath "s"
1.432 +#define PRFilePathLiteral "%s"
1.433 +#elif defined(OS_WIN)
1.434 +#define FILE_PATH_LITERAL(x) L ## x
1.435 +#define PRFilePath "ls"
1.436 +#define PRFilePathLiteral L"%ls"
1.437 +#endif // OS_WIN
1.438 +
1.439 +// Provide a hash function so that hash_sets and maps can contain FilePath
1.440 +// objects.
1.441 +namespace BASE_HASH_NAMESPACE {
1.442 +#if defined(COMPILER_GCC)
1.443 +
1.444 +template<>
1.445 +struct hash<base::FilePath> {
1.446 + size_t operator()(const base::FilePath& f) const {
1.447 + return hash<base::FilePath::StringType>()(f.value());
1.448 + }
1.449 +};
1.450 +
1.451 +#elif defined(COMPILER_MSVC)
1.452 +
1.453 +inline size_t hash_value(const base::FilePath& f) {
1.454 + return hash_value(f.value());
1.455 +}
1.456 +
1.457 +#endif // COMPILER
1.458 +
1.459 +} // namespace BASE_HASH_NAMESPACE
1.460 +
1.461 +#endif // BASE_FILES_FILE_PATH_H_
