[go: nahoru, domu]
Google Git
Sign in
chromium / chromium / src / bebb2006102cff96ecbe5e8ed40d3d52bc274206 / . / base / files / file_path.cc
blob: a1b8019d394ca8194811733b8b1d2f9b6ebca344 [file] [log] [blame]
kinuko@chromium.orgd9034ed22012-02-10 02:04:40[diff] [blame]1// Copyright (c) 2012 The Chromium Authors. All rights reserved.
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]2// Use of this source code is governed by a BSD-style license that can be
3// found in the LICENSE file.
4
brettw@chromium.org57999812013-02-24 05:40:52[diff] [blame]5#include "base/files/file_path.h"
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]6
Hans Wennborg5f5b451a2021-03-30 18:31:27[diff] [blame]7// file_path.h is a widely included header and its size has significant impact
8// on build time. Try not to raise this limit unless necessary. See
9// https://chromium.googlesource.com/chromium/src/+/HEAD/docs/wmax_tokens.md
10#ifndef NACL_TC_REV
11#pragma clang max_tokens_here 340000
12#endif
13
jhawkins@chromium.org2edc2862011-04-04 18:04:37[diff] [blame]14#include <string.h>
15#include <algorithm>
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]16
Hans Wennborgc3cffa62020-04-27 10:09:12[diff] [blame]17#include "base/check_op.h"
avi543540e2015-12-24 05:15:32[diff] [blame]18#include "base/macros.h"
finnur@chromium.org44ec9b3b2010-02-03 01:17:26[diff] [blame]19#include "base/pickle.h"
Jan Wilken Dörriefb1d7b662020-02-04 11:10:23[diff] [blame]20#include "base/strings/strcat.h"
tfarina@chromium.orgeb62f7262013-03-30 14:29:00[diff] [blame]21#include "base/strings/string_piece.h"
avi@chromium.org251cd6e52013-06-11 13:36:37[diff] [blame]22#include "base/strings/string_util.h"
brettw@chromium.org9fe1a5b2013-02-07 19:18:03[diff] [blame]23#include "base/strings/sys_string_conversions.h"
avi@chromium.orga4ea1f12013-06-07 18:37:07[diff] [blame]24#include "base/strings/utf_string_conversions.h"
Alexander Timin7372d5362021-03-09 00:55:31[diff] [blame]25#include "base/trace_event/base_tracing.h"
avi543540e2015-12-24 05:15:32[diff] [blame]26#include "build/build_config.h"
evanm@google.com640517f2008-10-30 23:54:04[diff] [blame]27
Avi Drissman5b286372020-07-28 21:59:38[diff] [blame]28#if defined(OS_APPLE)
brettw@chromium.orgdf0ca6c82010-10-17 04:09:06[diff] [blame]29#include "base/mac/scoped_cftyperef.h"
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]30#include "base/third_party/icu/icu_utf.h"
31#endif
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]32
jhawkins@chromium.org2edc2862011-04-04 18:04:37[diff] [blame]33#if defined(OS_WIN)
34#include <windows.h>
Cliff Smolinskyc5c52102019-05-03 20:51:54[diff] [blame]35#include "base/win/win_util.h"
Avi Drissman5b286372020-07-28 21:59:38[diff] [blame]36#elif defined(OS_APPLE)
jhawkins@chromium.org2edc2862011-04-04 18:04:37[diff] [blame]37#include <CoreFoundation/CoreFoundation.h>
38#endif
39
brettw@chromium.org04af979a2013-02-16 04:12:26[diff] [blame]40namespace base {
41
brettw89365dc2015-06-16 05:52:47[diff] [blame]42using StringType = FilePath::StringType;
43using StringPieceType = FilePath::StringPieceType;
erikkay@google.com235c17a2009-01-09 18:26:19[diff] [blame]44
mark@chromium.org1906c032008-12-08 16:49:08[diff] [blame]45namespace {
46
Lei Zhang3a430222020-01-10 22:48:18[diff] [blame]47const char* const kCommonDoubleExtensionSuffixes[] = {"gz", "xz", "bz2", "z",
48 "bz"};
thestig073d514d2014-10-21 03:11:21[diff] [blame]49const char* const kCommonDoubleExtensions[] = { "user.js" };
estade@chromium.org34322912010-07-27 21:24:39[diff] [blame]50
aedla@chromium.orgaeae59f2013-01-28 13:47:55[diff] [blame]51const FilePath::CharType kStringTerminator = FILE_PATH_LITERAL('\0');
52
mark@chromium.org1906c032008-12-08 16:49:08[diff] [blame]53// If this FilePath contains a drive letter specification, returns the
54// position of the last character of the drive letter specification,
55// otherwise returns npos. This can only be true on Windows, when a pathname
56// begins with a letter followed by a colon. On other platforms, this always
57// returns npos.
brettw89365dc2015-06-16 05:52:47[diff] [blame]58StringPieceType::size_type FindDriveLetter(StringPieceType path) {
mark@chromium.org1906c032008-12-08 16:49:08[diff] [blame]59#if defined(FILE_PATH_USES_DRIVE_LETTERS)
60 // This is dependent on an ASCII-based character set, but that's a
61 // reasonable assumption. iswalpha can be too inclusive here.
62 if (path.length() >= 2 && path[1] == L':' &&
63 ((path[0] >= L'A' && path[0] <= L'Z') ||
64 (path[0] >= L'a' && path[0] <= L'z'))) {
65 return 1;
66 }
mark@chromium.org1906c032008-12-08 16:49:08[diff] [blame]67#endif // FILE_PATH_USES_DRIVE_LETTERS
estade@chromium.org34322912010-07-27 21:24:39[diff] [blame]68 return StringType::npos;
mark@chromium.org1906c032008-12-08 16:49:08[diff] [blame]69}
70
rafaelw@chromium.org62e78f52009-07-01 01:03:34[diff] [blame]71#if defined(FILE_PATH_USES_DRIVE_LETTERS)
brettw89365dc2015-06-16 05:52:47[diff] [blame]72bool EqualDriveLetterCaseInsensitive(StringPieceType a, StringPieceType b) {
rafaelw@chromium.org62e78f52009-07-01 01:03:34[diff] [blame]73 size_t a_letter_pos = FindDriveLetter(a);
74 size_t b_letter_pos = FindDriveLetter(b);
75
estade@chromium.org34322912010-07-27 21:24:39[diff] [blame]76 if (a_letter_pos == StringType::npos || b_letter_pos == StringType::npos)
rafaelw@chromium.org62e78f52009-07-01 01:03:34[diff] [blame]77 return a == b;
78
brettw89365dc2015-06-16 05:52:47[diff] [blame]79 StringPieceType a_letter(a.substr(0, a_letter_pos + 1));
80 StringPieceType b_letter(b.substr(0, b_letter_pos + 1));
Jan Wilken Dörrieaf27f302020-05-17 16:35:28[diff] [blame]81 if (!StartsWith(a_letter, b_letter, CompareCase::INSENSITIVE_ASCII))
rafaelw@chromium.org62e78f52009-07-01 01:03:34[diff] [blame]82 return false;
83
brettw89365dc2015-06-16 05:52:47[diff] [blame]84 StringPieceType a_rest(a.substr(a_letter_pos + 1));
85 StringPieceType b_rest(b.substr(b_letter_pos + 1));
rafaelw@chromium.org62e78f52009-07-01 01:03:34[diff] [blame]86 return a_rest == b_rest;
87}
88#endif // defined(FILE_PATH_USES_DRIVE_LETTERS)
89
brettw89365dc2015-06-16 05:52:47[diff] [blame]90bool IsPathAbsolute(StringPieceType path) {
mark@chromium.org1906c032008-12-08 16:49:08[diff] [blame]91#if defined(FILE_PATH_USES_DRIVE_LETTERS)
estade@chromium.org34322912010-07-27 21:24:39[diff] [blame]92 StringType::size_type letter = FindDriveLetter(path);
93 if (letter != StringType::npos) {
mark@chromium.org1906c032008-12-08 16:49:08[diff] [blame]94 // Look for a separator right after the drive specification.
95 return path.length() > letter + 1 &&
96 FilePath::IsSeparator(path[letter + 1]);
97 }
98 // Look for a pair of leading separators.
99 return path.length() > 1 &&
100 FilePath::IsSeparator(path[0]) && FilePath::IsSeparator(path[1]);
101#else // FILE_PATH_USES_DRIVE_LETTERS
102 // Look for a separator in the first position.
103 return path.length() > 0 && FilePath::IsSeparator(path[0]);
104#endif // FILE_PATH_USES_DRIVE_LETTERS
105}
106
estade@chromium.org34322912010-07-27 21:24:39[diff] [blame]107bool AreAllSeparators(const StringType& input) {
jdoerrie6c6229352018-10-22 15:55:43[diff] [blame]108 for (auto it : input) {
109 if (!FilePath::IsSeparator(it))
rafaelw@chromium.org0e4f68db2009-06-24 21:28:30[diff] [blame]110 return false;
111 }
112
113 return true;
114}
115
estade@chromium.org34322912010-07-27 21:24:39[diff] [blame]116// Find the position of the '.' that separates the extension from the rest
117// of the file name. The position is relative to BaseName(), not value().
davidben@chromium.org18cbba0a2013-12-09 17:06:52[diff] [blame]118// Returns npos if it can't find an extension.
119StringType::size_type FinalExtensionSeparatorPosition(const StringType& path) {
estade@chromium.org34322912010-07-27 21:24:39[diff] [blame]120 // Special case "." and ".."
121 if (path == FilePath::kCurrentDirectory || path == FilePath::kParentDirectory)
122 return StringType::npos;
123
davidben@chromium.org18cbba0a2013-12-09 17:06:52[diff] [blame]124 return path.rfind(FilePath::kExtensionSeparator);
125}
126
127// Same as above, but allow a second extension component of up to 4
128// characters when the rightmost extension component is a common double
129// extension (gz, bz2, Z). For example, foo.tar.gz or foo.tar.Z would have
130// extension components of '.tar.gz' and '.tar.Z' respectively.
131StringType::size_type ExtensionSeparatorPosition(const StringType& path) {
132 const StringType::size_type last_dot = FinalExtensionSeparatorPosition(path);
estade@chromium.org34322912010-07-27 21:24:39[diff] [blame]133
134 // No extension, or the extension is the whole filename.
135 if (last_dot == StringType::npos || last_dot == 0U)
136 return last_dot;
137
estade@chromium.org34322912010-07-27 21:24:39[diff] [blame]138 const StringType::size_type penultimate_dot =
139 path.rfind(FilePath::kExtensionSeparator, last_dot - 1);
140 const StringType::size_type last_separator =
141 path.find_last_of(FilePath::kSeparators, last_dot - 1,
scottmg@chromium.orgf39e03812013-05-21 09:44:02[diff] [blame]142 FilePath::kSeparatorsLength - 1);
aa@chromium.org278f9ac612012-08-26 12:11:10[diff] [blame]143
144 if (penultimate_dot == StringType::npos ||
145 (last_separator != StringType::npos &&
146 penultimate_dot < last_separator)) {
147 return last_dot;
148 }
149
jdoerrie6c6229352018-10-22 15:55:43[diff] [blame]150 for (auto* i : kCommonDoubleExtensions) {
aa@chromium.org278f9ac612012-08-26 12:11:10[diff] [blame]151 StringType extension(path, penultimate_dot + 1);
Jan Wilken Dörrieaf27f302020-05-17 16:35:28[diff] [blame]152 if (LowerCaseEqualsASCII(extension, i))
aa@chromium.org278f9ac612012-08-26 12:11:10[diff] [blame]153 return penultimate_dot;
154 }
155
156 StringType extension(path, last_dot + 1);
jdoerrie6c6229352018-10-22 15:55:43[diff] [blame]157 for (auto* i : kCommonDoubleExtensionSuffixes) {
Jan Wilken Dörrieaf27f302020-05-17 16:35:28[diff] [blame]158 if (LowerCaseEqualsASCII(extension, i)) {
aa@chromium.org278f9ac612012-08-26 12:11:10[diff] [blame]159 if ((last_dot - penultimate_dot) <= 5U &&
160 (last_dot - penultimate_dot) > 1U) {
161 return penultimate_dot;
162 }
163 }
estade@chromium.org34322912010-07-27 21:24:39[diff] [blame]164 }
165
166 return last_dot;
167}
168
rdevlin.cronin@chromium.org922828b32012-04-25 01:53:44[diff] [blame]169// Returns true if path is "", ".", or "..".
170bool IsEmptyOrSpecialCase(const StringType& path) {
171 // Special cases "", ".", and ".."
172 if (path.empty() || path == FilePath::kCurrentDirectory ||
173 path == FilePath::kParentDirectory) {
174 return true;
175 }
176
177 return false;
178}
179
mark@chromium.org1906c032008-12-08 16:49:08[diff] [blame]180} // namespace
181
Chris Watkinsbb7211c2017-11-29 07:16:38[diff] [blame]182FilePath::FilePath() = default;
erg@chromium.org201366472010-07-16 17:22:49[diff] [blame]183
Chris Watkinsbb7211c2017-11-29 07:16:38[diff] [blame]184FilePath::FilePath(const FilePath& that) = default;
brettw4e53ee162017-03-29 18:41:18[diff] [blame]185FilePath::FilePath(FilePath&& that) noexcept = default;
erg@chromium.org201366472010-07-16 17:22:49[diff] [blame]186
Jan Wilken Dörriec15b4bc2020-01-30 07:04:12[diff] [blame]187FilePath::FilePath(StringPieceType path) : path_(path) {
aedla@chromium.orgaeae59f2013-01-28 13:47:55[diff] [blame]188 StringType::size_type nul_pos = path_.find(kStringTerminator);
189 if (nul_pos != StringType::npos)
190 path_.erase(nul_pos, StringType::npos);
erg@chromium.org201366472010-07-16 17:22:49[diff] [blame]191}
192
Chris Watkinsbb7211c2017-11-29 07:16:38[diff] [blame]193FilePath::~FilePath() = default;
erg@chromium.org201366472010-07-16 17:22:49[diff] [blame]194
Chris Watkinsbb7211c2017-11-29 07:16:38[diff] [blame]195FilePath& FilePath::operator=(const FilePath& that) = default;
erg@chromium.org201366472010-07-16 17:22:49[diff] [blame]196
Stephan Hartmannc65b4d02020-05-27 18:46:40[diff] [blame]197FilePath& FilePath::operator=(FilePath&& that) noexcept = default;
dmurph590f8e62017-01-12 02:54:12[diff] [blame]198
erg@google.com9989c9bb2011-01-07 20:23:43[diff] [blame]199bool FilePath::operator==(const FilePath& that) const {
200#if defined(FILE_PATH_USES_DRIVE_LETTERS)
201 return EqualDriveLetterCaseInsensitive(this->path_, that.path_);
202#else // defined(FILE_PATH_USES_DRIVE_LETTERS)
203 return path_ == that.path_;
204#endif // defined(FILE_PATH_USES_DRIVE_LETTERS)
205}
206
207bool FilePath::operator!=(const FilePath& that) const {
208#if defined(FILE_PATH_USES_DRIVE_LETTERS)
209 return !EqualDriveLetterCaseInsensitive(this->path_, that.path_);
210#else // defined(FILE_PATH_USES_DRIVE_LETTERS)
211 return path_ != that.path_;
212#endif // defined(FILE_PATH_USES_DRIVE_LETTERS)
213}
214
Mike Bjorge2cd6ba02017-05-08 22:45:20[diff] [blame]215std::ostream& operator<<(std::ostream& out, const FilePath& file_path) {
216 return out << file_path.value();
217}
218
erg@google.com9989c9bb2011-01-07 20:23:43[diff] [blame]219// static
estade@chromium.orgb9e04f02008-11-27 04:03:57[diff] [blame]220bool FilePath::IsSeparator(CharType character) {
scottmg@chromium.orgf39e03812013-05-21 09:44:02[diff] [blame]221 for (size_t i = 0; i < kSeparatorsLength - 1; ++i) {
estade@chromium.orgb9e04f02008-11-27 04:03:57[diff] [blame]222 if (character == kSeparators[i]) {
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]223 return true;
224 }
225 }
226
227 return false;
228}
229
estade@chromium.org34322912010-07-27 21:24:39[diff] [blame]230void FilePath::GetComponents(std::vector<StringType>* components) const {
rafaelw@chromium.org0e4f68db2009-06-24 21:28:30[diff] [blame]231 DCHECK(components);
232 if (!components)
233 return;
234 components->clear();
235 if (value().empty())
236 return;
237
estade@chromium.org34322912010-07-27 21:24:39[diff] [blame]238 std::vector<StringType> ret_val;
rafaelw@chromium.org0e4f68db2009-06-24 21:28:30[diff] [blame]239 FilePath current = *this;
240 FilePath base;
241
242 // Capture path components.
243 while (current != current.DirName()) {
244 base = current.BaseName();
245 if (!AreAllSeparators(base.value()))
246 ret_val.push_back(base.value());
247 current = current.DirName();
248 }
249
250 // Capture root, if any.
251 base = current.BaseName();
252 if (!base.value().empty() && base.value() != kCurrentDirectory)
253 ret_val.push_back(current.BaseName().value());
254
255 // Capture drive letter, if any.
256 FilePath dir = current.DirName();
257 StringType::size_type letter = FindDriveLetter(dir.value());
estade@chromium.org34322912010-07-27 21:24:39[diff] [blame]258 if (letter != StringType::npos) {
259 ret_val.push_back(StringType(dir.value(), 0, letter + 1));
rafaelw@chromium.org0e4f68db2009-06-24 21:28:30[diff] [blame]260 }
261
estade@chromium.org34322912010-07-27 21:24:39[diff] [blame]262 *components = std::vector<StringType>(ret_val.rbegin(), ret_val.rend());
rafaelw@chromium.org0e4f68db2009-06-24 21:28:30[diff] [blame]263}
264
265bool FilePath::IsParent(const FilePath& child) const {
Ivan Kotenkova16212a52017-11-08 12:37:33[diff] [blame]266 return AppendRelativePath(child, nullptr);
mark@chromium.org405a64b2009-09-16 21:03:44[diff] [blame]267}
268
269bool FilePath::AppendRelativePath(const FilePath& child,
270 FilePath* path) const {
estade@chromium.org34322912010-07-27 21:24:39[diff] [blame]271 std::vector<StringType> parent_components;
272 std::vector<StringType> child_components;
rafaelw@chromium.org0e4f68db2009-06-24 21:28:30[diff] [blame]273 GetComponents(&parent_components);
274 child.GetComponents(&child_components);
275
erg@google.com6775e40a2011-03-04 21:03:47[diff] [blame]276 if (parent_components.empty() ||
277 parent_components.size() >= child_components.size())
rafaelw@chromium.org0e4f68db2009-06-24 21:28:30[diff] [blame]278 return false;
279
estade@chromium.org34322912010-07-27 21:24:39[diff] [blame]280 std::vector<StringType>::const_iterator parent_comp =
rafaelw@chromium.org0e4f68db2009-06-24 21:28:30[diff] [blame]281 parent_components.begin();
estade@chromium.org34322912010-07-27 21:24:39[diff] [blame]282 std::vector<StringType>::const_iterator child_comp =
rafaelw@chromium.org0e4f68db2009-06-24 21:28:30[diff] [blame]283 child_components.begin();
rafaelw@chromium.org62e78f52009-07-01 01:03:34[diff] [blame]284
285#if defined(FILE_PATH_USES_DRIVE_LETTERS)
286 // Windows can access case sensitive filesystems, so component
287 // comparisions must be case sensitive, but drive letters are
288 // never case sensitive.
estade@chromium.org34322912010-07-27 21:24:39[diff] [blame]289 if ((FindDriveLetter(*parent_comp) != StringType::npos) &&
290 (FindDriveLetter(*child_comp) != StringType::npos)) {
Jan Wilken Dörrieaf27f302020-05-17 16:35:28[diff] [blame]291 if (!StartsWith(*parent_comp, *child_comp, CompareCase::INSENSITIVE_ASCII))
rafaelw@chromium.org62e78f52009-07-01 01:03:34[diff] [blame]292 return false;
293 ++parent_comp;
294 ++child_comp;
295 }
296#endif // defined(FILE_PATH_USES_DRIVE_LETTERS)
297
Joel Hockeybc3a2c32021-03-24 08:42:46[diff] [blame]298 // The first 2 components for network paths are [<2-Separators>, <hostname>].
299 // Use case-insensitive comparison for the hostname.
300 // https://tools.ietf.org/html/rfc3986#section-3.2.2
301 if (IsNetwork() && parent_components.size() > 1) {
302 if (*parent_comp++ != *child_comp++ ||
303 !base::EqualsCaseInsensitiveASCII(*parent_comp++, *child_comp++)) {
304 return false;
305 }
306 }
307
rafaelw@chromium.org0e4f68db2009-06-24 21:28:30[diff] [blame]308 while (parent_comp != parent_components.end()) {
309 if (*parent_comp != *child_comp)
310 return false;
311 ++parent_comp;
312 ++child_comp;
313 }
314
Ivan Kotenkova16212a52017-11-08 12:37:33[diff] [blame]315 if (path != nullptr) {
mark@chromium.org405a64b2009-09-16 21:03:44[diff] [blame]316 for (; child_comp != child_components.end(); ++child_comp) {
317 *path = path->Append(*child_comp);
318 }
319 }
rafaelw@chromium.org0e4f68db2009-06-24 21:28:30[diff] [blame]320 return true;
321}
322
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]323// libgen's dirname and basename aren't guaranteed to be thread-safe and aren't
avi@google.com690a99c2009-01-06 16:48:45[diff] [blame]324// guaranteed to not modify their input strings, and in fact are implemented
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]325// differently in this regard on different platforms. Don't use them, but
326// adhere to their behavior.
327FilePath FilePath::DirName() const {
328 FilePath new_path(path_);
avi@google.com0b733222008-12-11 14:55:12[diff] [blame]329 new_path.StripTrailingSeparatorsInternal();
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]330
331 // The drive letter, if any, always needs to remain in the output. If there
332 // is no drive letter, as will always be the case on platforms which do not
333 // support drive letters, letter will be npos, or -1, so the comparisons and
334 // resizes below using letter will still be valid.
mark@chromium.org1906c032008-12-08 16:49:08[diff] [blame]335 StringType::size_type letter = FindDriveLetter(new_path.path_);
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]336
337 StringType::size_type last_separator =
338 new_path.path_.find_last_of(kSeparators, StringType::npos,
scottmg@chromium.orgf39e03812013-05-21 09:44:02[diff] [blame]339 kSeparatorsLength - 1);
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]340 if (last_separator == StringType::npos) {
341 // path_ is in the current directory.
342 new_path.path_.resize(letter + 1);
343 } else if (last_separator == letter + 1) {
344 // path_ is in the root directory.
345 new_path.path_.resize(letter + 2);
346 } else if (last_separator == letter + 2 &&
347 IsSeparator(new_path.path_[letter + 1])) {
348 // path_ is in "//" (possibly with a drive letter); leave the double
349 // separator intact indicating alternate root.
350 new_path.path_.resize(letter + 3);
351 } else if (last_separator != 0) {
352 // path_ is somewhere else, trim the basename.
353 new_path.path_.resize(last_separator);
354 }
355
avi@google.com0b733222008-12-11 14:55:12[diff] [blame]356 new_path.StripTrailingSeparatorsInternal();
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]357 if (!new_path.path_.length())
358 new_path.path_ = kCurrentDirectory;
359
360 return new_path;
361}
362
estade@chromium.org5e207882009-01-05 23:59:36[diff] [blame]363FilePath FilePath::BaseName() const {
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]364 FilePath new_path(path_);
avi@google.com0b733222008-12-11 14:55:12[diff] [blame]365 new_path.StripTrailingSeparatorsInternal();
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]366
367 // The drive letter, if any, is always stripped.
mark@chromium.org1906c032008-12-08 16:49:08[diff] [blame]368 StringType::size_type letter = FindDriveLetter(new_path.path_);
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]369 if (letter != StringType::npos) {
370 new_path.path_.erase(0, letter + 1);
371 }
372
373 // Keep everything after the final separator, but if the pathname is only
374 // one character and it's a separator, leave it alone.
375 StringType::size_type last_separator =
376 new_path.path_.find_last_of(kSeparators, StringType::npos,
scottmg@chromium.orgf39e03812013-05-21 09:44:02[diff] [blame]377 kSeparatorsLength - 1);
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]378 if (last_separator != StringType::npos &&
379 last_separator < new_path.path_.length() - 1) {
380 new_path.path_.erase(0, last_separator + 1);
381 }
382
estade@chromium.org5e207882009-01-05 23:59:36[diff] [blame]383 return new_path;
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]384}
385
estade@chromium.org34322912010-07-27 21:24:39[diff] [blame]386StringType FilePath::Extension() const {
387 FilePath base(BaseName());
388 const StringType::size_type dot = ExtensionSeparatorPosition(base.path_);
389 if (dot == StringType::npos)
erikkay@google.com235c17a2009-01-09 18:26:19[diff] [blame]390 return StringType();
391
estade@chromium.org34322912010-07-27 21:24:39[diff] [blame]392 return base.path_.substr(dot, StringType::npos);
erikkay@google.com235c17a2009-01-09 18:26:19[diff] [blame]393}
394
davidben@chromium.org18cbba0a2013-12-09 17:06:52[diff] [blame]395StringType FilePath::FinalExtension() const {
396 FilePath base(BaseName());
397 const StringType::size_type dot = FinalExtensionSeparatorPosition(base.path_);
398 if (dot == StringType::npos)
399 return StringType();
400
401 return base.path_.substr(dot, StringType::npos);
402}
403
erikkay@google.com235c17a2009-01-09 18:26:19[diff] [blame]404FilePath FilePath::RemoveExtension() const {
estade@chromium.org34322912010-07-27 21:24:39[diff] [blame]405 if (Extension().empty())
406 return *this;
407
408 const StringType::size_type dot = ExtensionSeparatorPosition(path_);
409 if (dot == StringType::npos)
410 return *this;
411
412 return FilePath(path_.substr(0, dot));
erikkay@google.com235c17a2009-01-09 18:26:19[diff] [blame]413}
414
davidben@chromium.org18cbba0a2013-12-09 17:06:52[diff] [blame]415FilePath FilePath::RemoveFinalExtension() const {
416 if (FinalExtension().empty())
417 return *this;
418
419 const StringType::size_type dot = FinalExtensionSeparatorPosition(path_);
420 if (dot == StringType::npos)
421 return *this;
422
423 return FilePath(path_.substr(0, dot));
424}
425
brettw89365dc2015-06-16 05:52:47[diff] [blame]426FilePath FilePath::InsertBeforeExtension(StringPieceType suffix) const {
erikkay@google.com235c17a2009-01-09 18:26:19[diff] [blame]427 if (suffix.empty())
428 return FilePath(path_);
429
rdevlin.cronin@chromium.org922828b32012-04-25 01:53:44[diff] [blame]430 if (IsEmptyOrSpecialCase(BaseName().value()))
erikkay@google.com235c17a2009-01-09 18:26:19[diff] [blame]431 return FilePath();
432
Jan Wilken Dörriefb1d7b662020-02-04 11:10:23[diff] [blame]433 return FilePath(
434 base::StrCat({RemoveExtension().value(), suffix, Extension()}));
erikkay@google.com235c17a2009-01-09 18:26:19[diff] [blame]435}
436
brettw89365dc2015-06-16 05:52:47[diff] [blame]437FilePath FilePath::InsertBeforeExtensionASCII(StringPiece suffix)
tony@chromium.org8a16266e2009-09-10 21:08:39[diff] [blame]438 const {
estade@chromium.org73e23482009-08-13 00:26:58[diff] [blame]439 DCHECK(IsStringASCII(suffix));
440#if defined(OS_WIN)
Jan Wilken Dörrie9bb441d2019-11-01 17:48:40[diff] [blame]441 return InsertBeforeExtension(UTF8ToWide(suffix));
Fabrice de Gans-Riberi306871de2018-05-16 19:38:39[diff] [blame]442#elif defined(OS_POSIX) || defined(OS_FUCHSIA)
brettw89365dc2015-06-16 05:52:47[diff] [blame]443 return InsertBeforeExtension(suffix);
estade@chromium.org73e23482009-08-13 00:26:58[diff] [blame]444#endif
445}
446
brettw89365dc2015-06-16 05:52:47[diff] [blame]447FilePath FilePath::AddExtension(StringPieceType extension) const {
rdevlin.cronin@chromium.org922828b32012-04-25 01:53:44[diff] [blame]448 if (IsEmptyOrSpecialCase(BaseName().value()))
erikkay@google.com235c17a2009-01-09 18:26:19[diff] [blame]449 return FilePath();
450
rdevlin.cronin@chromium.org922828b32012-04-25 01:53:44[diff] [blame]451 // If the new extension is "" or ".", then just return the current FilePath.
brettw89365dc2015-06-16 05:52:47[diff] [blame]452 if (extension.empty() ||
453 (extension.size() == 1 && extension[0] == kExtensionSeparator))
rdevlin.cronin@chromium.org922828b32012-04-25 01:53:44[diff] [blame]454 return *this;
455
456 StringType str = path_;
457 if (extension[0] != kExtensionSeparator &&
458 *(str.end() - 1) != kExtensionSeparator) {
459 str.append(1, kExtensionSeparator);
erikkay@google.com235c17a2009-01-09 18:26:19[diff] [blame]460 }
Jan Wilken Dörriefb1d7b662020-02-04 11:10:23[diff] [blame]461 str.append(extension.data(), extension.size());
rdevlin.cronin@chromium.org922828b32012-04-25 01:53:44[diff] [blame]462 return FilePath(str);
463}
464
Erik Jensenb54fb9c2018-11-27 23:58:24[diff] [blame]465FilePath FilePath::AddExtensionASCII(StringPiece extension) const {
466 DCHECK(IsStringASCII(extension));
467#if defined(OS_WIN)
Jan Wilken Dörrie9bb441d2019-11-01 17:48:40[diff] [blame]468 return AddExtension(UTF8ToWide(extension));
Erik Jensenb54fb9c2018-11-27 23:58:24[diff] [blame]469#elif defined(OS_POSIX) || defined(OS_FUCHSIA)
470 return AddExtension(extension);
471#endif
472}
473
brettw89365dc2015-06-16 05:52:47[diff] [blame]474FilePath FilePath::ReplaceExtension(StringPieceType extension) const {
rdevlin.cronin@chromium.org922828b32012-04-25 01:53:44[diff] [blame]475 if (IsEmptyOrSpecialCase(BaseName().value()))
476 return FilePath();
erikkay@google.com235c17a2009-01-09 18:26:19[diff] [blame]477
478 FilePath no_ext = RemoveExtension();
479 // If the new extension is "" or ".", then just remove the current extension.
brettw89365dc2015-06-16 05:52:47[diff] [blame]480 if (extension.empty() ||
481 (extension.size() == 1 && extension[0] == kExtensionSeparator))
erikkay@google.com235c17a2009-01-09 18:26:19[diff] [blame]482 return no_ext;
483
484 StringType str = no_ext.value();
485 if (extension[0] != kExtensionSeparator)
486 str.append(1, kExtensionSeparator);
Jan Wilken Dörriefb1d7b662020-02-04 11:10:23[diff] [blame]487 str.append(extension.data(), extension.size());
erikkay@google.com235c17a2009-01-09 18:26:19[diff] [blame]488 return FilePath(str);
489}
490
brettw89365dc2015-06-16 05:52:47[diff] [blame]491bool FilePath::MatchesExtension(StringPieceType extension) const {
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]492 DCHECK(extension.empty() || extension[0] == kExtensionSeparator);
493
estade@chromium.org34322912010-07-27 21:24:39[diff] [blame]494 StringType current_extension = Extension();
avi@chromium.orgf1ce6e62009-06-29 20:31:29[diff] [blame]495
496 if (current_extension.length() != extension.length())
497 return false;
498
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]499 return FilePath::CompareEqualIgnoreCase(extension, current_extension);
avi@chromium.orgf1ce6e62009-06-29 20:31:29[diff] [blame]500}
501
brettw89365dc2015-06-16 05:52:47[diff] [blame]502FilePath FilePath::Append(StringPieceType component) const {
503 StringPieceType appended = component;
aedla@chromium.orgaeae59f2013-01-28 13:47:55[diff] [blame]504 StringType without_nuls;
505
506 StringType::size_type nul_pos = component.find(kStringTerminator);
brettw89365dc2015-06-16 05:52:47[diff] [blame]507 if (nul_pos != StringPieceType::npos) {
Jan Wilken Dörriec15b4bc2020-01-30 07:04:12[diff] [blame]508 without_nuls = StringType(component.substr(0, nul_pos));
brettw89365dc2015-06-16 05:52:47[diff] [blame]509 appended = StringPieceType(without_nuls);
aedla@chromium.orgaeae59f2013-01-28 13:47:55[diff] [blame]510 }
511
brettw89365dc2015-06-16 05:52:47[diff] [blame]512 DCHECK(!IsPathAbsolute(appended));
aedla@chromium.orgaeae59f2013-01-28 13:47:55[diff] [blame]513
Sami Kyostilaf161e9c52017-06-05 15:12:53[diff] [blame]514 if (path_.compare(kCurrentDirectory) == 0 && !appended.empty()) {
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]515 // Append normally doesn't do any normalization, but as a special case,
516 // when appending to kCurrentDirectory, just return a new path for the
517 // component argument. Appending component to kCurrentDirectory would
518 // serve no purpose other than needlessly lengthening the path, and
519 // it's likely in practice to wind up with FilePath objects containing
520 // only kCurrentDirectory when calling DirName on a single relative path
521 // component.
brettw89365dc2015-06-16 05:52:47[diff] [blame]522 return FilePath(appended);
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]523 }
524
525 FilePath new_path(path_);
avi@google.com0b733222008-12-11 14:55:12[diff] [blame]526 new_path.StripTrailingSeparatorsInternal();
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]527
528 // Don't append a separator if the path is empty (indicating the current
529 // directory) or if the path component is empty (indicating nothing to
530 // append).
pkasting9022cb42016-02-05 00:08:56[diff] [blame]531 if (!appended.empty() && !new_path.path_.empty()) {
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]532 // Don't append a separator if the path still ends with a trailing
533 // separator after stripping (indicating the root directory).
pkasting9022cb42016-02-05 00:08:56[diff] [blame]534 if (!IsSeparator(new_path.path_.back())) {
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]535 // Don't append a separator if the path is just a drive letter.
mark@chromium.org1906c032008-12-08 16:49:08[diff] [blame]536 if (FindDriveLetter(new_path.path_) + 1 != new_path.path_.length()) {
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]537 new_path.path_.append(1, kSeparators[0]);
538 }
539 }
540 }
541
Jan Wilken Dörriefb1d7b662020-02-04 11:10:23[diff] [blame]542 new_path.path_.append(appended.data(), appended.size());
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]543 return new_path;
544}
545
mark@chromium.org1906c032008-12-08 16:49:08[diff] [blame]546FilePath FilePath::Append(const FilePath& component) const {
547 return Append(component.value());
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]548}
549
brettw89365dc2015-06-16 05:52:47[diff] [blame]550FilePath FilePath::AppendASCII(StringPiece component) const {
brettw@chromium.org527965412014-05-07 14:38:26[diff] [blame]551 DCHECK(base::IsStringASCII(component));
erikkay@google.com835c74a32009-01-21 22:15:57[diff] [blame]552#if defined(OS_WIN)
Jan Wilken Dörrie9bb441d2019-11-01 17:48:40[diff] [blame]553 return Append(UTF8ToWide(component));
Fabrice de Gans-Riberi306871de2018-05-16 19:38:39[diff] [blame]554#elif defined(OS_POSIX) || defined(OS_FUCHSIA)
brettw89365dc2015-06-16 05:52:47[diff] [blame]555 return Append(component);
erikkay@google.com835c74a32009-01-21 22:15:57[diff] [blame]556#endif
557}
558
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]559bool FilePath::IsAbsolute() const {
mark@chromium.org1906c032008-12-08 16:49:08[diff] [blame]560 return IsPathAbsolute(path_);
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]561}
562
Joel Hockeybc3a2c32021-03-24 08:42:46[diff] [blame]563bool FilePath::IsNetwork() const {
564 return path_.length() > 1 && FilePath::IsSeparator(path_[0]) &&
565 FilePath::IsSeparator(path_[1]);
566}
567
brettw@chromium.org15476932013-04-12 05:17:15[diff] [blame]568bool FilePath::EndsWithSeparator() const {
569 if (empty())
570 return false;
thestig84abdbb72016-07-08 01:41:06[diff] [blame]571 return IsSeparator(path_.back());
brettw@chromium.org15476932013-04-12 05:17:15[diff] [blame]572}
573
574FilePath FilePath::AsEndingWithSeparator() const {
575 if (EndsWithSeparator() || path_.empty())
576 return *this;
577
578 StringType path_str;
579 path_str.reserve(path_.length() + 1); // Only allocate string once.
580
581 path_str = path_;
582 path_str.append(&kSeparators[0], 1);
583 return FilePath(path_str);
584}
585
erg@google.com9989c9bb2011-01-07 20:23:43[diff] [blame]586FilePath FilePath::StripTrailingSeparators() const {
587 FilePath new_path(path_);
588 new_path.StripTrailingSeparatorsInternal();
589
590 return new_path;
591}
592
593bool FilePath::ReferencesParent() const {
jkarlinf95b1d92016-11-11 20:44:33[diff] [blame]594 if (path_.find(kParentDirectory) == StringType::npos) {
595 // GetComponents is quite expensive, so avoid calling it in the majority
596 // of cases where there isn't a kParentDirectory anywhere in the path.
597 return false;
598 }
599
erg@google.com9989c9bb2011-01-07 20:23:43[diff] [blame]600 std::vector<StringType> components;
601 GetComponents(&components);
602
davidben19011ca2015-12-01 06:46:00[diff] [blame]603 std::vector<StringType>::const_iterator it = components.begin();
604 for (; it != components.end(); ++it) {
605 const StringType& component = *it;
jschuh@chromium.orga00f5452013-05-17 02:27:46[diff] [blame]606 // Windows has odd, undocumented behavior with path components containing
607 // only whitespace and . characters. So, if all we see is . and
608 // whitespace, then we treat any .. sequence as referencing parent.
davidben19011ca2015-12-01 06:46:00[diff] [blame]609 // For simplicity we enforce this on all platforms.
jschuh@chromium.orga00f5452013-05-17 02:27:46[diff] [blame]610 if (component.find_first_not_of(FILE_PATH_LITERAL(". \n\r\t")) ==
611 std::string::npos &&
612 component.find(kParentDirectory) != std::string::npos) {
erg@google.com9989c9bb2011-01-07 20:23:43[diff] [blame]613 return true;
jschuh@chromium.orga00f5452013-05-17 02:27:46[diff] [blame]614 }
erg@google.com9989c9bb2011-01-07 20:23:43[diff] [blame]615 }
616 return false;
617}
618
Fabrice de Gans-Riberi306871de2018-05-16 19:38:39[diff] [blame]619#if defined(OS_WIN)
620
Jan Wilken Dörrie085b2aa2021-03-12 16:26:57[diff] [blame]621std::u16string FilePath::LossyDisplayName() const {
Jan Wilken Dörrieaf27f302020-05-17 16:35:28[diff] [blame]622 return AsString16(path_);
Fabrice de Gans-Riberi306871de2018-05-16 19:38:39[diff] [blame]623}
624
625std::string FilePath::MaybeAsASCII() const {
Jan Wilken Dörrieaf27f302020-05-17 16:35:28[diff] [blame]626 return base::IsStringASCII(path_) ? WideToASCII(path_) : std::string();
Fabrice de Gans-Riberi306871de2018-05-16 19:38:39[diff] [blame]627}
628
629std::string FilePath::AsUTF8Unsafe() const {
Jan Wilken Dörrieaf27f302020-05-17 16:35:28[diff] [blame]630 return WideToUTF8(value());
Fabrice de Gans-Riberi306871de2018-05-16 19:38:39[diff] [blame]631}
632
Jan Wilken Dörrie085b2aa2021-03-12 16:26:57[diff] [blame]633std::u16string FilePath::AsUTF16Unsafe() const {
Jan Wilken Dörrieaf27f302020-05-17 16:35:28[diff] [blame]634 return WideToUTF16(value());
Fabrice de Gans-Riberi306871de2018-05-16 19:38:39[diff] [blame]635}
636
637// static
638FilePath FilePath::FromUTF8Unsafe(StringPiece utf8) {
Jan Wilken Dörrie9bb441d2019-11-01 17:48:40[diff] [blame]639 return FilePath(UTF8ToWide(utf8));
Fabrice de Gans-Riberi306871de2018-05-16 19:38:39[diff] [blame]640}
641
642// static
643FilePath FilePath::FromUTF16Unsafe(StringPiece16 utf16) {
Jan Wilken Dörrieaf27f302020-05-17 16:35:28[diff] [blame]644 return FilePath(AsWStringPiece(utf16));
Fabrice de Gans-Riberi306871de2018-05-16 19:38:39[diff] [blame]645}
646
647#elif defined(OS_POSIX) || defined(OS_FUCHSIA)
648
erg@google.com9989c9bb2011-01-07 20:23:43[diff] [blame]649// See file_path.h for a discussion of the encoding of paths on POSIX
evan@chromium.org8a205c02011-02-04 20:41:33[diff] [blame]650// platforms. These encoding conversion functions are not quite correct.
651
Jan Wilken Dörrie085b2aa2021-03-12 16:26:57[diff] [blame]652std::u16string FilePath::LossyDisplayName() const {
brettw@chromium.org04af979a2013-02-16 04:12:26[diff] [blame]653 return WideToUTF16(SysNativeMBToWide(path_));
evan@chromium.org8a205c02011-02-04 20:41:33[diff] [blame]654}
655
evan@chromium.org1f34afe42011-03-01 21:38:51[diff] [blame]656std::string FilePath::MaybeAsASCII() const {
brettw@chromium.org527965412014-05-07 14:38:26[diff] [blame]657 if (base::IsStringASCII(path_))
evan@chromium.org1f34afe42011-03-01 21:38:51[diff] [blame]658 return path_;
dcheng@chromium.org007b3f82013-04-09 08:46:45[diff] [blame]659 return std::string();
evan@chromium.org1f34afe42011-03-01 21:38:51[diff] [blame]660}
661
satorux@chromium.org45440512011-11-02 04:55:23[diff] [blame]662std::string FilePath::AsUTF8Unsafe() const {
halliwellbd52a8a12014-12-12 17:21:04[diff] [blame]663#if defined(SYSTEM_NATIVE_UTF8)
satorux@chromium.org45440512011-11-02 04:55:23[diff] [blame]664 return value();
665#else
brettw@chromium.org04af979a2013-02-16 04:12:26[diff] [blame]666 return WideToUTF8(SysNativeMBToWide(value()));
satorux@chromium.org45440512011-11-02 04:55:23[diff] [blame]667#endif
668}
669
Jan Wilken Dörrie085b2aa2021-03-12 16:26:57[diff] [blame]670std::u16string FilePath::AsUTF16Unsafe() const {
halliwellbd52a8a12014-12-12 17:21:04[diff] [blame]671#if defined(SYSTEM_NATIVE_UTF8)
darin@chromium.org65e269f2013-06-20 21:29:54[diff] [blame]672 return UTF8ToUTF16(value());
673#else
674 return WideToUTF16(SysNativeMBToWide(value()));
675#endif
676}
677
satorux@chromium.org45440512011-11-02 04:55:23[diff] [blame]678// static
kinukob473f002016-02-22 05:23:19[diff] [blame]679FilePath FilePath::FromUTF8Unsafe(StringPiece utf8) {
halliwellbd52a8a12014-12-12 17:21:04[diff] [blame]680#if defined(SYSTEM_NATIVE_UTF8)
satorux@chromium.org45440512011-11-02 04:55:23[diff] [blame]681 return FilePath(utf8);
682#else
brettw@chromium.org04af979a2013-02-16 04:12:26[diff] [blame]683 return FilePath(SysWideToNativeMB(UTF8ToWide(utf8)));
satorux@chromium.org45440512011-11-02 04:55:23[diff] [blame]684#endif
685}
686
darin@chromium.org65e269f2013-06-20 21:29:54[diff] [blame]687// static
kinukob473f002016-02-22 05:23:19[diff] [blame]688FilePath FilePath::FromUTF16Unsafe(StringPiece16 utf16) {
halliwellbd52a8a12014-12-12 17:21:04[diff] [blame]689#if defined(SYSTEM_NATIVE_UTF8)
darin@chromium.org65e269f2013-06-20 21:29:54[diff] [blame]690 return FilePath(UTF16ToUTF8(utf16));
691#else
Jan Wilken Dörrie4aa966b2020-08-24 18:20:29[diff] [blame]692 return FilePath(SysWideToNativeMB(UTF16ToWide(utf16)));
darin@chromium.org65e269f2013-06-20 21:29:54[diff] [blame]693#endif
694}
695
Fabrice de Gans-Riberi306871de2018-05-16 19:38:39[diff] [blame]696#endif // defined(OS_WIN)
erg@google.com9989c9bb2011-01-07 20:23:43[diff] [blame]697
aedla@chromium.orgaeae59f2013-01-28 13:47:55[diff] [blame]698void FilePath::WriteToPickle(Pickle* pickle) const {
evan@chromium.org8ce9e492011-05-31 18:04:33[diff] [blame]699#if defined(OS_WIN)
Jan Wilken Dörrieaf27f302020-05-17 16:35:28[diff] [blame]700 pickle->WriteString16(AsStringPiece16(path_));
Fabrice de Gans-Riberi306871de2018-05-16 19:38:39[diff] [blame]701#elif defined(OS_POSIX) || defined(OS_FUCHSIA)
evan@chromium.org8ce9e492011-05-31 18:04:33[diff] [blame]702 pickle->WriteString(path_);
Fabrice de Gans-Riberi306871de2018-05-16 19:38:39[diff] [blame]703#else
704#error Unsupported platform
evan@chromium.org8ce9e492011-05-31 18:04:33[diff] [blame]705#endif
erg@google.com9989c9bb2011-01-07 20:23:43[diff] [blame]706}
707
jbates@chromium.orgce208f872012-03-07 20:42:56[diff] [blame]708bool FilePath::ReadFromPickle(PickleIterator* iter) {
evan@chromium.org8ce9e492011-05-31 18:04:33[diff] [blame]709#if defined(OS_WIN)
Jan Wilken Dörrie85285b02021-03-11 23:38:47[diff] [blame]710 std::u16string path;
Jan Wilken Dörrie9bb441d2019-11-01 17:48:40[diff] [blame]711 if (!iter->ReadString16(&path))
evan@chromium.org8ce9e492011-05-31 18:04:33[diff] [blame]712 return false;
Jan Wilken Dörrie9bb441d2019-11-01 17:48:40[diff] [blame]713 path_ = UTF16ToWide(path);
Fabrice de Gans-Riberi306871de2018-05-16 19:38:39[diff] [blame]714#elif defined(OS_POSIX) || defined(OS_FUCHSIA)
jbates@chromium.orgce208f872012-03-07 20:42:56[diff] [blame]715 if (!iter->ReadString(&path_))
evan@chromium.org8ce9e492011-05-31 18:04:33[diff] [blame]716 return false;
Fabrice de Gans-Riberi306871de2018-05-16 19:38:39[diff] [blame]717#else
718#error Unsupported platform
evan@chromium.org8ce9e492011-05-31 18:04:33[diff] [blame]719#endif
720
aedla@chromium.orgaeae59f2013-01-28 13:47:55[diff] [blame]721 if (path_.find(kStringTerminator) != StringType::npos)
722 return false;
723
evan@chromium.org8ce9e492011-05-31 18:04:33[diff] [blame]724 return true;
erg@google.com9989c9bb2011-01-07 20:23:43[diff] [blame]725}
726
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]727#if defined(OS_WIN)
brettw89365dc2015-06-16 05:52:47[diff] [blame]728// Windows specific implementation of file string comparisons.
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]729
brettw89365dc2015-06-16 05:52:47[diff] [blame]730int FilePath::CompareIgnoreCase(StringPieceType string1,
731 StringPieceType string2) {
Cliff Smolinskyc5c52102019-05-03 20:51:54[diff] [blame]732 // CharUpperW within user32 is used here because it will provide unicode
733 // conversions regardless of locale. The STL alternative, towupper, has a
734 // locale consideration that prevents it from converting all characters by
735 // default.
736 CHECK(win::IsUser32AndGdi32Available());
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]737 // Perform character-wise upper case comparison rather than using the
738 // fully Unicode-aware CompareString(). For details see:
739 // http://blogs.msdn.com/michkap/archive/2005/10/17/481600.aspx
brettw89365dc2015-06-16 05:52:47[diff] [blame]740 StringPieceType::const_iterator i1 = string1.begin();
741 StringPieceType::const_iterator i2 = string2.begin();
742 StringPieceType::const_iterator string1end = string1.end();
743 StringPieceType::const_iterator string2end = string2.end();
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]744 for ( ; i1 != string1end && i2 != string2end; ++i1, ++i2) {
thakis@chromium.orgff5f0892014-08-20 20:12:12[diff] [blame]745 wchar_t c1 =
Cliff Smolinskyc5c52102019-05-03 20:51:54[diff] [blame]746 (wchar_t)LOWORD(::CharUpperW((LPWSTR)(DWORD_PTR)MAKELONG(*i1, 0)));
thakis@chromium.orgff5f0892014-08-20 20:12:12[diff] [blame]747 wchar_t c2 =
Cliff Smolinskyc5c52102019-05-03 20:51:54[diff] [blame]748 (wchar_t)LOWORD(::CharUpperW((LPWSTR)(DWORD_PTR)MAKELONG(*i2, 0)));
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]749 if (c1 < c2)
750 return -1;
751 if (c1 > c2)
752 return 1;
753 }
754 if (i1 != string1end)
755 return 1;
756 if (i2 != string2end)
757 return -1;
758 return 0;
759}
760
Avi Drissman5b286372020-07-28 21:59:38[diff] [blame]761#elif defined(OS_APPLE)
brettw89365dc2015-06-16 05:52:47[diff] [blame]762// Mac OS X specific implementation of file string comparisons.
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]763
764// cf. http://developer.apple.com/mac/library/technotes/tn/tn1150.html#UnicodeSubtleties
765//
766// "When using CreateTextEncoding to create a text encoding, you should set
767// the TextEncodingBase to kTextEncodingUnicodeV2_0, set the
768// TextEncodingVariant to kUnicodeCanonicalDecompVariant, and set the
769// TextEncodingFormat to kUnicode16BitFormat. Using these values ensures that
770// the Unicode will be in the same form as on an HFS Plus volume, even as the
771// Unicode standard evolves."
772//
773// Another technical article for X 10.4 updates this: one should use
774// the new (unambiguous) kUnicodeHFSPlusDecompVariant.
775// cf. http://developer.apple.com/mac/library/releasenotes/TextFonts/RN-TEC/index.html
776//
777// This implementation uses CFStringGetFileSystemRepresentation() to get the
778// decomposed form, and an adapted version of the FastUnicodeCompare as
779// described in the tech note to compare the strings.
780
781// Character conversion table for FastUnicodeCompare()
782//
783// The lower case table consists of a 256-entry high-byte table followed by
784// some number of 256-entry subtables. The high-byte table contains either an
785// offset to the subtable for characters with that high byte or zero, which
786// means that there are no case mappings or ignored characters in that block.
787// Ignored characters are mapped to zero.
788//
789// cf. downloadable file linked in
790// http://developer.apple.com/mac/library/technotes/tn/tn1150.html#StringComparisonAlgorithm
791
792namespace {
793
794const UInt16 lower_case_table[] = {
795 // High-byte indices ( == 0 iff no case mapping and no ignorables )
796
797 /* 0 */ 0x0100, 0x0200, 0x0000, 0x0300, 0x0400, 0x0500, 0x0000, 0x0000,
798 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
799 /* 1 */ 0x0600, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
800 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
801 /* 2 */ 0x0700, 0x0800, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
802 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
803 /* 3 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
804 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
805 /* 4 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
806 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
807 /* 5 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
808 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
809 /* 6 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
810 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
811 /* 7 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
812 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
813 /* 8 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
814 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
815 /* 9 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
816 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
817 /* A */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
818 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
819 /* B */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
820 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
821 /* C */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
822 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
823 /* D */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
824 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
825 /* E */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
826 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
827 /* F */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
828 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0900, 0x0A00,
829
830 // Table 1 (for high byte 0x00)
831
832 /* 0 */ 0xFFFF, 0x0001, 0x0002, 0x0003, 0x0004, 0x0005, 0x0006, 0x0007,
833 0x0008, 0x0009, 0x000A, 0x000B, 0x000C, 0x000D, 0x000E, 0x000F,
834 /* 1 */ 0x0010, 0x0011, 0x0012, 0x0013, 0x0014, 0x0015, 0x0016, 0x0017,
835 0x0018, 0x0019, 0x001A, 0x001B, 0x001C, 0x001D, 0x001E, 0x001F,
836 /* 2 */ 0x0020, 0x0021, 0x0022, 0x0023, 0x0024, 0x0025, 0x0026, 0x0027,
837 0x0028, 0x0029, 0x002A, 0x002B, 0x002C, 0x002D, 0x002E, 0x002F,
838 /* 3 */ 0x0030, 0x0031, 0x0032, 0x0033, 0x0034, 0x0035, 0x0036, 0x0037,
839 0x0038, 0x0039, 0x003A, 0x003B, 0x003C, 0x003D, 0x003E, 0x003F,
840 /* 4 */ 0x0040, 0x0061, 0x0062, 0x0063, 0x0064, 0x0065, 0x0066, 0x0067,
841 0x0068, 0x0069, 0x006A, 0x006B, 0x006C, 0x006D, 0x006E, 0x006F,
842 /* 5 */ 0x0070, 0x0071, 0x0072, 0x0073, 0x0074, 0x0075, 0x0076, 0x0077,
843 0x0078, 0x0079, 0x007A, 0x005B, 0x005C, 0x005D, 0x005E, 0x005F,
844 /* 6 */ 0x0060, 0x0061, 0x0062, 0x0063, 0x0064, 0x0065, 0x0066, 0x0067,
845 0x0068, 0x0069, 0x006A, 0x006B, 0x006C, 0x006D, 0x006E, 0x006F,
846 /* 7 */ 0x0070, 0x0071, 0x0072, 0x0073, 0x0074, 0x0075, 0x0076, 0x0077,
847 0x0078, 0x0079, 0x007A, 0x007B, 0x007C, 0x007D, 0x007E, 0x007F,
848 /* 8 */ 0x0080, 0x0081, 0x0082, 0x0083, 0x0084, 0x0085, 0x0086, 0x0087,
849 0x0088, 0x0089, 0x008A, 0x008B, 0x008C, 0x008D, 0x008E, 0x008F,
850 /* 9 */ 0x0090, 0x0091, 0x0092, 0x0093, 0x0094, 0x0095, 0x0096, 0x0097,
851 0x0098, 0x0099, 0x009A, 0x009B, 0x009C, 0x009D, 0x009E, 0x009F,
852 /* A */ 0x00A0, 0x00A1, 0x00A2, 0x00A3, 0x00A4, 0x00A5, 0x00A6, 0x00A7,
853 0x00A8, 0x00A9, 0x00AA, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00AF,
854 /* B */ 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x00B4, 0x00B5, 0x00B6, 0x00B7,
855 0x00B8, 0x00B9, 0x00BA, 0x00BB, 0x00BC, 0x00BD, 0x00BE, 0x00BF,
856 /* C */ 0x00C0, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00E6, 0x00C7,
857 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF,
858 /* D */ 0x00F0, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x00D7,
859 0x00F8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x00DD, 0x00FE, 0x00DF,
860 /* E */ 0x00E0, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x00E7,
861 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF,
862 /* F */ 0x00F0, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x00F7,
863 0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x00FD, 0x00FE, 0x00FF,
864
865 // Table 2 (for high byte 0x01)
866
867 /* 0 */ 0x0100, 0x0101, 0x0102, 0x0103, 0x0104, 0x0105, 0x0106, 0x0107,
868 0x0108, 0x0109, 0x010A, 0x010B, 0x010C, 0x010D, 0x010E, 0x010F,
869 /* 1 */ 0x0111, 0x0111, 0x0112, 0x0113, 0x0114, 0x0115, 0x0116, 0x0117,
870 0x0118, 0x0119, 0x011A, 0x011B, 0x011C, 0x011D, 0x011E, 0x011F,
871 /* 2 */ 0x0120, 0x0121, 0x0122, 0x0123, 0x0124, 0x0125, 0x0127, 0x0127,
872 0x0128, 0x0129, 0x012A, 0x012B, 0x012C, 0x012D, 0x012E, 0x012F,
873 /* 3 */ 0x0130, 0x0131, 0x0133, 0x0133, 0x0134, 0x0135, 0x0136, 0x0137,
874 0x0138, 0x0139, 0x013A, 0x013B, 0x013C, 0x013D, 0x013E, 0x0140,
875 /* 4 */ 0x0140, 0x0142, 0x0142, 0x0143, 0x0144, 0x0145, 0x0146, 0x0147,
876 0x0148, 0x0149, 0x014B, 0x014B, 0x014C, 0x014D, 0x014E, 0x014F,
877 /* 5 */ 0x0150, 0x0151, 0x0153, 0x0153, 0x0154, 0x0155, 0x0156, 0x0157,
878 0x0158, 0x0159, 0x015A, 0x015B, 0x015C, 0x015D, 0x015E, 0x015F,
879 /* 6 */ 0x0160, 0x0161, 0x0162, 0x0163, 0x0164, 0x0165, 0x0167, 0x0167,
880 0x0168, 0x0169, 0x016A, 0x016B, 0x016C, 0x016D, 0x016E, 0x016F,
881 /* 7 */ 0x0170, 0x0171, 0x0172, 0x0173, 0x0174, 0x0175, 0x0176, 0x0177,
882 0x0178, 0x0179, 0x017A, 0x017B, 0x017C, 0x017D, 0x017E, 0x017F,
883 /* 8 */ 0x0180, 0x0253, 0x0183, 0x0183, 0x0185, 0x0185, 0x0254, 0x0188,
884 0x0188, 0x0256, 0x0257, 0x018C, 0x018C, 0x018D, 0x01DD, 0x0259,
885 /* 9 */ 0x025B, 0x0192, 0x0192, 0x0260, 0x0263, 0x0195, 0x0269, 0x0268,
886 0x0199, 0x0199, 0x019A, 0x019B, 0x026F, 0x0272, 0x019E, 0x0275,
887 /* A */ 0x01A0, 0x01A1, 0x01A3, 0x01A3, 0x01A5, 0x01A5, 0x01A6, 0x01A8,
888 0x01A8, 0x0283, 0x01AA, 0x01AB, 0x01AD, 0x01AD, 0x0288, 0x01AF,
889 /* B */ 0x01B0, 0x028A, 0x028B, 0x01B4, 0x01B4, 0x01B6, 0x01B6, 0x0292,
890 0x01B9, 0x01B9, 0x01BA, 0x01BB, 0x01BD, 0x01BD, 0x01BE, 0x01BF,
891 /* C */ 0x01C0, 0x01C1, 0x01C2, 0x01C3, 0x01C6, 0x01C6, 0x01C6, 0x01C9,
892 0x01C9, 0x01C9, 0x01CC, 0x01CC, 0x01CC, 0x01CD, 0x01CE, 0x01CF,
893 /* D */ 0x01D0, 0x01D1, 0x01D2, 0x01D3, 0x01D4, 0x01D5, 0x01D6, 0x01D7,
894 0x01D8, 0x01D9, 0x01DA, 0x01DB, 0x01DC, 0x01DD, 0x01DE, 0x01DF,
895 /* E */ 0x01E0, 0x01E1, 0x01E2, 0x01E3, 0x01E5, 0x01E5, 0x01E6, 0x01E7,
896 0x01E8, 0x01E9, 0x01EA, 0x01EB, 0x01EC, 0x01ED, 0x01EE, 0x01EF,
897 /* F */ 0x01F0, 0x01F3, 0x01F3, 0x01F3, 0x01F4, 0x01F5, 0x01F6, 0x01F7,
898 0x01F8, 0x01F9, 0x01FA, 0x01FB, 0x01FC, 0x01FD, 0x01FE, 0x01FF,
899
900 // Table 3 (for high byte 0x03)
901
902 /* 0 */ 0x0300, 0x0301, 0x0302, 0x0303, 0x0304, 0x0305, 0x0306, 0x0307,
903 0x0308, 0x0309, 0x030A, 0x030B, 0x030C, 0x030D, 0x030E, 0x030F,
904 /* 1 */ 0x0310, 0x0311, 0x0312, 0x0313, 0x0314, 0x0315, 0x0316, 0x0317,
905 0x0318, 0x0319, 0x031A, 0x031B, 0x031C, 0x031D, 0x031E, 0x031F,
906 /* 2 */ 0x0320, 0x0321, 0x0322, 0x0323, 0x0324, 0x0325, 0x0326, 0x0327,
907 0x0328, 0x0329, 0x032A, 0x032B, 0x032C, 0x032D, 0x032E, 0x032F,
908 /* 3 */ 0x0330, 0x0331, 0x0332, 0x0333, 0x0334, 0x0335, 0x0336, 0x0337,
909 0x0338, 0x0339, 0x033A, 0x033B, 0x033C, 0x033D, 0x033E, 0x033F,
910 /* 4 */ 0x0340, 0x0341, 0x0342, 0x0343, 0x0344, 0x0345, 0x0346, 0x0347,
911 0x0348, 0x0349, 0x034A, 0x034B, 0x034C, 0x034D, 0x034E, 0x034F,
912 /* 5 */ 0x0350, 0x0351, 0x0352, 0x0353, 0x0354, 0x0355, 0x0356, 0x0357,
913 0x0358, 0x0359, 0x035A, 0x035B, 0x035C, 0x035D, 0x035E, 0x035F,
914 /* 6 */ 0x0360, 0x0361, 0x0362, 0x0363, 0x0364, 0x0365, 0x0366, 0x0367,
915 0x0368, 0x0369, 0x036A, 0x036B, 0x036C, 0x036D, 0x036E, 0x036F,
916 /* 7 */ 0x0370, 0x0371, 0x0372, 0x0373, 0x0374, 0x0375, 0x0376, 0x0377,
917 0x0378, 0x0379, 0x037A, 0x037B, 0x037C, 0x037D, 0x037E, 0x037F,
918 /* 8 */ 0x0380, 0x0381, 0x0382, 0x0383, 0x0384, 0x0385, 0x0386, 0x0387,
919 0x0388, 0x0389, 0x038A, 0x038B, 0x038C, 0x038D, 0x038E, 0x038F,
920 /* 9 */ 0x0390, 0x03B1, 0x03B2, 0x03B3, 0x03B4, 0x03B5, 0x03B6, 0x03B7,
921 0x03B8, 0x03B9, 0x03BA, 0x03BB, 0x03BC, 0x03BD, 0x03BE, 0x03BF,
922 /* A */ 0x03C0, 0x03C1, 0x03A2, 0x03C3, 0x03C4, 0x03C5, 0x03C6, 0x03C7,
923 0x03C8, 0x03C9, 0x03AA, 0x03AB, 0x03AC, 0x03AD, 0x03AE, 0x03AF,
924 /* B */ 0x03B0, 0x03B1, 0x03B2, 0x03B3, 0x03B4, 0x03B5, 0x03B6, 0x03B7,
925 0x03B8, 0x03B9, 0x03BA, 0x03BB, 0x03BC, 0x03BD, 0x03BE, 0x03BF,
926 /* C */ 0x03C0, 0x03C1, 0x03C2, 0x03C3, 0x03C4, 0x03C5, 0x03C6, 0x03C7,
927 0x03C8, 0x03C9, 0x03CA, 0x03CB, 0x03CC, 0x03CD, 0x03CE, 0x03CF,
928 /* D */ 0x03D0, 0x03D1, 0x03D2, 0x03D3, 0x03D4, 0x03D5, 0x03D6, 0x03D7,
929 0x03D8, 0x03D9, 0x03DA, 0x03DB, 0x03DC, 0x03DD, 0x03DE, 0x03DF,
930 /* E */ 0x03E0, 0x03E1, 0x03E3, 0x03E3, 0x03E5, 0x03E5, 0x03E7, 0x03E7,
931 0x03E9, 0x03E9, 0x03EB, 0x03EB, 0x03ED, 0x03ED, 0x03EF, 0x03EF,
932 /* F */ 0x03F0, 0x03F1, 0x03F2, 0x03F3, 0x03F4, 0x03F5, 0x03F6, 0x03F7,
933 0x03F8, 0x03F9, 0x03FA, 0x03FB, 0x03FC, 0x03FD, 0x03FE, 0x03FF,
934
935 // Table 4 (for high byte 0x04)
936
937 /* 0 */ 0x0400, 0x0401, 0x0452, 0x0403, 0x0454, 0x0455, 0x0456, 0x0407,
938 0x0458, 0x0459, 0x045A, 0x045B, 0x040C, 0x040D, 0x040E, 0x045F,
939 /* 1 */ 0x0430, 0x0431, 0x0432, 0x0433, 0x0434, 0x0435, 0x0436, 0x0437,
940 0x0438, 0x0419, 0x043A, 0x043B, 0x043C, 0x043D, 0x043E, 0x043F,
941 /* 2 */ 0x0440, 0x0441, 0x0442, 0x0443, 0x0444, 0x0445, 0x0446, 0x0447,
942 0x0448, 0x0449, 0x044A, 0x044B, 0x044C, 0x044D, 0x044E, 0x044F,
943 /* 3 */ 0x0430, 0x0431, 0x0432, 0x0433, 0x0434, 0x0435, 0x0436, 0x0437,
944 0x0438, 0x0439, 0x043A, 0x043B, 0x043C, 0x043D, 0x043E, 0x043F,
945 /* 4 */ 0x0440, 0x0441, 0x0442, 0x0443, 0x0444, 0x0445, 0x0446, 0x0447,
946 0x0448, 0x0449, 0x044A, 0x044B, 0x044C, 0x044D, 0x044E, 0x044F,
947 /* 5 */ 0x0450, 0x0451, 0x0452, 0x0453, 0x0454, 0x0455, 0x0456, 0x0457,
948 0x0458, 0x0459, 0x045A, 0x045B, 0x045C, 0x045D, 0x045E, 0x045F,
949 /* 6 */ 0x0461, 0x0461, 0x0463, 0x0463, 0x0465, 0x0465, 0x0467, 0x0467,
950 0x0469, 0x0469, 0x046B, 0x046B, 0x046D, 0x046D, 0x046F, 0x046F,
951 /* 7 */ 0x0471, 0x0471, 0x0473, 0x0473, 0x0475, 0x0475, 0x0476, 0x0477,
952 0x0479, 0x0479, 0x047B, 0x047B, 0x047D, 0x047D, 0x047F, 0x047F,
953 /* 8 */ 0x0481, 0x0481, 0x0482, 0x0483, 0x0484, 0x0485, 0x0486, 0x0487,
954 0x0488, 0x0489, 0x048A, 0x048B, 0x048C, 0x048D, 0x048E, 0x048F,
955 /* 9 */ 0x0491, 0x0491, 0x0493, 0x0493, 0x0495, 0x0495, 0x0497, 0x0497,
956 0x0499, 0x0499, 0x049B, 0x049B, 0x049D, 0x049D, 0x049F, 0x049F,
957 /* A */ 0x04A1, 0x04A1, 0x04A3, 0x04A3, 0x04A5, 0x04A5, 0x04A7, 0x04A7,
958 0x04A9, 0x04A9, 0x04AB, 0x04AB, 0x04AD, 0x04AD, 0x04AF, 0x04AF,
959 /* B */ 0x04B1, 0x04B1, 0x04B3, 0x04B3, 0x04B5, 0x04B5, 0x04B7, 0x04B7,
960 0x04B9, 0x04B9, 0x04BB, 0x04BB, 0x04BD, 0x04BD, 0x04BF, 0x04BF,
961 /* C */ 0x04C0, 0x04C1, 0x04C2, 0x04C4, 0x04C4, 0x04C5, 0x04C6, 0x04C8,
962 0x04C8, 0x04C9, 0x04CA, 0x04CC, 0x04CC, 0x04CD, 0x04CE, 0x04CF,
963 /* D */ 0x04D0, 0x04D1, 0x04D2, 0x04D3, 0x04D4, 0x04D5, 0x04D6, 0x04D7,
964 0x04D8, 0x04D9, 0x04DA, 0x04DB, 0x04DC, 0x04DD, 0x04DE, 0x04DF,
965 /* E */ 0x04E0, 0x04E1, 0x04E2, 0x04E3, 0x04E4, 0x04E5, 0x04E6, 0x04E7,
966 0x04E8, 0x04E9, 0x04EA, 0x04EB, 0x04EC, 0x04ED, 0x04EE, 0x04EF,
967 /* F */ 0x04F0, 0x04F1, 0x04F2, 0x04F3, 0x04F4, 0x04F5, 0x04F6, 0x04F7,
968 0x04F8, 0x04F9, 0x04FA, 0x04FB, 0x04FC, 0x04FD, 0x04FE, 0x04FF,
969
970 // Table 5 (for high byte 0x05)
971
972 /* 0 */ 0x0500, 0x0501, 0x0502, 0x0503, 0x0504, 0x0505, 0x0506, 0x0507,
973 0x0508, 0x0509, 0x050A, 0x050B, 0x050C, 0x050D, 0x050E, 0x050F,
974 /* 1 */ 0x0510, 0x0511, 0x0512, 0x0513, 0x0514, 0x0515, 0x0516, 0x0517,
975 0x0518, 0x0519, 0x051A, 0x051B, 0x051C, 0x051D, 0x051E, 0x051F,
976 /* 2 */ 0x0520, 0x0521, 0x0522, 0x0523, 0x0524, 0x0525, 0x0526, 0x0527,
977 0x0528, 0x0529, 0x052A, 0x052B, 0x052C, 0x052D, 0x052E, 0x052F,
978 /* 3 */ 0x0530, 0x0561, 0x0562, 0x0563, 0x0564, 0x0565, 0x0566, 0x0567,
979 0x0568, 0x0569, 0x056A, 0x056B, 0x056C, 0x056D, 0x056E, 0x056F,
980 /* 4 */ 0x0570, 0x0571, 0x0572, 0x0573, 0x0574, 0x0575, 0x0576, 0x0577,
981 0x0578, 0x0579, 0x057A, 0x057B, 0x057C, 0x057D, 0x057E, 0x057F,
982 /* 5 */ 0x0580, 0x0581, 0x0582, 0x0583, 0x0584, 0x0585, 0x0586, 0x0557,
983 0x0558, 0x0559, 0x055A, 0x055B, 0x055C, 0x055D, 0x055E, 0x055F,
984 /* 6 */ 0x0560, 0x0561, 0x0562, 0x0563, 0x0564, 0x0565, 0x0566, 0x0567,
985 0x0568, 0x0569, 0x056A, 0x056B, 0x056C, 0x056D, 0x056E, 0x056F,
986 /* 7 */ 0x0570, 0x0571, 0x0572, 0x0573, 0x0574, 0x0575, 0x0576, 0x0577,
987 0x0578, 0x0579, 0x057A, 0x057B, 0x057C, 0x057D, 0x057E, 0x057F,
988 /* 8 */ 0x0580, 0x0581, 0x0582, 0x0583, 0x0584, 0x0585, 0x0586, 0x0587,
989 0x0588, 0x0589, 0x058A, 0x058B, 0x058C, 0x058D, 0x058E, 0x058F,
990 /* 9 */ 0x0590, 0x0591, 0x0592, 0x0593, 0x0594, 0x0595, 0x0596, 0x0597,
991 0x0598, 0x0599, 0x059A, 0x059B, 0x059C, 0x059D, 0x059E, 0x059F,
992 /* A */ 0x05A0, 0x05A1, 0x05A2, 0x05A3, 0x05A4, 0x05A5, 0x05A6, 0x05A7,
993 0x05A8, 0x05A9, 0x05AA, 0x05AB, 0x05AC, 0x05AD, 0x05AE, 0x05AF,
994 /* B */ 0x05B0, 0x05B1, 0x05B2, 0x05B3, 0x05B4, 0x05B5, 0x05B6, 0x05B7,
995 0x05B8, 0x05B9, 0x05BA, 0x05BB, 0x05BC, 0x05BD, 0x05BE, 0x05BF,
996 /* C */ 0x05C0, 0x05C1, 0x05C2, 0x05C3, 0x05C4, 0x05C5, 0x05C6, 0x05C7,
997 0x05C8, 0x05C9, 0x05CA, 0x05CB, 0x05CC, 0x05CD, 0x05CE, 0x05CF,
998 /* D */ 0x05D0, 0x05D1, 0x05D2, 0x05D3, 0x05D4, 0x05D5, 0x05D6, 0x05D7,
999 0x05D8, 0x05D9, 0x05DA, 0x05DB, 0x05DC, 0x05DD, 0x05DE, 0x05DF,
1000 /* E */ 0x05E0, 0x05E1, 0x05E2, 0x05E3, 0x05E4, 0x05E5, 0x05E6, 0x05E7,
1001 0x05E8, 0x05E9, 0x05EA, 0x05EB, 0x05EC, 0x05ED, 0x05EE, 0x05EF,
1002 /* F */ 0x05F0, 0x05F1, 0x05F2, 0x05F3, 0x05F4, 0x05F5, 0x05F6, 0x05F7,
1003 0x05F8, 0x05F9, 0x05FA, 0x05FB, 0x05FC, 0x05FD, 0x05FE, 0x05FF,
1004
1005 // Table 6 (for high byte 0x10)
1006
1007 /* 0 */ 0x1000, 0x1001, 0x1002, 0x1003, 0x1004, 0x1005, 0x1006, 0x1007,
1008 0x1008, 0x1009, 0x100A, 0x100B, 0x100C, 0x100D, 0x100E, 0x100F,
1009 /* 1 */ 0x1010, 0x1011, 0x1012, 0x1013, 0x1014, 0x1015, 0x1016, 0x1017,
1010 0x1018, 0x1019, 0x101A, 0x101B, 0x101C, 0x101D, 0x101E, 0x101F,
1011 /* 2 */ 0x1020, 0x1021, 0x1022, 0x1023, 0x1024, 0x1025, 0x1026, 0x1027,
1012 0x1028, 0x1029, 0x102A, 0x102B, 0x102C, 0x102D, 0x102E, 0x102F,
1013 /* 3 */ 0x1030, 0x1031, 0x1032, 0x1033, 0x1034, 0x1035, 0x1036, 0x1037,
1014 0x1038, 0x1039, 0x103A, 0x103B, 0x103C, 0x103D, 0x103E, 0x103F,
1015 /* 4 */ 0x1040, 0x1041, 0x1042, 0x1043, 0x1044, 0x1045, 0x1046, 0x1047,
1016 0x1048, 0x1049, 0x104A, 0x104B, 0x104C, 0x104D, 0x104E, 0x104F,
1017 /* 5 */ 0x1050, 0x1051, 0x1052, 0x1053, 0x1054, 0x1055, 0x1056, 0x1057,
1018 0x1058, 0x1059, 0x105A, 0x105B, 0x105C, 0x105D, 0x105E, 0x105F,
1019 /* 6 */ 0x1060, 0x1061, 0x1062, 0x1063, 0x1064, 0x1065, 0x1066, 0x1067,
1020 0x1068, 0x1069, 0x106A, 0x106B, 0x106C, 0x106D, 0x106E, 0x106F,
1021 /* 7 */ 0x1070, 0x1071, 0x1072, 0x1073, 0x1074, 0x1075, 0x1076, 0x1077,
1022 0x1078, 0x1079, 0x107A, 0x107B, 0x107C, 0x107D, 0x107E, 0x107F,
1023 /* 8 */ 0x1080, 0x1081, 0x1082, 0x1083, 0x1084, 0x1085, 0x1086, 0x1087,
1024 0x1088, 0x1089, 0x108A, 0x108B, 0x108C, 0x108D, 0x108E, 0x108F,
1025 /* 9 */ 0x1090, 0x1091, 0x1092, 0x1093, 0x1094, 0x1095, 0x1096, 0x1097,
1026 0x1098, 0x1099, 0x109A, 0x109B, 0x109C, 0x109D, 0x109E, 0x109F,
1027 /* A */ 0x10D0, 0x10D1, 0x10D2, 0x10D3, 0x10D4, 0x10D5, 0x10D6, 0x10D7,
1028 0x10D8, 0x10D9, 0x10DA, 0x10DB, 0x10DC, 0x10DD, 0x10DE, 0x10DF,
1029 /* B */ 0x10E0, 0x10E1, 0x10E2, 0x10E3, 0x10E4, 0x10E5, 0x10E6, 0x10E7,
1030 0x10E8, 0x10E9, 0x10EA, 0x10EB, 0x10EC, 0x10ED, 0x10EE, 0x10EF,
1031 /* C */ 0x10F0, 0x10F1, 0x10F2, 0x10F3, 0x10F4, 0x10F5, 0x10C6, 0x10C7,
1032 0x10C8, 0x10C9, 0x10CA, 0x10CB, 0x10CC, 0x10CD, 0x10CE, 0x10CF,
1033 /* D */ 0x10D0, 0x10D1, 0x10D2, 0x10D3, 0x10D4, 0x10D5, 0x10D6, 0x10D7,
1034 0x10D8, 0x10D9, 0x10DA, 0x10DB, 0x10DC, 0x10DD, 0x10DE, 0x10DF,
1035 /* E */ 0x10E0, 0x10E1, 0x10E2, 0x10E3, 0x10E4, 0x10E5, 0x10E6, 0x10E7,
1036 0x10E8, 0x10E9, 0x10EA, 0x10EB, 0x10EC, 0x10ED, 0x10EE, 0x10EF,
1037 /* F */ 0x10F0, 0x10F1, 0x10F2, 0x10F3, 0x10F4, 0x10F5, 0x10F6, 0x10F7,
1038 0x10F8, 0x10F9, 0x10FA, 0x10FB, 0x10FC, 0x10FD, 0x10FE, 0x10FF,
1039
1040 // Table 7 (for high byte 0x20)
1041
1042 /* 0 */ 0x2000, 0x2001, 0x2002, 0x2003, 0x2004, 0x2005, 0x2006, 0x2007,
1043 0x2008, 0x2009, 0x200A, 0x200B, 0x0000, 0x0000, 0x0000, 0x0000,
1044 /* 1 */ 0x2010, 0x2011, 0x2012, 0x2013, 0x2014, 0x2015, 0x2016, 0x2017,
1045 0x2018, 0x2019, 0x201A, 0x201B, 0x201C, 0x201D, 0x201E, 0x201F,
1046 /* 2 */ 0x2020, 0x2021, 0x2022, 0x2023, 0x2024, 0x2025, 0x2026, 0x2027,
1047 0x2028, 0x2029, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x202F,
1048 /* 3 */ 0x2030, 0x2031, 0x2032, 0x2033, 0x2034, 0x2035, 0x2036, 0x2037,
1049 0x2038, 0x2039, 0x203A, 0x203B, 0x203C, 0x203D, 0x203E, 0x203F,
1050 /* 4 */ 0x2040, 0x2041, 0x2042, 0x2043, 0x2044, 0x2045, 0x2046, 0x2047,
1051 0x2048, 0x2049, 0x204A, 0x204B, 0x204C, 0x204D, 0x204E, 0x204F,
1052 /* 5 */ 0x2050, 0x2051, 0x2052, 0x2053, 0x2054, 0x2055, 0x2056, 0x2057,
1053 0x2058, 0x2059, 0x205A, 0x205B, 0x205C, 0x205D, 0x205E, 0x205F,
1054 /* 6 */ 0x2060, 0x2061, 0x2062, 0x2063, 0x2064, 0x2065, 0x2066, 0x2067,
1055 0x2068, 0x2069, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
1056 /* 7 */ 0x2070, 0x2071, 0x2072, 0x2073, 0x2074, 0x2075, 0x2076, 0x2077,
1057 0x2078, 0x2079, 0x207A, 0x207B, 0x207C, 0x207D, 0x207E, 0x207F,
1058 /* 8 */ 0x2080, 0x2081, 0x2082, 0x2083, 0x2084, 0x2085, 0x2086, 0x2087,
1059 0x2088, 0x2089, 0x208A, 0x208B, 0x208C, 0x208D, 0x208E, 0x208F,
1060 /* 9 */ 0x2090, 0x2091, 0x2092, 0x2093, 0x2094, 0x2095, 0x2096, 0x2097,
1061 0x2098, 0x2099, 0x209A, 0x209B, 0x209C, 0x209D, 0x209E, 0x209F,
1062 /* A */ 0x20A0, 0x20A1, 0x20A2, 0x20A3, 0x20A4, 0x20A5, 0x20A6, 0x20A7,
1063 0x20A8, 0x20A9, 0x20AA, 0x20AB, 0x20AC, 0x20AD, 0x20AE, 0x20AF,
1064 /* B */ 0x20B0, 0x20B1, 0x20B2, 0x20B3, 0x20B4, 0x20B5, 0x20B6, 0x20B7,
1065 0x20B8, 0x20B9, 0x20BA, 0x20BB, 0x20BC, 0x20BD, 0x20BE, 0x20BF,
1066 /* C */ 0x20C0, 0x20C1, 0x20C2, 0x20C3, 0x20C4, 0x20C5, 0x20C6, 0x20C7,
1067 0x20C8, 0x20C9, 0x20CA, 0x20CB, 0x20CC, 0x20CD, 0x20CE, 0x20CF,
1068 /* D */ 0x20D0, 0x20D1, 0x20D2, 0x20D3, 0x20D4, 0x20D5, 0x20D6, 0x20D7,
1069 0x20D8, 0x20D9, 0x20DA, 0x20DB, 0x20DC, 0x20DD, 0x20DE, 0x20DF,
1070 /* E */ 0x20E0, 0x20E1, 0x20E2, 0x20E3, 0x20E4, 0x20E5, 0x20E6, 0x20E7,
1071 0x20E8, 0x20E9, 0x20EA, 0x20EB, 0x20EC, 0x20ED, 0x20EE, 0x20EF,
1072 /* F */ 0x20F0, 0x20F1, 0x20F2, 0x20F3, 0x20F4, 0x20F5, 0x20F6, 0x20F7,
1073 0x20F8, 0x20F9, 0x20FA, 0x20FB, 0x20FC, 0x20FD, 0x20FE, 0x20FF,
1074
1075 // Table 8 (for high byte 0x21)
1076
1077 /* 0 */ 0x2100, 0x2101, 0x2102, 0x2103, 0x2104, 0x2105, 0x2106, 0x2107,
1078 0x2108, 0x2109, 0x210A, 0x210B, 0x210C, 0x210D, 0x210E, 0x210F,
1079 /* 1 */ 0x2110, 0x2111, 0x2112, 0x2113, 0x2114, 0x2115, 0x2116, 0x2117,
1080 0x2118, 0x2119, 0x211A, 0x211B, 0x211C, 0x211D, 0x211E, 0x211F,
1081 /* 2 */ 0x2120, 0x2121, 0x2122, 0x2123, 0x2124, 0x2125, 0x2126, 0x2127,
1082 0x2128, 0x2129, 0x212A, 0x212B, 0x212C, 0x212D, 0x212E, 0x212F,
1083 /* 3 */ 0x2130, 0x2131, 0x2132, 0x2133, 0x2134, 0x2135, 0x2136, 0x2137,
1084 0x2138, 0x2139, 0x213A, 0x213B, 0x213C, 0x213D, 0x213E, 0x213F,
1085 /* 4 */ 0x2140, 0x2141, 0x2142, 0x2143, 0x2144, 0x2145, 0x2146, 0x2147,
1086 0x2148, 0x2149, 0x214A, 0x214B, 0x214C, 0x214D, 0x214E, 0x214F,
1087 /* 5 */ 0x2150, 0x2151, 0x2152, 0x2153, 0x2154, 0x2155, 0x2156, 0x2157,
1088 0x2158, 0x2159, 0x215A, 0x215B, 0x215C, 0x215D, 0x215E, 0x215F,
1089 /* 6 */ 0x2170, 0x2171, 0x2172, 0x2173, 0x2174, 0x2175, 0x2176, 0x2177,
1090 0x2178, 0x2179, 0x217A, 0x217B, 0x217C, 0x217D, 0x217E, 0x217F,
1091 /* 7 */ 0x2170, 0x2171, 0x2172, 0x2173, 0x2174, 0x2175, 0x2176, 0x2177,
1092 0x2178, 0x2179, 0x217A, 0x217B, 0x217C, 0x217D, 0x217E, 0x217F,
1093 /* 8 */ 0x2180, 0x2181, 0x2182, 0x2183, 0x2184, 0x2185, 0x2186, 0x2187,
1094 0x2188, 0x2189, 0x218A, 0x218B, 0x218C, 0x218D, 0x218E, 0x218F,
1095 /* 9 */ 0x2190, 0x2191, 0x2192, 0x2193, 0x2194, 0x2195, 0x2196, 0x2197,
1096 0x2198, 0x2199, 0x219A, 0x219B, 0x219C, 0x219D, 0x219E, 0x219F,
1097 /* A */ 0x21A0, 0x21A1, 0x21A2, 0x21A3, 0x21A4, 0x21A5, 0x21A6, 0x21A7,
1098 0x21A8, 0x21A9, 0x21AA, 0x21AB, 0x21AC, 0x21AD, 0x21AE, 0x21AF,
1099 /* B */ 0x21B0, 0x21B1, 0x21B2, 0x21B3, 0x21B4, 0x21B5, 0x21B6, 0x21B7,
1100 0x21B8, 0x21B9, 0x21BA, 0x21BB, 0x21BC, 0x21BD, 0x21BE, 0x21BF,
1101 /* C */ 0x21C0, 0x21C1, 0x21C2, 0x21C3, 0x21C4, 0x21C5, 0x21C6, 0x21C7,
1102 0x21C8, 0x21C9, 0x21CA, 0x21CB, 0x21CC, 0x21CD, 0x21CE, 0x21CF,
1103 /* D */ 0x21D0, 0x21D1, 0x21D2, 0x21D3, 0x21D4, 0x21D5, 0x21D6, 0x21D7,
1104 0x21D8, 0x21D9, 0x21DA, 0x21DB, 0x21DC, 0x21DD, 0x21DE, 0x21DF,
1105 /* E */ 0x21E0, 0x21E1, 0x21E2, 0x21E3, 0x21E4, 0x21E5, 0x21E6, 0x21E7,
1106 0x21E8, 0x21E9, 0x21EA, 0x21EB, 0x21EC, 0x21ED, 0x21EE, 0x21EF,
1107 /* F */ 0x21F0, 0x21F1, 0x21F2, 0x21F3, 0x21F4, 0x21F5, 0x21F6, 0x21F7,
1108 0x21F8, 0x21F9, 0x21FA, 0x21FB, 0x21FC, 0x21FD, 0x21FE, 0x21FF,
1109
1110 // Table 9 (for high byte 0xFE)
1111
1112 /* 0 */ 0xFE00, 0xFE01, 0xFE02, 0xFE03, 0xFE04, 0xFE05, 0xFE06, 0xFE07,
1113 0xFE08, 0xFE09, 0xFE0A, 0xFE0B, 0xFE0C, 0xFE0D, 0xFE0E, 0xFE0F,
1114 /* 1 */ 0xFE10, 0xFE11, 0xFE12, 0xFE13, 0xFE14, 0xFE15, 0xFE16, 0xFE17,
1115 0xFE18, 0xFE19, 0xFE1A, 0xFE1B, 0xFE1C, 0xFE1D, 0xFE1E, 0xFE1F,
1116 /* 2 */ 0xFE20, 0xFE21, 0xFE22, 0xFE23, 0xFE24, 0xFE25, 0xFE26, 0xFE27,
1117 0xFE28, 0xFE29, 0xFE2A, 0xFE2B, 0xFE2C, 0xFE2D, 0xFE2E, 0xFE2F,
1118 /* 3 */ 0xFE30, 0xFE31, 0xFE32, 0xFE33, 0xFE34, 0xFE35, 0xFE36, 0xFE37,
1119 0xFE38, 0xFE39, 0xFE3A, 0xFE3B, 0xFE3C, 0xFE3D, 0xFE3E, 0xFE3F,
1120 /* 4 */ 0xFE40, 0xFE41, 0xFE42, 0xFE43, 0xFE44, 0xFE45, 0xFE46, 0xFE47,
1121 0xFE48, 0xFE49, 0xFE4A, 0xFE4B, 0xFE4C, 0xFE4D, 0xFE4E, 0xFE4F,
1122 /* 5 */ 0xFE50, 0xFE51, 0xFE52, 0xFE53, 0xFE54, 0xFE55, 0xFE56, 0xFE57,
1123 0xFE58, 0xFE59, 0xFE5A, 0xFE5B, 0xFE5C, 0xFE5D, 0xFE5E, 0xFE5F,
1124 /* 6 */ 0xFE60, 0xFE61, 0xFE62, 0xFE63, 0xFE64, 0xFE65, 0xFE66, 0xFE67,
1125 0xFE68, 0xFE69, 0xFE6A, 0xFE6B, 0xFE6C, 0xFE6D, 0xFE6E, 0xFE6F,
1126 /* 7 */ 0xFE70, 0xFE71, 0xFE72, 0xFE73, 0xFE74, 0xFE75, 0xFE76, 0xFE77,
1127 0xFE78, 0xFE79, 0xFE7A, 0xFE7B, 0xFE7C, 0xFE7D, 0xFE7E, 0xFE7F,
1128 /* 8 */ 0xFE80, 0xFE81, 0xFE82, 0xFE83, 0xFE84, 0xFE85, 0xFE86, 0xFE87,
1129 0xFE88, 0xFE89, 0xFE8A, 0xFE8B, 0xFE8C, 0xFE8D, 0xFE8E, 0xFE8F,
1130 /* 9 */ 0xFE90, 0xFE91, 0xFE92, 0xFE93, 0xFE94, 0xFE95, 0xFE96, 0xFE97,
1131 0xFE98, 0xFE99, 0xFE9A, 0xFE9B, 0xFE9C, 0xFE9D, 0xFE9E, 0xFE9F,
1132 /* A */ 0xFEA0, 0xFEA1, 0xFEA2, 0xFEA3, 0xFEA4, 0xFEA5, 0xFEA6, 0xFEA7,
1133 0xFEA8, 0xFEA9, 0xFEAA, 0xFEAB, 0xFEAC, 0xFEAD, 0xFEAE, 0xFEAF,
1134 /* B */ 0xFEB0, 0xFEB1, 0xFEB2, 0xFEB3, 0xFEB4, 0xFEB5, 0xFEB6, 0xFEB7,
1135 0xFEB8, 0xFEB9, 0xFEBA, 0xFEBB, 0xFEBC, 0xFEBD, 0xFEBE, 0xFEBF,
1136 /* C */ 0xFEC0, 0xFEC1, 0xFEC2, 0xFEC3, 0xFEC4, 0xFEC5, 0xFEC6, 0xFEC7,
1137 0xFEC8, 0xFEC9, 0xFECA, 0xFECB, 0xFECC, 0xFECD, 0xFECE, 0xFECF,
1138 /* D */ 0xFED0, 0xFED1, 0xFED2, 0xFED3, 0xFED4, 0xFED5, 0xFED6, 0xFED7,
1139 0xFED8, 0xFED9, 0xFEDA, 0xFEDB, 0xFEDC, 0xFEDD, 0xFEDE, 0xFEDF,
1140 /* E */ 0xFEE0, 0xFEE1, 0xFEE2, 0xFEE3, 0xFEE4, 0xFEE5, 0xFEE6, 0xFEE7,
1141 0xFEE8, 0xFEE9, 0xFEEA, 0xFEEB, 0xFEEC, 0xFEED, 0xFEEE, 0xFEEF,
1142 /* F */ 0xFEF0, 0xFEF1, 0xFEF2, 0xFEF3, 0xFEF4, 0xFEF5, 0xFEF6, 0xFEF7,
1143 0xFEF8, 0xFEF9, 0xFEFA, 0xFEFB, 0xFEFC, 0xFEFD, 0xFEFE, 0x0000,
1144
1145 // Table 10 (for high byte 0xFF)
1146
1147 /* 0 */ 0xFF00, 0xFF01, 0xFF02, 0xFF03, 0xFF04, 0xFF05, 0xFF06, 0xFF07,
1148 0xFF08, 0xFF09, 0xFF0A, 0xFF0B, 0xFF0C, 0xFF0D, 0xFF0E, 0xFF0F,
1149 /* 1 */ 0xFF10, 0xFF11, 0xFF12, 0xFF13, 0xFF14, 0xFF15, 0xFF16, 0xFF17,
1150 0xFF18, 0xFF19, 0xFF1A, 0xFF1B, 0xFF1C, 0xFF1D, 0xFF1E, 0xFF1F,
1151 /* 2 */ 0xFF20, 0xFF41, 0xFF42, 0xFF43, 0xFF44, 0xFF45, 0xFF46, 0xFF47,
1152 0xFF48, 0xFF49, 0xFF4A, 0xFF4B, 0xFF4C, 0xFF4D, 0xFF4E, 0xFF4F,
1153 /* 3 */ 0xFF50, 0xFF51, 0xFF52, 0xFF53, 0xFF54, 0xFF55, 0xFF56, 0xFF57,
1154 0xFF58, 0xFF59, 0xFF5A, 0xFF3B, 0xFF3C, 0xFF3D, 0xFF3E, 0xFF3F,
1155 /* 4 */ 0xFF40, 0xFF41, 0xFF42, 0xFF43, 0xFF44, 0xFF45, 0xFF46, 0xFF47,
1156 0xFF48, 0xFF49, 0xFF4A, 0xFF4B, 0xFF4C, 0xFF4D, 0xFF4E, 0xFF4F,
1157 /* 5 */ 0xFF50, 0xFF51, 0xFF52, 0xFF53, 0xFF54, 0xFF55, 0xFF56, 0xFF57,
1158 0xFF58, 0xFF59, 0xFF5A, 0xFF5B, 0xFF5C, 0xFF5D, 0xFF5E, 0xFF5F,
1159 /* 6 */ 0xFF60, 0xFF61, 0xFF62, 0xFF63, 0xFF64, 0xFF65, 0xFF66, 0xFF67,
1160 0xFF68, 0xFF69, 0xFF6A, 0xFF6B, 0xFF6C, 0xFF6D, 0xFF6E, 0xFF6F,
1161 /* 7 */ 0xFF70, 0xFF71, 0xFF72, 0xFF73, 0xFF74, 0xFF75, 0xFF76, 0xFF77,
1162 0xFF78, 0xFF79, 0xFF7A, 0xFF7B, 0xFF7C, 0xFF7D, 0xFF7E, 0xFF7F,
1163 /* 8 */ 0xFF80, 0xFF81, 0xFF82, 0xFF83, 0xFF84, 0xFF85, 0xFF86, 0xFF87,
1164 0xFF88, 0xFF89, 0xFF8A, 0xFF8B, 0xFF8C, 0xFF8D, 0xFF8E, 0xFF8F,
1165 /* 9 */ 0xFF90, 0xFF91, 0xFF92, 0xFF93, 0xFF94, 0xFF95, 0xFF96, 0xFF97,
1166 0xFF98, 0xFF99, 0xFF9A, 0xFF9B, 0xFF9C, 0xFF9D, 0xFF9E, 0xFF9F,
1167 /* A */ 0xFFA0, 0xFFA1, 0xFFA2, 0xFFA3, 0xFFA4, 0xFFA5, 0xFFA6, 0xFFA7,
1168 0xFFA8, 0xFFA9, 0xFFAA, 0xFFAB, 0xFFAC, 0xFFAD, 0xFFAE, 0xFFAF,
1169 /* B */ 0xFFB0, 0xFFB1, 0xFFB2, 0xFFB3, 0xFFB4, 0xFFB5, 0xFFB6, 0xFFB7,
1170 0xFFB8, 0xFFB9, 0xFFBA, 0xFFBB, 0xFFBC, 0xFFBD, 0xFFBE, 0xFFBF,
1171 /* C */ 0xFFC0, 0xFFC1, 0xFFC2, 0xFFC3, 0xFFC4, 0xFFC5, 0xFFC6, 0xFFC7,
1172 0xFFC8, 0xFFC9, 0xFFCA, 0xFFCB, 0xFFCC, 0xFFCD, 0xFFCE, 0xFFCF,
1173 /* D */ 0xFFD0, 0xFFD1, 0xFFD2, 0xFFD3, 0xFFD4, 0xFFD5, 0xFFD6, 0xFFD7,
1174 0xFFD8, 0xFFD9, 0xFFDA, 0xFFDB, 0xFFDC, 0xFFDD, 0xFFDE, 0xFFDF,
1175 /* E */ 0xFFE0, 0xFFE1, 0xFFE2, 0xFFE3, 0xFFE4, 0xFFE5, 0xFFE6, 0xFFE7,
1176 0xFFE8, 0xFFE9, 0xFFEA, 0xFFEB, 0xFFEC, 0xFFED, 0xFFEE, 0xFFEF,
1177 /* F */ 0xFFF0, 0xFFF1, 0xFFF2, 0xFFF3, 0xFFF4, 0xFFF5, 0xFFF6, 0xFFF7,
1178 0xFFF8, 0xFFF9, 0xFFFA, 0xFFFB, 0xFFFC, 0xFFFD, 0xFFFE, 0xFFFF,
1179};
1180
1181// Returns the next non-ignorable codepoint within string starting from the
1182// position indicated by index, or zero if there are no more.
1183// The passed-in index is automatically advanced as the characters in the input
1184// HFS-decomposed UTF-8 strings are read.
1185inline int HFSReadNextNonIgnorableCodepoint(const char* string,
1186 int length,
1187 int* index) {
1188 int codepoint = 0;
1189 while (*index < length && codepoint == 0) {
1190 // CBU8_NEXT returns a value < 0 in error cases. For purposes of string
1191 // comparison, we just use that value and flag it with DCHECK.
1192 CBU8_NEXT(string, *index, length, codepoint);
1193 DCHECK_GT(codepoint, 0);
1194 if (codepoint > 0) {
1195 // Check if there is a subtable for this upper byte.
1196 int lookup_offset = lower_case_table[codepoint >> 8];
1197 if (lookup_offset != 0)
1198 codepoint = lower_case_table[lookup_offset + (codepoint & 0x00FF)];
1199 // Note: codepoint1 may be again 0 at this point if the character was
1200 // an ignorable.
1201 }
1202 }
1203 return codepoint;
1204}
1205
brettw89365dc2015-06-16 05:52:47[diff] [blame]1206} // namespace
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]1207
1208// Special UTF-8 version of FastUnicodeCompare. Cf:
1209// http://developer.apple.com/mac/library/technotes/tn/tn1150.html#StringComparisonAlgorithm
1210// The input strings must be in the special HFS decomposed form.
brettw89365dc2015-06-16 05:52:47[diff] [blame]1211int FilePath::HFSFastUnicodeCompare(StringPieceType string1,
1212 StringPieceType string2) {
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]1213 int length1 = string1.length();
1214 int length2 = string2.length();
1215 int index1 = 0;
1216 int index2 = 0;
1217
1218 for (;;) {
brettw89365dc2015-06-16 05:52:47[diff] [blame]1219 int codepoint1 = HFSReadNextNonIgnorableCodepoint(string1.data(),
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]1220 length1,
1221 &index1);
brettw89365dc2015-06-16 05:52:47[diff] [blame]1222 int codepoint2 = HFSReadNextNonIgnorableCodepoint(string2.data(),
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]1223 length2,
1224 &index2);
1225 if (codepoint1 != codepoint2)
1226 return (codepoint1 < codepoint2) ? -1 : 1;
1227 if (codepoint1 == 0) {
1228 DCHECK_EQ(index1, length1);
1229 DCHECK_EQ(index2, length2);
1230 return 0;
1231 }
1232 }
1233}
1234
brettw89365dc2015-06-16 05:52:47[diff] [blame]1235StringType FilePath::GetHFSDecomposedForm(StringPieceType string) {
jialiul03478a112016-03-31 18:38:43[diff] [blame]1236 StringType result;
thakis@chromium.orgdc4f4c22013-06-24 23:14:17[diff] [blame]1237 ScopedCFTypeRef<CFStringRef> cfstring(
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]1238 CFStringCreateWithBytesNoCopy(
1239 NULL,
brettw89365dc2015-06-16 05:52:47[diff] [blame]1240 reinterpret_cast<const UInt8*>(string.data()),
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]1241 string.length(),
1242 kCFStringEncodingUTF8,
1243 false,
1244 kCFAllocatorNull));
jialiul03478a112016-03-31 18:38:43[diff] [blame]1245 if (cfstring) {
1246 // Query the maximum length needed to store the result. In most cases this
1247 // will overestimate the required space. The return value also already
1248 // includes the space needed for a terminating 0.
1249 CFIndex length = CFStringGetMaximumSizeOfFileSystemRepresentation(cfstring);
1250 DCHECK_GT(length, 0); // should be at least 1 for the 0-terminator.
1251 // Reserve enough space for CFStringGetFileSystemRepresentation to write
1252 // into. Also set the length to the maximum so that we can shrink it later.
1253 // (Increasing rather than decreasing it would clobber the string contents!)
1254 result.reserve(length);
1255 result.resize(length - 1);
1256 Boolean success = CFStringGetFileSystemRepresentation(cfstring,
1257 &result[0],
1258 length);
1259 if (success) {
1260 // Reduce result.length() to actual string length.
1261 result.resize(strlen(result.c_str()));
1262 } else {
1263 // An error occurred -> clear result.
1264 result.clear();
1265 }
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]1266 }
1267 return result;
1268}
1269
brettw89365dc2015-06-16 05:52:47[diff] [blame]1270int FilePath::CompareIgnoreCase(StringPieceType string1,
1271 StringPieceType string2) {
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]1272 // Quick checks for empty strings - these speed things up a bit and make the
1273 // following code cleaner.
1274 if (string1.empty())
1275 return string2.empty() ? 0 : -1;
1276 if (string2.empty())
1277 return 1;
1278
1279 StringType hfs1 = GetHFSDecomposedForm(string1);
1280 StringType hfs2 = GetHFSDecomposedForm(string2);
1281
1282 // GetHFSDecomposedForm() returns an empty string in an error case.
1283 if (hfs1.empty() || hfs2.empty()) {
thakis@chromium.orgdc4f4c22013-06-24 23:14:17[diff] [blame]1284 ScopedCFTypeRef<CFStringRef> cfstring1(
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]1285 CFStringCreateWithBytesNoCopy(
1286 NULL,
brettw89365dc2015-06-16 05:52:47[diff] [blame]1287 reinterpret_cast<const UInt8*>(string1.data()),
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]1288 string1.length(),
1289 kCFStringEncodingUTF8,
1290 false,
1291 kCFAllocatorNull));
thakis@chromium.orgdc4f4c22013-06-24 23:14:17[diff] [blame]1292 ScopedCFTypeRef<CFStringRef> cfstring2(
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]1293 CFStringCreateWithBytesNoCopy(
1294 NULL,
brettw89365dc2015-06-16 05:52:47[diff] [blame]1295 reinterpret_cast<const UInt8*>(string2.data()),
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]1296 string2.length(),
1297 kCFStringEncodingUTF8,
1298 false,
1299 kCFAllocatorNull));
Daniel Ruberyab28fb42019-06-05 00:02:45[diff] [blame]1300 // If neither GetHFSDecomposedForm nor CFStringCreateWithBytesNoCopy
1301 // succeed, fall back to strcmp. This can occur when the input string is
1302 // invalid UTF-8.
1303 if (!cfstring1 || !cfstring2) {
Jan Wilken Dörrie5db50ac2021-02-15 11:43:16[diff] [blame]1304 int comparison = memcmp(string1.data(), string2.data(),
1305 std::min(string1.length(), string2.length()));
Daniel Ruberyab28fb42019-06-05 00:02:45[diff] [blame]1306 if (comparison < 0)
1307 return -1;
1308 if (comparison > 0)
1309 return 1;
1310 return 0;
1311 }
1312
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]1313 return CFStringCompare(cfstring1,
1314 cfstring2,
1315 kCFCompareCaseInsensitive);
1316 }
1317
1318 return HFSFastUnicodeCompare(hfs1, hfs2);
1319}
1320
Fabrice de Gans-Riberi306871de2018-05-16 19:38:39[diff] [blame]1321#elif defined(OS_POSIX) || defined(OS_FUCHSIA)
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]1322
brettw8a800902015-07-10 18:28:33[diff] [blame]1323// Generic Posix system comparisons.
brettw89365dc2015-06-16 05:52:47[diff] [blame]1324int FilePath::CompareIgnoreCase(StringPieceType string1,
1325 StringPieceType string2) {
Jan Wilken Dörrie5db50ac2021-02-15 11:43:16[diff] [blame]1326 size_t rlen = std::min(string1.size(), string2.size());
1327 int comparison = strncasecmp(string1.data(), string2.data(), rlen);
1328 if (comparison < 0 || (comparison == 0 && string1.size() < string2.size()))
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]1329 return -1;
Jan Wilken Dörrie5db50ac2021-02-15 11:43:16[diff] [blame]1330 if (comparison > 0 || (comparison == 0 && string1.size() > string2.size()))
rolandsteiner@chromium.orgeccb9d12009-10-28 05:40:09[diff] [blame]1331 return 1;
1332 return 0;
1333}
1334
1335#endif // OS versions of CompareIgnoreCase()
1336
finnur@chromium.org44ec9b3b2010-02-03 01:17:26[diff] [blame]1337
avi@google.com0b733222008-12-11 14:55:12[diff] [blame]1338void FilePath::StripTrailingSeparatorsInternal() {
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]1339 // If there is no drive letter, start will be 1, which will prevent stripping
1340 // the leading separator if there is only one separator. If there is a drive
1341 // letter, start will be set appropriately to prevent stripping the first
1342 // separator following the drive letter, if a separator immediately follows
1343 // the drive letter.
mark@chromium.org1906c032008-12-08 16:49:08[diff] [blame]1344 StringType::size_type start = FindDriveLetter(path_) + 2;
mark@chromium.org65d00aff2008-10-03 18:21:47[diff] [blame]1345
1346 StringType::size_type last_stripped = StringType::npos;
1347 for (StringType::size_type pos = path_.length();
1348 pos > start && IsSeparator(path_[pos - 1]);
1349 --pos) {
1350 // If the string only has two separators and they're at the beginning,
1351 // don't strip them, unless the string began with more than two separators.
1352 if (pos != start + 1 || last_stripped == start + 2 ||
1353 !IsSeparator(path_[start - 1])) {
1354 path_.resize(pos - 1);
1355 last_stripped = pos;
1356 }
1357 }
1358}
cevans@chromium.org06708612009-08-15 21:44:31[diff] [blame]1359
kinuko@chromium.orgd9034ed22012-02-10 02:04:40[diff] [blame]1360FilePath FilePath::NormalizePathSeparators() const {
scottmg@chromium.orgee55eace2014-04-03 07:45:42[diff] [blame]1361 return NormalizePathSeparatorsTo(kSeparators[0]);
1362}
1363
Alexander Timinbebb2002021-04-20 15:42:24[diff] [blame^]1364void FilePath::WriteIntoTrace(perfetto::TracedValue context) const {
Alexander Timin7372d5362021-03-09 00:55:31[diff] [blame]1365 perfetto::WriteIntoTracedValue(std::move(context), value());
1366}
1367
scottmg@chromium.orgee55eace2014-04-03 07:45:42[diff] [blame]1368FilePath FilePath::NormalizePathSeparatorsTo(CharType separator) const {
tony@chromium.org7c373772010-06-25 05:33:17[diff] [blame]1369#if defined(FILE_PATH_USES_WIN_SEPARATORS)
scottmg@chromium.orgee55eace2014-04-03 07:45:42[diff] [blame]1370 DCHECK_NE(kSeparators + kSeparatorsLength,
1371 std::find(kSeparators, kSeparators + kSeparatorsLength, separator));
tony@chromium.org7c373772010-06-25 05:33:17[diff] [blame]1372 StringType copy = path_;
scottmg@chromium.orgee55eace2014-04-03 07:45:42[diff] [blame]1373 for (size_t i = 0; i < kSeparatorsLength; ++i) {
1374 std::replace(copy.begin(), copy.end(), kSeparators[i], separator);
tony@chromium.org7c373772010-06-25 05:33:17[diff] [blame]1375 }
1376 return FilePath(copy);
kinuko@chromium.orgd9034ed22012-02-10 02:04:40[diff] [blame]1377#else
1378 return *this;
tony@chromium.org7c373772010-06-25 05:33:17[diff] [blame]1379#endif
kinuko@chromium.orgd9034ed22012-02-10 02:04:40[diff] [blame]1380}
gab@chromium.org1a0af402012-09-26 21:21:13[diff] [blame]1381
qinmin@chromium.orgf12d1e12013-11-20 07:04:55[diff] [blame]1382#if defined(OS_ANDROID)
1383bool FilePath::IsContentUri() const {
brettw95509312015-07-16 23:57:33[diff] [blame]1384 return StartsWith(path_, "content://", base::CompareCase::INSENSITIVE_ASCII);
qinmin@chromium.orgf12d1e12013-11-20 07:04:55[diff] [blame]1385}
1386#endif
1387
brettw@chromium.org04af979a2013-02-16 04:12:26[diff] [blame]1388} // namespace base