1 // Copyright 2012 The Chromium Authors
2 // Use of this source code is governed by a BSD-style license that can be
3 // found in the LICENSE file.
4
5 #include "base/files/file_path.h"
6
7 #include <string.h>
8
9 #include <algorithm>
10
11 #include "base/check_op.h"
12 #include "base/files/safe_base_name.h"
13 #include "base/numerics/safe_conversions.h"
14 #include "base/pickle.h"
15 #include "base/ranges/algorithm.h"
16 #include "base/strings/strcat.h"
17 #include "base/strings/string_piece.h"
18 #include "base/strings/string_util.h"
19 #include "base/strings/sys_string_conversions.h"
20 #include "base/strings/utf_ostream_operators.h"
21 #include "base/strings/utf_string_conversions.h"
22 #include "base/trace_event/base_tracing.h"
23
24 #if BUILDFLAG(IS_APPLE)
25 #include "base/apple/scoped_cftyperef.h"
26 #include "base/third_party/icu/icu_utf.h"
27 #endif
28
29 #if BUILDFLAG(IS_WIN)
30 #include <windows.h>
31
32 #include "base/win/win_util.h"
33 #elif BUILDFLAG(IS_APPLE)
34 #include <CoreFoundation/CoreFoundation.h>
35 #endif
36
37 namespace base {
38
39 using StringType = FilePath::StringType;
40 using StringPieceType = FilePath::StringPieceType;
41
42 namespace {
43
44 const char* const kCommonDoubleExtensionSuffixes[] = {
45 "bz", "bz2", "gz", "lz", "lzma", "lzo", "xz", "z", "zst"};
46 const char* const kCommonDoubleExtensions[] = {"user.js"};
47
48 const FilePath::CharType kStringTerminator = FILE_PATH_LITERAL('\0');
49
50 // If this FilePath contains a drive letter specification, returns the
51 // position of the last character of the drive letter specification,
52 // otherwise returns npos. This can only be true on Windows, when a pathname
53 // begins with a letter followed by a colon. On other platforms, this always
54 // returns npos.
FindDriveLetter(StringPieceType path)55 StringPieceType::size_type FindDriveLetter(StringPieceType path) {
56 #if defined(FILE_PATH_USES_DRIVE_LETTERS)
57 // This is dependent on an ASCII-based character set, but that's a
58 // reasonable assumption. iswalpha can be too inclusive here.
59 if (path.length() >= 2 && path[1] == L':' &&
60 ((path[0] >= L'A' && path[0] <= L'Z') ||
61 (path[0] >= L'a' && path[0] <= L'z'))) {
62 return 1;
63 }
64 #endif // FILE_PATH_USES_DRIVE_LETTERS
65 return StringType::npos;
66 }
67
68 #if defined(FILE_PATH_USES_DRIVE_LETTERS)
EqualDriveLetterCaseInsensitive(StringPieceType a,StringPieceType b)69 bool EqualDriveLetterCaseInsensitive(StringPieceType a, StringPieceType b) {
70 size_t a_letter_pos = FindDriveLetter(a);
71 size_t b_letter_pos = FindDriveLetter(b);
72
73 if (a_letter_pos == StringType::npos || b_letter_pos == StringType::npos)
74 return a == b;
75
76 StringPieceType a_letter(a.substr(0, a_letter_pos + 1));
77 StringPieceType b_letter(b.substr(0, b_letter_pos + 1));
78 if (!StartsWith(a_letter, b_letter, CompareCase::INSENSITIVE_ASCII))
79 return false;
80
81 StringPieceType a_rest(a.substr(a_letter_pos + 1));
82 StringPieceType b_rest(b.substr(b_letter_pos + 1));
83 return a_rest == b_rest;
84 }
85 #endif // defined(FILE_PATH_USES_DRIVE_LETTERS)
86
IsPathAbsolute(StringPieceType path)87 bool IsPathAbsolute(StringPieceType path) {
88 #if defined(FILE_PATH_USES_DRIVE_LETTERS)
89 StringType::size_type letter = FindDriveLetter(path);
90 if (letter != StringType::npos) {
91 // Look for a separator right after the drive specification.
92 return path.length() > letter + 1 &&
93 FilePath::IsSeparator(path[letter + 1]);
94 }
95 // Look for a pair of leading separators.
96 return path.length() > 1 &&
97 FilePath::IsSeparator(path[0]) && FilePath::IsSeparator(path[1]);
98 #else // FILE_PATH_USES_DRIVE_LETTERS
99 // Look for a separator in the first position.
100 return path.length() > 0 && FilePath::IsSeparator(path[0]);
101 #endif // FILE_PATH_USES_DRIVE_LETTERS
102 }
103
AreAllSeparators(const StringType & input)104 bool AreAllSeparators(const StringType& input) {
105 for (auto it : input) {
106 if (!FilePath::IsSeparator(it))
107 return false;
108 }
109
110 return true;
111 }
112
113 // Find the position of the '.' that separates the extension from the rest
114 // of the file name. The position is relative to BaseName(), not value().
115 // Returns npos if it can't find an extension.
FinalExtensionSeparatorPosition(const StringType & path)116 StringType::size_type FinalExtensionSeparatorPosition(const StringType& path) {
117 // Special case "." and ".."
118 if (path == FilePath::kCurrentDirectory || path == FilePath::kParentDirectory)
119 return StringType::npos;
120
121 return path.rfind(FilePath::kExtensionSeparator);
122 }
123
124 // Same as above, but allow a second extension component of up to 4
125 // characters when the rightmost extension component is a common double
126 // extension (gz, bz2, Z). For example, foo.tar.gz or foo.tar.Z would have
127 // extension components of '.tar.gz' and '.tar.Z' respectively.
ExtensionSeparatorPosition(const StringType & path)128 StringType::size_type ExtensionSeparatorPosition(const StringType& path) {
129 const StringType::size_type last_dot = FinalExtensionSeparatorPosition(path);
130
131 // No extension, or the extension is the whole filename.
132 if (last_dot == StringType::npos || last_dot == 0U)
133 return last_dot;
134
135 const StringType::size_type penultimate_dot =
136 path.rfind(FilePath::kExtensionSeparator, last_dot - 1);
137 const StringType::size_type last_separator =
138 path.find_last_of(FilePath::kSeparators, last_dot - 1,
139 FilePath::kSeparatorsLength - 1);
140
141 if (penultimate_dot == StringType::npos ||
142 (last_separator != StringType::npos &&
143 penultimate_dot < last_separator)) {
144 return last_dot;
145 }
146
147 for (auto* i : kCommonDoubleExtensions) {
148 StringType extension(path, penultimate_dot + 1);
149 if (EqualsCaseInsensitiveASCII(extension, i))
150 return penultimate_dot;
151 }
152
153 StringType extension(path, last_dot + 1);
154 for (auto* i : kCommonDoubleExtensionSuffixes) {
155 if (EqualsCaseInsensitiveASCII(extension, i)) {
156 if ((last_dot - penultimate_dot) <= 5U &&
157 (last_dot - penultimate_dot) > 1U) {
158 return penultimate_dot;
159 }
160 }
161 }
162
163 return last_dot;
164 }
165
166 // Returns true if path is "", ".", or "..".
IsEmptyOrSpecialCase(const StringType & path)167 bool IsEmptyOrSpecialCase(const StringType& path) {
168 // Special cases "", ".", and ".."
169 if (path.empty() || path == FilePath::kCurrentDirectory ||
170 path == FilePath::kParentDirectory) {
171 return true;
172 }
173
174 return false;
175 }
176
177 } // namespace
178
179 FilePath::FilePath() = default;
180
181 FilePath::FilePath(const FilePath& that) = default;
182 FilePath::FilePath(FilePath&& that) noexcept = default;
183
FilePath(StringPieceType path)184 FilePath::FilePath(StringPieceType path) : path_(path) {
185 StringType::size_type nul_pos = path_.find(kStringTerminator);
186 if (nul_pos != StringType::npos)
187 path_.erase(nul_pos, StringType::npos);
188 }
189
190 FilePath::~FilePath() = default;
191
192 FilePath& FilePath::operator=(const FilePath& that) = default;
193
194 FilePath& FilePath::operator=(FilePath&& that) noexcept = default;
195
operator ==(const FilePath & that) const196 bool FilePath::operator==(const FilePath& that) const {
197 #if defined(FILE_PATH_USES_DRIVE_LETTERS)
198 return EqualDriveLetterCaseInsensitive(this->path_, that.path_);
199 #else // defined(FILE_PATH_USES_DRIVE_LETTERS)
200 return path_ == that.path_;
201 #endif // defined(FILE_PATH_USES_DRIVE_LETTERS)
202 }
203
operator !=(const FilePath & that) const204 bool FilePath::operator!=(const FilePath& that) const {
205 #if defined(FILE_PATH_USES_DRIVE_LETTERS)
206 return !EqualDriveLetterCaseInsensitive(this->path_, that.path_);
207 #else // defined(FILE_PATH_USES_DRIVE_LETTERS)
208 return path_ != that.path_;
209 #endif // defined(FILE_PATH_USES_DRIVE_LETTERS)
210 }
211
operator <<(std::ostream & out,const FilePath & file_path)212 std::ostream& operator<<(std::ostream& out, const FilePath& file_path) {
213 return out << file_path.value();
214 }
215
216 // static
IsSeparator(CharType character)217 bool FilePath::IsSeparator(CharType character) {
218 for (size_t i = 0; i < kSeparatorsLength - 1; ++i) {
219 if (character == kSeparators[i]) {
220 return true;
221 }
222 }
223
224 return false;
225 }
226
GetComponents() const227 std::vector<FilePath::StringType> FilePath::GetComponents() const {
228 std::vector<StringType> ret_val;
229 if (value().empty())
230 return ret_val;
231
232 FilePath current = *this;
233 FilePath base;
234
235 // Capture path components.
236 while (current != current.DirName()) {
237 base = current.BaseName();
238 if (!AreAllSeparators(base.value()))
239 ret_val.push_back(base.value());
240 current = current.DirName();
241 }
242
243 // Capture root, if any.
244 base = current.BaseName();
245 if (!base.value().empty() && base.value() != kCurrentDirectory)
246 ret_val.push_back(current.BaseName().value());
247
248 // Capture drive letter, if any.
249 FilePath dir = current.DirName();
250 StringType::size_type letter = FindDriveLetter(dir.value());
251 if (letter != StringType::npos)
252 ret_val.emplace_back(dir.value(), 0, letter + 1);
253
254 ranges::reverse(ret_val);
255 return ret_val;
256 }
257
IsParent(const FilePath & child) const258 bool FilePath::IsParent(const FilePath& child) const {
259 return AppendRelativePath(child, nullptr);
260 }
261
AppendRelativePath(const FilePath & child,FilePath * path) const262 bool FilePath::AppendRelativePath(const FilePath& child,
263 FilePath* path) const {
264 std::vector<StringType> parent_components = GetComponents();
265 std::vector<StringType> child_components = child.GetComponents();
266
267 if (parent_components.empty() ||
268 parent_components.size() >= child_components.size())
269 return false;
270
271 std::vector<StringType>::const_iterator parent_comp =
272 parent_components.begin();
273 std::vector<StringType>::const_iterator child_comp =
274 child_components.begin();
275
276 #if defined(FILE_PATH_USES_DRIVE_LETTERS)
277 // Windows can access case sensitive filesystems, so component
278 // comparisions must be case sensitive, but drive letters are
279 // never case sensitive.
280 if ((FindDriveLetter(*parent_comp) != StringType::npos) &&
281 (FindDriveLetter(*child_comp) != StringType::npos)) {
282 if (!StartsWith(*parent_comp, *child_comp, CompareCase::INSENSITIVE_ASCII))
283 return false;
284 ++parent_comp;
285 ++child_comp;
286 }
287 #endif // defined(FILE_PATH_USES_DRIVE_LETTERS)
288
289 // The first 2 components for network paths are [<2-Separators>, <hostname>].
290 // Use case-insensitive comparison for the hostname.
291 // https://tools.ietf.org/html/rfc3986#section-3.2.2
292 if (IsNetwork() && parent_components.size() > 1) {
293 if (*parent_comp++ != *child_comp++ ||
294 !base::EqualsCaseInsensitiveASCII(*parent_comp++, *child_comp++)) {
295 return false;
296 }
297 }
298
299 while (parent_comp != parent_components.end()) {
300 if (*parent_comp != *child_comp)
301 return false;
302 ++parent_comp;
303 ++child_comp;
304 }
305
306 if (path != nullptr) {
307 for (; child_comp != child_components.end(); ++child_comp) {
308 *path = path->Append(*child_comp);
309 }
310 }
311 return true;
312 }
313
314 // libgen's dirname and basename aren't guaranteed to be thread-safe and aren't
315 // guaranteed to not modify their input strings, and in fact are implemented
316 // differently in this regard on different platforms. Don't use them, but
317 // adhere to their behavior.
DirName() const318 FilePath FilePath::DirName() const {
319 FilePath new_path(path_);
320 new_path.StripTrailingSeparatorsInternal();
321
322 // The drive letter, if any, always needs to remain in the output. If there
323 // is no drive letter, as will always be the case on platforms which do not
324 // support drive letters, letter will be npos, or -1, so the comparisons and
325 // resizes below using letter will still be valid.
326 StringType::size_type letter = FindDriveLetter(new_path.path_);
327
328 StringType::size_type last_separator =
329 new_path.path_.find_last_of(kSeparators, StringType::npos,
330 kSeparatorsLength - 1);
331 if (last_separator == StringType::npos) {
332 // path_ is in the current directory.
333 new_path.path_.resize(letter + 1);
334 } else if (last_separator == letter + 1) {
335 // path_ is in the root directory.
336 new_path.path_.resize(letter + 2);
337 } else if (last_separator == letter + 2 &&
338 IsSeparator(new_path.path_[letter + 1])) {
339 // path_ is in "//" (possibly with a drive letter); leave the double
340 // separator intact indicating alternate root.
341 new_path.path_.resize(letter + 3);
342 } else if (last_separator != 0) {
343 bool trim_to_basename = true;
344 #if BUILDFLAG(IS_POSIX)
345 // On Posix, more than two leading separators are always collapsed to one.
346 // See
347 // https://pubs.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap04.html#tag_04_13
348 // So, do not strip any of the separators, let
349 // StripTrailingSeparatorsInternal() take care of the extra.
350 if (AreAllSeparators(new_path.path_.substr(0, last_separator + 1))) {
351 new_path.path_.resize(last_separator + 1);
352 trim_to_basename = false;
353 }
354 #endif // BUILDFLAG(IS_POSIX)
355 if (trim_to_basename) {
356 // path_ is somewhere else, trim the basename.
357 new_path.path_.resize(last_separator);
358 }
359 }
360
361 new_path.StripTrailingSeparatorsInternal();
362 if (!new_path.path_.length())
363 new_path.path_ = kCurrentDirectory;
364
365 return new_path;
366 }
367
BaseName() const368 FilePath FilePath::BaseName() const {
369 FilePath new_path(path_);
370 new_path.StripTrailingSeparatorsInternal();
371
372 // The drive letter, if any, is always stripped.
373 StringType::size_type letter = FindDriveLetter(new_path.path_);
374 if (letter != StringType::npos) {
375 new_path.path_.erase(0, letter + 1);
376 }
377
378 // Keep everything after the final separator, but if the pathname is only
379 // one character and it's a separator, leave it alone.
380 StringType::size_type last_separator =
381 new_path.path_.find_last_of(kSeparators, StringType::npos,
382 kSeparatorsLength - 1);
383 if (last_separator != StringType::npos &&
384 last_separator < new_path.path_.length() - 1) {
385 new_path.path_.erase(0, last_separator + 1);
386 }
387
388 return new_path;
389 }
390
Extension() const391 StringType FilePath::Extension() const {
392 FilePath base(BaseName());
393 const StringType::size_type dot = ExtensionSeparatorPosition(base.path_);
394 if (dot == StringType::npos)
395 return StringType();
396
397 return base.path_.substr(dot, StringType::npos);
398 }
399
FinalExtension() const400 StringType FilePath::FinalExtension() const {
401 FilePath base(BaseName());
402 const StringType::size_type dot = FinalExtensionSeparatorPosition(base.path_);
403 if (dot == StringType::npos)
404 return StringType();
405
406 return base.path_.substr(dot, StringType::npos);
407 }
408
RemoveExtension() const409 FilePath FilePath::RemoveExtension() const {
410 if (Extension().empty())
411 return *this;
412
413 const StringType::size_type dot = ExtensionSeparatorPosition(path_);
414 if (dot == StringType::npos)
415 return *this;
416
417 return FilePath(path_.substr(0, dot));
418 }
419
RemoveFinalExtension() const420 FilePath FilePath::RemoveFinalExtension() const {
421 if (FinalExtension().empty())
422 return *this;
423
424 const StringType::size_type dot = FinalExtensionSeparatorPosition(path_);
425 if (dot == StringType::npos)
426 return *this;
427
428 return FilePath(path_.substr(0, dot));
429 }
430
InsertBeforeExtension(StringPieceType suffix) const431 FilePath FilePath::InsertBeforeExtension(StringPieceType suffix) const {
432 if (suffix.empty())
433 return FilePath(path_);
434
435 if (IsEmptyOrSpecialCase(BaseName().value()))
436 return FilePath();
437
438 return FilePath(
439 base::StrCat({RemoveExtension().value(), suffix, Extension()}));
440 }
441
InsertBeforeExtensionASCII(StringPiece suffix) const442 FilePath FilePath::InsertBeforeExtensionASCII(StringPiece suffix)
443 const {
444 DCHECK(IsStringASCII(suffix));
445 #if BUILDFLAG(IS_WIN)
446 return InsertBeforeExtension(UTF8ToWide(suffix));
447 #elif BUILDFLAG(IS_POSIX) || BUILDFLAG(IS_FUCHSIA)
448 return InsertBeforeExtension(suffix);
449 #endif
450 }
451
AddExtension(StringPieceType extension) const452 FilePath FilePath::AddExtension(StringPieceType extension) const {
453 if (IsEmptyOrSpecialCase(BaseName().value()))
454 return FilePath();
455
456 // If the new extension is "" or ".", then just return the current FilePath.
457 if (extension.empty() ||
458 (extension.size() == 1 && extension[0] == kExtensionSeparator))
459 return *this;
460
461 StringType str = path_;
462 if (extension[0] != kExtensionSeparator &&
463 *(str.end() - 1) != kExtensionSeparator) {
464 str.append(1, kExtensionSeparator);
465 }
466 str.append(extension);
467 return FilePath(str);
468 }
469
AddExtensionASCII(StringPiece extension) const470 FilePath FilePath::AddExtensionASCII(StringPiece extension) const {
471 DCHECK(IsStringASCII(extension));
472 #if BUILDFLAG(IS_WIN)
473 return AddExtension(UTF8ToWide(extension));
474 #elif BUILDFLAG(IS_POSIX) || BUILDFLAG(IS_FUCHSIA)
475 return AddExtension(extension);
476 #endif
477 }
478
ReplaceExtension(StringPieceType extension) const479 FilePath FilePath::ReplaceExtension(StringPieceType extension) const {
480 if (IsEmptyOrSpecialCase(BaseName().value()))
481 return FilePath();
482
483 FilePath no_ext = RemoveExtension();
484 // If the new extension is "" or ".", then just remove the current extension.
485 if (extension.empty() ||
486 (extension.size() == 1 && extension[0] == kExtensionSeparator))
487 return no_ext;
488
489 StringType str = no_ext.value();
490 if (extension[0] != kExtensionSeparator)
491 str.append(1, kExtensionSeparator);
492 str.append(extension);
493 return FilePath(str);
494 }
495
MatchesExtension(StringPieceType extension) const496 bool FilePath::MatchesExtension(StringPieceType extension) const {
497 DCHECK(extension.empty() || extension[0] == kExtensionSeparator);
498
499 StringType current_extension = Extension();
500
501 if (current_extension.length() != extension.length())
502 return false;
503
504 return FilePath::CompareEqualIgnoreCase(extension, current_extension);
505 }
506
MatchesFinalExtension(StringPieceType extension) const507 bool FilePath::MatchesFinalExtension(StringPieceType extension) const {
508 DCHECK(extension.empty() || extension[0] == kExtensionSeparator);
509
510 StringType current_final_extension = FinalExtension();
511
512 if (current_final_extension.length() != extension.length())
513 return false;
514
515 return FilePath::CompareEqualIgnoreCase(extension, current_final_extension);
516 }
517
Append(StringPieceType component) const518 FilePath FilePath::Append(StringPieceType component) const {
519 StringPieceType appended = component;
520 StringType without_nuls;
521
522 StringType::size_type nul_pos = component.find(kStringTerminator);
523 if (nul_pos != StringPieceType::npos) {
524 without_nuls = StringType(component.substr(0, nul_pos));
525 appended = StringPieceType(without_nuls);
526 }
527
528 DCHECK(!IsPathAbsolute(appended));
529
530 if (path_.compare(kCurrentDirectory) == 0 && !appended.empty()) {
531 // Append normally doesn't do any normalization, but as a special case,
532 // when appending to kCurrentDirectory, just return a new path for the
533 // component argument. Appending component to kCurrentDirectory would
534 // serve no purpose other than needlessly lengthening the path, and
535 // it's likely in practice to wind up with FilePath objects containing
536 // only kCurrentDirectory when calling DirName on a single relative path
537 // component.
538 return FilePath(appended);
539 }
540
541 FilePath new_path(path_);
542 new_path.StripTrailingSeparatorsInternal();
543
544 // Don't append a separator if the path is empty (indicating the current
545 // directory) or if the path component is empty (indicating nothing to
546 // append).
547 if (!appended.empty() && !new_path.path_.empty()) {
548 // Don't append a separator if the path still ends with a trailing
549 // separator after stripping (indicating the root directory).
550 if (!IsSeparator(new_path.path_.back())) {
551 // Don't append a separator if the path is just a drive letter.
552 if (FindDriveLetter(new_path.path_) + 1 != new_path.path_.length()) {
553 new_path.path_.append(1, kSeparators[0]);
554 }
555 }
556 }
557
558 new_path.path_.append(appended);
559 return new_path;
560 }
561
Append(const FilePath & component) const562 FilePath FilePath::Append(const FilePath& component) const {
563 return Append(component.value());
564 }
565
Append(const SafeBaseName & component) const566 FilePath FilePath::Append(const SafeBaseName& component) const {
567 return Append(component.path().value());
568 }
569
AppendASCII(StringPiece component) const570 FilePath FilePath::AppendASCII(StringPiece component) const {
571 DCHECK(base::IsStringASCII(component));
572 #if BUILDFLAG(IS_WIN)
573 return Append(UTF8ToWide(component));
574 #elif BUILDFLAG(IS_POSIX) || BUILDFLAG(IS_FUCHSIA)
575 return Append(component);
576 #endif
577 }
578
IsAbsolute() const579 bool FilePath::IsAbsolute() const {
580 return IsPathAbsolute(path_);
581 }
582
IsNetwork() const583 bool FilePath::IsNetwork() const {
584 return path_.length() > 1 && FilePath::IsSeparator(path_[0]) &&
585 FilePath::IsSeparator(path_[1]);
586 }
587
EndsWithSeparator() const588 bool FilePath::EndsWithSeparator() const {
589 if (empty())
590 return false;
591 return IsSeparator(path_.back());
592 }
593
AsEndingWithSeparator() const594 FilePath FilePath::AsEndingWithSeparator() const {
595 if (EndsWithSeparator() || path_.empty())
596 return *this;
597
598 StringType path_str;
599 path_str.reserve(path_.length() + 1); // Only allocate string once.
600
601 path_str = path_;
602 path_str.append(&kSeparators[0], 1);
603 return FilePath(path_str);
604 }
605
StripTrailingSeparators() const606 FilePath FilePath::StripTrailingSeparators() const {
607 FilePath new_path(path_);
608 new_path.StripTrailingSeparatorsInternal();
609
610 return new_path;
611 }
612
ReferencesParent() const613 bool FilePath::ReferencesParent() const {
614 if (path_.find(kParentDirectory) == StringType::npos) {
615 // GetComponents is quite expensive, so avoid calling it in the majority
616 // of cases where there isn't a kParentDirectory anywhere in the path.
617 return false;
618 }
619
620 std::vector<StringType> components = GetComponents();
621 std::vector<StringType>::const_iterator it = components.begin();
622 for (; it != components.end(); ++it) {
623 const StringType& component = *it;
624 // Windows has odd, undocumented behavior with path components containing
625 // only whitespace and . characters. So, if all we see is . and
626 // whitespace, then we treat any .. sequence as referencing parent.
627 // For simplicity we enforce this on all platforms.
628 if (component.find_first_not_of(FILE_PATH_LITERAL(". \n\r\t")) ==
629 std::string::npos &&
630 component.find(kParentDirectory) != std::string::npos) {
631 return true;
632 }
633 }
634 return false;
635 }
636
637 #if BUILDFLAG(IS_WIN)
638
LossyDisplayName() const639 std::u16string FilePath::LossyDisplayName() const {
640 return AsString16(path_);
641 }
642
MaybeAsASCII() const643 std::string FilePath::MaybeAsASCII() const {
644 return base::IsStringASCII(path_) ? WideToASCII(path_) : std::string();
645 }
646
AsUTF8Unsafe() const647 std::string FilePath::AsUTF8Unsafe() const {
648 return WideToUTF8(value());
649 }
650
AsUTF16Unsafe() const651 std::u16string FilePath::AsUTF16Unsafe() const {
652 return WideToUTF16(value());
653 }
654
655 // static
FromASCII(StringPiece ascii)656 FilePath FilePath::FromASCII(StringPiece ascii) {
657 DCHECK(base::IsStringASCII(ascii));
658 return FilePath(ASCIIToWide(ascii));
659 }
660
661 // static
FromUTF8Unsafe(StringPiece utf8)662 FilePath FilePath::FromUTF8Unsafe(StringPiece utf8) {
663 return FilePath(UTF8ToWide(utf8));
664 }
665
666 // static
FromUTF16Unsafe(StringPiece16 utf16)667 FilePath FilePath::FromUTF16Unsafe(StringPiece16 utf16) {
668 return FilePath(AsWStringView(utf16));
669 }
670
671 #elif BUILDFLAG(IS_POSIX) || BUILDFLAG(IS_FUCHSIA)
672
673 // See file_path.h for a discussion of the encoding of paths on POSIX
674 // platforms. These encoding conversion functions are not quite correct.
675
LossyDisplayName() const676 std::u16string FilePath::LossyDisplayName() const {
677 return WideToUTF16(SysNativeMBToWide(path_));
678 }
679
MaybeAsASCII() const680 std::string FilePath::MaybeAsASCII() const {
681 if (base::IsStringASCII(path_))
682 return path_;
683 return std::string();
684 }
685
AsUTF8Unsafe() const686 std::string FilePath::AsUTF8Unsafe() const {
687 #if defined(SYSTEM_NATIVE_UTF8)
688 return value();
689 #else
690 return WideToUTF8(SysNativeMBToWide(value()));
691 #endif
692 }
693
AsUTF16Unsafe() const694 std::u16string FilePath::AsUTF16Unsafe() const {
695 #if defined(SYSTEM_NATIVE_UTF8)
696 return UTF8ToUTF16(value());
697 #else
698 return WideToUTF16(SysNativeMBToWide(value()));
699 #endif
700 }
701
702 // static
FromASCII(StringPiece ascii)703 FilePath FilePath::FromASCII(StringPiece ascii) {
704 DCHECK(base::IsStringASCII(ascii));
705 return FilePath(ascii);
706 }
707
708 // static
FromUTF8Unsafe(StringPiece utf8)709 FilePath FilePath::FromUTF8Unsafe(StringPiece utf8) {
710 #if defined(SYSTEM_NATIVE_UTF8)
711 return FilePath(utf8);
712 #else
713 return FilePath(SysWideToNativeMB(UTF8ToWide(utf8)));
714 #endif
715 }
716
717 // static
FromUTF16Unsafe(StringPiece16 utf16)718 FilePath FilePath::FromUTF16Unsafe(StringPiece16 utf16) {
719 #if defined(SYSTEM_NATIVE_UTF8)
720 return FilePath(UTF16ToUTF8(utf16));
721 #else
722 return FilePath(SysWideToNativeMB(UTF16ToWide(utf16)));
723 #endif
724 }
725
726 #endif // BUILDFLAG(IS_WIN)
727
WriteToPickle(Pickle * pickle) const728 void FilePath::WriteToPickle(Pickle* pickle) const {
729 #if BUILDFLAG(IS_WIN)
730 pickle->WriteString16(AsStringPiece16(path_));
731 #elif BUILDFLAG(IS_POSIX) || BUILDFLAG(IS_FUCHSIA)
732 pickle->WriteString(path_);
733 #else
734 #error Unsupported platform
735 #endif
736 }
737
ReadFromPickle(PickleIterator * iter)738 bool FilePath::ReadFromPickle(PickleIterator* iter) {
739 #if BUILDFLAG(IS_WIN)
740 std::u16string path;
741 if (!iter->ReadString16(&path))
742 return false;
743 path_ = UTF16ToWide(path);
744 #elif BUILDFLAG(IS_POSIX) || BUILDFLAG(IS_FUCHSIA)
745 if (!iter->ReadString(&path_))
746 return false;
747 #else
748 #error Unsupported platform
749 #endif
750
751 if (path_.find(kStringTerminator) != StringType::npos)
752 return false;
753
754 return true;
755 }
756
757 #if BUILDFLAG(IS_WIN)
758 // Windows specific implementation of file string comparisons.
759
CompareIgnoreCase(StringPieceType string1,StringPieceType string2)760 int FilePath::CompareIgnoreCase(StringPieceType string1,
761 StringPieceType string2) {
762 // CharUpperW within user32 is used here because it will provide unicode
763 // conversions regardless of locale. The STL alternative, towupper, has a
764 // locale consideration that prevents it from converting all characters by
765 // default.
766 CHECK(win::IsUser32AndGdi32Available());
767 // Perform character-wise upper case comparison rather than using the
768 // fully Unicode-aware CompareString(). For details see:
769 // http://blogs.msdn.com/michkap/archive/2005/10/17/481600.aspx
770 StringPieceType::const_iterator i1 = string1.begin();
771 StringPieceType::const_iterator i2 = string2.begin();
772 StringPieceType::const_iterator string1end = string1.end();
773 StringPieceType::const_iterator string2end = string2.end();
774 for ( ; i1 != string1end && i2 != string2end; ++i1, ++i2) {
775 wchar_t c1 =
776 (wchar_t)LOWORD(::CharUpperW((LPWSTR)(DWORD_PTR)MAKELONG(*i1, 0)));
777 wchar_t c2 =
778 (wchar_t)LOWORD(::CharUpperW((LPWSTR)(DWORD_PTR)MAKELONG(*i2, 0)));
779 if (c1 < c2)
780 return -1;
781 if (c1 > c2)
782 return 1;
783 }
784 if (i1 != string1end)
785 return 1;
786 if (i2 != string2end)
787 return -1;
788 return 0;
789 }
790
791 #elif BUILDFLAG(IS_APPLE)
792 // Mac OS X specific implementation of file string comparisons.
793
794 // cf. https://developer.apple.com/library/archive/technotes/tn/tn1150.html#UnicodeSubtleties
795 //
796 // "When using CreateTextEncoding to create a text encoding, you should set
797 // the TextEncodingBase to kTextEncodingUnicodeV2_0, set the
798 // TextEncodingVariant to kUnicodeCanonicalDecompVariant, and set the
799 // TextEncodingFormat to kUnicode16BitFormat. Using these values ensures that
800 // the Unicode will be in the same form as on an HFS Plus volume, even as the
801 // Unicode standard evolves."
802 //
803 // Another technical article for X 10.4 updates this: one should use
804 // the new (unambiguous) kUnicodeHFSPlusDecompVariant.
805 // cf. http://developer.apple.com/mac/library/releasenotes/TextFonts/RN-TEC/index.html
806 //
807 // This implementation uses CFStringGetFileSystemRepresentation() to get the
808 // decomposed form, and an adapted version of the FastUnicodeCompare as
809 // described in the tech note to compare the strings.
810
811 // Character conversion table for FastUnicodeCompare()
812 //
813 // The lower case table consists of a 256-entry high-byte table followed by
814 // some number of 256-entry subtables. The high-byte table contains either an
815 // offset to the subtable for characters with that high byte or zero, which
816 // means that there are no case mappings or ignored characters in that block.
817 // Ignored characters are mapped to zero.
818 //
819 // cf. downloadable file linked in
820 // https://developer.apple.com/library/archive/technotes/tn/tn1150.html#Downloads
821
822 namespace {
823
824 // clang-format off
825 const UInt16 lower_case_table[11 * 256] = {
826 // High-byte indices ( == 0 iff no case mapping and no ignorables )
827
828 /* 0 */ 0x0100, 0x0200, 0x0000, 0x0300, 0x0400, 0x0500, 0x0000, 0x0000,
829 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
830 /* 1 */ 0x0600, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
831 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
832 /* 2 */ 0x0700, 0x0800, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
833 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
834 /* 3 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
835 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
836 /* 4 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
837 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
838 /* 5 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
839 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
840 /* 6 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
841 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
842 /* 7 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
843 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
844 /* 8 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
845 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
846 /* 9 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
847 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
848 /* A */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
849 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
850 /* B */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
851 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
852 /* C */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
853 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
854 /* D */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
855 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
856 /* E */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
857 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
858 /* F */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
859 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0900, 0x0A00,
860
861 // Table 1 (for high byte 0x00)
862
863 /* 0 */ 0xFFFF, 0x0001, 0x0002, 0x0003, 0x0004, 0x0005, 0x0006, 0x0007,
864 0x0008, 0x0009, 0x000A, 0x000B, 0x000C, 0x000D, 0x000E, 0x000F,
865 /* 1 */ 0x0010, 0x0011, 0x0012, 0x0013, 0x0014, 0x0015, 0x0016, 0x0017,
866 0x0018, 0x0019, 0x001A, 0x001B, 0x001C, 0x001D, 0x001E, 0x001F,
867 /* 2 */ 0x0020, 0x0021, 0x0022, 0x0023, 0x0024, 0x0025, 0x0026, 0x0027,
868 0x0028, 0x0029, 0x002A, 0x002B, 0x002C, 0x002D, 0x002E, 0x002F,
869 /* 3 */ 0x0030, 0x0031, 0x0032, 0x0033, 0x0034, 0x0035, 0x0036, 0x0037,
870 0x0038, 0x0039, 0x003A, 0x003B, 0x003C, 0x003D, 0x003E, 0x003F,
871 /* 4 */ 0x0040, 0x0061, 0x0062, 0x0063, 0x0064, 0x0065, 0x0066, 0x0067,
872 0x0068, 0x0069, 0x006A, 0x006B, 0x006C, 0x006D, 0x006E, 0x006F,
873 /* 5 */ 0x0070, 0x0071, 0x0072, 0x0073, 0x0074, 0x0075, 0x0076, 0x0077,
874 0x0078, 0x0079, 0x007A, 0x005B, 0x005C, 0x005D, 0x005E, 0x005F,
875 /* 6 */ 0x0060, 0x0061, 0x0062, 0x0063, 0x0064, 0x0065, 0x0066, 0x0067,
876 0x0068, 0x0069, 0x006A, 0x006B, 0x006C, 0x006D, 0x006E, 0x006F,
877 /* 7 */ 0x0070, 0x0071, 0x0072, 0x0073, 0x0074, 0x0075, 0x0076, 0x0077,
878 0x0078, 0x0079, 0x007A, 0x007B, 0x007C, 0x007D, 0x007E, 0x007F,
879 /* 8 */ 0x0080, 0x0081, 0x0082, 0x0083, 0x0084, 0x0085, 0x0086, 0x0087,
880 0x0088, 0x0089, 0x008A, 0x008B, 0x008C, 0x008D, 0x008E, 0x008F,
881 /* 9 */ 0x0090, 0x0091, 0x0092, 0x0093, 0x0094, 0x0095, 0x0096, 0x0097,
882 0x0098, 0x0099, 0x009A, 0x009B, 0x009C, 0x009D, 0x009E, 0x009F,
883 /* A */ 0x00A0, 0x00A1, 0x00A2, 0x00A3, 0x00A4, 0x00A5, 0x00A6, 0x00A7,
884 0x00A8, 0x00A9, 0x00AA, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00AF,
885 /* B */ 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x00B4, 0x00B5, 0x00B6, 0x00B7,
886 0x00B8, 0x00B9, 0x00BA, 0x00BB, 0x00BC, 0x00BD, 0x00BE, 0x00BF,
887 /* C */ 0x00C0, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00E6, 0x00C7,
888 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF,
889 /* D */ 0x00F0, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x00D7,
890 0x00F8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x00DD, 0x00FE, 0x00DF,
891 /* E */ 0x00E0, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x00E7,
892 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF,
893 /* F */ 0x00F0, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x00F7,
894 0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x00FD, 0x00FE, 0x00FF,
895
896 // Table 2 (for high byte 0x01)
897
898 /* 0 */ 0x0100, 0x0101, 0x0102, 0x0103, 0x0104, 0x0105, 0x0106, 0x0107,
899 0x0108, 0x0109, 0x010A, 0x010B, 0x010C, 0x010D, 0x010E, 0x010F,
900 /* 1 */ 0x0111, 0x0111, 0x0112, 0x0113, 0x0114, 0x0115, 0x0116, 0x0117,
901 0x0118, 0x0119, 0x011A, 0x011B, 0x011C, 0x011D, 0x011E, 0x011F,
902 /* 2 */ 0x0120, 0x0121, 0x0122, 0x0123, 0x0124, 0x0125, 0x0127, 0x0127,
903 0x0128, 0x0129, 0x012A, 0x012B, 0x012C, 0x012D, 0x012E, 0x012F,
904 /* 3 */ 0x0130, 0x0131, 0x0133, 0x0133, 0x0134, 0x0135, 0x0136, 0x0137,
905 0x0138, 0x0139, 0x013A, 0x013B, 0x013C, 0x013D, 0x013E, 0x0140,
906 /* 4 */ 0x0140, 0x0142, 0x0142, 0x0143, 0x0144, 0x0145, 0x0146, 0x0147,
907 0x0148, 0x0149, 0x014B, 0x014B, 0x014C, 0x014D, 0x014E, 0x014F,
908 /* 5 */ 0x0150, 0x0151, 0x0153, 0x0153, 0x0154, 0x0155, 0x0156, 0x0157,
909 0x0158, 0x0159, 0x015A, 0x015B, 0x015C, 0x015D, 0x015E, 0x015F,
910 /* 6 */ 0x0160, 0x0161, 0x0162, 0x0163, 0x0164, 0x0165, 0x0167, 0x0167,
911 0x0168, 0x0169, 0x016A, 0x016B, 0x016C, 0x016D, 0x016E, 0x016F,
912 /* 7 */ 0x0170, 0x0171, 0x0172, 0x0173, 0x0174, 0x0175, 0x0176, 0x0177,
913 0x0178, 0x0179, 0x017A, 0x017B, 0x017C, 0x017D, 0x017E, 0x017F,
914 /* 8 */ 0x0180, 0x0253, 0x0183, 0x0183, 0x0185, 0x0185, 0x0254, 0x0188,
915 0x0188, 0x0256, 0x0257, 0x018C, 0x018C, 0x018D, 0x01DD, 0x0259,
916 /* 9 */ 0x025B, 0x0192, 0x0192, 0x0260, 0x0263, 0x0195, 0x0269, 0x0268,
917 0x0199, 0x0199, 0x019A, 0x019B, 0x026F, 0x0272, 0x019E, 0x0275,
918 /* A */ 0x01A0, 0x01A1, 0x01A3, 0x01A3, 0x01A5, 0x01A5, 0x01A6, 0x01A8,
919 0x01A8, 0x0283, 0x01AA, 0x01AB, 0x01AD, 0x01AD, 0x0288, 0x01AF,
920 /* B */ 0x01B0, 0x028A, 0x028B, 0x01B4, 0x01B4, 0x01B6, 0x01B6, 0x0292,
921 0x01B9, 0x01B9, 0x01BA, 0x01BB, 0x01BD, 0x01BD, 0x01BE, 0x01BF,
922 /* C */ 0x01C0, 0x01C1, 0x01C2, 0x01C3, 0x01C6, 0x01C6, 0x01C6, 0x01C9,
923 0x01C9, 0x01C9, 0x01CC, 0x01CC, 0x01CC, 0x01CD, 0x01CE, 0x01CF,
924 /* D */ 0x01D0, 0x01D1, 0x01D2, 0x01D3, 0x01D4, 0x01D5, 0x01D6, 0x01D7,
925 0x01D8, 0x01D9, 0x01DA, 0x01DB, 0x01DC, 0x01DD, 0x01DE, 0x01DF,
926 /* E */ 0x01E0, 0x01E1, 0x01E2, 0x01E3, 0x01E5, 0x01E5, 0x01E6, 0x01E7,
927 0x01E8, 0x01E9, 0x01EA, 0x01EB, 0x01EC, 0x01ED, 0x01EE, 0x01EF,
928 /* F */ 0x01F0, 0x01F3, 0x01F3, 0x01F3, 0x01F4, 0x01F5, 0x01F6, 0x01F7,
929 0x01F8, 0x01F9, 0x01FA, 0x01FB, 0x01FC, 0x01FD, 0x01FE, 0x01FF,
930
931 // Table 3 (for high byte 0x03)
932
933 /* 0 */ 0x0300, 0x0301, 0x0302, 0x0303, 0x0304, 0x0305, 0x0306, 0x0307,
934 0x0308, 0x0309, 0x030A, 0x030B, 0x030C, 0x030D, 0x030E, 0x030F,
935 /* 1 */ 0x0310, 0x0311, 0x0312, 0x0313, 0x0314, 0x0315, 0x0316, 0x0317,
936 0x0318, 0x0319, 0x031A, 0x031B, 0x031C, 0x031D, 0x031E, 0x031F,
937 /* 2 */ 0x0320, 0x0321, 0x0322, 0x0323, 0x0324, 0x0325, 0x0326, 0x0327,
938 0x0328, 0x0329, 0x032A, 0x032B, 0x032C, 0x032D, 0x032E, 0x032F,
939 /* 3 */ 0x0330, 0x0331, 0x0332, 0x0333, 0x0334, 0x0335, 0x0336, 0x0337,
940 0x0338, 0x0339, 0x033A, 0x033B, 0x033C, 0x033D, 0x033E, 0x033F,
941 /* 4 */ 0x0340, 0x0341, 0x0342, 0x0343, 0x0344, 0x0345, 0x0346, 0x0347,
942 0x0348, 0x0349, 0x034A, 0x034B, 0x034C, 0x034D, 0x034E, 0x034F,
943 /* 5 */ 0x0350, 0x0351, 0x0352, 0x0353, 0x0354, 0x0355, 0x0356, 0x0357,
944 0x0358, 0x0359, 0x035A, 0x035B, 0x035C, 0x035D, 0x035E, 0x035F,
945 /* 6 */ 0x0360, 0x0361, 0x0362, 0x0363, 0x0364, 0x0365, 0x0366, 0x0367,
946 0x0368, 0x0369, 0x036A, 0x036B, 0x036C, 0x036D, 0x036E, 0x036F,
947 /* 7 */ 0x0370, 0x0371, 0x0372, 0x0373, 0x0374, 0x0375, 0x0376, 0x0377,
948 0x0378, 0x0379, 0x037A, 0x037B, 0x037C, 0x037D, 0x037E, 0x037F,
949 /* 8 */ 0x0380, 0x0381, 0x0382, 0x0383, 0x0384, 0x0385, 0x0386, 0x0387,
950 0x0388, 0x0389, 0x038A, 0x038B, 0x038C, 0x038D, 0x038E, 0x038F,
951 /* 9 */ 0x0390, 0x03B1, 0x03B2, 0x03B3, 0x03B4, 0x03B5, 0x03B6, 0x03B7,
952 0x03B8, 0x03B9, 0x03BA, 0x03BB, 0x03BC, 0x03BD, 0x03BE, 0x03BF,
953 /* A */ 0x03C0, 0x03C1, 0x03A2, 0x03C3, 0x03C4, 0x03C5, 0x03C6, 0x03C7,
954 0x03C8, 0x03C9, 0x03AA, 0x03AB, 0x03AC, 0x03AD, 0x03AE, 0x03AF,
955 /* B */ 0x03B0, 0x03B1, 0x03B2, 0x03B3, 0x03B4, 0x03B5, 0x03B6, 0x03B7,
956 0x03B8, 0x03B9, 0x03BA, 0x03BB, 0x03BC, 0x03BD, 0x03BE, 0x03BF,
957 /* C */ 0x03C0, 0x03C1, 0x03C2, 0x03C3, 0x03C4, 0x03C5, 0x03C6, 0x03C7,
958 0x03C8, 0x03C9, 0x03CA, 0x03CB, 0x03CC, 0x03CD, 0x03CE, 0x03CF,
959 /* D */ 0x03D0, 0x03D1, 0x03D2, 0x03D3, 0x03D4, 0x03D5, 0x03D6, 0x03D7,
960 0x03D8, 0x03D9, 0x03DA, 0x03DB, 0x03DC, 0x03DD, 0x03DE, 0x03DF,
961 /* E */ 0x03E0, 0x03E1, 0x03E3, 0x03E3, 0x03E5, 0x03E5, 0x03E7, 0x03E7,
962 0x03E9, 0x03E9, 0x03EB, 0x03EB, 0x03ED, 0x03ED, 0x03EF, 0x03EF,
963 /* F */ 0x03F0, 0x03F1, 0x03F2, 0x03F3, 0x03F4, 0x03F5, 0x03F6, 0x03F7,
964 0x03F8, 0x03F9, 0x03FA, 0x03FB, 0x03FC, 0x03FD, 0x03FE, 0x03FF,
965
966 // Table 4 (for high byte 0x04)
967
968 /* 0 */ 0x0400, 0x0401, 0x0452, 0x0403, 0x0454, 0x0455, 0x0456, 0x0407,
969 0x0458, 0x0459, 0x045A, 0x045B, 0x040C, 0x040D, 0x040E, 0x045F,
970 /* 1 */ 0x0430, 0x0431, 0x0432, 0x0433, 0x0434, 0x0435, 0x0436, 0x0437,
971 0x0438, 0x0419, 0x043A, 0x043B, 0x043C, 0x043D, 0x043E, 0x043F,
972 /* 2 */ 0x0440, 0x0441, 0x0442, 0x0443, 0x0444, 0x0445, 0x0446, 0x0447,
973 0x0448, 0x0449, 0x044A, 0x044B, 0x044C, 0x044D, 0x044E, 0x044F,
974 /* 3 */ 0x0430, 0x0431, 0x0432, 0x0433, 0x0434, 0x0435, 0x0436, 0x0437,
975 0x0438, 0x0439, 0x043A, 0x043B, 0x043C, 0x043D, 0x043E, 0x043F,
976 /* 4 */ 0x0440, 0x0441, 0x0442, 0x0443, 0x0444, 0x0445, 0x0446, 0x0447,
977 0x0448, 0x0449, 0x044A, 0x044B, 0x044C, 0x044D, 0x044E, 0x044F,
978 /* 5 */ 0x0450, 0x0451, 0x0452, 0x0453, 0x0454, 0x0455, 0x0456, 0x0457,
979 0x0458, 0x0459, 0x045A, 0x045B, 0x045C, 0x045D, 0x045E, 0x045F,
980 /* 6 */ 0x0461, 0x0461, 0x0463, 0x0463, 0x0465, 0x0465, 0x0467, 0x0467,
981 0x0469, 0x0469, 0x046B, 0x046B, 0x046D, 0x046D, 0x046F, 0x046F,
982 /* 7 */ 0x0471, 0x0471, 0x0473, 0x0473, 0x0475, 0x0475, 0x0476, 0x0477,
983 0x0479, 0x0479, 0x047B, 0x047B, 0x047D, 0x047D, 0x047F, 0x047F,
984 /* 8 */ 0x0481, 0x0481, 0x0482, 0x0483, 0x0484, 0x0485, 0x0486, 0x0487,
985 0x0488, 0x0489, 0x048A, 0x048B, 0x048C, 0x048D, 0x048E, 0x048F,
986 /* 9 */ 0x0491, 0x0491, 0x0493, 0x0493, 0x0495, 0x0495, 0x0497, 0x0497,
987 0x0499, 0x0499, 0x049B, 0x049B, 0x049D, 0x049D, 0x049F, 0x049F,
988 /* A */ 0x04A1, 0x04A1, 0x04A3, 0x04A3, 0x04A5, 0x04A5, 0x04A7, 0x04A7,
989 0x04A9, 0x04A9, 0x04AB, 0x04AB, 0x04AD, 0x04AD, 0x04AF, 0x04AF,
990 /* B */ 0x04B1, 0x04B1, 0x04B3, 0x04B3, 0x04B5, 0x04B5, 0x04B7, 0x04B7,
991 0x04B9, 0x04B9, 0x04BB, 0x04BB, 0x04BD, 0x04BD, 0x04BF, 0x04BF,
992 /* C */ 0x04C0, 0x04C1, 0x04C2, 0x04C4, 0x04C4, 0x04C5, 0x04C6, 0x04C8,
993 0x04C8, 0x04C9, 0x04CA, 0x04CC, 0x04CC, 0x04CD, 0x04CE, 0x04CF,
994 /* D */ 0x04D0, 0x04D1, 0x04D2, 0x04D3, 0x04D4, 0x04D5, 0x04D6, 0x04D7,
995 0x04D8, 0x04D9, 0x04DA, 0x04DB, 0x04DC, 0x04DD, 0x04DE, 0x04DF,
996 /* E */ 0x04E0, 0x04E1, 0x04E2, 0x04E3, 0x04E4, 0x04E5, 0x04E6, 0x04E7,
997 0x04E8, 0x04E9, 0x04EA, 0x04EB, 0x04EC, 0x04ED, 0x04EE, 0x04EF,
998 /* F */ 0x04F0, 0x04F1, 0x04F2, 0x04F3, 0x04F4, 0x04F5, 0x04F6, 0x04F7,
999 0x04F8, 0x04F9, 0x04FA, 0x04FB, 0x04FC, 0x04FD, 0x04FE, 0x04FF,
1000
1001 // Table 5 (for high byte 0x05)
1002
1003 /* 0 */ 0x0500, 0x0501, 0x0502, 0x0503, 0x0504, 0x0505, 0x0506, 0x0507,
1004 0x0508, 0x0509, 0x050A, 0x050B, 0x050C, 0x050D, 0x050E, 0x050F,
1005 /* 1 */ 0x0510, 0x0511, 0x0512, 0x0513, 0x0514, 0x0515, 0x0516, 0x0517,
1006 0x0518, 0x0519, 0x051A, 0x051B, 0x051C, 0x051D, 0x051E, 0x051F,
1007 /* 2 */ 0x0520, 0x0521, 0x0522, 0x0523, 0x0524, 0x0525, 0x0526, 0x0527,
1008 0x0528, 0x0529, 0x052A, 0x052B, 0x052C, 0x052D, 0x052E, 0x052F,
1009 /* 3 */ 0x0530, 0x0561, 0x0562, 0x0563, 0x0564, 0x0565, 0x0566, 0x0567,
1010 0x0568, 0x0569, 0x056A, 0x056B, 0x056C, 0x056D, 0x056E, 0x056F,
1011 /* 4 */ 0x0570, 0x0571, 0x0572, 0x0573, 0x0574, 0x0575, 0x0576, 0x0577,
1012 0x0578, 0x0579, 0x057A, 0x057B, 0x057C, 0x057D, 0x057E, 0x057F,
1013 /* 5 */ 0x0580, 0x0581, 0x0582, 0x0583, 0x0584, 0x0585, 0x0586, 0x0557,
1014 0x0558, 0x0559, 0x055A, 0x055B, 0x055C, 0x055D, 0x055E, 0x055F,
1015 /* 6 */ 0x0560, 0x0561, 0x0562, 0x0563, 0x0564, 0x0565, 0x0566, 0x0567,
1016 0x0568, 0x0569, 0x056A, 0x056B, 0x056C, 0x056D, 0x056E, 0x056F,
1017 /* 7 */ 0x0570, 0x0571, 0x0572, 0x0573, 0x0574, 0x0575, 0x0576, 0x0577,
1018 0x0578, 0x0579, 0x057A, 0x057B, 0x057C, 0x057D, 0x057E, 0x057F,
1019 /* 8 */ 0x0580, 0x0581, 0x0582, 0x0583, 0x0584, 0x0585, 0x0586, 0x0587,
1020 0x0588, 0x0589, 0x058A, 0x058B, 0x058C, 0x058D, 0x058E, 0x058F,
1021 /* 9 */ 0x0590, 0x0591, 0x0592, 0x0593, 0x0594, 0x0595, 0x0596, 0x0597,
1022 0x0598, 0x0599, 0x059A, 0x059B, 0x059C, 0x059D, 0x059E, 0x059F,
1023 /* A */ 0x05A0, 0x05A1, 0x05A2, 0x05A3, 0x05A4, 0x05A5, 0x05A6, 0x05A7,
1024 0x05A8, 0x05A9, 0x05AA, 0x05AB, 0x05AC, 0x05AD, 0x05AE, 0x05AF,
1025 /* B */ 0x05B0, 0x05B1, 0x05B2, 0x05B3, 0x05B4, 0x05B5, 0x05B6, 0x05B7,
1026 0x05B8, 0x05B9, 0x05BA, 0x05BB, 0x05BC, 0x05BD, 0x05BE, 0x05BF,
1027 /* C */ 0x05C0, 0x05C1, 0x05C2, 0x05C3, 0x05C4, 0x05C5, 0x05C6, 0x05C7,
1028 0x05C8, 0x05C9, 0x05CA, 0x05CB, 0x05CC, 0x05CD, 0x05CE, 0x05CF,
1029 /* D */ 0x05D0, 0x05D1, 0x05D2, 0x05D3, 0x05D4, 0x05D5, 0x05D6, 0x05D7,
1030 0x05D8, 0x05D9, 0x05DA, 0x05DB, 0x05DC, 0x05DD, 0x05DE, 0x05DF,
1031 /* E */ 0x05E0, 0x05E1, 0x05E2, 0x05E3, 0x05E4, 0x05E5, 0x05E6, 0x05E7,
1032 0x05E8, 0x05E9, 0x05EA, 0x05EB, 0x05EC, 0x05ED, 0x05EE, 0x05EF,
1033 /* F */ 0x05F0, 0x05F1, 0x05F2, 0x05F3, 0x05F4, 0x05F5, 0x05F6, 0x05F7,
1034 0x05F8, 0x05F9, 0x05FA, 0x05FB, 0x05FC, 0x05FD, 0x05FE, 0x05FF,
1035
1036 // Table 6 (for high byte 0x10)
1037
1038 /* 0 */ 0x1000, 0x1001, 0x1002, 0x1003, 0x1004, 0x1005, 0x1006, 0x1007,
1039 0x1008, 0x1009, 0x100A, 0x100B, 0x100C, 0x100D, 0x100E, 0x100F,
1040 /* 1 */ 0x1010, 0x1011, 0x1012, 0x1013, 0x1014, 0x1015, 0x1016, 0x1017,
1041 0x1018, 0x1019, 0x101A, 0x101B, 0x101C, 0x101D, 0x101E, 0x101F,
1042 /* 2 */ 0x1020, 0x1021, 0x1022, 0x1023, 0x1024, 0x1025, 0x1026, 0x1027,
1043 0x1028, 0x1029, 0x102A, 0x102B, 0x102C, 0x102D, 0x102E, 0x102F,
1044 /* 3 */ 0x1030, 0x1031, 0x1032, 0x1033, 0x1034, 0x1035, 0x1036, 0x1037,
1045 0x1038, 0x1039, 0x103A, 0x103B, 0x103C, 0x103D, 0x103E, 0x103F,
1046 /* 4 */ 0x1040, 0x1041, 0x1042, 0x1043, 0x1044, 0x1045, 0x1046, 0x1047,
1047 0x1048, 0x1049, 0x104A, 0x104B, 0x104C, 0x104D, 0x104E, 0x104F,
1048 /* 5 */ 0x1050, 0x1051, 0x1052, 0x1053, 0x1054, 0x1055, 0x1056, 0x1057,
1049 0x1058, 0x1059, 0x105A, 0x105B, 0x105C, 0x105D, 0x105E, 0x105F,
1050 /* 6 */ 0x1060, 0x1061, 0x1062, 0x1063, 0x1064, 0x1065, 0x1066, 0x1067,
1051 0x1068, 0x1069, 0x106A, 0x106B, 0x106C, 0x106D, 0x106E, 0x106F,
1052 /* 7 */ 0x1070, 0x1071, 0x1072, 0x1073, 0x1074, 0x1075, 0x1076, 0x1077,
1053 0x1078, 0x1079, 0x107A, 0x107B, 0x107C, 0x107D, 0x107E, 0x107F,
1054 /* 8 */ 0x1080, 0x1081, 0x1082, 0x1083, 0x1084, 0x1085, 0x1086, 0x1087,
1055 0x1088, 0x1089, 0x108A, 0x108B, 0x108C, 0x108D, 0x108E, 0x108F,
1056 /* 9 */ 0x1090, 0x1091, 0x1092, 0x1093, 0x1094, 0x1095, 0x1096, 0x1097,
1057 0x1098, 0x1099, 0x109A, 0x109B, 0x109C, 0x109D, 0x109E, 0x109F,
1058 /* A */ 0x10D0, 0x10D1, 0x10D2, 0x10D3, 0x10D4, 0x10D5, 0x10D6, 0x10D7,
1059 0x10D8, 0x10D9, 0x10DA, 0x10DB, 0x10DC, 0x10DD, 0x10DE, 0x10DF,
1060 /* B */ 0x10E0, 0x10E1, 0x10E2, 0x10E3, 0x10E4, 0x10E5, 0x10E6, 0x10E7,
1061 0x10E8, 0x10E9, 0x10EA, 0x10EB, 0x10EC, 0x10ED, 0x10EE, 0x10EF,
1062 /* C */ 0x10F0, 0x10F1, 0x10F2, 0x10F3, 0x10F4, 0x10F5, 0x10C6, 0x10C7,
1063 0x10C8, 0x10C9, 0x10CA, 0x10CB, 0x10CC, 0x10CD, 0x10CE, 0x10CF,
1064 /* D */ 0x10D0, 0x10D1, 0x10D2, 0x10D3, 0x10D4, 0x10D5, 0x10D6, 0x10D7,
1065 0x10D8, 0x10D9, 0x10DA, 0x10DB, 0x10DC, 0x10DD, 0x10DE, 0x10DF,
1066 /* E */ 0x10E0, 0x10E1, 0x10E2, 0x10E3, 0x10E4, 0x10E5, 0x10E6, 0x10E7,
1067 0x10E8, 0x10E9, 0x10EA, 0x10EB, 0x10EC, 0x10ED, 0x10EE, 0x10EF,
1068 /* F */ 0x10F0, 0x10F1, 0x10F2, 0x10F3, 0x10F4, 0x10F5, 0x10F6, 0x10F7,
1069 0x10F8, 0x10F9, 0x10FA, 0x10FB, 0x10FC, 0x10FD, 0x10FE, 0x10FF,
1070
1071 // Table 7 (for high byte 0x20)
1072
1073 /* 0 */ 0x2000, 0x2001, 0x2002, 0x2003, 0x2004, 0x2005, 0x2006, 0x2007,
1074 0x2008, 0x2009, 0x200A, 0x200B, 0x0000, 0x0000, 0x0000, 0x0000,
1075 /* 1 */ 0x2010, 0x2011, 0x2012, 0x2013, 0x2014, 0x2015, 0x2016, 0x2017,
1076 0x2018, 0x2019, 0x201A, 0x201B, 0x201C, 0x201D, 0x201E, 0x201F,
1077 /* 2 */ 0x2020, 0x2021, 0x2022, 0x2023, 0x2024, 0x2025, 0x2026, 0x2027,
1078 0x2028, 0x2029, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x202F,
1079 /* 3 */ 0x2030, 0x2031, 0x2032, 0x2033, 0x2034, 0x2035, 0x2036, 0x2037,
1080 0x2038, 0x2039, 0x203A, 0x203B, 0x203C, 0x203D, 0x203E, 0x203F,
1081 /* 4 */ 0x2040, 0x2041, 0x2042, 0x2043, 0x2044, 0x2045, 0x2046, 0x2047,
1082 0x2048, 0x2049, 0x204A, 0x204B, 0x204C, 0x204D, 0x204E, 0x204F,
1083 /* 5 */ 0x2050, 0x2051, 0x2052, 0x2053, 0x2054, 0x2055, 0x2056, 0x2057,
1084 0x2058, 0x2059, 0x205A, 0x205B, 0x205C, 0x205D, 0x205E, 0x205F,
1085 /* 6 */ 0x2060, 0x2061, 0x2062, 0x2063, 0x2064, 0x2065, 0x2066, 0x2067,
1086 0x2068, 0x2069, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
1087 /* 7 */ 0x2070, 0x2071, 0x2072, 0x2073, 0x2074, 0x2075, 0x2076, 0x2077,
1088 0x2078, 0x2079, 0x207A, 0x207B, 0x207C, 0x207D, 0x207E, 0x207F,
1089 /* 8 */ 0x2080, 0x2081, 0x2082, 0x2083, 0x2084, 0x2085, 0x2086, 0x2087,
1090 0x2088, 0x2089, 0x208A, 0x208B, 0x208C, 0x208D, 0x208E, 0x208F,
1091 /* 9 */ 0x2090, 0x2091, 0x2092, 0x2093, 0x2094, 0x2095, 0x2096, 0x2097,
1092 0x2098, 0x2099, 0x209A, 0x209B, 0x209C, 0x209D, 0x209E, 0x209F,
1093 /* A */ 0x20A0, 0x20A1, 0x20A2, 0x20A3, 0x20A4, 0x20A5, 0x20A6, 0x20A7,
1094 0x20A8, 0x20A9, 0x20AA, 0x20AB, 0x20AC, 0x20AD, 0x20AE, 0x20AF,
1095 /* B */ 0x20B0, 0x20B1, 0x20B2, 0x20B3, 0x20B4, 0x20B5, 0x20B6, 0x20B7,
1096 0x20B8, 0x20B9, 0x20BA, 0x20BB, 0x20BC, 0x20BD, 0x20BE, 0x20BF,
1097 /* C */ 0x20C0, 0x20C1, 0x20C2, 0x20C3, 0x20C4, 0x20C5, 0x20C6, 0x20C7,
1098 0x20C8, 0x20C9, 0x20CA, 0x20CB, 0x20CC, 0x20CD, 0x20CE, 0x20CF,
1099 /* D */ 0x20D0, 0x20D1, 0x20D2, 0x20D3, 0x20D4, 0x20D5, 0x20D6, 0x20D7,
1100 0x20D8, 0x20D9, 0x20DA, 0x20DB, 0x20DC, 0x20DD, 0x20DE, 0x20DF,
1101 /* E */ 0x20E0, 0x20E1, 0x20E2, 0x20E3, 0x20E4, 0x20E5, 0x20E6, 0x20E7,
1102 0x20E8, 0x20E9, 0x20EA, 0x20EB, 0x20EC, 0x20ED, 0x20EE, 0x20EF,
1103 /* F */ 0x20F0, 0x20F1, 0x20F2, 0x20F3, 0x20F4, 0x20F5, 0x20F6, 0x20F7,
1104 0x20F8, 0x20F9, 0x20FA, 0x20FB, 0x20FC, 0x20FD, 0x20FE, 0x20FF,
1105
1106 // Table 8 (for high byte 0x21)
1107
1108 /* 0 */ 0x2100, 0x2101, 0x2102, 0x2103, 0x2104, 0x2105, 0x2106, 0x2107,
1109 0x2108, 0x2109, 0x210A, 0x210B, 0x210C, 0x210D, 0x210E, 0x210F,
1110 /* 1 */ 0x2110, 0x2111, 0x2112, 0x2113, 0x2114, 0x2115, 0x2116, 0x2117,
1111 0x2118, 0x2119, 0x211A, 0x211B, 0x211C, 0x211D, 0x211E, 0x211F,
1112 /* 2 */ 0x2120, 0x2121, 0x2122, 0x2123, 0x2124, 0x2125, 0x2126, 0x2127,
1113 0x2128, 0x2129, 0x212A, 0x212B, 0x212C, 0x212D, 0x212E, 0x212F,
1114 /* 3 */ 0x2130, 0x2131, 0x2132, 0x2133, 0x2134, 0x2135, 0x2136, 0x2137,
1115 0x2138, 0x2139, 0x213A, 0x213B, 0x213C, 0x213D, 0x213E, 0x213F,
1116 /* 4 */ 0x2140, 0x2141, 0x2142, 0x2143, 0x2144, 0x2145, 0x2146, 0x2147,
1117 0x2148, 0x2149, 0x214A, 0x214B, 0x214C, 0x214D, 0x214E, 0x214F,
1118 /* 5 */ 0x2150, 0x2151, 0x2152, 0x2153, 0x2154, 0x2155, 0x2156, 0x2157,
1119 0x2158, 0x2159, 0x215A, 0x215B, 0x215C, 0x215D, 0x215E, 0x215F,
1120 /* 6 */ 0x2170, 0x2171, 0x2172, 0x2173, 0x2174, 0x2175, 0x2176, 0x2177,
1121 0x2178, 0x2179, 0x217A, 0x217B, 0x217C, 0x217D, 0x217E, 0x217F,
1122 /* 7 */ 0x2170, 0x2171, 0x2172, 0x2173, 0x2174, 0x2175, 0x2176, 0x2177,
1123 0x2178, 0x2179, 0x217A, 0x217B, 0x217C, 0x217D, 0x217E, 0x217F,
1124 /* 8 */ 0x2180, 0x2181, 0x2182, 0x2183, 0x2184, 0x2185, 0x2186, 0x2187,
1125 0x2188, 0x2189, 0x218A, 0x218B, 0x218C, 0x218D, 0x218E, 0x218F,
1126 /* 9 */ 0x2190, 0x2191, 0x2192, 0x2193, 0x2194, 0x2195, 0x2196, 0x2197,
1127 0x2198, 0x2199, 0x219A, 0x219B, 0x219C, 0x219D, 0x219E, 0x219F,
1128 /* A */ 0x21A0, 0x21A1, 0x21A2, 0x21A3, 0x21A4, 0x21A5, 0x21A6, 0x21A7,
1129 0x21A8, 0x21A9, 0x21AA, 0x21AB, 0x21AC, 0x21AD, 0x21AE, 0x21AF,
1130 /* B */ 0x21B0, 0x21B1, 0x21B2, 0x21B3, 0x21B4, 0x21B5, 0x21B6, 0x21B7,
1131 0x21B8, 0x21B9, 0x21BA, 0x21BB, 0x21BC, 0x21BD, 0x21BE, 0x21BF,
1132 /* C */ 0x21C0, 0x21C1, 0x21C2, 0x21C3, 0x21C4, 0x21C5, 0x21C6, 0x21C7,
1133 0x21C8, 0x21C9, 0x21CA, 0x21CB, 0x21CC, 0x21CD, 0x21CE, 0x21CF,
1134 /* D */ 0x21D0, 0x21D1, 0x21D2, 0x21D3, 0x21D4, 0x21D5, 0x21D6, 0x21D7,
1135 0x21D8, 0x21D9, 0x21DA, 0x21DB, 0x21DC, 0x21DD, 0x21DE, 0x21DF,
1136 /* E */ 0x21E0, 0x21E1, 0x21E2, 0x21E3, 0x21E4, 0x21E5, 0x21E6, 0x21E7,
1137 0x21E8, 0x21E9, 0x21EA, 0x21EB, 0x21EC, 0x21ED, 0x21EE, 0x21EF,
1138 /* F */ 0x21F0, 0x21F1, 0x21F2, 0x21F3, 0x21F4, 0x21F5, 0x21F6, 0x21F7,
1139 0x21F8, 0x21F9, 0x21FA, 0x21FB, 0x21FC, 0x21FD, 0x21FE, 0x21FF,
1140
1141 // Table 9 (for high byte 0xFE)
1142
1143 /* 0 */ 0xFE00, 0xFE01, 0xFE02, 0xFE03, 0xFE04, 0xFE05, 0xFE06, 0xFE07,
1144 0xFE08, 0xFE09, 0xFE0A, 0xFE0B, 0xFE0C, 0xFE0D, 0xFE0E, 0xFE0F,
1145 /* 1 */ 0xFE10, 0xFE11, 0xFE12, 0xFE13, 0xFE14, 0xFE15, 0xFE16, 0xFE17,
1146 0xFE18, 0xFE19, 0xFE1A, 0xFE1B, 0xFE1C, 0xFE1D, 0xFE1E, 0xFE1F,
1147 /* 2 */ 0xFE20, 0xFE21, 0xFE22, 0xFE23, 0xFE24, 0xFE25, 0xFE26, 0xFE27,
1148 0xFE28, 0xFE29, 0xFE2A, 0xFE2B, 0xFE2C, 0xFE2D, 0xFE2E, 0xFE2F,
1149 /* 3 */ 0xFE30, 0xFE31, 0xFE32, 0xFE33, 0xFE34, 0xFE35, 0xFE36, 0xFE37,
1150 0xFE38, 0xFE39, 0xFE3A, 0xFE3B, 0xFE3C, 0xFE3D, 0xFE3E, 0xFE3F,
1151 /* 4 */ 0xFE40, 0xFE41, 0xFE42, 0xFE43, 0xFE44, 0xFE45, 0xFE46, 0xFE47,
1152 0xFE48, 0xFE49, 0xFE4A, 0xFE4B, 0xFE4C, 0xFE4D, 0xFE4E, 0xFE4F,
1153 /* 5 */ 0xFE50, 0xFE51, 0xFE52, 0xFE53, 0xFE54, 0xFE55, 0xFE56, 0xFE57,
1154 0xFE58, 0xFE59, 0xFE5A, 0xFE5B, 0xFE5C, 0xFE5D, 0xFE5E, 0xFE5F,
1155 /* 6 */ 0xFE60, 0xFE61, 0xFE62, 0xFE63, 0xFE64, 0xFE65, 0xFE66, 0xFE67,
1156 0xFE68, 0xFE69, 0xFE6A, 0xFE6B, 0xFE6C, 0xFE6D, 0xFE6E, 0xFE6F,
1157 /* 7 */ 0xFE70, 0xFE71, 0xFE72, 0xFE73, 0xFE74, 0xFE75, 0xFE76, 0xFE77,
1158 0xFE78, 0xFE79, 0xFE7A, 0xFE7B, 0xFE7C, 0xFE7D, 0xFE7E, 0xFE7F,
1159 /* 8 */ 0xFE80, 0xFE81, 0xFE82, 0xFE83, 0xFE84, 0xFE85, 0xFE86, 0xFE87,
1160 0xFE88, 0xFE89, 0xFE8A, 0xFE8B, 0xFE8C, 0xFE8D, 0xFE8E, 0xFE8F,
1161 /* 9 */ 0xFE90, 0xFE91, 0xFE92, 0xFE93, 0xFE94, 0xFE95, 0xFE96, 0xFE97,
1162 0xFE98, 0xFE99, 0xFE9A, 0xFE9B, 0xFE9C, 0xFE9D, 0xFE9E, 0xFE9F,
1163 /* A */ 0xFEA0, 0xFEA1, 0xFEA2, 0xFEA3, 0xFEA4, 0xFEA5, 0xFEA6, 0xFEA7,
1164 0xFEA8, 0xFEA9, 0xFEAA, 0xFEAB, 0xFEAC, 0xFEAD, 0xFEAE, 0xFEAF,
1165 /* B */ 0xFEB0, 0xFEB1, 0xFEB2, 0xFEB3, 0xFEB4, 0xFEB5, 0xFEB6, 0xFEB7,
1166 0xFEB8, 0xFEB9, 0xFEBA, 0xFEBB, 0xFEBC, 0xFEBD, 0xFEBE, 0xFEBF,
1167 /* C */ 0xFEC0, 0xFEC1, 0xFEC2, 0xFEC3, 0xFEC4, 0xFEC5, 0xFEC6, 0xFEC7,
1168 0xFEC8, 0xFEC9, 0xFECA, 0xFECB, 0xFECC, 0xFECD, 0xFECE, 0xFECF,
1169 /* D */ 0xFED0, 0xFED1, 0xFED2, 0xFED3, 0xFED4, 0xFED5, 0xFED6, 0xFED7,
1170 0xFED8, 0xFED9, 0xFEDA, 0xFEDB, 0xFEDC, 0xFEDD, 0xFEDE, 0xFEDF,
1171 /* E */ 0xFEE0, 0xFEE1, 0xFEE2, 0xFEE3, 0xFEE4, 0xFEE5, 0xFEE6, 0xFEE7,
1172 0xFEE8, 0xFEE9, 0xFEEA, 0xFEEB, 0xFEEC, 0xFEED, 0xFEEE, 0xFEEF,
1173 /* F */ 0xFEF0, 0xFEF1, 0xFEF2, 0xFEF3, 0xFEF4, 0xFEF5, 0xFEF6, 0xFEF7,
1174 0xFEF8, 0xFEF9, 0xFEFA, 0xFEFB, 0xFEFC, 0xFEFD, 0xFEFE, 0x0000,
1175
1176 // Table 10 (for high byte 0xFF)
1177
1178 /* 0 */ 0xFF00, 0xFF01, 0xFF02, 0xFF03, 0xFF04, 0xFF05, 0xFF06, 0xFF07,
1179 0xFF08, 0xFF09, 0xFF0A, 0xFF0B, 0xFF0C, 0xFF0D, 0xFF0E, 0xFF0F,
1180 /* 1 */ 0xFF10, 0xFF11, 0xFF12, 0xFF13, 0xFF14, 0xFF15, 0xFF16, 0xFF17,
1181 0xFF18, 0xFF19, 0xFF1A, 0xFF1B, 0xFF1C, 0xFF1D, 0xFF1E, 0xFF1F,
1182 /* 2 */ 0xFF20, 0xFF41, 0xFF42, 0xFF43, 0xFF44, 0xFF45, 0xFF46, 0xFF47,
1183 0xFF48, 0xFF49, 0xFF4A, 0xFF4B, 0xFF4C, 0xFF4D, 0xFF4E, 0xFF4F,
1184 /* 3 */ 0xFF50, 0xFF51, 0xFF52, 0xFF53, 0xFF54, 0xFF55, 0xFF56, 0xFF57,
1185 0xFF58, 0xFF59, 0xFF5A, 0xFF3B, 0xFF3C, 0xFF3D, 0xFF3E, 0xFF3F,
1186 /* 4 */ 0xFF40, 0xFF41, 0xFF42, 0xFF43, 0xFF44, 0xFF45, 0xFF46, 0xFF47,
1187 0xFF48, 0xFF49, 0xFF4A, 0xFF4B, 0xFF4C, 0xFF4D, 0xFF4E, 0xFF4F,
1188 /* 5 */ 0xFF50, 0xFF51, 0xFF52, 0xFF53, 0xFF54, 0xFF55, 0xFF56, 0xFF57,
1189 0xFF58, 0xFF59, 0xFF5A, 0xFF5B, 0xFF5C, 0xFF5D, 0xFF5E, 0xFF5F,
1190 /* 6 */ 0xFF60, 0xFF61, 0xFF62, 0xFF63, 0xFF64, 0xFF65, 0xFF66, 0xFF67,
1191 0xFF68, 0xFF69, 0xFF6A, 0xFF6B, 0xFF6C, 0xFF6D, 0xFF6E, 0xFF6F,
1192 /* 7 */ 0xFF70, 0xFF71, 0xFF72, 0xFF73, 0xFF74, 0xFF75, 0xFF76, 0xFF77,
1193 0xFF78, 0xFF79, 0xFF7A, 0xFF7B, 0xFF7C, 0xFF7D, 0xFF7E, 0xFF7F,
1194 /* 8 */ 0xFF80, 0xFF81, 0xFF82, 0xFF83, 0xFF84, 0xFF85, 0xFF86, 0xFF87,
1195 0xFF88, 0xFF89, 0xFF8A, 0xFF8B, 0xFF8C, 0xFF8D, 0xFF8E, 0xFF8F,
1196 /* 9 */ 0xFF90, 0xFF91, 0xFF92, 0xFF93, 0xFF94, 0xFF95, 0xFF96, 0xFF97,
1197 0xFF98, 0xFF99, 0xFF9A, 0xFF9B, 0xFF9C, 0xFF9D, 0xFF9E, 0xFF9F,
1198 /* A */ 0xFFA0, 0xFFA1, 0xFFA2, 0xFFA3, 0xFFA4, 0xFFA5, 0xFFA6, 0xFFA7,
1199 0xFFA8, 0xFFA9, 0xFFAA, 0xFFAB, 0xFFAC, 0xFFAD, 0xFFAE, 0xFFAF,
1200 /* B */ 0xFFB0, 0xFFB1, 0xFFB2, 0xFFB3, 0xFFB4, 0xFFB5, 0xFFB6, 0xFFB7,
1201 0xFFB8, 0xFFB9, 0xFFBA, 0xFFBB, 0xFFBC, 0xFFBD, 0xFFBE, 0xFFBF,
1202 /* C */ 0xFFC0, 0xFFC1, 0xFFC2, 0xFFC3, 0xFFC4, 0xFFC5, 0xFFC6, 0xFFC7,
1203 0xFFC8, 0xFFC9, 0xFFCA, 0xFFCB, 0xFFCC, 0xFFCD, 0xFFCE, 0xFFCF,
1204 /* D */ 0xFFD0, 0xFFD1, 0xFFD2, 0xFFD3, 0xFFD4, 0xFFD5, 0xFFD6, 0xFFD7,
1205 0xFFD8, 0xFFD9, 0xFFDA, 0xFFDB, 0xFFDC, 0xFFDD, 0xFFDE, 0xFFDF,
1206 /* E */ 0xFFE0, 0xFFE1, 0xFFE2, 0xFFE3, 0xFFE4, 0xFFE5, 0xFFE6, 0xFFE7,
1207 0xFFE8, 0xFFE9, 0xFFEA, 0xFFEB, 0xFFEC, 0xFFED, 0xFFEE, 0xFFEF,
1208 /* F */ 0xFFF0, 0xFFF1, 0xFFF2, 0xFFF3, 0xFFF4, 0xFFF5, 0xFFF6, 0xFFF7,
1209 0xFFF8, 0xFFF9, 0xFFFA, 0xFFFB, 0xFFFC, 0xFFFD, 0xFFFE, 0xFFFF,
1210 };
1211 // clang-format on
1212
1213 // Returns the next non-ignorable codepoint within `string` starting from the
1214 // position indicated by `index`, or zero if there are no more.
1215 // The passed-in `index` is automatically advanced as the characters in the
1216 // input HFS-decomposed UTF-8 strings are read.
HFSReadNextNonIgnorableCodepoint(const char * string,size_t length,size_t * index)1217 inline base_icu::UChar32 HFSReadNextNonIgnorableCodepoint(const char* string,
1218 size_t length,
1219 size_t* index) {
1220 base_icu::UChar32 codepoint = 0;
1221 while (*index < length && codepoint == 0) {
1222 // CBU8_NEXT returns a value < 0 in error cases. For purposes of string
1223 // comparison, we just use that value and flag it with DCHECK.
1224 CBU8_NEXT(reinterpret_cast<const uint8_t*>(string), *index, length,
1225 codepoint);
1226 DCHECK_GT(codepoint, 0);
1227
1228 // Note: Here, there are no lower case conversion implemented in the
1229 // Supplementary Multilingual Plane (codepoint > 0xFFFF).
1230
1231 if (codepoint > 0 && codepoint <= 0xFFFF) {
1232 // Check if there is a subtable for this upper byte.
1233 int lookup_offset = lower_case_table[codepoint >> 8];
1234 if (lookup_offset != 0)
1235 codepoint = lower_case_table[lookup_offset + (codepoint & 0x00FF)];
1236 // Note: `codepoint` may be again 0 at this point if the character was
1237 // an ignorable.
1238 }
1239 }
1240 return codepoint;
1241 }
1242
1243 } // namespace
1244
1245 // Special UTF-8 version of FastUnicodeCompare. Cf:
1246 // http://developer.apple.com/mac/library/technotes/tn/tn1150.html#StringComparisonAlgorithm
1247 // The input strings must be in the special HFS decomposed form.
HFSFastUnicodeCompare(StringPieceType string1,StringPieceType string2)1248 int FilePath::HFSFastUnicodeCompare(StringPieceType string1,
1249 StringPieceType string2) {
1250 size_t length1 = string1.length();
1251 size_t length2 = string2.length();
1252 size_t index1 = 0;
1253 size_t index2 = 0;
1254
1255 for (;;) {
1256 base_icu::UChar32 codepoint1 =
1257 HFSReadNextNonIgnorableCodepoint(string1.data(), length1, &index1);
1258 base_icu::UChar32 codepoint2 =
1259 HFSReadNextNonIgnorableCodepoint(string2.data(), length2, &index2);
1260 if (codepoint1 != codepoint2)
1261 return (codepoint1 < codepoint2) ? -1 : 1;
1262 if (codepoint1 == 0) {
1263 DCHECK_EQ(index1, length1);
1264 DCHECK_EQ(index2, length2);
1265 return 0;
1266 }
1267 }
1268 }
1269
GetHFSDecomposedForm(StringPieceType string)1270 StringType FilePath::GetHFSDecomposedForm(StringPieceType string) {
1271 StringType result;
1272 apple::ScopedCFTypeRef<CFStringRef> cfstring(CFStringCreateWithBytesNoCopy(
1273 NULL, reinterpret_cast<const UInt8*>(string.data()),
1274 checked_cast<CFIndex>(string.length()), kCFStringEncodingUTF8, false,
1275 kCFAllocatorNull));
1276 if (cfstring) {
1277 // Query the maximum length needed to store the result. In most cases this
1278 // will overestimate the required space. The return value also already
1279 // includes the space needed for a terminating 0.
1280 CFIndex length =
1281 CFStringGetMaximumSizeOfFileSystemRepresentation(cfstring.get());
1282 DCHECK_GT(length, 0); // should be at least 1 for the 0-terminator.
1283 // Reserve enough space for CFStringGetFileSystemRepresentation to write
1284 // into. Also set the length to the maximum so that we can shrink it later.
1285 // (Increasing rather than decreasing it would clobber the string contents!)
1286 result.reserve(static_cast<size_t>(length));
1287 result.resize(static_cast<size_t>(length) - 1);
1288 Boolean success =
1289 CFStringGetFileSystemRepresentation(cfstring.get(), &result[0], length);
1290 if (success) {
1291 // Reduce result.length() to actual string length.
1292 result.resize(strlen(result.c_str()));
1293 } else {
1294 // An error occurred -> clear result.
1295 result.clear();
1296 }
1297 }
1298 return result;
1299 }
1300
CompareIgnoreCase(StringPieceType string1,StringPieceType string2)1301 int FilePath::CompareIgnoreCase(StringPieceType string1,
1302 StringPieceType string2) {
1303 // Quick checks for empty strings - these speed things up a bit and make the
1304 // following code cleaner.
1305 if (string1.empty())
1306 return string2.empty() ? 0 : -1;
1307 if (string2.empty())
1308 return 1;
1309
1310 StringType hfs1 = GetHFSDecomposedForm(string1);
1311 StringType hfs2 = GetHFSDecomposedForm(string2);
1312
1313 // GetHFSDecomposedForm() returns an empty string in an error case.
1314 if (hfs1.empty() || hfs2.empty()) {
1315 apple::ScopedCFTypeRef<CFStringRef> cfstring1(CFStringCreateWithBytesNoCopy(
1316 NULL, reinterpret_cast<const UInt8*>(string1.data()),
1317 checked_cast<CFIndex>(string1.length()), kCFStringEncodingUTF8, false,
1318 kCFAllocatorNull));
1319 apple::ScopedCFTypeRef<CFStringRef> cfstring2(CFStringCreateWithBytesNoCopy(
1320 NULL, reinterpret_cast<const UInt8*>(string2.data()),
1321 checked_cast<CFIndex>(string2.length()), kCFStringEncodingUTF8, false,
1322 kCFAllocatorNull));
1323 // If neither GetHFSDecomposedForm nor CFStringCreateWithBytesNoCopy
1324 // succeed, fall back to strcmp. This can occur when the input string is
1325 // invalid UTF-8.
1326 if (!cfstring1 || !cfstring2) {
1327 int comparison = memcmp(string1.data(), string2.data(),
1328 std::min(string1.length(), string2.length()));
1329 if (comparison < 0)
1330 return -1;
1331 if (comparison > 0)
1332 return 1;
1333 return 0;
1334 }
1335
1336 return static_cast<int>(CFStringCompare(cfstring1.get(), cfstring2.get(),
1337 kCFCompareCaseInsensitive));
1338 }
1339
1340 return HFSFastUnicodeCompare(hfs1, hfs2);
1341 }
1342
1343 #elif BUILDFLAG(IS_POSIX) || BUILDFLAG(IS_FUCHSIA)
1344
1345 // Generic Posix system comparisons.
CompareIgnoreCase(StringPieceType string1,StringPieceType string2)1346 int FilePath::CompareIgnoreCase(StringPieceType string1,
1347 StringPieceType string2) {
1348 size_t rlen = std::min(string1.size(), string2.size());
1349 int comparison = strncasecmp(string1.data(), string2.data(), rlen);
1350 if (comparison < 0 || (comparison == 0 && string1.size() < string2.size()))
1351 return -1;
1352 if (comparison > 0 || (comparison == 0 && string1.size() > string2.size()))
1353 return 1;
1354 return 0;
1355 }
1356
1357 #endif // OS versions of CompareIgnoreCase()
1358
1359
StripTrailingSeparatorsInternal()1360 void FilePath::StripTrailingSeparatorsInternal() {
1361 // If there is no drive letter, start will be 1, which will prevent stripping
1362 // the leading separator if there is only one separator. If there is a drive
1363 // letter, start will be set appropriately to prevent stripping the first
1364 // separator following the drive letter, if a separator immediately follows
1365 // the drive letter.
1366 StringType::size_type start = FindDriveLetter(path_) + 2;
1367
1368 StringType::size_type last_stripped = StringType::npos;
1369 for (StringType::size_type pos = path_.length();
1370 pos > start && IsSeparator(path_[pos - 1]);
1371 --pos) {
1372 // If the string only has two separators and they're at the beginning,
1373 // don't strip them, unless the string began with more than two separators.
1374 if (pos != start + 1 || last_stripped == start + 2 ||
1375 !IsSeparator(path_[start - 1])) {
1376 path_.resize(pos - 1);
1377 last_stripped = pos;
1378 }
1379 }
1380 }
1381
NormalizePathSeparators() const1382 FilePath FilePath::NormalizePathSeparators() const {
1383 return NormalizePathSeparatorsTo(kSeparators[0]);
1384 }
1385
WriteIntoTrace(perfetto::TracedValue context) const1386 void FilePath::WriteIntoTrace(perfetto::TracedValue context) const {
1387 perfetto::WriteIntoTracedValue(std::move(context), value());
1388 }
1389
NormalizePathSeparatorsTo(CharType separator) const1390 FilePath FilePath::NormalizePathSeparatorsTo(CharType separator) const {
1391 #if defined(FILE_PATH_USES_WIN_SEPARATORS)
1392 DCHECK_NE(kSeparators + kSeparatorsLength,
1393 std::find(kSeparators, kSeparators + kSeparatorsLength, separator));
1394 StringType copy = path_;
1395 for (size_t i = 0; i < kSeparatorsLength; ++i) {
1396 std::replace(copy.begin(), copy.end(), kSeparators[i], separator);
1397 }
1398 return FilePath(copy);
1399 #else
1400 return *this;
1401 #endif
1402 }
1403
1404 #if BUILDFLAG(IS_ANDROID)
IsContentUri() const1405 bool FilePath::IsContentUri() const {
1406 return StartsWith(path_, "content://", base::CompareCase::INSENSITIVE_ASCII);
1407 }
1408 #endif
1409
1410 } // namespace base
1411