base/file_path.cc

   1 // Copyright (c) 2012 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #include "base/file_path.h"
   6
   7 #include <string.h>
   8 #include <algorithm>
   9
  10 #include "base/basictypes.h"
  11 #include "base/logging.h"
  12 #include "base/pickle.h"
  13
  14 // These includes are just for the *Hack functions, and should be removed
  15 // when those functions are removed.
  16 #include "base/string_piece.h"
  17 #include "base/string_util.h"
  18 #include "base/sys_string_conversions.h"
  19 #include "base/utf_string_conversions.h"
  20
  21 #if defined(OS_MACOSX)
  22 #include "base/mac/scoped_cftyperef.h"
  23 #include "base/third_party/icu/icu_utf.h"
  24 #endif
  25
  26 #if defined(OS_WIN)
  27 #include <windows.h>
  28 #elif defined(OS_MACOSX)
  29 #include <CoreFoundation/CoreFoundation.h>
  30 #endif
  31
  32 #if defined(FILE_PATH_USES_WIN_SEPARATORS)
  33 const FilePath::CharType FilePath::kSeparators[] = FILE_PATH_LITERAL("\\/");
  34 #else  // FILE_PATH_USES_WIN_SEPARATORS
  35 const FilePath::CharType FilePath::kSeparators[] = FILE_PATH_LITERAL("/");
  36 #endif  // FILE_PATH_USES_WIN_SEPARATORS
  37
  38 const FilePath::CharType FilePath::kCurrentDirectory[] = FILE_PATH_LITERAL(".");
  39 const FilePath::CharType FilePath::kParentDirectory[] = FILE_PATH_LITERAL("..");
  40
  41 const FilePath::CharType FilePath::kExtensionSeparator = FILE_PATH_LITERAL('.');
  42
  43 typedef FilePath::StringType StringType;
  44
  45 namespace {
  46
  47 const char* kCommonDoubleExtensionSuffixes[] = { "gz", "z", "bz2" };
  48 const char* kCommonDoubleExtensions[] = { "user.js" };
  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.
  55 StringType::size_type FindDriveLetter(const StringType& 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)
  69 bool EqualDriveLetterCaseInsensitive(const StringType& a,
  70                                      const StringType& b) {
  71   size_t a_letter_pos = FindDriveLetter(a);
  72   size_t b_letter_pos = FindDriveLetter(b);
  73
  74   if (a_letter_pos == StringType::npos || b_letter_pos == StringType::npos)
  75     return a == b;
  76
  77   StringType a_letter(a.substr(0, a_letter_pos + 1));
  78   StringType b_letter(b.substr(0, b_letter_pos + 1));
  79   if (!StartsWith(a_letter, b_letter, false))
  80     return false;
  81
  82   StringType a_rest(a.substr(a_letter_pos + 1));
  83   StringType b_rest(b.substr(b_letter_pos + 1));
  84   return a_rest == b_rest;
  85 }
  86 #endif  // defined(FILE_PATH_USES_DRIVE_LETTERS)
  87
  88 bool IsPathAbsolute(const StringType& path) {
  89 #if defined(FILE_PATH_USES_DRIVE_LETTERS)
  90   StringType::size_type letter = FindDriveLetter(path);
  91   if (letter != StringType::npos) {
  92     // Look for a separator right after the drive specification.
  93     return path.length() > letter + 1 &&
  94         FilePath::IsSeparator(path[letter + 1]);
  95   }
  96   // Look for a pair of leading separators.
  97   return path.length() > 1 &&
  98       FilePath::IsSeparator(path[0]) && FilePath::IsSeparator(path[1]);
  99 #else  // FILE_PATH_USES_DRIVE_LETTERS
 100   // Look for a separator in the first position.
 101   return path.length() > 0 && FilePath::IsSeparator(path[0]);
 102 #endif  // FILE_PATH_USES_DRIVE_LETTERS
 103 }
 104
 105 bool AreAllSeparators(const StringType& input) {
 106   for (StringType::const_iterator it = input.begin();
 107       it != input.end(); ++it) {
 108     if (!FilePath::IsSeparator(*it))
 109       return false;
 110   }
 111
 112   return true;
 113 }
 114
 115 // Find the position of the '.' that separates the extension from the rest
 116 // of the file name. The position is relative to BaseName(), not value().
 117 // This allows a second extension component of up to 4 characters when the
 118 // rightmost extension component is a common double extension (gz, bz2, Z).
 119 // For example, foo.tar.gz or foo.tar.Z would have extension components of
 120 // '.tar.gz' and '.tar.Z' respectively. Returns npos if it can't find an
 121 // extension.
 122 StringType::size_type ExtensionSeparatorPosition(const StringType& path) {
 123   // Special case "." and ".."
 124   if (path == FilePath::kCurrentDirectory || path == FilePath::kParentDirectory)
 125     return StringType::npos;
 126
 127   const StringType::size_type last_dot =
 128       path.rfind(FilePath::kExtensionSeparator);
 129
 130   // No extension, or the extension is the whole filename.
 131   if (last_dot == StringType::npos || last_dot == 0U)
 132     return last_dot;
 133
 134   const StringType::size_type penultimate_dot =
 135       path.rfind(FilePath::kExtensionSeparator, last_dot - 1);
 136   const StringType::size_type last_separator =
 137       path.find_last_of(FilePath::kSeparators, last_dot - 1,
 138                         arraysize(FilePath::kSeparators) - 1);
 139
 140   if (penultimate_dot == StringType::npos ||
 141       (last_separator != StringType::npos &&
 142        penultimate_dot < last_separator)) {
 143     return last_dot;
 144   }
 145
 146   for (size_t i = 0; i < arraysize(kCommonDoubleExtensions); ++i) {
 147     StringType extension(path, penultimate_dot + 1);
 148     if (LowerCaseEqualsASCII(extension, kCommonDoubleExtensions[i]))
 149       return penultimate_dot;
 150   }
 151
 152   StringType extension(path, last_dot + 1);
 153   for (size_t i = 0; i < arraysize(kCommonDoubleExtensionSuffixes); ++i) {
 154     if (LowerCaseEqualsASCII(extension, kCommonDoubleExtensionSuffixes[i])) {
 155       if ((last_dot - penultimate_dot) <= 5U &&
 156           (last_dot - penultimate_dot) > 1U) {
 157         return penultimate_dot;
 158       }
 159     }
 160   }
 161
 162   return last_dot;
 163 }
 164
 165 // Returns true if path is "", ".", or "..".
 166 bool IsEmptyOrSpecialCase(const StringType& path) {
 167   // Special cases "", ".", and ".."
 168   if (path.empty() || path == FilePath::kCurrentDirectory ||
 169       path == FilePath::kParentDirectory) {
 170     return true;
 171   }
 172
 173   return false;
 174 }
 175
 176 }  // namespace
 177
 178 FilePath::FilePath() {
 179 }
 180
 181 FilePath::FilePath(const FilePath& that) : path_(that.path_) {
 182 }
 183
 184 FilePath::FilePath(const StringType& path) : path_(path) {
 185 }
 186
 187 FilePath::~FilePath() {
 188 }
 189
 190 FilePath& FilePath::operator=(const FilePath& that) {
 191   path_ = that.path_;
 192   return *this;
 193 }
 194
 195 bool FilePath::operator==(const FilePath& that) const {
 196 #if defined(FILE_PATH_USES_DRIVE_LETTERS)
 197   return EqualDriveLetterCaseInsensitive(this->path_, that.path_);
 198 #else  // defined(FILE_PATH_USES_DRIVE_LETTERS)
 199   return path_ == that.path_;
 200 #endif  // defined(FILE_PATH_USES_DRIVE_LETTERS)
 201 }
 202
 203 bool FilePath::operator!=(const FilePath& that) const {
 204 #if defined(FILE_PATH_USES_DRIVE_LETTERS)
 205   return !EqualDriveLetterCaseInsensitive(this->path_, that.path_);
 206 #else  // defined(FILE_PATH_USES_DRIVE_LETTERS)
 207   return path_ != that.path_;
 208 #endif  // defined(FILE_PATH_USES_DRIVE_LETTERS)
 209 }
 210
 211 // static
 212 bool FilePath::IsSeparator(CharType character) {
 213   for (size_t i = 0; i < arraysize(kSeparators) - 1; ++i) {
 214     if (character == kSeparators[i]) {
 215       return true;
 216     }
 217   }
 218
 219   return false;
 220 }
 221
 222 void FilePath::GetComponents(std::vector<StringType>* components) const {
 223   DCHECK(components);
 224   if (!components)
 225     return;
 226   components->clear();
 227   if (value().empty())
 228     return;
 229
 230   std::vector<StringType> ret_val;
 231   FilePath current = *this;
 232   FilePath base;
 233
 234   // Capture path components.
 235   while (current != current.DirName()) {
 236     base = current.BaseName();
 237     if (!AreAllSeparators(base.value()))
 238       ret_val.push_back(base.value());
 239     current = current.DirName();
 240   }
 241
 242   // Capture root, if any.
 243   base = current.BaseName();
 244   if (!base.value().empty() && base.value() != kCurrentDirectory)
 245     ret_val.push_back(current.BaseName().value());
 246
 247   // Capture drive letter, if any.
 248   FilePath dir = current.DirName();
 249   StringType::size_type letter = FindDriveLetter(dir.value());
 250   if (letter != StringType::npos) {
 251     ret_val.push_back(StringType(dir.value(), 0, letter + 1));
 252   }
 253
 254   *components = std::vector<StringType>(ret_val.rbegin(), ret_val.rend());
 255 }
 256
 257 bool FilePath::IsParent(const FilePath& child) const {
 258   return AppendRelativePath(child, NULL);
 259 }
 260
 261 bool FilePath::AppendRelativePath(const FilePath& child,
 262                                   FilePath* path) const {
 263   std::vector<StringType> parent_components;
 264   std::vector<StringType> child_components;
 265   GetComponents(&parent_components);
 266   child.GetComponents(&child_components);
 267
 268   if (parent_components.empty() ||
 269       parent_components.size() >= child_components.size())
 270     return false;
 271
 272   std::vector<StringType>::const_iterator parent_comp =
 273       parent_components.begin();
 274   std::vector<StringType>::const_iterator child_comp =
 275       child_components.begin();
 276
 277 #if defined(FILE_PATH_USES_DRIVE_LETTERS)
 278   // Windows can access case sensitive filesystems, so component
 279   // comparisions must be case sensitive, but drive letters are
 280   // never case sensitive.
 281   if ((FindDriveLetter(*parent_comp) != StringType::npos) &&
 282       (FindDriveLetter(*child_comp) != StringType::npos)) {
 283     if (!StartsWith(*parent_comp, *child_comp, false))
 284       return false;
 285     ++parent_comp;
 286     ++child_comp;
 287   }
 288 #endif  // defined(FILE_PATH_USES_DRIVE_LETTERS)
 289
 290   while (parent_comp != parent_components.end()) {
 291     if (*parent_comp != *child_comp)
 292       return false;
 293     ++parent_comp;
 294     ++child_comp;
 295   }
 296
 297   if (path != NULL) {
 298     for (; child_comp != child_components.end(); ++child_comp) {
 299       *path = path->Append(*child_comp);
 300     }
 301   }
 302   return true;
 303 }
 304
 305 // libgen's dirname and basename aren't guaranteed to be thread-safe and aren't
 306 // guaranteed to not modify their input strings, and in fact are implemented
 307 // differently in this regard on different platforms.  Don't use them, but
 308 // adhere to their behavior.
 309 FilePath FilePath::DirName() const {
 310   FilePath new_path(path_);
 311   new_path.StripTrailingSeparatorsInternal();
 312
 313   // The drive letter, if any, always needs to remain in the output.  If there
 314   // is no drive letter, as will always be the case on platforms which do not
 315   // support drive letters, letter will be npos, or -1, so the comparisons and
 316   // resizes below using letter will still be valid.
 317   StringType::size_type letter = FindDriveLetter(new_path.path_);
 318
 319   StringType::size_type last_separator =
 320       new_path.path_.find_last_of(kSeparators, StringType::npos,
 321                                   arraysize(kSeparators) - 1);
 322   if (last_separator == StringType::npos) {
 323     // path_ is in the current directory.
 324     new_path.path_.resize(letter + 1);
 325   } else if (last_separator == letter + 1) {
 326     // path_ is in the root directory.
 327     new_path.path_.resize(letter + 2);
 328   } else if (last_separator == letter + 2 &&
 329              IsSeparator(new_path.path_[letter + 1])) {
 330     // path_ is in "//" (possibly with a drive letter); leave the double
 331     // separator intact indicating alternate root.
 332     new_path.path_.resize(letter + 3);
 333   } else if (last_separator != 0) {
 334     // path_ is somewhere else, trim the basename.
 335     new_path.path_.resize(last_separator);
 336   }
 337
 338   new_path.StripTrailingSeparatorsInternal();
 339   if (!new_path.path_.length())
 340     new_path.path_ = kCurrentDirectory;
 341
 342   return new_path;
 343 }
 344
 345 FilePath FilePath::BaseName() const {
 346   FilePath new_path(path_);
 347   new_path.StripTrailingSeparatorsInternal();
 348
 349   // The drive letter, if any, is always stripped.
 350   StringType::size_type letter = FindDriveLetter(new_path.path_);
 351   if (letter != StringType::npos) {
 352     new_path.path_.erase(0, letter + 1);
 353   }
 354
 355   // Keep everything after the final separator, but if the pathname is only
 356   // one character and it's a separator, leave it alone.
 357   StringType::size_type last_separator =
 358       new_path.path_.find_last_of(kSeparators, StringType::npos,
 359                                   arraysize(kSeparators) - 1);
 360   if (last_separator != StringType::npos &&
 361       last_separator < new_path.path_.length() - 1) {
 362     new_path.path_.erase(0, last_separator + 1);
 363   }
 364
 365   return new_path;
 366 }
 367
 368 StringType FilePath::Extension() const {
 369   FilePath base(BaseName());
 370   const StringType::size_type dot = ExtensionSeparatorPosition(base.path_);
 371   if (dot == StringType::npos)
 372     return StringType();
 373
 374   return base.path_.substr(dot, StringType::npos);
 375 }
 376
 377 FilePath FilePath::RemoveExtension() const {
 378   if (Extension().empty())
 379     return *this;
 380
 381   const StringType::size_type dot = ExtensionSeparatorPosition(path_);
 382   if (dot == StringType::npos)
 383     return *this;
 384
 385   return FilePath(path_.substr(0, dot));
 386 }
 387
 388 FilePath FilePath::InsertBeforeExtension(const StringType& suffix) const {
 389   if (suffix.empty())
 390     return FilePath(path_);
 391
 392   if (IsEmptyOrSpecialCase(BaseName().value()))
 393     return FilePath();
 394
 395   StringType ext = Extension();
 396   StringType ret = RemoveExtension().value();
 397   ret.append(suffix);
 398   ret.append(ext);
 399   return FilePath(ret);
 400 }
 401
 402 FilePath FilePath::InsertBeforeExtensionASCII(const base::StringPiece& suffix)
 403     const {
 404   DCHECK(IsStringASCII(suffix));
 405 #if defined(OS_WIN)
 406   return InsertBeforeExtension(ASCIIToUTF16(suffix.as_string()));
 407 #elif defined(OS_POSIX)
 408   return InsertBeforeExtension(suffix.as_string());
 409 #endif
 410 }
 411
 412 FilePath FilePath::AddExtension(const StringType& extension) const {
 413   if (IsEmptyOrSpecialCase(BaseName().value()))
 414     return FilePath();
 415
 416   // If the new extension is "" or ".", then just return the current FilePath.
 417   if (extension.empty() || extension == StringType(1, kExtensionSeparator))
 418     return *this;
 419
 420   StringType str = path_;
 421   if (extension[0] != kExtensionSeparator &&
 422       *(str.end() - 1) != kExtensionSeparator) {
 423     str.append(1, kExtensionSeparator);
 424   }
 425   str.append(extension);
 426   return FilePath(str);
 427 }
 428
 429 FilePath FilePath::ReplaceExtension(const StringType& extension) const {
 430   if (IsEmptyOrSpecialCase(BaseName().value()))
 431     return FilePath();
 432
 433   FilePath no_ext = RemoveExtension();
 434   // If the new extension is "" or ".", then just remove the current extension.
 435   if (extension.empty() || extension == StringType(1, kExtensionSeparator))
 436     return no_ext;
 437
 438   StringType str = no_ext.value();
 439   if (extension[0] != kExtensionSeparator)
 440     str.append(1, kExtensionSeparator);
 441   str.append(extension);
 442   return FilePath(str);
 443 }
 444
 445 bool FilePath::MatchesExtension(const StringType& extension) const {
 446   DCHECK(extension.empty() || extension[0] == kExtensionSeparator);
 447
 448   StringType current_extension = Extension();
 449
 450   if (current_extension.length() != extension.length())
 451     return false;
 452
 453   return FilePath::CompareEqualIgnoreCase(extension, current_extension);
 454 }
 455
 456 FilePath FilePath::Append(const StringType& component) const {
 457   DCHECK(!IsPathAbsolute(component));
 458   if (path_.compare(kCurrentDirectory) == 0) {
 459     // Append normally doesn't do any normalization, but as a special case,
 460     // when appending to kCurrentDirectory, just return a new path for the
 461     // component argument.  Appending component to kCurrentDirectory would
 462     // serve no purpose other than needlessly lengthening the path, and
 463     // it's likely in practice to wind up with FilePath objects containing
 464     // only kCurrentDirectory when calling DirName on a single relative path
 465     // component.
 466     return FilePath(component);
 467   }
 468
 469   FilePath new_path(path_);
 470   new_path.StripTrailingSeparatorsInternal();
 471
 472   // Don't append a separator if the path is empty (indicating the current
 473   // directory) or if the path component is empty (indicating nothing to
 474   // append).
 475   if (component.length() > 0 && new_path.path_.length() > 0) {
 476     // Don't append a separator if the path still ends with a trailing
 477     // separator after stripping (indicating the root directory).
 478     if (!IsSeparator(new_path.path_[new_path.path_.length() - 1])) {
 479       // Don't append a separator if the path is just a drive letter.
 480       if (FindDriveLetter(new_path.path_) + 1 != new_path.path_.length()) {
 481         new_path.path_.append(1, kSeparators[0]);
 482       }
 483     }
 484   }
 485
 486   new_path.path_.append(component);
 487   return new_path;
 488 }
 489
 490 FilePath FilePath::Append(const FilePath& component) const {
 491   return Append(component.value());
 492 }
 493
 494 FilePath FilePath::AppendASCII(const base::StringPiece& component) const {
 495   DCHECK(IsStringASCII(component));
 496 #if defined(OS_WIN)
 497   return Append(ASCIIToUTF16(component.as_string()));
 498 #elif defined(OS_POSIX)
 499   return Append(component.as_string());
 500 #endif
 501 }
 502
 503 bool FilePath::IsAbsolute() const {
 504   return IsPathAbsolute(path_);
 505 }
 506
 507 FilePath FilePath::StripTrailingSeparators() const {
 508   FilePath new_path(path_);
 509   new_path.StripTrailingSeparatorsInternal();
 510
 511   return new_path;
 512 }
 513
 514 bool FilePath::ReferencesParent() const {
 515   std::vector<StringType> components;
 516   GetComponents(&components);
 517
 518   std::vector<StringType>::const_iterator it = components.begin();
 519   for (; it != components.end(); ++it) {
 520     const StringType& component = *it;
 521     if (component == kParentDirectory)
 522       return true;
 523   }
 524   return false;
 525 }
 526
 527 #if defined(OS_POSIX)
 528 // See file_path.h for a discussion of the encoding of paths on POSIX
 529 // platforms.  These encoding conversion functions are not quite correct.
 530
 531 string16 FilePath::LossyDisplayName() const {
 532   return WideToUTF16(base::SysNativeMBToWide(path_));
 533 }
 534
 535 std::string FilePath::MaybeAsASCII() const {
 536   if (IsStringASCII(path_))
 537     return path_;
 538   return "";
 539 }
 540
 541 std::string FilePath::AsUTF8Unsafe() const {
 542 #if defined(OS_MACOSX) || defined(OS_CHROMEOS)
 543   return value();
 544 #else
 545   return WideToUTF8(base::SysNativeMBToWide(value()));
 546 #endif
 547 }
 548
 549 // The *Hack functions are temporary while we fix the remainder of the code.
 550 // Remember to remove the #includes at the top when you remove these.
 551
 552 // static
 553 FilePath FilePath::FromWStringHack(const std::wstring& wstring) {
 554   return FilePath(base::SysWideToNativeMB(wstring));
 555 }
 556
 557 // static
 558 FilePath FilePath::FromUTF8Unsafe(const std::string& utf8) {
 559 #if defined(OS_MACOSX) || defined(OS_CHROMEOS)
 560   return FilePath(utf8);
 561 #else
 562   return FilePath(base::SysWideToNativeMB(UTF8ToWide(utf8)));
 563 #endif
 564 }
 565
 566 #elif defined(OS_WIN)
 567 string16 FilePath::LossyDisplayName() const {
 568   return path_;
 569 }
 570
 571 std::string FilePath::MaybeAsASCII() const {
 572   if (IsStringASCII(path_))
 573     return WideToASCII(path_);
 574   return "";
 575 }
 576
 577 std::string FilePath::AsUTF8Unsafe() const {
 578   return WideToUTF8(value());
 579 }
 580
 581 // static
 582 FilePath FilePath::FromWStringHack(const std::wstring& wstring) {
 583   return FilePath(wstring);
 584 }
 585
 586 // static
 587 FilePath FilePath::FromUTF8Unsafe(const std::string& utf8) {
 588   return FilePath(UTF8ToWide(utf8));
 589 }
 590 #endif
 591
 592 void FilePath::WriteToPickle(Pickle* pickle) {
 593 #if defined(OS_WIN)
 594   pickle->WriteString16(path_);
 595 #else
 596   pickle->WriteString(path_);
 597 #endif
 598 }
 599
 600 bool FilePath::ReadFromPickle(PickleIterator* iter) {
 601 #if defined(OS_WIN)
 602   if (!iter->ReadString16(&path_))
 603     return false;
 604 #else
 605   if (!iter->ReadString(&path_))
 606     return false;
 607 #endif
 608
 609   return true;
 610 }
 611
 612 #if defined(OS_WIN)
 613 // Windows specific implementation of file string comparisons
 614
 615 int FilePath::CompareIgnoreCase(const StringType& string1,
 616                                 const StringType& string2) {
 617   // Perform character-wise upper case comparison rather than using the
 618   // fully Unicode-aware CompareString(). For details see:
 619   // http://blogs.msdn.com/michkap/archive/2005/10/17/481600.aspx
 620   StringType::const_iterator i1 = string1.begin();
 621   StringType::const_iterator i2 = string2.begin();
 622   StringType::const_iterator string1end = string1.end();
 623   StringType::const_iterator string2end = string2.end();
 624   for ( ; i1 != string1end && i2 != string2end; ++i1, ++i2) {
 625     wchar_t c1 = (wchar_t)LOWORD(::CharUpperW((LPWSTR)MAKELONG(*i1, 0)));
 626     wchar_t c2 = (wchar_t)LOWORD(::CharUpperW((LPWSTR)MAKELONG(*i2, 0)));
 627     if (c1 < c2)
 628       return -1;
 629     if (c1 > c2)
 630       return 1;
 631   }
 632   if (i1 != string1end)
 633     return 1;
 634   if (i2 != string2end)
 635     return -1;
 636   return 0;
 637 }
 638
 639 #elif defined(OS_MACOSX)
 640 // Mac OS X specific implementation of file string comparisons
 641
 642 // cf. http://developer.apple.com/mac/library/technotes/tn/tn1150.html#UnicodeSubtleties
 643 //
 644 // "When using CreateTextEncoding to create a text encoding, you should set
 645 // the TextEncodingBase to kTextEncodingUnicodeV2_0, set the
 646 // TextEncodingVariant to kUnicodeCanonicalDecompVariant, and set the
 647 // TextEncodingFormat to kUnicode16BitFormat. Using these values ensures that
 648 // the Unicode will be in the same form as on an HFS Plus volume, even as the
 649 // Unicode standard evolves."
 650 //
 651 // Another technical article for X 10.4 updates this: one should use
 652 // the new (unambiguous) kUnicodeHFSPlusDecompVariant.
 653 // cf. http://developer.apple.com/mac/library/releasenotes/TextFonts/RN-TEC/index.html
 654 //
 655 // This implementation uses CFStringGetFileSystemRepresentation() to get the
 656 // decomposed form, and an adapted version of the FastUnicodeCompare as
 657 // described in the tech note to compare the strings.
 658
 659 // Character conversion table for FastUnicodeCompare()
 660 //
 661 // The lower case table consists of a 256-entry high-byte table followed by
 662 // some number of 256-entry subtables. The high-byte table contains either an
 663 // offset to the subtable for characters with that high byte or zero, which
 664 // means that there are no case mappings or ignored characters in that block.
 665 // Ignored characters are mapped to zero.
 666 //
 667 // cf. downloadable file linked in
 668 // http://developer.apple.com/mac/library/technotes/tn/tn1150.html#StringComparisonAlgorithm
 669
 670 namespace {
 671
 672 const UInt16 lower_case_table[] = {
 673   // High-byte indices ( == 0 iff no case mapping and no ignorables )
 674
 675   /* 0 */ 0x0100, 0x0200, 0x0000, 0x0300, 0x0400, 0x0500, 0x0000, 0x0000,
 676           0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 677   /* 1 */ 0x0600, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 678           0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 679   /* 2 */ 0x0700, 0x0800, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 680           0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 681   /* 3 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 682           0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 683   /* 4 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 684           0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 685   /* 5 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 686           0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 687   /* 6 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 688           0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 689   /* 7 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 690           0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 691   /* 8 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 692           0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 693   /* 9 */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 694           0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 695   /* A */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 696           0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 697   /* B */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 698           0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 699   /* C */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 700           0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 701   /* D */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 702           0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 703   /* E */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 704           0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 705   /* F */ 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 706           0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0900, 0x0A00,
 707
 708   // Table 1 (for high byte 0x00)
 709
 710   /* 0 */ 0xFFFF, 0x0001, 0x0002, 0x0003, 0x0004, 0x0005, 0x0006, 0x0007,
 711           0x0008, 0x0009, 0x000A, 0x000B, 0x000C, 0x000D, 0x000E, 0x000F,
 712   /* 1 */ 0x0010, 0x0011, 0x0012, 0x0013, 0x0014, 0x0015, 0x0016, 0x0017,
 713           0x0018, 0x0019, 0x001A, 0x001B, 0x001C, 0x001D, 0x001E, 0x001F,
 714   /* 2 */ 0x0020, 0x0021, 0x0022, 0x0023, 0x0024, 0x0025, 0x0026, 0x0027,
 715           0x0028, 0x0029, 0x002A, 0x002B, 0x002C, 0x002D, 0x002E, 0x002F,
 716   /* 3 */ 0x0030, 0x0031, 0x0032, 0x0033, 0x0034, 0x0035, 0x0036, 0x0037,
 717           0x0038, 0x0039, 0x003A, 0x003B, 0x003C, 0x003D, 0x003E, 0x003F,
 718   /* 4 */ 0x0040, 0x0061, 0x0062, 0x0063, 0x0064, 0x0065, 0x0066, 0x0067,
 719           0x0068, 0x0069, 0x006A, 0x006B, 0x006C, 0x006D, 0x006E, 0x006F,
 720   /* 5 */ 0x0070, 0x0071, 0x0072, 0x0073, 0x0074, 0x0075, 0x0076, 0x0077,
 721           0x0078, 0x0079, 0x007A, 0x005B, 0x005C, 0x005D, 0x005E, 0x005F,
 722   /* 6 */ 0x0060, 0x0061, 0x0062, 0x0063, 0x0064, 0x0065, 0x0066, 0x0067,
 723           0x0068, 0x0069, 0x006A, 0x006B, 0x006C, 0x006D, 0x006E, 0x006F,
 724   /* 7 */ 0x0070, 0x0071, 0x0072, 0x0073, 0x0074, 0x0075, 0x0076, 0x0077,
 725           0x0078, 0x0079, 0x007A, 0x007B, 0x007C, 0x007D, 0x007E, 0x007F,
 726   /* 8 */ 0x0080, 0x0081, 0x0082, 0x0083, 0x0084, 0x0085, 0x0086, 0x0087,
 727           0x0088, 0x0089, 0x008A, 0x008B, 0x008C, 0x008D, 0x008E, 0x008F,
 728   /* 9 */ 0x0090, 0x0091, 0x0092, 0x0093, 0x0094, 0x0095, 0x0096, 0x0097,
 729           0x0098, 0x0099, 0x009A, 0x009B, 0x009C, 0x009D, 0x009E, 0x009F,
 730   /* A */ 0x00A0, 0x00A1, 0x00A2, 0x00A3, 0x00A4, 0x00A5, 0x00A6, 0x00A7,
 731           0x00A8, 0x00A9, 0x00AA, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00AF,
 732   /* B */ 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x00B4, 0x00B5, 0x00B6, 0x00B7,
 733           0x00B8, 0x00B9, 0x00BA, 0x00BB, 0x00BC, 0x00BD, 0x00BE, 0x00BF,
 734   /* C */ 0x00C0, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00E6, 0x00C7,
 735           0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF,
 736   /* D */ 0x00F0, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x00D7,
 737           0x00F8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x00DD, 0x00FE, 0x00DF,
 738   /* E */ 0x00E0, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x00E7,
 739           0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF,
 740   /* F */ 0x00F0, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x00F7,
 741           0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x00FD, 0x00FE, 0x00FF,
 742
 743   // Table 2 (for high byte 0x01)
 744
 745   /* 0 */ 0x0100, 0x0101, 0x0102, 0x0103, 0x0104, 0x0105, 0x0106, 0x0107,
 746           0x0108, 0x0109, 0x010A, 0x010B, 0x010C, 0x010D, 0x010E, 0x010F,
 747   /* 1 */ 0x0111, 0x0111, 0x0112, 0x0113, 0x0114, 0x0115, 0x0116, 0x0117,
 748           0x0118, 0x0119, 0x011A, 0x011B, 0x011C, 0x011D, 0x011E, 0x011F,
 749   /* 2 */ 0x0120, 0x0121, 0x0122, 0x0123, 0x0124, 0x0125, 0x0127, 0x0127,
 750           0x0128, 0x0129, 0x012A, 0x012B, 0x012C, 0x012D, 0x012E, 0x012F,
 751   /* 3 */ 0x0130, 0x0131, 0x0133, 0x0133, 0x0134, 0x0135, 0x0136, 0x0137,
 752           0x0138, 0x0139, 0x013A, 0x013B, 0x013C, 0x013D, 0x013E, 0x0140,
 753   /* 4 */ 0x0140, 0x0142, 0x0142, 0x0143, 0x0144, 0x0145, 0x0146, 0x0147,
 754           0x0148, 0x0149, 0x014B, 0x014B, 0x014C, 0x014D, 0x014E, 0x014F,
 755   /* 5 */ 0x0150, 0x0151, 0x0153, 0x0153, 0x0154, 0x0155, 0x0156, 0x0157,
 756           0x0158, 0x0159, 0x015A, 0x015B, 0x015C, 0x015D, 0x015E, 0x015F,
 757   /* 6 */ 0x0160, 0x0161, 0x0162, 0x0163, 0x0164, 0x0165, 0x0167, 0x0167,
 758           0x0168, 0x0169, 0x016A, 0x016B, 0x016C, 0x016D, 0x016E, 0x016F,
 759   /* 7 */ 0x0170, 0x0171, 0x0172, 0x0173, 0x0174, 0x0175, 0x0176, 0x0177,
 760           0x0178, 0x0179, 0x017A, 0x017B, 0x017C, 0x017D, 0x017E, 0x017F,
 761   /* 8 */ 0x0180, 0x0253, 0x0183, 0x0183, 0x0185, 0x0185, 0x0254, 0x0188,
 762           0x0188, 0x0256, 0x0257, 0x018C, 0x018C, 0x018D, 0x01DD, 0x0259,
 763   /* 9 */ 0x025B, 0x0192, 0x0192, 0x0260, 0x0263, 0x0195, 0x0269, 0x0268,
 764           0x0199, 0x0199, 0x019A, 0x019B, 0x026F, 0x0272, 0x019E, 0x0275,
 765   /* A */ 0x01A0, 0x01A1, 0x01A3, 0x01A3, 0x01A5, 0x01A5, 0x01A6, 0x01A8,
 766           0x01A8, 0x0283, 0x01AA, 0x01AB, 0x01AD, 0x01AD, 0x0288, 0x01AF,
 767   /* B */ 0x01B0, 0x028A, 0x028B, 0x01B4, 0x01B4, 0x01B6, 0x01B6, 0x0292,
 768           0x01B9, 0x01B9, 0x01BA, 0x01BB, 0x01BD, 0x01BD, 0x01BE, 0x01BF,
 769   /* C */ 0x01C0, 0x01C1, 0x01C2, 0x01C3, 0x01C6, 0x01C6, 0x01C6, 0x01C9,
 770           0x01C9, 0x01C9, 0x01CC, 0x01CC, 0x01CC, 0x01CD, 0x01CE, 0x01CF,
 771   /* D */ 0x01D0, 0x01D1, 0x01D2, 0x01D3, 0x01D4, 0x01D5, 0x01D6, 0x01D7,
 772           0x01D8, 0x01D9, 0x01DA, 0x01DB, 0x01DC, 0x01DD, 0x01DE, 0x01DF,
 773   /* E */ 0x01E0, 0x01E1, 0x01E2, 0x01E3, 0x01E5, 0x01E5, 0x01E6, 0x01E7,
 774           0x01E8, 0x01E9, 0x01EA, 0x01EB, 0x01EC, 0x01ED, 0x01EE, 0x01EF,
 775   /* F */ 0x01F0, 0x01F3, 0x01F3, 0x01F3, 0x01F4, 0x01F5, 0x01F6, 0x01F7,
 776           0x01F8, 0x01F9, 0x01FA, 0x01FB, 0x01FC, 0x01FD, 0x01FE, 0x01FF,
 777
 778   // Table 3 (for high byte 0x03)
 779
 780   /* 0 */ 0x0300, 0x0301, 0x0302, 0x0303, 0x0304, 0x0305, 0x0306, 0x0307,
 781           0x0308, 0x0309, 0x030A, 0x030B, 0x030C, 0x030D, 0x030E, 0x030F,
 782   /* 1 */ 0x0310, 0x0311, 0x0312, 0x0313, 0x0314, 0x0315, 0x0316, 0x0317,
 783           0x0318, 0x0319, 0x031A, 0x031B, 0x031C, 0x031D, 0x031E, 0x031F,
 784   /* 2 */ 0x0320, 0x0321, 0x0322, 0x0323, 0x0324, 0x0325, 0x0326, 0x0327,
 785           0x0328, 0x0329, 0x032A, 0x032B, 0x032C, 0x032D, 0x032E, 0x032F,
 786   /* 3 */ 0x0330, 0x0331, 0x0332, 0x0333, 0x0334, 0x0335, 0x0336, 0x0337,
 787           0x0338, 0x0339, 0x033A, 0x033B, 0x033C, 0x033D, 0x033E, 0x033F,
 788   /* 4 */ 0x0340, 0x0341, 0x0342, 0x0343, 0x0344, 0x0345, 0x0346, 0x0347,
 789           0x0348, 0x0349, 0x034A, 0x034B, 0x034C, 0x034D, 0x034E, 0x034F,
 790   /* 5 */ 0x0350, 0x0351, 0x0352, 0x0353, 0x0354, 0x0355, 0x0356, 0x0357,
 791           0x0358, 0x0359, 0x035A, 0x035B, 0x035C, 0x035D, 0x035E, 0x035F,
 792   /* 6 */ 0x0360, 0x0361, 0x0362, 0x0363, 0x0364, 0x0365, 0x0366, 0x0367,
 793           0x0368, 0x0369, 0x036A, 0x036B, 0x036C, 0x036D, 0x036E, 0x036F,
 794   /* 7 */ 0x0370, 0x0371, 0x0372, 0x0373, 0x0374, 0x0375, 0x0376, 0x0377,
 795           0x0378, 0x0379, 0x037A, 0x037B, 0x037C, 0x037D, 0x037E, 0x037F,
 796   /* 8 */ 0x0380, 0x0381, 0x0382, 0x0383, 0x0384, 0x0385, 0x0386, 0x0387,
 797           0x0388, 0x0389, 0x038A, 0x038B, 0x038C, 0x038D, 0x038E, 0x038F,
 798   /* 9 */ 0x0390, 0x03B1, 0x03B2, 0x03B3, 0x03B4, 0x03B5, 0x03B6, 0x03B7,
 799           0x03B8, 0x03B9, 0x03BA, 0x03BB, 0x03BC, 0x03BD, 0x03BE, 0x03BF,
 800   /* A */ 0x03C0, 0x03C1, 0x03A2, 0x03C3, 0x03C4, 0x03C5, 0x03C6, 0x03C7,
 801           0x03C8, 0x03C9, 0x03AA, 0x03AB, 0x03AC, 0x03AD, 0x03AE, 0x03AF,
 802   /* B */ 0x03B0, 0x03B1, 0x03B2, 0x03B3, 0x03B4, 0x03B5, 0x03B6, 0x03B7,
 803           0x03B8, 0x03B9, 0x03BA, 0x03BB, 0x03BC, 0x03BD, 0x03BE, 0x03BF,
 804   /* C */ 0x03C0, 0x03C1, 0x03C2, 0x03C3, 0x03C4, 0x03C5, 0x03C6, 0x03C7,
 805           0x03C8, 0x03C9, 0x03CA, 0x03CB, 0x03CC, 0x03CD, 0x03CE, 0x03CF,
 806   /* D */ 0x03D0, 0x03D1, 0x03D2, 0x03D3, 0x03D4, 0x03D5, 0x03D6, 0x03D7,
 807           0x03D8, 0x03D9, 0x03DA, 0x03DB, 0x03DC, 0x03DD, 0x03DE, 0x03DF,
 808   /* E */ 0x03E0, 0x03E1, 0x03E3, 0x03E3, 0x03E5, 0x03E5, 0x03E7, 0x03E7,
 809           0x03E9, 0x03E9, 0x03EB, 0x03EB, 0x03ED, 0x03ED, 0x03EF, 0x03EF,
 810   /* F */ 0x03F0, 0x03F1, 0x03F2, 0x03F3, 0x03F4, 0x03F5, 0x03F6, 0x03F7,
 811           0x03F8, 0x03F9, 0x03FA, 0x03FB, 0x03FC, 0x03FD, 0x03FE, 0x03FF,
 812
 813   // Table 4 (for high byte 0x04)
 814
 815   /* 0 */ 0x0400, 0x0401, 0x0452, 0x0403, 0x0454, 0x0455, 0x0456, 0x0407,
 816           0x0458, 0x0459, 0x045A, 0x045B, 0x040C, 0x040D, 0x040E, 0x045F,
 817   /* 1 */ 0x0430, 0x0431, 0x0432, 0x0433, 0x0434, 0x0435, 0x0436, 0x0437,
 818           0x0438, 0x0419, 0x043A, 0x043B, 0x043C, 0x043D, 0x043E, 0x043F,
 819   /* 2 */ 0x0440, 0x0441, 0x0442, 0x0443, 0x0444, 0x0445, 0x0446, 0x0447,
 820           0x0448, 0x0449, 0x044A, 0x044B, 0x044C, 0x044D, 0x044E, 0x044F,
 821   /* 3 */ 0x0430, 0x0431, 0x0432, 0x0433, 0x0434, 0x0435, 0x0436, 0x0437,
 822           0x0438, 0x0439, 0x043A, 0x043B, 0x043C, 0x043D, 0x043E, 0x043F,
 823   /* 4 */ 0x0440, 0x0441, 0x0442, 0x0443, 0x0444, 0x0445, 0x0446, 0x0447,
 824           0x0448, 0x0449, 0x044A, 0x044B, 0x044C, 0x044D, 0x044E, 0x044F,
 825   /* 5 */ 0x0450, 0x0451, 0x0452, 0x0453, 0x0454, 0x0455, 0x0456, 0x0457,
 826           0x0458, 0x0459, 0x045A, 0x045B, 0x045C, 0x045D, 0x045E, 0x045F,
 827   /* 6 */ 0x0461, 0x0461, 0x0463, 0x0463, 0x0465, 0x0465, 0x0467, 0x0467,
 828           0x0469, 0x0469, 0x046B, 0x046B, 0x046D, 0x046D, 0x046F, 0x046F,
 829   /* 7 */ 0x0471, 0x0471, 0x0473, 0x0473, 0x0475, 0x0475, 0x0476, 0x0477,
 830           0x0479, 0x0479, 0x047B, 0x047B, 0x047D, 0x047D, 0x047F, 0x047F,
 831   /* 8 */ 0x0481, 0x0481, 0x0482, 0x0483, 0x0484, 0x0485, 0x0486, 0x0487,
 832           0x0488, 0x0489, 0x048A, 0x048B, 0x048C, 0x048D, 0x048E, 0x048F,
 833   /* 9 */ 0x0491, 0x0491, 0x0493, 0x0493, 0x0495, 0x0495, 0x0497, 0x0497,
 834           0x0499, 0x0499, 0x049B, 0x049B, 0x049D, 0x049D, 0x049F, 0x049F,
 835   /* A */ 0x04A1, 0x04A1, 0x04A3, 0x04A3, 0x04A5, 0x04A5, 0x04A7, 0x04A7,
 836           0x04A9, 0x04A9, 0x04AB, 0x04AB, 0x04AD, 0x04AD, 0x04AF, 0x04AF,
 837   /* B */ 0x04B1, 0x04B1, 0x04B3, 0x04B3, 0x04B5, 0x04B5, 0x04B7, 0x04B7,
 838           0x04B9, 0x04B9, 0x04BB, 0x04BB, 0x04BD, 0x04BD, 0x04BF, 0x04BF,
 839   /* C */ 0x04C0, 0x04C1, 0x04C2, 0x04C4, 0x04C4, 0x04C5, 0x04C6, 0x04C8,
 840           0x04C8, 0x04C9, 0x04CA, 0x04CC, 0x04CC, 0x04CD, 0x04CE, 0x04CF,
 841   /* D */ 0x04D0, 0x04D1, 0x04D2, 0x04D3, 0x04D4, 0x04D5, 0x04D6, 0x04D7,
 842           0x04D8, 0x04D9, 0x04DA, 0x04DB, 0x04DC, 0x04DD, 0x04DE, 0x04DF,
 843   /* E */ 0x04E0, 0x04E1, 0x04E2, 0x04E3, 0x04E4, 0x04E5, 0x04E6, 0x04E7,
 844           0x04E8, 0x04E9, 0x04EA, 0x04EB, 0x04EC, 0x04ED, 0x04EE, 0x04EF,
 845   /* F */ 0x04F0, 0x04F1, 0x04F2, 0x04F3, 0x04F4, 0x04F5, 0x04F6, 0x04F7,
 846           0x04F8, 0x04F9, 0x04FA, 0x04FB, 0x04FC, 0x04FD, 0x04FE, 0x04FF,
 847
 848   // Table 5 (for high byte 0x05)
 849
 850   /* 0 */ 0x0500, 0x0501, 0x0502, 0x0503, 0x0504, 0x0505, 0x0506, 0x0507,
 851           0x0508, 0x0509, 0x050A, 0x050B, 0x050C, 0x050D, 0x050E, 0x050F,
 852   /* 1 */ 0x0510, 0x0511, 0x0512, 0x0513, 0x0514, 0x0515, 0x0516, 0x0517,
 853           0x0518, 0x0519, 0x051A, 0x051B, 0x051C, 0x051D, 0x051E, 0x051F,
 854   /* 2 */ 0x0520, 0x0521, 0x0522, 0x0523, 0x0524, 0x0525, 0x0526, 0x0527,
 855           0x0528, 0x0529, 0x052A, 0x052B, 0x052C, 0x052D, 0x052E, 0x052F,
 856   /* 3 */ 0x0530, 0x0561, 0x0562, 0x0563, 0x0564, 0x0565, 0x0566, 0x0567,
 857           0x0568, 0x0569, 0x056A, 0x056B, 0x056C, 0x056D, 0x056E, 0x056F,
 858   /* 4 */ 0x0570, 0x0571, 0x0572, 0x0573, 0x0574, 0x0575, 0x0576, 0x0577,
 859           0x0578, 0x0579, 0x057A, 0x057B, 0x057C, 0x057D, 0x057E, 0x057F,
 860   /* 5 */ 0x0580, 0x0581, 0x0582, 0x0583, 0x0584, 0x0585, 0x0586, 0x0557,
 861           0x0558, 0x0559, 0x055A, 0x055B, 0x055C, 0x055D, 0x055E, 0x055F,
 862   /* 6 */ 0x0560, 0x0561, 0x0562, 0x0563, 0x0564, 0x0565, 0x0566, 0x0567,
 863           0x0568, 0x0569, 0x056A, 0x056B, 0x056C, 0x056D, 0x056E, 0x056F,
 864   /* 7 */ 0x0570, 0x0571, 0x0572, 0x0573, 0x0574, 0x0575, 0x0576, 0x0577,
 865           0x0578, 0x0579, 0x057A, 0x057B, 0x057C, 0x057D, 0x057E, 0x057F,
 866   /* 8 */ 0x0580, 0x0581, 0x0582, 0x0583, 0x0584, 0x0585, 0x0586, 0x0587,
 867           0x0588, 0x0589, 0x058A, 0x058B, 0x058C, 0x058D, 0x058E, 0x058F,
 868   /* 9 */ 0x0590, 0x0591, 0x0592, 0x0593, 0x0594, 0x0595, 0x0596, 0x0597,
 869           0x0598, 0x0599, 0x059A, 0x059B, 0x059C, 0x059D, 0x059E, 0x059F,
 870   /* A */ 0x05A0, 0x05A1, 0x05A2, 0x05A3, 0x05A4, 0x05A5, 0x05A6, 0x05A7,
 871           0x05A8, 0x05A9, 0x05AA, 0x05AB, 0x05AC, 0x05AD, 0x05AE, 0x05AF,
 872   /* B */ 0x05B0, 0x05B1, 0x05B2, 0x05B3, 0x05B4, 0x05B5, 0x05B6, 0x05B7,
 873           0x05B8, 0x05B9, 0x05BA, 0x05BB, 0x05BC, 0x05BD, 0x05BE, 0x05BF,
 874   /* C */ 0x05C0, 0x05C1, 0x05C2, 0x05C3, 0x05C4, 0x05C5, 0x05C6, 0x05C7,
 875           0x05C8, 0x05C9, 0x05CA, 0x05CB, 0x05CC, 0x05CD, 0x05CE, 0x05CF,
 876   /* D */ 0x05D0, 0x05D1, 0x05D2, 0x05D3, 0x05D4, 0x05D5, 0x05D6, 0x05D7,
 877           0x05D8, 0x05D9, 0x05DA, 0x05DB, 0x05DC, 0x05DD, 0x05DE, 0x05DF,
 878   /* E */ 0x05E0, 0x05E1, 0x05E2, 0x05E3, 0x05E4, 0x05E5, 0x05E6, 0x05E7,
 879           0x05E8, 0x05E9, 0x05EA, 0x05EB, 0x05EC, 0x05ED, 0x05EE, 0x05EF,
 880   /* F */ 0x05F0, 0x05F1, 0x05F2, 0x05F3, 0x05F4, 0x05F5, 0x05F6, 0x05F7,
 881           0x05F8, 0x05F9, 0x05FA, 0x05FB, 0x05FC, 0x05FD, 0x05FE, 0x05FF,
 882
 883   // Table 6 (for high byte 0x10)
 884
 885   /* 0 */ 0x1000, 0x1001, 0x1002, 0x1003, 0x1004, 0x1005, 0x1006, 0x1007,
 886           0x1008, 0x1009, 0x100A, 0x100B, 0x100C, 0x100D, 0x100E, 0x100F,
 887   /* 1 */ 0x1010, 0x1011, 0x1012, 0x1013, 0x1014, 0x1015, 0x1016, 0x1017,
 888           0x1018, 0x1019, 0x101A, 0x101B, 0x101C, 0x101D, 0x101E, 0x101F,
 889   /* 2 */ 0x1020, 0x1021, 0x1022, 0x1023, 0x1024, 0x1025, 0x1026, 0x1027,
 890           0x1028, 0x1029, 0x102A, 0x102B, 0x102C, 0x102D, 0x102E, 0x102F,
 891   /* 3 */ 0x1030, 0x1031, 0x1032, 0x1033, 0x1034, 0x1035, 0x1036, 0x1037,
 892           0x1038, 0x1039, 0x103A, 0x103B, 0x103C, 0x103D, 0x103E, 0x103F,
 893   /* 4 */ 0x1040, 0x1041, 0x1042, 0x1043, 0x1044, 0x1045, 0x1046, 0x1047,
 894           0x1048, 0x1049, 0x104A, 0x104B, 0x104C, 0x104D, 0x104E, 0x104F,
 895   /* 5 */ 0x1050, 0x1051, 0x1052, 0x1053, 0x1054, 0x1055, 0x1056, 0x1057,
 896           0x1058, 0x1059, 0x105A, 0x105B, 0x105C, 0x105D, 0x105E, 0x105F,
 897   /* 6 */ 0x1060, 0x1061, 0x1062, 0x1063, 0x1064, 0x1065, 0x1066, 0x1067,
 898           0x1068, 0x1069, 0x106A, 0x106B, 0x106C, 0x106D, 0x106E, 0x106F,
 899   /* 7 */ 0x1070, 0x1071, 0x1072, 0x1073, 0x1074, 0x1075, 0x1076, 0x1077,
 900           0x1078, 0x1079, 0x107A, 0x107B, 0x107C, 0x107D, 0x107E, 0x107F,
 901   /* 8 */ 0x1080, 0x1081, 0x1082, 0x1083, 0x1084, 0x1085, 0x1086, 0x1087,
 902           0x1088, 0x1089, 0x108A, 0x108B, 0x108C, 0x108D, 0x108E, 0x108F,
 903   /* 9 */ 0x1090, 0x1091, 0x1092, 0x1093, 0x1094, 0x1095, 0x1096, 0x1097,
 904           0x1098, 0x1099, 0x109A, 0x109B, 0x109C, 0x109D, 0x109E, 0x109F,
 905   /* A */ 0x10D0, 0x10D1, 0x10D2, 0x10D3, 0x10D4, 0x10D5, 0x10D6, 0x10D7,
 906           0x10D8, 0x10D9, 0x10DA, 0x10DB, 0x10DC, 0x10DD, 0x10DE, 0x10DF,
 907   /* B */ 0x10E0, 0x10E1, 0x10E2, 0x10E3, 0x10E4, 0x10E5, 0x10E6, 0x10E7,
 908           0x10E8, 0x10E9, 0x10EA, 0x10EB, 0x10EC, 0x10ED, 0x10EE, 0x10EF,
 909   /* C */ 0x10F0, 0x10F1, 0x10F2, 0x10F3, 0x10F4, 0x10F5, 0x10C6, 0x10C7,
 910           0x10C8, 0x10C9, 0x10CA, 0x10CB, 0x10CC, 0x10CD, 0x10CE, 0x10CF,
 911   /* D */ 0x10D0, 0x10D1, 0x10D2, 0x10D3, 0x10D4, 0x10D5, 0x10D6, 0x10D7,
 912           0x10D8, 0x10D9, 0x10DA, 0x10DB, 0x10DC, 0x10DD, 0x10DE, 0x10DF,
 913   /* E */ 0x10E0, 0x10E1, 0x10E2, 0x10E3, 0x10E4, 0x10E5, 0x10E6, 0x10E7,
 914           0x10E8, 0x10E9, 0x10EA, 0x10EB, 0x10EC, 0x10ED, 0x10EE, 0x10EF,
 915   /* F */ 0x10F0, 0x10F1, 0x10F2, 0x10F3, 0x10F4, 0x10F5, 0x10F6, 0x10F7,
 916           0x10F8, 0x10F9, 0x10FA, 0x10FB, 0x10FC, 0x10FD, 0x10FE, 0x10FF,
 917
 918   // Table 7 (for high byte 0x20)
 919
 920   /* 0 */ 0x2000, 0x2001, 0x2002, 0x2003, 0x2004, 0x2005, 0x2006, 0x2007,
 921           0x2008, 0x2009, 0x200A, 0x200B, 0x0000, 0x0000, 0x0000, 0x0000,
 922   /* 1 */ 0x2010, 0x2011, 0x2012, 0x2013, 0x2014, 0x2015, 0x2016, 0x2017,
 923           0x2018, 0x2019, 0x201A, 0x201B, 0x201C, 0x201D, 0x201E, 0x201F,
 924   /* 2 */ 0x2020, 0x2021, 0x2022, 0x2023, 0x2024, 0x2025, 0x2026, 0x2027,
 925           0x2028, 0x2029, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x202F,
 926   /* 3 */ 0x2030, 0x2031, 0x2032, 0x2033, 0x2034, 0x2035, 0x2036, 0x2037,
 927           0x2038, 0x2039, 0x203A, 0x203B, 0x203C, 0x203D, 0x203E, 0x203F,
 928   /* 4 */ 0x2040, 0x2041, 0x2042, 0x2043, 0x2044, 0x2045, 0x2046, 0x2047,
 929           0x2048, 0x2049, 0x204A, 0x204B, 0x204C, 0x204D, 0x204E, 0x204F,
 930   /* 5 */ 0x2050, 0x2051, 0x2052, 0x2053, 0x2054, 0x2055, 0x2056, 0x2057,
 931           0x2058, 0x2059, 0x205A, 0x205B, 0x205C, 0x205D, 0x205E, 0x205F,
 932   /* 6 */ 0x2060, 0x2061, 0x2062, 0x2063, 0x2064, 0x2065, 0x2066, 0x2067,
 933           0x2068, 0x2069, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
 934   /* 7 */ 0x2070, 0x2071, 0x2072, 0x2073, 0x2074, 0x2075, 0x2076, 0x2077,
 935           0x2078, 0x2079, 0x207A, 0x207B, 0x207C, 0x207D, 0x207E, 0x207F,
 936   /* 8 */ 0x2080, 0x2081, 0x2082, 0x2083, 0x2084, 0x2085, 0x2086, 0x2087,
 937           0x2088, 0x2089, 0x208A, 0x208B, 0x208C, 0x208D, 0x208E, 0x208F,
 938   /* 9 */ 0x2090, 0x2091, 0x2092, 0x2093, 0x2094, 0x2095, 0x2096, 0x2097,
 939           0x2098, 0x2099, 0x209A, 0x209B, 0x209C, 0x209D, 0x209E, 0x209F,
 940   /* A */ 0x20A0, 0x20A1, 0x20A2, 0x20A3, 0x20A4, 0x20A5, 0x20A6, 0x20A7,
 941           0x20A8, 0x20A9, 0x20AA, 0x20AB, 0x20AC, 0x20AD, 0x20AE, 0x20AF,
 942   /* B */ 0x20B0, 0x20B1, 0x20B2, 0x20B3, 0x20B4, 0x20B5, 0x20B6, 0x20B7,
 943           0x20B8, 0x20B9, 0x20BA, 0x20BB, 0x20BC, 0x20BD, 0x20BE, 0x20BF,
 944   /* C */ 0x20C0, 0x20C1, 0x20C2, 0x20C3, 0x20C4, 0x20C5, 0x20C6, 0x20C7,
 945           0x20C8, 0x20C9, 0x20CA, 0x20CB, 0x20CC, 0x20CD, 0x20CE, 0x20CF,
 946   /* D */ 0x20D0, 0x20D1, 0x20D2, 0x20D3, 0x20D4, 0x20D5, 0x20D6, 0x20D7,
 947           0x20D8, 0x20D9, 0x20DA, 0x20DB, 0x20DC, 0x20DD, 0x20DE, 0x20DF,
 948   /* E */ 0x20E0, 0x20E1, 0x20E2, 0x20E3, 0x20E4, 0x20E5, 0x20E6, 0x20E7,
 949           0x20E8, 0x20E9, 0x20EA, 0x20EB, 0x20EC, 0x20ED, 0x20EE, 0x20EF,
 950   /* F */ 0x20F0, 0x20F1, 0x20F2, 0x20F3, 0x20F4, 0x20F5, 0x20F6, 0x20F7,
 951           0x20F8, 0x20F9, 0x20FA, 0x20FB, 0x20FC, 0x20FD, 0x20FE, 0x20FF,
 952
 953   // Table 8 (for high byte 0x21)
 954
 955   /* 0 */ 0x2100, 0x2101, 0x2102, 0x2103, 0x2104, 0x2105, 0x2106, 0x2107,
 956           0x2108, 0x2109, 0x210A, 0x210B, 0x210C, 0x210D, 0x210E, 0x210F,
 957   /* 1 */ 0x2110, 0x2111, 0x2112, 0x2113, 0x2114, 0x2115, 0x2116, 0x2117,
 958           0x2118, 0x2119, 0x211A, 0x211B, 0x211C, 0x211D, 0x211E, 0x211F,
 959   /* 2 */ 0x2120, 0x2121, 0x2122, 0x2123, 0x2124, 0x2125, 0x2126, 0x2127,
 960           0x2128, 0x2129, 0x212A, 0x212B, 0x212C, 0x212D, 0x212E, 0x212F,
 961   /* 3 */ 0x2130, 0x2131, 0x2132, 0x2133, 0x2134, 0x2135, 0x2136, 0x2137,
 962           0x2138, 0x2139, 0x213A, 0x213B, 0x213C, 0x213D, 0x213E, 0x213F,
 963   /* 4 */ 0x2140, 0x2141, 0x2142, 0x2143, 0x2144, 0x2145, 0x2146, 0x2147,
 964           0x2148, 0x2149, 0x214A, 0x214B, 0x214C, 0x214D, 0x214E, 0x214F,
 965   /* 5 */ 0x2150, 0x2151, 0x2152, 0x2153, 0x2154, 0x2155, 0x2156, 0x2157,
 966           0x2158, 0x2159, 0x215A, 0x215B, 0x215C, 0x215D, 0x215E, 0x215F,
 967   /* 6 */ 0x2170, 0x2171, 0x2172, 0x2173, 0x2174, 0x2175, 0x2176, 0x2177,
 968           0x2178, 0x2179, 0x217A, 0x217B, 0x217C, 0x217D, 0x217E, 0x217F,
 969   /* 7 */ 0x2170, 0x2171, 0x2172, 0x2173, 0x2174, 0x2175, 0x2176, 0x2177,
 970           0x2178, 0x2179, 0x217A, 0x217B, 0x217C, 0x217D, 0x217E, 0x217F,
 971   /* 8 */ 0x2180, 0x2181, 0x2182, 0x2183, 0x2184, 0x2185, 0x2186, 0x2187,
 972           0x2188, 0x2189, 0x218A, 0x218B, 0x218C, 0x218D, 0x218E, 0x218F,
 973   /* 9 */ 0x2190, 0x2191, 0x2192, 0x2193, 0x2194, 0x2195, 0x2196, 0x2197,
 974           0x2198, 0x2199, 0x219A, 0x219B, 0x219C, 0x219D, 0x219E, 0x219F,
 975   /* A */ 0x21A0, 0x21A1, 0x21A2, 0x21A3, 0x21A4, 0x21A5, 0x21A6, 0x21A7,
 976           0x21A8, 0x21A9, 0x21AA, 0x21AB, 0x21AC, 0x21AD, 0x21AE, 0x21AF,
 977   /* B */ 0x21B0, 0x21B1, 0x21B2, 0x21B3, 0x21B4, 0x21B5, 0x21B6, 0x21B7,
 978           0x21B8, 0x21B9, 0x21BA, 0x21BB, 0x21BC, 0x21BD, 0x21BE, 0x21BF,
 979   /* C */ 0x21C0, 0x21C1, 0x21C2, 0x21C3, 0x21C4, 0x21C5, 0x21C6, 0x21C7,
 980           0x21C8, 0x21C9, 0x21CA, 0x21CB, 0x21CC, 0x21CD, 0x21CE, 0x21CF,
 981   /* D */ 0x21D0, 0x21D1, 0x21D2, 0x21D3, 0x21D4, 0x21D5, 0x21D6, 0x21D7,
 982           0x21D8, 0x21D9, 0x21DA, 0x21DB, 0x21DC, 0x21DD, 0x21DE, 0x21DF,
 983   /* E */ 0x21E0, 0x21E1, 0x21E2, 0x21E3, 0x21E4, 0x21E5, 0x21E6, 0x21E7,
 984           0x21E8, 0x21E9, 0x21EA, 0x21EB, 0x21EC, 0x21ED, 0x21EE, 0x21EF,
 985   /* F */ 0x21F0, 0x21F1, 0x21F2, 0x21F3, 0x21F4, 0x21F5, 0x21F6, 0x21F7,
 986           0x21F8, 0x21F9, 0x21FA, 0x21FB, 0x21FC, 0x21FD, 0x21FE, 0x21FF,
 987
 988   // Table 9 (for high byte 0xFE)
 989
 990   /* 0 */ 0xFE00, 0xFE01, 0xFE02, 0xFE03, 0xFE04, 0xFE05, 0xFE06, 0xFE07,
 991           0xFE08, 0xFE09, 0xFE0A, 0xFE0B, 0xFE0C, 0xFE0D, 0xFE0E, 0xFE0F,
 992   /* 1 */ 0xFE10, 0xFE11, 0xFE12, 0xFE13, 0xFE14, 0xFE15, 0xFE16, 0xFE17,
 993           0xFE18, 0xFE19, 0xFE1A, 0xFE1B, 0xFE1C, 0xFE1D, 0xFE1E, 0xFE1F,
 994   /* 2 */ 0xFE20, 0xFE21, 0xFE22, 0xFE23, 0xFE24, 0xFE25, 0xFE26, 0xFE27,
 995           0xFE28, 0xFE29, 0xFE2A, 0xFE2B, 0xFE2C, 0xFE2D, 0xFE2E, 0xFE2F,
 996   /* 3 */ 0xFE30, 0xFE31, 0xFE32, 0xFE33, 0xFE34, 0xFE35, 0xFE36, 0xFE37,
 997           0xFE38, 0xFE39, 0xFE3A, 0xFE3B, 0xFE3C, 0xFE3D, 0xFE3E, 0xFE3F,
 998   /* 4 */ 0xFE40, 0xFE41, 0xFE42, 0xFE43, 0xFE44, 0xFE45, 0xFE46, 0xFE47,
 999           0xFE48, 0xFE49, 0xFE4A, 0xFE4B, 0xFE4C, 0xFE4D, 0xFE4E, 0xFE4F,
1000   /* 5 */ 0xFE50, 0xFE51, 0xFE52, 0xFE53, 0xFE54, 0xFE55, 0xFE56, 0xFE57,
1001           0xFE58, 0xFE59, 0xFE5A, 0xFE5B, 0xFE5C, 0xFE5D, 0xFE5E, 0xFE5F,
1002   /* 6 */ 0xFE60, 0xFE61, 0xFE62, 0xFE63, 0xFE64, 0xFE65, 0xFE66, 0xFE67,
1003           0xFE68, 0xFE69, 0xFE6A, 0xFE6B, 0xFE6C, 0xFE6D, 0xFE6E, 0xFE6F,
1004   /* 7 */ 0xFE70, 0xFE71, 0xFE72, 0xFE73, 0xFE74, 0xFE75, 0xFE76, 0xFE77,
1005           0xFE78, 0xFE79, 0xFE7A, 0xFE7B, 0xFE7C, 0xFE7D, 0xFE7E, 0xFE7F,
1006   /* 8 */ 0xFE80, 0xFE81, 0xFE82, 0xFE83, 0xFE84, 0xFE85, 0xFE86, 0xFE87,
1007           0xFE88, 0xFE89, 0xFE8A, 0xFE8B, 0xFE8C, 0xFE8D, 0xFE8E, 0xFE8F,
1008   /* 9 */ 0xFE90, 0xFE91, 0xFE92, 0xFE93, 0xFE94, 0xFE95, 0xFE96, 0xFE97,
1009           0xFE98, 0xFE99, 0xFE9A, 0xFE9B, 0xFE9C, 0xFE9D, 0xFE9E, 0xFE9F,
1010   /* A */ 0xFEA0, 0xFEA1, 0xFEA2, 0xFEA3, 0xFEA4, 0xFEA5, 0xFEA6, 0xFEA7,
1011           0xFEA8, 0xFEA9, 0xFEAA, 0xFEAB, 0xFEAC, 0xFEAD, 0xFEAE, 0xFEAF,
1012   /* B */ 0xFEB0, 0xFEB1, 0xFEB2, 0xFEB3, 0xFEB4, 0xFEB5, 0xFEB6, 0xFEB7,
1013           0xFEB8, 0xFEB9, 0xFEBA, 0xFEBB, 0xFEBC, 0xFEBD, 0xFEBE, 0xFEBF,
1014   /* C */ 0xFEC0, 0xFEC1, 0xFEC2, 0xFEC3, 0xFEC4, 0xFEC5, 0xFEC6, 0xFEC7,
1015           0xFEC8, 0xFEC9, 0xFECA, 0xFECB, 0xFECC, 0xFECD, 0xFECE, 0xFECF,
1016   /* D */ 0xFED0, 0xFED1, 0xFED2, 0xFED3, 0xFED4, 0xFED5, 0xFED6, 0xFED7,
1017           0xFED8, 0xFED9, 0xFEDA, 0xFEDB, 0xFEDC, 0xFEDD, 0xFEDE, 0xFEDF,
1018   /* E */ 0xFEE0, 0xFEE1, 0xFEE2, 0xFEE3, 0xFEE4, 0xFEE5, 0xFEE6, 0xFEE7,
1019           0xFEE8, 0xFEE9, 0xFEEA, 0xFEEB, 0xFEEC, 0xFEED, 0xFEEE, 0xFEEF,
1020   /* F */ 0xFEF0, 0xFEF1, 0xFEF2, 0xFEF3, 0xFEF4, 0xFEF5, 0xFEF6, 0xFEF7,
1021           0xFEF8, 0xFEF9, 0xFEFA, 0xFEFB, 0xFEFC, 0xFEFD, 0xFEFE, 0x0000,
1022
1023   // Table 10 (for high byte 0xFF)
1024
1025   /* 0 */ 0xFF00, 0xFF01, 0xFF02, 0xFF03, 0xFF04, 0xFF05, 0xFF06, 0xFF07,
1026           0xFF08, 0xFF09, 0xFF0A, 0xFF0B, 0xFF0C, 0xFF0D, 0xFF0E, 0xFF0F,
1027   /* 1 */ 0xFF10, 0xFF11, 0xFF12, 0xFF13, 0xFF14, 0xFF15, 0xFF16, 0xFF17,
1028           0xFF18, 0xFF19, 0xFF1A, 0xFF1B, 0xFF1C, 0xFF1D, 0xFF1E, 0xFF1F,
1029   /* 2 */ 0xFF20, 0xFF41, 0xFF42, 0xFF43, 0xFF44, 0xFF45, 0xFF46, 0xFF47,
1030           0xFF48, 0xFF49, 0xFF4A, 0xFF4B, 0xFF4C, 0xFF4D, 0xFF4E, 0xFF4F,
1031   /* 3 */ 0xFF50, 0xFF51, 0xFF52, 0xFF53, 0xFF54, 0xFF55, 0xFF56, 0xFF57,
1032           0xFF58, 0xFF59, 0xFF5A, 0xFF3B, 0xFF3C, 0xFF3D, 0xFF3E, 0xFF3F,
1033   /* 4 */ 0xFF40, 0xFF41, 0xFF42, 0xFF43, 0xFF44, 0xFF45, 0xFF46, 0xFF47,
1034           0xFF48, 0xFF49, 0xFF4A, 0xFF4B, 0xFF4C, 0xFF4D, 0xFF4E, 0xFF4F,
1035   /* 5 */ 0xFF50, 0xFF51, 0xFF52, 0xFF53, 0xFF54, 0xFF55, 0xFF56, 0xFF57,
1036           0xFF58, 0xFF59, 0xFF5A, 0xFF5B, 0xFF5C, 0xFF5D, 0xFF5E, 0xFF5F,
1037   /* 6 */ 0xFF60, 0xFF61, 0xFF62, 0xFF63, 0xFF64, 0xFF65, 0xFF66, 0xFF67,
1038           0xFF68, 0xFF69, 0xFF6A, 0xFF6B, 0xFF6C, 0xFF6D, 0xFF6E, 0xFF6F,
1039   /* 7 */ 0xFF70, 0xFF71, 0xFF72, 0xFF73, 0xFF74, 0xFF75, 0xFF76, 0xFF77,
1040           0xFF78, 0xFF79, 0xFF7A, 0xFF7B, 0xFF7C, 0xFF7D, 0xFF7E, 0xFF7F,
1041   /* 8 */ 0xFF80, 0xFF81, 0xFF82, 0xFF83, 0xFF84, 0xFF85, 0xFF86, 0xFF87,
1042           0xFF88, 0xFF89, 0xFF8A, 0xFF8B, 0xFF8C, 0xFF8D, 0xFF8E, 0xFF8F,
1043   /* 9 */ 0xFF90, 0xFF91, 0xFF92, 0xFF93, 0xFF94, 0xFF95, 0xFF96, 0xFF97,
1044           0xFF98, 0xFF99, 0xFF9A, 0xFF9B, 0xFF9C, 0xFF9D, 0xFF9E, 0xFF9F,
1045   /* A */ 0xFFA0, 0xFFA1, 0xFFA2, 0xFFA3, 0xFFA4, 0xFFA5, 0xFFA6, 0xFFA7,
1046           0xFFA8, 0xFFA9, 0xFFAA, 0xFFAB, 0xFFAC, 0xFFAD, 0xFFAE, 0xFFAF,
1047   /* B */ 0xFFB0, 0xFFB1, 0xFFB2, 0xFFB3, 0xFFB4, 0xFFB5, 0xFFB6, 0xFFB7,
1048           0xFFB8, 0xFFB9, 0xFFBA, 0xFFBB, 0xFFBC, 0xFFBD, 0xFFBE, 0xFFBF,
1049   /* C */ 0xFFC0, 0xFFC1, 0xFFC2, 0xFFC3, 0xFFC4, 0xFFC5, 0xFFC6, 0xFFC7,
1050           0xFFC8, 0xFFC9, 0xFFCA, 0xFFCB, 0xFFCC, 0xFFCD, 0xFFCE, 0xFFCF,
1051   /* D */ 0xFFD0, 0xFFD1, 0xFFD2, 0xFFD3, 0xFFD4, 0xFFD5, 0xFFD6, 0xFFD7,
1052           0xFFD8, 0xFFD9, 0xFFDA, 0xFFDB, 0xFFDC, 0xFFDD, 0xFFDE, 0xFFDF,
1053   /* E */ 0xFFE0, 0xFFE1, 0xFFE2, 0xFFE3, 0xFFE4, 0xFFE5, 0xFFE6, 0xFFE7,
1054           0xFFE8, 0xFFE9, 0xFFEA, 0xFFEB, 0xFFEC, 0xFFED, 0xFFEE, 0xFFEF,
1055   /* F */ 0xFFF0, 0xFFF1, 0xFFF2, 0xFFF3, 0xFFF4, 0xFFF5, 0xFFF6, 0xFFF7,
1056           0xFFF8, 0xFFF9, 0xFFFA, 0xFFFB, 0xFFFC, 0xFFFD, 0xFFFE, 0xFFFF,
1057 };
1058
1059 // Returns the next non-ignorable codepoint within string starting from the
1060 // position indicated by index, or zero if there are no more.
1061 // The passed-in index is automatically advanced as the characters in the input
1062 // HFS-decomposed UTF-8 strings are read.
1063 inline int HFSReadNextNonIgnorableCodepoint(const char* string,
1064                                             int length,
1065                                             int* index) {
1066   int codepoint = 0;
1067   while (*index < length && codepoint == 0) {
1068     // CBU8_NEXT returns a value < 0 in error cases. For purposes of string
1069     // comparison, we just use that value and flag it with DCHECK.
1070     CBU8_NEXT(string, *index, length, codepoint);
1071     DCHECK_GT(codepoint, 0);
1072     if (codepoint > 0) {
1073       // Check if there is a subtable for this upper byte.
1074       int lookup_offset = lower_case_table[codepoint >> 8];
1075       if (lookup_offset != 0)
1076         codepoint = lower_case_table[lookup_offset + (codepoint & 0x00FF)];
1077       // Note: codepoint1 may be again 0 at this point if the character was
1078       // an ignorable.
1079     }
1080   }
1081   return codepoint;
1082 }
1083
1084 }  // anonymous namespace
1085
1086 // Special UTF-8 version of FastUnicodeCompare. Cf:
1087 // http://developer.apple.com/mac/library/technotes/tn/tn1150.html#StringComparisonAlgorithm
1088 // The input strings must be in the special HFS decomposed form.
1089 int FilePath::HFSFastUnicodeCompare(const StringType& string1,
1090                                     const StringType& string2) {
1091   int length1 = string1.length();
1092   int length2 = string2.length();
1093   int index1 = 0;
1094   int index2 = 0;
1095
1096   for (;;) {
1097     int codepoint1 = HFSReadNextNonIgnorableCodepoint(string1.c_str(),
1098                                                       length1,
1099                                                       &index1);
1100     int codepoint2 = HFSReadNextNonIgnorableCodepoint(string2.c_str(),
1101                                                       length2,
1102                                                       &index2);
1103     if (codepoint1 != codepoint2)
1104       return (codepoint1 < codepoint2) ? -1 : 1;
1105     if (codepoint1 == 0) {
1106       DCHECK_EQ(index1, length1);
1107       DCHECK_EQ(index2, length2);
1108       return 0;
1109     }
1110   }
1111 }
1112
1113 StringType FilePath::GetHFSDecomposedForm(const StringType& string) {
1114   base::mac::ScopedCFTypeRef<CFStringRef> cfstring(
1115       CFStringCreateWithBytesNoCopy(
1116           NULL,
1117           reinterpret_cast<const UInt8*>(string.c_str()),
1118           string.length(),
1119           kCFStringEncodingUTF8,
1120           false,
1121           kCFAllocatorNull));
1122   // Query the maximum length needed to store the result. In most cases this
1123   // will overestimate the required space. The return value also already
1124   // includes the space needed for a terminating 0.
1125   CFIndex length = CFStringGetMaximumSizeOfFileSystemRepresentation(cfstring);
1126   DCHECK_GT(length, 0);  // should be at least 1 for the 0-terminator.
1127   // Reserve enough space for CFStringGetFileSystemRepresentation to write into.
1128   // Also set the length to the maximum so that we can shrink it later.
1129   // (Increasing rather than decreasing it would clobber the string contents!)
1130   StringType result;
1131   result.reserve(length);
1132   result.resize(length - 1);
1133   Boolean success = CFStringGetFileSystemRepresentation(cfstring,
1134                                                         &result[0],
1135                                                         length);
1136   if (success) {
1137     // Reduce result.length() to actual string length.
1138     result.resize(strlen(result.c_str()));
1139   } else {
1140     // An error occurred -> clear result.
1141     result.clear();
1142   }
1143   return result;
1144 }
1145
1146 int FilePath::CompareIgnoreCase(const StringType& string1,
1147                                 const StringType& string2) {
1148   // Quick checks for empty strings - these speed things up a bit and make the
1149   // following code cleaner.
1150   if (string1.empty())
1151     return string2.empty() ? 0 : -1;
1152   if (string2.empty())
1153     return 1;
1154
1155   StringType hfs1 = GetHFSDecomposedForm(string1);
1156   StringType hfs2 = GetHFSDecomposedForm(string2);
1157
1158   // GetHFSDecomposedForm() returns an empty string in an error case.
1159   if (hfs1.empty() || hfs2.empty()) {
1160     NOTREACHED();
1161     base::mac::ScopedCFTypeRef<CFStringRef> cfstring1(
1162         CFStringCreateWithBytesNoCopy(
1163             NULL,
1164             reinterpret_cast<const UInt8*>(string1.c_str()),
1165             string1.length(),
1166             kCFStringEncodingUTF8,
1167             false,
1168             kCFAllocatorNull));
1169     base::mac::ScopedCFTypeRef<CFStringRef> cfstring2(
1170         CFStringCreateWithBytesNoCopy(
1171             NULL,
1172             reinterpret_cast<const UInt8*>(string2.c_str()),
1173             string2.length(),
1174             kCFStringEncodingUTF8,
1175             false,
1176             kCFAllocatorNull));
1177     return CFStringCompare(cfstring1,
1178                            cfstring2,
1179                            kCFCompareCaseInsensitive);
1180   }
1181
1182   return HFSFastUnicodeCompare(hfs1, hfs2);
1183 }
1184
1185 #else  // << WIN. MACOSX | other (POSIX) >>
1186
1187 // Generic (POSIX) implementation of file string comparison.
1188 // TODO(rolandsteiner) check if this is sufficient/correct.
1189 int FilePath::CompareIgnoreCase(const StringType& string1,
1190                                 const StringType& string2) {
1191   int comparison = strcasecmp(string1.c_str(), string2.c_str());
1192   if (comparison < 0)
1193     return -1;
1194   if (comparison > 0)
1195     return 1;
1196   return 0;
1197 }
1198
1199 #endif  // OS versions of CompareIgnoreCase()
1200
1201
1202 void FilePath::StripTrailingSeparatorsInternal() {
1203   // If there is no drive letter, start will be 1, which will prevent stripping
1204   // the leading separator if there is only one separator.  If there is a drive
1205   // letter, start will be set appropriately to prevent stripping the first
1206   // separator following the drive letter, if a separator immediately follows
1207   // the drive letter.
1208   StringType::size_type start = FindDriveLetter(path_) + 2;
1209
1210   StringType::size_type last_stripped = StringType::npos;
1211   for (StringType::size_type pos = path_.length();
1212        pos > start && IsSeparator(path_[pos - 1]);
1213        --pos) {
1214     // If the string only has two separators and they're at the beginning,
1215     // don't strip them, unless the string began with more than two separators.
1216     if (pos != start + 1 || last_stripped == start + 2 ||
1217         !IsSeparator(path_[start - 1])) {
1218       path_.resize(pos - 1);
1219       last_stripped = pos;
1220     }
1221   }
1222 }
1223
1224 FilePath FilePath::NormalizePathSeparators() const {
1225 #if defined(FILE_PATH_USES_WIN_SEPARATORS)
1226   StringType copy = path_;
1227   for (size_t i = 1; i < arraysize(kSeparators); ++i) {
1228     std::replace(copy.begin(), copy.end(), kSeparators[i], kSeparators[0]);
1229   }
1230   return FilePath(copy);
1231 #else
1232   return *this;
1233 #endif
1234 }
1235
1236 void PrintTo(const FilePath& path, std::ostream* out) {
1237   *out << path.value();
1238 }