/usr/include/OGRE/OgreUTFString.h is in libogre-1.9-dev 1.9.0+dfsg1-4.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 | // Modified from OpenGUI under lenient license
// Original copyright details and licensing below:
// OpenGUI (http://opengui.sourceforge.net)
// This source code is released under the BSD License
// Permission is given to the Ogre project to use the contents of file within its
// source and binary applications, as well as any derivative works, in accordance
// with the terms of any license under which Ogre is or will be distributed.
//
// Ogre may relicense its copy of this file, as well as any OpenGUI released updates
// to this file, under any terms that it deems fit, and is not required to maintain
// the original BSD licensing terms of this file, however OpenGUI retains the right
// to present its copy of this file under the terms of any license under which
// OpenGUI is distributed.
//
// Ogre is not required to release to OpenGUI any future changes that it makes to
// this file, and understands and agrees that any such changes that are released
// back to OpenGUI will become available under the terms of any license under which
// OpenGUI is distributed.
//
// For brevity, this permission text may be removed from this file if desired.
// The original record kept within the SourceForge (http://sourceforge.net/) tracker
// is sufficient.
//
// - Eric Shorkey (zero/zeroskill) <opengui@rightbracket.com> [January 20th, 2007]
#ifndef __OGRE_UTFSTRING_H__
#define __OGRE_UTFSTRING_H__
#include "OgrePrerequisites.h"
#include "OgreHeaderPrefix.h"
#if OGRE_UNICODE_SUPPORT
// these are explained later
#include <iterator>
#include <string>
#include <stdexcept>
// Workaround for VC7/7.1/8.0/9.0 (2003 - 2008):
// when build with /MD or /MDd, VC have both std::basic_string<unsigned short> and
// basic_string<__wchar_t> instantiated in msvcprt[d].lib/MSVCP71[D].dll, but the header
// files tells compiler that only one of them is over there (based on /Zc:wchar_t compile
// option). And since this file used both of them, causing compiler instantiating another
// one in user object code, which lead to duplicate symbols with msvcprt.lib/MSVCP71[D].dll.
//
#if OGRE_COMPILER == OGRE_COMPILER_MSVC && (OGRE_COMP_VER >= 1300 && OGRE_COMP_VER < 1600)
# if defined(_DLL_CPPLIB)
namespace std
{
template class _CRTIMP2 basic_string<unsigned short, char_traits<unsigned short>,
allocator<unsigned short> >;
template class _CRTIMP2 basic_string<__wchar_t, char_traits<__wchar_t>,
allocator<__wchar_t> >;
}
# endif // defined(_DLL_CPPLIB)
#endif // OGRE_COMPILER == OGRE_COMPILER_MSVC && OGRE_COMP_VER == 1300
namespace Ogre {
/** \addtogroup Core
* @{
*/
/** \addtogroup Overlays
* @{
*/
/* READ THIS NOTICE BEFORE USING IN YOUR OWN APPLICATIONS
=NOTICE=
This class is not a complete Unicode solution. It purposefully does not
provide certain functionality, such as proper lexical sorting for
Unicode values. It does provide comparison operators for the sole purpose
of using UTFString as an index with std::map and other operator< sorted
containers, but it should NOT be relied upon for meaningful lexical
operations, such as alphabetical sorts. If you need this type of
functionality, look into using ICU instead (http://icu.sourceforge.net/).
=REQUIREMENTS=
There are a few requirements for proper operation. They are fairly small,
and shouldn't restrict usage on any reasonable target.
* Compiler must support unsigned 16-bit integer types
* Compiler must support signed 32-bit integer types
* wchar_t must be either UTF-16 or UTF-32 encoding, and specified as such
using the WCHAR_UTF16 macro as outlined below.
* You must include <iterator>, <string>, and <wchar>. Probably more, but
these are the most obvious.
=REQUIRED PREPROCESSOR MACROS=
This class requires two preprocessor macros to be defined in order to
work as advertised.
INT32 - must be mapped to a signed 32 bit integer (ex. #define INT32 int)
UINT16 - must be mapped to an unsigned 16 bit integer (ex. #define UINT32 unsigned short)
Additionally, a third macro should be defined to control the evaluation of wchar_t:
WCHAR_UTF16 - should be defined when wchar_t represents UTF-16 code points,
such as in Windows. Otherwise it is assumed that wchar_t is a 32-bit
integer representing UTF-32 code points.
*/
// THIS IS A VERY BRIEF AUTO DETECTION. YOU MAY NEED TO TWEAK THIS
#ifdef __STDC_ISO_10646__
// for any compiler that provides this, wchar_t is guaranteed to hold any Unicode value with a single code point (32-bit or larger)
// so we can safely skip the rest of the testing
#else // #ifdef __STDC_ISO_10646__
#if defined( __WIN32__ ) || defined( _WIN32 ) || !defined(ANDROID)
#define WCHAR_UTF16 // All currently known Windows platforms utilize UTF-16 encoding in wchar_t
#else // #if defined( __WIN32__ ) || defined( _WIN32 )
#if OGRE_COMPILER != OGRE_COMPILER_GCCE
#if WCHAR_MAX <= 0xFFFF // this is a last resort fall back test; WCHAR_MAX is defined in <wchar.h>
#define WCHAR_UTF16 // best we can tell, wchar_t is not larger than 16-bit
#endif // #if WCHAR_MAX <= 0xFFFF
#endif
#endif // #if defined( __WIN32__ ) || defined( _WIN32 )
#endif // #ifdef __STDC_ISO_10646__
// OGRE_IS_NATIVE_WCHAR_T means that wchar_t isn't a typedef of
// uint16 or uint32.
#if OGRE_COMPILER == OGRE_COMPILER_MSVC
// Don't define wchar_t related functions since it'll duplicate
// with UTFString::code_point related functions when compile
// without /Zc:wchar_t, because in this case both of them are
// a typedef of uint16.
# if defined(_NATIVE_WCHAR_T_DEFINED)
# define OGRE_IS_NATIVE_WCHAR_T 1
# else
# define OGRE_IS_NATIVE_WCHAR_T 0
# endif
#else // OGRE_COMPILER != OGRE_COMPILER_MSVC
// Assumed wchar_t is natively for other compilers
# define OGRE_IS_NATIVE_WCHAR_T 1
#endif // OGRE_COMPILER == OGRE_COMPILER_MSVC
//! A UTF-16 string with implicit conversion to/from std::string and std::wstring
/*! This class provides a complete 1 to 1 map of most std::string functions (at least to my
knowledge). Implicit conversions allow this string class to work with all common C++ string
formats, with specialty functions defined where implicit conversion would cause potential
problems or is otherwise unavailable.
Some additional functionality is present to assist in working with characters using the
32-bit UTF-32 encoding. (Which is guaranteed to fit any Unicode character into a single
code point.) \b Note: Reverse iterators do not have this functionality due to the
ambiguity that surrounds working with UTF-16 in reverse. (Such as, where should an
iterator point to represent the beginning of a surrogate pair?)
\par Supported Input Types
The supported string types for input, and their assumed encoding schemes, are:
- std::string (UTF-8)
- char* (UTF-8)
- std::wstring (autodetected UTF-16 / UTF-32 based on compiler)
- wchar_t* (autodetected UTF-16 / UTF-32 based on compiler)
\see
- For additional information on UTF-16 encoding: http://en.wikipedia.org/wiki/UTF-16
- For additional information on UTF-8 encoding: http://en.wikipedia.org/wiki/UTF-8
- For additional information on UTF-32 encoding: http://en.wikipedia.org/wiki/UTF-32
*/
class _OgreExport UTFString {
// constants used in UTF-8 conversions
static const unsigned char _lead1 = 0xC0; //110xxxxx
static const unsigned char _lead1_mask = 0x1F; //00011111
static const unsigned char _lead2 = 0xE0; //1110xxxx
static const unsigned char _lead2_mask = 0x0F; //00001111
static const unsigned char _lead3 = 0xF0; //11110xxx
static const unsigned char _lead3_mask = 0x07; //00000111
static const unsigned char _lead4 = 0xF8; //111110xx
static const unsigned char _lead4_mask = 0x03; //00000011
static const unsigned char _lead5 = 0xFC; //1111110x
static const unsigned char _lead5_mask = 0x01; //00000001
static const unsigned char _cont = 0x80; //10xxxxxx
static const unsigned char _cont_mask = 0x3F; //00111111
public:
//! size type used to indicate string size and character positions within the string
typedef size_t size_type;
//! the usual constant representing: not found, no limit, etc
static const size_type npos = static_cast<size_type>(~0);
//! a single 32-bit Unicode character
typedef uint32 unicode_char;
//! a single UTF-16 code point
typedef uint16 code_point;
//! value type typedef for use in iterators
typedef code_point value_type;
typedef std::basic_string<code_point> dstring; // data string
//! string type used for returning UTF-32 formatted data
typedef std::basic_string<unicode_char> utf32string;
//! This exception is used when invalid data streams are encountered
class _OgreExport invalid_data: public std::runtime_error { /* i don't know why the beautifier is freaking out on this line */
public:
//! constructor takes a string message that can be later retrieved by the what() function
explicit invalid_data( const std::string& _Message ): std::runtime_error( _Message ) {
/* The thing is, Bob, it's not that I'm lazy, it's that I just don't care. */
}
};
//#########################################################################
//! base iterator class for UTFString
class _OgreExport _base_iterator: public std::iterator<std::random_access_iterator_tag, value_type> { /* i don't know why the beautifier is freaking out on this line */
friend class UTFString;
protected:
_base_iterator();
void _seekFwd( size_type c );
void _seekRev( size_type c );
void _become( const _base_iterator& i );
bool _test_begin() const;
bool _test_end() const;
size_type _get_index() const;
void _jump_to( size_type index );
unicode_char _getCharacter() const;
int _setCharacter( unicode_char uc );
void _moveNext();
void _movePrev();
dstring::iterator mIter;
UTFString* mString;
};
//#########################################################################
// FORWARD ITERATORS
//#########################################################################
class _const_fwd_iterator; // forward declaration
//! forward iterator for UTFString
class _OgreExport _fwd_iterator: public _base_iterator { /* i don't know why the beautifier is freaking out on this line */
friend class _const_fwd_iterator;
public:
_fwd_iterator();
_fwd_iterator( const _fwd_iterator& i );
//! pre-increment
_fwd_iterator& operator++();
//! post-increment
_fwd_iterator operator++( int );
//! pre-decrement
_fwd_iterator& operator--();
//! post-decrement
_fwd_iterator operator--( int );
//! addition operator
_fwd_iterator operator+( difference_type n );
//! subtraction operator
_fwd_iterator operator-( difference_type n );
//! addition assignment operator
_fwd_iterator& operator+=( difference_type n );
//! subtraction assignment operator
_fwd_iterator& operator-=( difference_type n );
//! dereference operator
value_type& operator*() const;
//! dereference at offset operator
value_type& operator[]( difference_type n ) const;
//! advances to the next Unicode character, honoring surrogate pairs in the UTF-16 stream
_fwd_iterator& moveNext();
//! rewinds to the previous Unicode character, honoring surrogate pairs in the UTF-16 stream
_fwd_iterator& movePrev();
//! Returns the Unicode value of the character at the current position (decodes surrogate pairs if needed)
unicode_char getCharacter() const;
//! Sets the Unicode value of the character at the current position (adding a surrogate pair if needed); returns the amount of string length change caused by the operation
int setCharacter( unicode_char uc );
};
//#########################################################################
//! const forward iterator for UTFString
class _OgreExport _const_fwd_iterator: public _base_iterator { /* i don't know why the beautifier is freaking out on this line */
public:
_const_fwd_iterator();
_const_fwd_iterator( const _const_fwd_iterator& i );
_const_fwd_iterator( const _fwd_iterator& i );
//! pre-increment
_const_fwd_iterator& operator++();
//! post-increment
_const_fwd_iterator operator++( int );
//! pre-decrement
_const_fwd_iterator& operator--();
//! post-decrement
_const_fwd_iterator operator--( int );
//! addition operator
_const_fwd_iterator operator+( difference_type n );
//! subtraction operator
_const_fwd_iterator operator-( difference_type n );
//! addition assignment operator
_const_fwd_iterator& operator+=( difference_type n );
//! subtraction assignment operator
_const_fwd_iterator& operator-=( difference_type n );
//! dereference operator
const value_type& operator*() const;
//! dereference at offset operator
const value_type& operator[]( difference_type n ) const;
//! advances to the next Unicode character, honoring surrogate pairs in the UTF-16 stream
_const_fwd_iterator& moveNext();
//! rewinds to the previous Unicode character, honoring surrogate pairs in the UTF-16 stream
_const_fwd_iterator& movePrev();
//! Returns the Unicode value of the character at the current position (decodes surrogate pairs if needed)
unicode_char getCharacter() const;
//! difference operator
friend size_type operator-( const _const_fwd_iterator& left, const _const_fwd_iterator& right );
//! equality operator
friend bool operator==( const _const_fwd_iterator& left, const _const_fwd_iterator& right );
//! inequality operator
friend bool operator!=( const _const_fwd_iterator& left, const _const_fwd_iterator& right );
//! less than
friend bool operator<( const _const_fwd_iterator& left, const _const_fwd_iterator& right );
//! less than or equal
friend bool operator<=( const _const_fwd_iterator& left, const _const_fwd_iterator& right );
//! greater than
friend bool operator>( const _const_fwd_iterator& left, const _const_fwd_iterator& right );
//! greater than or equal
friend bool operator>=( const _const_fwd_iterator& left, const _const_fwd_iterator& right );
};
//#########################################################################
// REVERSE ITERATORS
//#########################################################################
class _const_rev_iterator; // forward declaration
//! forward iterator for UTFString
class _OgreExport _rev_iterator: public _base_iterator { /* i don't know why the beautifier is freaking out on this line */
friend class _const_rev_iterator;
public:
_rev_iterator();
_rev_iterator( const _rev_iterator& i );
//! pre-increment
_rev_iterator& operator++();
//! post-increment
_rev_iterator operator++( int );
//! pre-decrement
_rev_iterator& operator--();
//! post-decrement
_rev_iterator operator--( int );
//! addition operator
_rev_iterator operator+( difference_type n );
//! subtraction operator
_rev_iterator operator-( difference_type n );
//! addition assignment operator
_rev_iterator& operator+=( difference_type n );
//! subtraction assignment operator
_rev_iterator& operator-=( difference_type n );
//! dereference operator
value_type& operator*() const;
//! dereference at offset operator
value_type& operator[]( difference_type n ) const;
};
//#########################################################################
//! const reverse iterator for UTFString
class _OgreExport _const_rev_iterator: public _base_iterator { /* i don't know why the beautifier is freaking out on this line */
public:
_const_rev_iterator();
_const_rev_iterator( const _const_rev_iterator& i );
_const_rev_iterator( const _rev_iterator& i );
//! pre-increment
_const_rev_iterator& operator++();
//! post-increment
_const_rev_iterator operator++( int );
//! pre-decrement
_const_rev_iterator& operator--();
//! post-decrement
_const_rev_iterator operator--( int );
//! addition operator
_const_rev_iterator operator+( difference_type n );
//! subtraction operator
_const_rev_iterator operator-( difference_type n );
//! addition assignment operator
_const_rev_iterator& operator+=( difference_type n );
//! subtraction assignment operator
_const_rev_iterator& operator-=( difference_type n );
//! dereference operator
const value_type& operator*() const;
//! dereference at offset operator
const value_type& operator[]( difference_type n ) const;
//! difference operator
friend size_type operator-( const _const_rev_iterator& left, const _const_rev_iterator& right );
//! equality operator
friend bool operator==( const _const_rev_iterator& left, const _const_rev_iterator& right );
//! inequality operator
friend bool operator!=( const _const_rev_iterator& left, const _const_rev_iterator& right );
//! less than
friend bool operator<( const _const_rev_iterator& left, const _const_rev_iterator& right );
//! less than or equal
friend bool operator<=( const _const_rev_iterator& left, const _const_rev_iterator& right );
//! greater than
friend bool operator>( const _const_rev_iterator& left, const _const_rev_iterator& right );
//! greater than or equal
friend bool operator>=( const _const_rev_iterator& left, const _const_rev_iterator& right );
};
//#########################################################################
typedef _fwd_iterator iterator; //!< iterator
typedef _rev_iterator reverse_iterator; //!< reverse iterator
typedef _const_fwd_iterator const_iterator; //!< const iterator
typedef _const_rev_iterator const_reverse_iterator; //!< const reverse iterator
//!\name Constructors/Destructor
//@{
//! default constructor, creates an empty string
UTFString();
//! copy constructor
UTFString( const UTFString& copy );
//! \a length copies of \a ch
UTFString( size_type length, const code_point& ch );
//! duplicate of nul-terminated sequence \a str
UTFString( const code_point* str );
//! duplicate of \a str, \a length code points long
UTFString( const code_point* str, size_type length );
//! substring of \a str starting at \a index and \a length code points long
UTFString( const UTFString& str, size_type index, size_type length );
#if OGRE_IS_NATIVE_WCHAR_T
//! duplicate of nul-terminated \c wchar_t array
UTFString( const wchar_t* w_str );
//! duplicate of \a w_str, \a length characters long
UTFString( const wchar_t* w_str, size_type length );
#endif
//! duplicate of \a wstr
UTFString( const std::wstring& wstr );
//! duplicate of nul-terminated C-string \a c_str (UTF-8 encoding)
UTFString( const char* c_str );
//! duplicate of \a c_str, \a length characters long (UTF-8 encoding)
UTFString( const char* c_str, size_type length );
//! duplicate of \a str (UTF-8 encoding)
UTFString( const std::string& str );
#if OGRE_STRING_USE_CUSTOM_MEMORY_ALLOCATOR
UTFString( const Ogre::String& str );
#endif
//! destructor
~UTFString();
//@}
//////////////////////////////////////////////////////////////////////////
//!\name Utility functions
//@{
//! Returns the number of code points in the current string
size_type size() const;
//! Returns the number of code points in the current string
size_type length() const;
//! Returns the number of Unicode characters in the string
/*! Executes in linear time. */
size_type length_Characters() const;
//! returns the maximum number of UTF-16 code points that the string can hold
size_type max_size() const;
//! sets the capacity of the string to at least \a size code points
void reserve( size_type size );
//! changes the size of the string to \a size, filling in any new area with \a val
void resize( size_type num, const code_point& val = 0 );
//! exchanges the elements of the current string with those of \a from
void swap( UTFString& from );
//! returns \c true if the string has no elements, \c false otherwise
bool empty() const;
//! returns a pointer to the first character in the current string
const code_point* c_str() const;
//! returns a pointer to the first character in the current string
const code_point* data() const;
//! returns the number of elements that the string can hold before it will need to allocate more space
size_type capacity() const;
//! deletes all of the elements in the string
void clear();
//! returns a substring of the current string, starting at \a index, and \a num characters long.
/*! If \a num is omitted, it will default to \c UTFString::npos, and the substr() function will simply return the remainder of the string starting at \a index. */
UTFString substr( size_type index, size_type num = npos ) const;
//! appends \a val to the end of the string
void push_back( unicode_char val );
#if OGRE_IS_NATIVE_WCHAR_T
//! appends \a val to the end of the string
void push_back( wchar_t val );
#endif
//! appends \a val to the end of the string
/*! This can be used to push surrogate pair code points, you'll just need to push them
one after the other. */
void push_back( code_point val );
//! appends \a val to the end of the string
/*! Limited to characters under the 127 value barrier. */
void push_back( char val );
//! returns \c true if the given Unicode character \a ch is in this string
bool inString( unicode_char ch ) const;
//@}
//////////////////////////////////////////////////////////////////////////
//!\name Stream variations
//@{
//! returns the current string in UTF-8 form within a std::string
const std::string& asUTF8() const;
//! returns the current string in UTF-8 form as a nul-terminated char array
const char* asUTF8_c_str() const;
//! returns the current string in UTF-32 form within a utf32string
const utf32string& asUTF32() const;
//! returns the current string in UTF-32 form as a nul-terminated unicode_char array
const unicode_char* asUTF32_c_str() const;
//! returns the current string in the native form of std::wstring
const std::wstring& asWStr() const;
//! returns the current string in the native form of a nul-terminated wchar_t array
const wchar_t* asWStr_c_str() const;
//@}
//////////////////////////////////////////////////////////////////////////
//!\name Single Character Access
//@{
//! returns a reference to the element in the string at index \c loc
code_point& at( size_type loc );
//! returns a reference to the element in the string at index \c loc
const code_point& at( size_type loc ) const;
//! returns the data point \a loc evaluated as a UTF-32 value
/*! This function will will only properly decode surrogate pairs when \a loc points to the index
of a lead code point that is followed by a trailing code point. Evaluating the trailing code point
itself, or pointing to a code point that is a sentinel value (part of a broken pair) will return
the value of just that code point (not a valid Unicode value, but useful as a sentinel value). */
unicode_char getChar( size_type loc ) const;
//! sets the value of the character at \a loc to the Unicode value \a ch (UTF-32)
/*! Providing sentinel values (values between U+D800-U+DFFF) are accepted, but you should be aware
that you can also unwittingly create a valid surrogate pair if you don't pay attention to what you
are doing. @note This operation may also lengthen the string if a surrogate pair is needed to
represent the value given, but one is not available to replace; or alternatively shorten the string
if an existing surrogate pair is replaced with a character that is representable without a surrogate
pair. The return value will signify any lengthening or shortening performed, returning 0 if no change
was made, -1 if the string was shortened, or 1 if the string was lengthened. Any single call can
only change the string length by + or - 1. */
int setChar( size_type loc, unicode_char ch );
//@}
//////////////////////////////////////////////////////////////////////////
//!\name iterator acquisition
//@{
//! returns an iterator to the first element of the string
iterator begin();
//! returns an iterator to the first element of the string
const_iterator begin() const;
//! returns an iterator just past the end of the string
iterator end();
//! returns an iterator just past the end of the string
const_iterator end() const;
//! returns a reverse iterator to the last element of the string
reverse_iterator rbegin();
//! returns a reverse iterator to the last element of the string
const_reverse_iterator rbegin() const;
//! returns a reverse iterator just past the beginning of the string
reverse_iterator rend();
//! returns a reverse iterator just past the beginning of the string
const_reverse_iterator rend() const;
//@}
//////////////////////////////////////////////////////////////////////////
//!\name assign
//@{
//! gives the current string the values from \a start to \a end
UTFString& assign( iterator start, iterator end );
//! assign \a str to the current string
UTFString& assign( const UTFString& str );
//! assign the nul-terminated \a str to the current string
UTFString& assign( const code_point* str );
//! assign the first \a num characters of \a str to the current string
UTFString& assign( const code_point* str, size_type num );
//! assign \a len entries from \a str to the current string, starting at \a index
UTFString& assign( const UTFString& str, size_type index, size_type len );
//! assign \a num copies of \a ch to the current string
UTFString& assign( size_type num, const code_point& ch );
//! assign \a wstr to the current string (\a wstr is treated as a UTF-16 stream)
UTFString& assign( const std::wstring& wstr );
#if OGRE_IS_NATIVE_WCHAR_T
//! assign \a w_str to the current string
UTFString& assign( const wchar_t* w_str );
//! assign the first \a num characters of \a w_str to the current string
UTFString& assign( const wchar_t* w_str, size_type num );
#endif
//! assign \a str to the current string (\a str is treated as a UTF-8 stream)
UTFString& assign( const std::string& str );
//! assign \a c_str to the current string (\a c_str is treated as a UTF-8 stream)
UTFString& assign( const char* c_str );
//! assign the first \a num characters of \a c_str to the current string (\a c_str is treated as a UTF-8 stream)
UTFString& assign( const char* c_str, size_type num );
//@}
//////////////////////////////////////////////////////////////////////////
//!\name append
//@{
//! appends \a str on to the end of the current string
UTFString& append( const UTFString& str );
//! appends \a str on to the end of the current string
UTFString& append( const code_point* str );
//! appends a substring of \a str starting at \a index that is \a len characters long on to the end of the current string
UTFString& append( const UTFString& str, size_type index, size_type len );
//! appends \a num characters of \a str on to the end of the current string
UTFString& append( const code_point* str, size_type num );
//! appends \a num repetitions of \a ch on to the end of the current string
UTFString& append( size_type num, code_point ch );
//! appends the sequence denoted by \a start and \a end on to the end of the current string
UTFString& append( iterator start, iterator end );
#if OGRE_IS_NATIVE_WCHAR_T
//! appends \a num characters of \a str on to the end of the current string
UTFString& append( const wchar_t* w_str, size_type num );
//! appends \a num repetitions of \a ch on to the end of the current string
UTFString& append( size_type num, wchar_t ch );
#endif
//! appends \a num characters of \a str on to the end of the current string (UTF-8 encoding)
UTFString& append( const char* c_str, size_type num );
//! appends \a num repetitions of \a ch on to the end of the current string (Unicode values less than 128)
UTFString& append( size_type num, char ch );
//! appends \a num repetitions of \a ch on to the end of the current string (Full Unicode spectrum)
UTFString& append( size_type num, unicode_char ch );
//@}
//////////////////////////////////////////////////////////////////////////
//!\name insert
//@{
//! inserts \a ch before the code point denoted by \a i
iterator insert( iterator i, const code_point& ch );
//! inserts \a str into the current string, at location \a index
UTFString& insert( size_type index, const UTFString& str );
//! inserts \a str into the current string, at location \a index
UTFString& insert( size_type index, const code_point* str ) {
mData.insert( index, str );
return *this;
}
//! inserts a substring of \a str (starting at \a index2 and \a num code points long) into the current string, at location \a index1
UTFString& insert( size_type index1, const UTFString& str, size_type index2, size_type num );
//! inserts the code points denoted by \a start and \a end into the current string, before the code point specified by \a i
void insert( iterator i, iterator start, iterator end );
//! inserts \a num code points of \a str into the current string, at location \a index
UTFString& insert( size_type index, const code_point* str, size_type num );
#if OGRE_IS_NATIVE_WCHAR_T
//! inserts \a num code points of \a str into the current string, at location \a index
UTFString& insert( size_type index, const wchar_t* w_str, size_type num );
#endif
//! inserts \a num code points of \a str into the current string, at location \a index
UTFString& insert( size_type index, const char* c_str, size_type num );
//! inserts \a num copies of \a ch into the current string, at location \a index
UTFString& insert( size_type index, size_type num, code_point ch );
#if OGRE_IS_NATIVE_WCHAR_T
//! inserts \a num copies of \a ch into the current string, at location \a index
UTFString& insert( size_type index, size_type num, wchar_t ch );
#endif
//! inserts \a num copies of \a ch into the current string, at location \a index
UTFString& insert( size_type index, size_type num, char ch );
//! inserts \a num copies of \a ch into the current string, at location \a index
UTFString& insert( size_type index, size_type num, unicode_char ch );
//! inserts \a num copies of \a ch into the current string, before the code point denoted by \a i
void insert( iterator i, size_type num, const code_point& ch );
#if OGRE_IS_NATIVE_WCHAR_T
//! inserts \a num copies of \a ch into the current string, before the code point denoted by \a i
void insert( iterator i, size_type num, const wchar_t& ch );
#endif
//! inserts \a num copies of \a ch into the current string, before the code point denoted by \a i
void insert( iterator i, size_type num, const char& ch );
//! inserts \a num copies of \a ch into the current string, before the code point denoted by \a i
void insert( iterator i, size_type num, const unicode_char& ch );
//@}
//////////////////////////////////////////////////////////////////////////
//!\name erase
//@{
//! removes the code point pointed to by \a loc, returning an iterator to the next character
iterator erase( iterator loc );
//! removes the code points between \a start and \a end (including the one at \a start but not the one at \a end), returning an iterator to the code point after the last code point removed
iterator erase( iterator start, iterator end );
//! removes \a num code points from the current string, starting at \a index
UTFString& erase( size_type index = 0, size_type num = npos );
//@}
//////////////////////////////////////////////////////////////////////////
//!\name replace
//@{
//! replaces up to \a num1 code points of the current string (starting at \a index1) with \a str
UTFString& replace( size_type index1, size_type num1, const UTFString& str );
//! replaces up to \a num1 code points of the current string (starting at \a index1) with up to \a num2 code points from \a str
UTFString& replace( size_type index1, size_type num1, const UTFString& str, size_type num2 );
//! replaces up to \a num1 code points of the current string (starting at \a index1) with up to \a num2 code points from \a str beginning at \a index2
UTFString& replace( size_type index1, size_type num1, const UTFString& str, size_type index2, size_type num2 );
//! replaces code points in the current string from \a start to \a end with \a num code points from \a str
UTFString& replace( iterator start, iterator end, const UTFString& str, size_type num = npos );
//! replaces up to \a num1 code points in the current string (beginning at \a index) with \c num2 copies of \c ch
UTFString& replace( size_type index, size_type num1, size_type num2, code_point ch );
//! replaces the code points in the current string from \a start to \a end with \a num copies of \a ch
UTFString& replace( iterator start, iterator end, size_type num, code_point ch );
//@}
//////////////////////////////////////////////////////////////////////////
//!\name compare
//@{
//! compare \a str to the current string
int compare( const UTFString& str ) const;
//! compare \a str to the current string
int compare( const code_point* str ) const;
//! compare \a str to a substring of the current string, starting at \a index for \a length characters
int compare( size_type index, size_type length, const UTFString& str ) const;
//! compare a substring of \a str to a substring of the current string, where \a index2 and \a length2 refer to \a str and \a index and \a length refer to the current string
int compare( size_type index, size_type length, const UTFString& str, size_type index2, size_type length2 ) const;
//! compare a substring of \a str to a substring of the current string, where the substring of \a str begins at zero and is \a length2 characters long, and the substring of the current string begins at \a index and is \a length characters long
int compare( size_type index, size_type length, const code_point* str, size_type length2 ) const;
#if OGRE_IS_NATIVE_WCHAR_T
//! compare a substring of \a str to a substring of the current string, where the substring of \a str begins at zero and is \a length2 elements long, and the substring of the current string begins at \a index and is \a length characters long
int compare( size_type index, size_type length, const wchar_t* w_str, size_type length2 ) const;
#endif
//! compare a substring of \a str to a substring of the current string, where the substring of \a str begins at zero and is \a length2 <b>UTF-8 code points</b> long, and the substring of the current string begins at \a index and is \a length characters long
int compare( size_type index, size_type length, const char* c_str, size_type length2 ) const;
//@}
//////////////////////////////////////////////////////////////////////////
//!\name find & rfind
//@{
//! returns the index of the first occurrence of \a str within the current string, starting at \a index; returns \c UTFString::npos if nothing is found
/*! \a str is a UTF-16 encoded string, but through implicit casting can also be a UTF-8 encoded string (const char* or std::string) */
size_type find( const UTFString& str, size_type index = 0 ) const;
//! returns the index of the first occurrence of \a str within the current string and within \a length code points, starting at \a index; returns \c UTFString::npos if nothing is found
/*! \a cp_str is a UTF-16 encoded string */
size_type find( const code_point* cp_str, size_type index, size_type length ) const;
//! returns the index of the first occurrence of \a str within the current string and within \a length code points, starting at \a index; returns \c UTFString::npos if nothing is found
/*! \a cp_str is a UTF-8 encoded string */
size_type find( const char* c_str, size_type index, size_type length ) const;
#if OGRE_IS_NATIVE_WCHAR_T
//! returns the index of the first occurrence of \a str within the current string and within \a length code points, starting at \a index; returns \c UTFString::npos if nothing is found
/*! \a cp_str is a UTF-16 encoded string */
size_type find( const wchar_t* w_str, size_type index, size_type length ) const;
#endif
//! returns the index of the first occurrence \a ch within the current string, starting at \a index; returns \c UTFString::npos if nothing is found
/*! \a ch is only capable of representing Unicode values up to U+007F (127) */
size_type find( char ch, size_type index = 0 ) const;
//! returns the index of the first occurrence \a ch within the current string, starting at \a index; returns \c UTFString::npos if nothing is found
/*! \a ch is only capable of representing Unicode values up to U+FFFF (65535) */
size_type find( code_point ch, size_type index = 0 ) const;
#if OGRE_IS_NATIVE_WCHAR_T
//! returns the index of the first occurrence \a ch within the current string, starting at \a index; returns \c UTFString::npos if nothing is found
/*! \a ch is only capable of representing Unicode values up to U+FFFF (65535) */
size_type find( wchar_t ch, size_type index = 0 ) const;
#endif
//! returns the index of the first occurrence \a ch within the current string, starting at \a index; returns \c UTFString::npos if nothing is found
/*! \a ch can fully represent any Unicode character */
size_type find( unicode_char ch, size_type index = 0 ) const;
//! returns the location of the first occurrence of \a str in the current string, doing a reverse search from \a index; returns \c UTFString::npos if nothing is found
size_type rfind( const UTFString& str, size_type index = 0 ) const;
//! returns the location of the first occurrence of \a str in the current string, doing a reverse search from \a index, searching at most \a num characters; returns \c UTFString::npos if nothing is found
size_type rfind( const code_point* cp_str, size_type index, size_type num ) const;
//! returns the location of the first occurrence of \a str in the current string, doing a reverse search from \a index, searching at most \a num characters; returns \c UTFString::npos if nothing is found
size_type rfind( const char* c_str, size_type index, size_type num ) const;
#if OGRE_IS_NATIVE_WCHAR_T
//! returns the location of the first occurrence of \a str in the current string, doing a reverse search from \a index, searching at most \a num characters; returns \c UTFString::npos if nothing is found
size_type rfind( const wchar_t* w_str, size_type index, size_type num ) const;
#endif
//! returns the location of the first occurrence of \a ch in the current string, doing a reverse search from \a index; returns \c UTFString::npos if nothing is found
size_type rfind( char ch, size_type index = 0 ) const;
//! returns the location of the first occurrence of \a ch in the current string, doing a reverse search from \a index; returns \c UTFString::npos if nothing is found
size_type rfind( code_point ch, size_type index ) const;
#if OGRE_IS_NATIVE_WCHAR_T
//! returns the location of the first occurrence of \a ch in the current string, doing a reverse search from \a index; returns \c UTFString::npos if nothing is found
size_type rfind( wchar_t ch, size_type index = 0 ) const;
#endif
//! returns the location of the first occurrence of \a ch in the current string, doing a reverse search from \a index; returns \c UTFString::npos if nothing is found
size_type rfind( unicode_char ch, size_type index = 0 ) const;
//@}
//////////////////////////////////////////////////////////////////////////
//!\name find_first/last_(not)_of
//@{
//! Returns the index of the first character within the current string that matches \b any character in \a str, beginning the search at \a index and searching at most \a num characters; returns \c UTFString::npos if nothing is found
size_type find_first_of( const UTFString &str, size_type index = 0, size_type num = npos ) const;
//! returns the index of the first occurrence of \a ch in the current string, starting the search at \a index; returns \c UTFString::npos if nothing is found
size_type find_first_of( code_point ch, size_type index = 0 ) const;
//! returns the index of the first occurrence of \a ch in the current string, starting the search at \a index; returns \c UTFString::npos if nothing is found
size_type find_first_of( char ch, size_type index = 0 ) const;
#if OGRE_IS_NATIVE_WCHAR_T
//! returns the index of the first occurrence of \a ch in the current string, starting the search at \a index; returns \c UTFString::npos if nothing is found
size_type find_first_of( wchar_t ch, size_type index = 0 ) const;
#endif
//! returns the index of the first occurrence of \a ch in the current string, starting the search at \a index; returns \c UTFString::npos if nothing is found
size_type find_first_of( unicode_char ch, size_type index = 0 ) const;
//! returns the index of the first character within the current string that does not match any character in \a str, beginning the search at \a index and searching at most \a num characters; returns \c UTFString::npos if nothing is found
size_type find_first_not_of( const UTFString& str, size_type index = 0, size_type num = npos ) const;
//! returns the index of the first character within the current string that does not match \a ch, starting the search at \a index; returns \c UTFString::npos if nothing is found
size_type find_first_not_of( code_point ch, size_type index = 0 ) const;
//! returns the index of the first character within the current string that does not match \a ch, starting the search at \a index; returns \c UTFString::npos if nothing is found
size_type find_first_not_of( char ch, size_type index = 0 ) const;
#if OGRE_IS_NATIVE_WCHAR_T
//! returns the index of the first character within the current string that does not match \a ch, starting the search at \a index; returns \c UTFString::npos if nothing is found
size_type find_first_not_of( wchar_t ch, size_type index = 0 ) const;
#endif
//! returns the index of the first character within the current string that does not match \a ch, starting the search at \a index; returns \c UTFString::npos if nothing is found
size_type find_first_not_of( unicode_char ch, size_type index = 0 ) const;
//! returns the index of the first character within the current string that matches any character in \a str, doing a reverse search from \a index and searching at most \a num characters; returns \c UTFString::npos if nothing is found
size_type find_last_of( const UTFString& str, size_type index = npos, size_type num = npos ) const;
//! returns the index of the first occurrence of \a ch in the current string, doing a reverse search from \a index; returns \c UTFString::npos if nothing is found
size_type find_last_of( code_point ch, size_type index = npos ) const;
//! returns the index of the first occurrence of \a ch in the current string, doing a reverse search from \a index; returns \c UTFString::npos if nothing is found
size_type find_last_of( char ch, size_type index = npos ) const {
return find_last_of( static_cast<code_point>( ch ), index );
}
#if OGRE_IS_NATIVE_WCHAR_T
//! returns the index of the first occurrence of \a ch in the current string, doing a reverse search from \a index; returns \c UTFString::npos if nothing is found
size_type find_last_of( wchar_t ch, size_type index = npos ) const;
#endif
//! returns the index of the first occurrence of \a ch in the current string, doing a reverse search from \a index; returns \c UTFString::npos if nothing is found
size_type find_last_of( unicode_char ch, size_type index = npos ) const;
//! returns the index of the last character within the current string that does not match any character in \a str, doing a reverse search from \a index; returns \c UTFString::npos if nothing is found
size_type find_last_not_of( const UTFString& str, size_type index = npos, size_type num = npos ) const;
//! returns the index of the last occurrence of a character that does not match \a ch in the current string, doing a reverse search from \a index; returns \c UTFString::npos if nothing is found
size_type find_last_not_of( code_point ch, size_type index = npos ) const;
//! returns the index of the last occurrence of a character that does not match \a ch in the current string, doing a reverse search from \a index; returns \c UTFString::npos if nothing is found
size_type find_last_not_of( char ch, size_type index = npos ) const;
#if OGRE_IS_NATIVE_WCHAR_T
//! returns the index of the last occurrence of a character that does not match \a ch in the current string, doing a reverse search from \a index; returns \c UTFString::npos if nothing is found
size_type find_last_not_of( wchar_t ch, size_type index = npos ) const;
#endif
//! returns the index of the last occurrence of a character that does not match \a ch in the current string, doing a reverse search from \a index; returns \c UTFString::npos if nothing is found
size_type find_last_not_of( unicode_char ch, size_type index = npos ) const;
//@}
//////////////////////////////////////////////////////////////////////////
//!\name Operators
//@{
//! less than operator
bool operator<( const UTFString& right ) const;
//! less than or equal operator
bool operator<=( const UTFString& right ) const;
//! greater than operator
bool operator>( const UTFString& right ) const;
//! greater than or equal operator
bool operator>=( const UTFString& right ) const;
//! equality operator
bool operator==( const UTFString& right ) const;
//! inequality operator
bool operator!=( const UTFString& right ) const;
//! assignment operator, implicitly casts all compatible types
UTFString& operator=( const UTFString& s );
//! assignment operator
UTFString& operator=( code_point ch );
//! assignment operator
UTFString& operator=( char ch );
#if OGRE_IS_NATIVE_WCHAR_T
//! assignment operator
UTFString& operator=( wchar_t ch );
#endif
//! assignment operator
UTFString& operator=( unicode_char ch );
//! code point dereference operator
code_point& operator[]( size_type index );
//! code point dereference operator
const code_point& operator[]( size_type index ) const;
//@}
//////////////////////////////////////////////////////////////////////////
//!\name Implicit Cast Operators
//@{
//! implicit cast to std::string
operator std::string() const;
//! implicit cast to std::wstring
operator std::wstring() const;
#if OGRE_STRING_USE_CUSTOM_MEMORY_ALLOCATOR
//! implicit cast to Ogre::String
operator Ogre::String() const;
#endif
//@}
//////////////////////////////////////////////////////////////////////////
//!\name UTF-16 character encoding/decoding
//@{
//! returns \c true if \a cp does not match the signature for the lead of follow code point of a surrogate pair in a UTF-16 sequence
static bool _utf16_independent_char( code_point cp );
//! returns \c true if \a cp matches the signature of a surrogate pair lead character
static bool _utf16_surrogate_lead( code_point cp );
//! returns \c true if \a cp matches the signature of a surrogate pair following character
static bool _utf16_surrogate_follow( code_point cp );
//! estimates the number of UTF-16 code points in the sequence starting with \a cp
static size_t _utf16_char_length( code_point cp );
//! returns the number of UTF-16 code points needed to represent the given UTF-32 character \a cp
static size_t _utf16_char_length( unicode_char uc );
//! converts the given UTF-16 character buffer \a in_cp to a single UTF-32 Unicode character \a out_uc, returns the number of code points used to create the output character (2 for surrogate pairs, otherwise 1)
/*! This function does it's best to prevent error conditions, verifying complete
surrogate pairs before applying the algorithm. In the event that half of a pair
is found it will happily generate a value in the 0xD800 - 0xDFFF range, which is
normally an invalid Unicode value but we preserve them for use as sentinel values. */
static size_t _utf16_to_utf32( const code_point in_cp[2], unicode_char& out_uc );
//! writes the given UTF-32 \a uc_in to the buffer location \a out_cp using UTF-16 encoding, returns the number of code points used to encode the input (always 1 or 2)
/*! This function, like its counterpart, will happily create invalid UTF-16 surrogate pairs. These
invalid entries will be created for any value of \c in_uc that falls in the range U+D800 - U+DFFF.
These are generally useful as sentinel values to represent various program specific conditions.
@note This function will also pass through any single UTF-16 code point without modification,
making it a safe method of ensuring a stream that is unknown UTF-32 or UTF-16 is truly UTF-16.*/
static size_t _utf32_to_utf16( const unicode_char& in_uc, code_point out_cp[2] );
//@}
//////////////////////////////////////////////////////////////////////////
//!\name UTF-8 character encoding/decoding
//@{
//! returns \c true if \a cp is the beginning of a UTF-8 sequence
static bool _utf8_start_char( unsigned char cp );
//! estimates the number of UTF-8 code points in the sequence starting with \a cp
static size_t _utf8_char_length( unsigned char cp );
//! returns the number of UTF-8 code points needed to represent the given UTF-32 character \a cp
static size_t _utf8_char_length( unicode_char uc );
//! converts the given UTF-8 character buffer to a single UTF-32 Unicode character, returns the number of bytes used to create the output character (maximum of 6)
static size_t _utf8_to_utf32( const unsigned char in_cp[6], unicode_char& out_uc );
//! writes the given UTF-32 \a uc_in to the buffer location \a out_cp using UTF-8 encoding, returns the number of bytes used to encode the input
static size_t _utf32_to_utf8( const unicode_char& in_uc, unsigned char out_cp[6] );
//! verifies a UTF-8 stream, returning the total number of Unicode characters found
static size_type _verifyUTF8( const unsigned char* c_str );
//! verifies a UTF-8 stream, returning the total number of Unicode characters found
static size_type _verifyUTF8( const std::string& str );
//@}
private:
//template<class ITER_TYPE> friend class _iterator;
dstring mData;
//! buffer data type identifier
enum BufferType {
bt_none,
bt_string,
bt_wstring,
bt_utf32string
};
//! common constructor operations
void _init();
///////////////////////////////////////////////////////////////////////
// Scratch buffer
//! auto cleans the scratch buffer using the proper delete for the stored type
void _cleanBuffer() const;
//! create a std::string in the scratch buffer area
void _getBufferStr() const;
//! create a std::wstring in the scratch buffer area
void _getBufferWStr() const;
//! create a utf32string in the scratch buffer area
void _getBufferUTF32Str() const;
void _load_buffer_UTF8() const;
void _load_buffer_WStr() const;
void _load_buffer_UTF32() const;
mutable BufferType mBufferType; // identifies the data type held in mBuffer
mutable size_t mBufferSize; // size of the CString buffer
// multi-purpose buffer used everywhere we need a throw-away buffer
union {
mutable void* mVoidBuffer;
mutable std::string* mStrBuffer;
mutable std::wstring* mWStrBuffer;
mutable utf32string* mUTF32StrBuffer;
}
mBuffer;
};
//! string addition operator \relates UTFString
inline UTFString operator+( const UTFString& s1, const UTFString& s2 ) {
return UTFString( s1 ).append( s2 );
}
//! string addition operator \relates UTFString
inline UTFString operator+( const UTFString& s1, UTFString::code_point c ) {
return UTFString( s1 ).append( 1, c );
}
//! string addition operator \relates UTFString
inline UTFString operator+( const UTFString& s1, UTFString::unicode_char c ) {
return UTFString( s1 ).append( 1, c );
}
//! string addition operator \relates UTFString
inline UTFString operator+( const UTFString& s1, char c ) {
return UTFString( s1 ).append( 1, c );
}
#if OGRE_IS_NATIVE_WCHAR_T
//! string addition operator \relates UTFString
inline UTFString operator+( const UTFString& s1, wchar_t c ) {
return UTFString( s1 ).append( 1, c );
}
#endif
//! string addition operator \relates UTFString
inline UTFString operator+( UTFString::code_point c, const UTFString& s2 ) {
return UTFString().append( 1, c ).append( s2 );
}
//! string addition operator \relates UTFString
inline UTFString operator+( UTFString::unicode_char c, const UTFString& s2 ) {
return UTFString().append( 1, c ).append( s2 );
}
//! string addition operator \relates UTFString
inline UTFString operator+( char c, const UTFString& s2 ) {
return UTFString().append( 1, c ).append( s2 );
}
#if OGRE_IS_NATIVE_WCHAR_T
//! string addition operator \relates UTFString
inline UTFString operator+( wchar_t c, const UTFString& s2 ) {
return UTFString().append( 1, c ).append( s2 );
}
#endif
// (const) forward iterator common operators
inline UTFString::size_type operator-( const UTFString::_const_fwd_iterator& left, const UTFString::_const_fwd_iterator& right ) {
return ( left.mIter - right.mIter );
}
inline bool operator==( const UTFString::_const_fwd_iterator& left, const UTFString::_const_fwd_iterator& right ) {
return left.mIter == right.mIter;
}
inline bool operator!=( const UTFString::_const_fwd_iterator& left, const UTFString::_const_fwd_iterator& right ) {
return left.mIter != right.mIter;
}
inline bool operator<( const UTFString::_const_fwd_iterator& left, const UTFString::_const_fwd_iterator& right ) {
return left.mIter < right.mIter;
}
inline bool operator<=( const UTFString::_const_fwd_iterator& left, const UTFString::_const_fwd_iterator& right ) {
return left.mIter <= right.mIter;
}
inline bool operator>( const UTFString::_const_fwd_iterator& left, const UTFString::_const_fwd_iterator& right ) {
return left.mIter > right.mIter;
}
inline bool operator>=( const UTFString::_const_fwd_iterator& left, const UTFString::_const_fwd_iterator& right ) {
return left.mIter >= right.mIter;
}
// (const) reverse iterator common operators
// NB: many of these operations are evaluated in reverse because this is a reverse iterator wrapping a forward iterator
inline UTFString::size_type operator-( const UTFString::_const_rev_iterator& left, const UTFString::_const_rev_iterator& right ) {
return ( right.mIter - left.mIter );
}
inline bool operator==( const UTFString::_const_rev_iterator& left, const UTFString::_const_rev_iterator& right ) {
return left.mIter == right.mIter;
}
inline bool operator!=( const UTFString::_const_rev_iterator& left, const UTFString::_const_rev_iterator& right ) {
return left.mIter != right.mIter;
}
inline bool operator<( const UTFString::_const_rev_iterator& left, const UTFString::_const_rev_iterator& right ) {
return right.mIter < left.mIter;
}
inline bool operator<=( const UTFString::_const_rev_iterator& left, const UTFString::_const_rev_iterator& right ) {
return right.mIter <= left.mIter;
}
inline bool operator>( const UTFString::_const_rev_iterator& left, const UTFString::_const_rev_iterator& right ) {
return right.mIter > left.mIter;
}
inline bool operator>=( const UTFString::_const_rev_iterator& left, const UTFString::_const_rev_iterator& right ) {
return right.mIter >= left.mIter;
}
//! std::ostream write operator \relates UTFString
inline std::ostream& operator << ( std::ostream& os, const UTFString& s ) {
return os << s.asUTF8();
}
//! std::wostream write operator \relates UTFString
inline std::wostream& operator << ( std::wostream& os, const UTFString& s ) {
return os << s.asWStr();
}
/** @} */
/** @} */
} // namespace Ogre{
#endif // OGRE_UNICODE_SUPPORT
#include "OgreHeaderSuffix.h"
#endif
|