Unicode versus non Unicode. Things to remember

Initial Steps for Unicode-enabling Microsoft C/C++ Source

• Define _UNICODE, undefine _MBCS if defined.


• Convert literal strings to use L or _T


• Convert string functions to use Wide or TCHAR versions.


• Clarify string lengths in API as byte or character counts. For character-based display or printing (as opposed to GUI which is pixel-based) use column counts, not byte or character.


• Replace character pointer arithmetic with GetNext style, as characters may consist of more than one Unicode code unit.


• Watch buffer size and buffer overflows- changing encodings may require either larger buffers or limiting string lengths. If character size changes from 1 byte to as many as 4 bytes, and string length was formerly 20 characters and 20 bytes, either expand the string buffer(s) from 20 to 80 bytes or limit the string to 5 characters (and therefore 20 bytes). Note maximum buffer expansion may be constrained (for example to 65 KB). Reducing string length to a fixed number of characters may break existing applications. Limiting strings to a fixed byte length is dangerous. For example, allowing any string that fits into 20 bytes. Simple operations such as uppercasing a string may cause it to grow and exceed the byte length.


• Replace functions that accept or return arguments of a single character, with functions that use strings instead. (International) Operations on a single character may result in more than one code point being returned. For example, upper('ß') returns "SS".


• Use wmain instead of main. The environment variable is then _wenviron instead of _environ.
  
  wmain( int argc, wchar_t *argv[ ], wchar_t *envp[ ] ).


• MFC Unicode applications use wWinMain as the entry point.
  
  In the Output page of the Linker folder in the project's Property Pages dialog box, set the Entry Point symbol to wWinMainCRTStartup.


• Consider fonts. Identify the fonts that will render each language or script used.

File I/O, Database, Transfer Protocol Considerations

• Consider whether to read/write UTF-8 or UTF-16 in files, databases, and for data exchange.


• Consider Endian-ness in UTF-16 files.
  
  Read/Write Big-Endian on networks. Use Big-Endian if you don't produce a BOM.
  
  Endian-ness of files will depend on the file format and/or the architecture of the source or target machine.
  
  When reading files encoded in UTF-16 or UTF-32, be prepared to swap-bytes to convert endian-ness.
  
  Also consider streams and transfer protocols and the encoding used in each.


• Label files or protocols for data exchange with the correct character encoding. E.g. set HTTP, HTML, XML to UTF-8 or UTF-16.


• Consider Unicode BOM (Byte Order Marker) and whether it should be written with data. Remove it when reading data.


• Consider encoding conversion of legacy data and files, import and export, transfer protocols. (MultiByteToWideChar, WideCharToMultiByte, mbtowc, wctomb, wctombs, mbstowcs )


• Consider writing to the Clipboard-
  
  use CF_TEXT format and write native character encoding (ANSI) text, and
  
  use CF_UNICODETEXT format and write Unicode text.


• Database applications should consider Data Type (NCHAR, NVARCHAR) and Schema Changes, Triggers, Stored Procedures, and Queries. Data Storage growth, Indexes and Performance.
  
  Note that the Unicode schema changes will have different impacts and concerns on different vendors' databases. If database portability is a requirement, the features and behaviors of each database need to be taken into account.
  
  (I know this item is seriously understated. To be expanded sometime in the future.)

Stream I/O

Streams are difficult in Microsoft C++. You may run into 3 types of problems:

1. Unicode filenames are not supported. The workaround is to use FILE * _wfopen and if needed, use the FILE handle in subsequent stream I/O.
   
   std::ifstream stm(_wfopen(pFilename, L"r"));
   
   


2. Stream I/O will convert Unicode data from/to native (ANSI) code page on read/write, not UTF-8 or UTF-16. However the stream class can be modified to read/write UTF-8. You can implement a facet to convert between Unicode and UTF-8.
   
   codecvt <wchar_t, char_traits <wchar_t> >
   
   


3. To read/write UTF-16 with stream I/O, use binary opens and binary I/O. To set binary I/O:
   
   _setmode( _fileno( stdin ), _O_BINARY );
   
   
   
   Also see the Microsoft run-time library reference: "Unicode Stream I/O in Text and Binary Modes".

Note: There aren't TCHAR equivalents for cout/wcout, cin/wcin, etc. You may want to make your own preprocessor definition for "tout", if you are compiling code both ways.

Internationalization, Advanced Unicode, Platform and Other Considerations

• Consider using locale-based routines and further internationalization.


• For Windows 95, 98 and ME, consider using the Microsoft MSLU (Microsoft Layer for Unicode)


• Consider string compares and sorting, Unicode Collation Algorithm


• Consider Unicode Normalization


• Consider Character Folding


• Reconsider doing this on your own. Bring in an experienced Unicode consultant, and deploy your existing resources on the tasks they do best. (Hey, an I18nGuy's gotta earn a living...)

Unicode BOM Encoding Values

Encoding Form BOM Encoding UTF-8 EF BB BF UTF-16 (big-endian) FE FF UTF-16 (little-endian) FF FE UTF-16BE, UTF-32BE (big-endian) No BOM! UTF-16LE, UTF-32LE (little-endian) No BOM! UTF-32 (big-endian) 00 00 FE FF UTF-32 (little-endian) FF FE 00 00 SCSU (compression) 0E FE FF

The Byte Order Marker (BOM) is Unicode character U+FEFF. (It can also represent a Zero Width No-break Space.) The code point U+FFFE is illegal in Unicode, and should never appear in a Unicode character stream. Therefore the BOM can be used in the first character of a file (or more generally a string), as an indicator of endian-ness. With UTF-16, if the first character is read as bytes FE FF then the text has the same endian-ness as the machine reading it. If the character is read as bytes FF FE, then the endian-ness is reversed and all 16-bit words should be byte-swapped as they are read-in. In the same way, the BOM indicates the endian-ness of text encoded with UTF-32. Note that not all files start with a BOM however. In fact, the Unicode Standard says that text that does not begin with a BOM MUST be interpreted in big-endian form. The character U+FEFF also serves as an encoding signature for the Unicode Encoding Forms. The table shows the encoding of U+FEFF in each of the Unicode encoding forms. Note that by definition, text labeled as UTF-16BE, UTF-32BE, UTF-32LE or UTF-16LE should not have a BOM. The endian-ness is indicated in the label. For text that is compressed with the SCSU (Standard Compression Scheme for Unicode) algorithm, there is also a recommended signature.

Constant and Global Variables

ANSI	Wide	TCHAR
EOF	WEOF	_TEOF
_environ	_wenviron	_tenviron
_pgmptr	_wpgmptr	_tpgmptr

Data Types

ANSI	Wide	TCHAR
char	wchar_t	_TCHAR
_finddata_t	_wfinddata_t	_tfinddata_t
__finddata64_t	__wfinddata64_t	_tfinddata64_t
_finddatai64_t	_wfinddatai64_t	_tfinddatai64_t
int	wint_t	_TINT
signed char	wchar_t	_TSCHAR
unsigned char	wchar_t	_TUCHAR
char	wchar_t	_TXCHAR
	L	_T or _TEXT
LPSTR (char *)	LPWSTR (wchar_t *)	LPTSTR (_TCHAR *)
LPCSTR (const char *)	LPCWSTR (const wchar_t *)	LPCTSTR (const _TCHAR *)
LPOLESTR (For OLE)	LPWSTR	LPTSTR

Platform SDK String Functions

There are many Windows API that compile into ANSI or Wide forms, depending on whether the symbol UNICODE is defined. Modules that operate on both ANSI and Wide characters, need to be aware of this. Otherwise, using the Character Data Type-independent name requires no changes, just compile with the symbol UNICODE defined.

The following list is by no means all of the Character Data Type-dependent API, just some character and string related ones. Look in WinNLS.h for some code page and locale related API.

ANSI	Wide	Character Data Type- Independent Name
CharLowerA	CharLowerW	CharLower
CharLowerBuffA	CharLowerBuffW	CharLowerBuff
CharNextA	CharNextW	CharNext
CharNextExA	CharNextExW	CharNextEx
CharPrevA	CharPrevW	CharPrev
CharPrevExA	CharPrevExW	CharPrevEx
CharToOemA	CharToOemW	CharToOem
CharToOemBuffA	CharToOemBuffW	CharToOemBuff
CharUpperA	CharUpperW	CharUpper
CharUpperBuffA	CharUpperBuffW	CharUpperBuff
CompareStringA	CompareStringW	CompareString
FoldStringA	FoldStringW	FoldString
GetStringTypeA	GetStringTypeW	GetStringType
GetStringTypeExA	GetStringTypeExW	GetStringTypeEx
IsCharAlphaA	IsCharAlphaW	IsCharAlpha
IsCharAlphaNumericA	IsCharAlphaNumericW	IsCharAlphaNumeric
IsCharLowerA	IsCharLowerW	IsCharLower
IsCharUpperA	IsCharUpperW	IsCharUpper
LoadStringA	LoadStringW	LoadString
lstrcatA	lstrcatW	lstrcat
lstrcmpA	lstrcmpW	lstrcmp
lstrcmpiA	lstrcmpiW	lstrcmpi
lstrcpyA	lstrcpyW	lstrcpy
lstrcpynA	lstrcpynW	lstrcpyn
lstrlenA	lstrlenW	lstrlen
OemToCharA	OemToCharW	OemToChar
OemToCharBuffA	OemToCharBuffW	OemToCharBuff
wsprintfA	wsprintfW	wsprintf
wvsprintfA	wvsprintfW	wvsprintf

TCHAR String Functions

Functions sorted by ANSI name, for ease of converting to Unicode.

ANSI	Wide	TCHAR
_access	_waccess	_taccess
_atoi64	_wtoi64	_tstoi64
_atoi64	_wtoi64	_ttoi64
_cgets	_cgetws	cgetts
_chdir	_wchdir	_tchdir
_chmod	_wchmod	_tchmod
_cprintf	_cwprintf	_tcprintf
_cputs	_cputws	_cputts
_creat	_wcreat	_tcreat
_cscanf	_cwscanf	_tcscanf
_ctime64	_wctime64	_tctime64
_execl	_wexecl	_texecl
_execle	_wexecle	_texecle
_execlp	_wexeclp	_texeclp
_execlpe	_wexeclpe	_texeclpe
_execv	_wexecv	_texecv
_execve	_wexecve	_texecve
_execvp	_wexecvp	_texecvp
_execvpe	_wexecvpe	_texecvpe
_fdopen	_wfdopen	_tfdopen
_fgetchar	_fgetwchar	_fgettchar
_findfirst	_wfindfirst	_tfindfirst
_findnext64	_wfindnext64	_tfindnext64
_findnext	_wfindnext	_tfindnext
_findnexti64	_wfindnexti64	_tfindnexti64
_fputchar	_fputwchar	_fputtchar
_fsopen	_wfsopen	_tfsopen
_fullpath	_wfullpath	_tfullpath
_getch	_getwch	_gettch
_getche	_getwche	_gettche
_getcwd	_wgetcwd	_tgetcwd
_getdcwd	_wgetdcwd	_tgetdcwd
_ltoa	_ltow	_ltot
_makepath	_wmakepath	_tmakepath
_mkdir	_wmkdir	_tmkdir
_mktemp	_wmktemp	_tmktemp
_open	_wopen	_topen
_popen	_wpopen	_tpopen
_putch	_putwch	_puttch
_putenv	_wputenv	_tputenv
_rmdir	_wrmdir	_trmdir
_scprintf	_scwprintf	_sctprintf
_searchenv	_wsearchenv	_tsearchenv
_snprintf	_snwprintf	_sntprintf
_snscanf	_snwscanf	_sntscanf
_sopen	_wsopen	_tsopen
_spawnl	_wspawnl	_tspawnl
_spawnle	_wspawnle	_tspawnle
_spawnlp	_wspawnlp	_tspawnlp
_spawnlpe	_wspawnlpe	_tspawnlpe
_spawnv	_wspawnv	_tspawnv
_spawnve	_wspawnve	_tspawnve
_spawnvp	_wspawnvp	_tspawnvp
_spawnvpe	_wspawnvpe	_tspawnvpe
_splitpath	_wsplitpath	_tsplitpath
_stat64	_wstat64	_tstat64
_stat	_wstat	_tstat
_stati64	_wstati64	_tstati64
_strdate	_wstrdate	_tstrdate
_strdec	_wcsdec	_tcsdec
_strdup	_wcsdup	_tcsdup
_stricmp	_wcsicmp	_tcsicmp
_stricoll	_wcsicoll	_tcsicoll
_strinc	_wcsinc	_tcsinc
_strlwr	_wcslwr	_tcslwr
_strncnt	_wcsncnt	_tcsnbcnt
_strncnt	_wcsncnt	_tcsnccnt
_strncnt	_wcsncnt	_tcsnccnt
_strncoll	_wcsncoll	_tcsnccoll
_strnextc	_wcsnextc	_tcsnextc
_strnicmp	_wcsnicmp	_tcsncicmp
_strnicmp	_wcsnicmp	_tcsnicmp
_strnicoll	_wcsnicoll	_tcsncicoll
_strnicoll	_wcsnicoll	_tcsnicoll
_strninc	_wcsninc	_tcsninc
_strnset	_wcsnset	_tcsncset
_strnset	_wcsnset	_tcsnset
_strrev	_wcsrev	_tcsrev
_strset	_wcsset	_tcsset
_strspnp	_wcsspnp	_tcsspnp
_strtime	_wstrtime	_tstrtime
_strtoi64	_wcstoi64	_tcstoi64
_strtoui64	_wcstoui64	_tcstoui64
_strupr	_wcsupr	_tcsupr
_tempnam	_wtempnam	_ttempnam
_ui64toa	_ui64tow	_ui64tot
_ultoa	_ultow	_ultot
_ungetch	_ungetwch	_ungettch
_unlink	_wunlink	_tunlink
_utime64	_wutime64	_tutime64
_utime	_wutime	_tutime
_vscprintf	_vscwprintf	_vsctprintf
_vsnprintf	_vsnwprintf	_vsntprintf
asctime	_wasctime	_tasctime
atof	_wtof	_tstof
atoi	_wtoi	_tstoi
atoi	_wtoi	_ttoi
atol	_wtol	_tstol
atol	_wtol	_ttol
character compare	Maps to macro or inline function	_tccmp
character copy	Maps to macro or inline function	_tccpy
character length	Maps to macro or inline function	_tclen
ctime	_wctime	_tctime
fgetc	fgetwc	_fgettc
fgets	fgetws	_fgetts
fopen	_wfopen	_tfopen
fprintf	fwprintf	_ftprintf
fputc	fputwc	_fputtc
fputs	fputws	_fputts
freopen	_wfreopen	_tfreopen
fscanf	fwscanf	_ftscanf
getc	getwc	_gettc
getchar	getwchar	_gettchar
getenv	_wgetenv	_tgetenv
gets	getws	_getts
isalnum	iswalnum	_istalnum
isalpha	iswalpha	_istalpha
isascii	iswascii	_istascii
iscntrl	iswcntrl	_istcntrl
isdigit	iswdigit	_istdigit
isgraph	iswgraph	_istgraph
islead (Always FALSE)	(Always FALSE)	_istlead
isleadbyte (Always FALSE)	isleadbyte (Always FALSE)	_istleadbyte
islegal (Always TRUE)	(Always TRUE)	_istlegal
islower	iswlower	_istlower
isprint	iswprint	_istprint
ispunct	iswpunct	_istpunct
isspace	iswspace	_istspace
isupper	iswupper	_istupper
isxdigit	iswxdigit	_istxdigit
main	wmain	_tmain
perror	_wperror	_tperror
printf	wprintf	_tprintf
putc	putwc	_puttc
putchar	putwchar	_puttchar
puts	_putws	_putts
remove	_wremove	_tremove
rename	_wrename	_trename
scanf	wscanf	_tscanf
setlocale	_wsetlocale	_tsetlocale
sprintf	swprintf	_stprintf
sscanf	swscanf	_stscanf
strcat	wcscat	_tcscat
strchr	wcschr	_tcschr
strcmp	wcscmp	_tcscmp
strcoll	wcscoll	_tcscoll
strcpy	wcscpy	_tcscpy
strcspn	wcscspn	_tcscspn
strerror	_wcserror	_tcserror
strftime	wcsftime	_tcsftime
strlen	wcslen	_tcsclen
strlen	wcslen	_tcslen
strncat	wcsncat	_tcsncat
strncat	wcsncat	_tcsnccat
strncmp	wcsncmp	_tcsnccmp
strncmp	wcsncmp	_tcsncmp
strncpy	wcsncpy	_tcsnccpy
strncpy	wcsncpy	_tcsncpy
strpbrk	wcspbrk	_tcspbrk
strrchr	wcsrchr	_tcsrchr
strspn	wcsspn	_tcsspn
strstr	wcsstr	_tcsstr
strtod	wcstod	_tcstod
strtok	wcstok	_tcstok
strtol	wcstol	_tcstol
strtoul	wcstoul	_tcstoul
strxfrm	wcsxfrm	_tcsxfrm
system	_wsystem	_tsystem
tmpnam	_wtmpnam	_ttmpnam
tolower	towlower	_totlower
toupper	towupper	_totupper
ungetc	ungetwc	_ungettc
vfprintf	vfwprintf	_vftprintf
vprintf	vwprintf	_vtprintf
vsprintf	vswprintf	_vstprintf
WinMain	wWinMain	_tWinMain