Functions for manipulating UTF and various types of text. More...

#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <iconv.h>
#include <wchar.h>
#include <errno.h>
#include <limits.h>
#include <math.h>
#include "uemf_utf.h"

Functions
void	wchar8show (const char *src)
	Dump a UTF8 string. Not for use in production code. More...

void	wchar16show (const uint16_t *src)
	Dump a UTF16 string. Not for use in production code. More...

void	wchar32show (const uint32_t *src)
	Dump a UTF32 string. Not for use in production code.

void	wchartshow (const wchar_t *src)
	Dump a wchar_t string. Not for use in production code. More...

size_t	wchar16len (const uint16_t *src)
	Find the number of (storage) characters in a 16 bit character string, not including terminator. More...

size_t	wchar32len (const uint32_t *src)
	Find the number of (storage) characters in a 32 bit character string, not including terminator. More...

void	wchar16strncpy (uint16_t dst, const uint16_t src, size_t nchars)
	Strncpy for wchar16 (UTF16). More...

void	wchar16strncpypad (uint16_t dst, const uint16_t src, size_t nchars)
	Fill the output string with N characters, if the input string is shorter than N, pad with nulls. More...

uint16_t *	U_Utf32leToUtf16le (const uint32_t src, size_t max, size_t len)
	Convert a UTF32LE string to a UTF16LE string. More...

uint32_t *	U_Utf16leToUtf32le (const uint16_t src, size_t max, size_t len)
	Convert a UTF16LE string to a UTF32LE string. More...

uint32_t *	U_Latin1ToUtf32le (const char src, size_t max, size_t len)
	Convert a Latin1 string to a UTF32LE string. More...

uint32_t *	U_Utf8ToUtf32le (const char src, size_t max, size_t len)
	Convert a UTF8 string to a UTF32LE string. More...

char *	U_Utf32leToUtf8 (const uint32_t src, size_t max, size_t len)
	Convert a UTF32LE string to a UTF8 string. More...

uint16_t *	U_Utf8ToUtf16le (const char src, size_t max, size_t len)
	Convert a UTF-8 string to a UTF16-LE string. More...

char *	U_Utf16leToUtf8 (const uint16_t src, size_t max, size_t len)
	Convert a UTF16LE string to a UTF8 string. More...

char *	U_Utf16leToLatin1 (const uint16_t src, size_t max, size_t len)
	Convert a UTF16LE string to a LATIN1 string. More...

uint16_t	U_Utf16le (const uint16_t src)
	Put a single 16 bit character into UTF-16LE form. More...

char *	U_Utf8ToLatin1 (const char src, size_t max, size_t len)
	Convert a UTF8 string to a Latin1 string. More...

char *	U_Latin1ToUtf8 (const char src, size_t max, size_t len)
	Convert a Latin1 string to a UTF8 string. More...

int	U_Utf16leEdit (uint16_t *src, uint16_t find, uint16_t replace)
	Single character replacement in a UTF-16LE string. More...

char *	U_strdup (const char *s)
	strdup for when strict C99 compliance is enforced More...

Detailed Description

Functions for manipulating UTF and various types of text.

Compile with "U_VALGRIND" defined defined to enable code which lets valgrind check each record for uninitialized data.

Compile with "SOL8" defined for Solaris 8 or 9 (Sparc).

Function Documentation

uint32_t* U_Latin1ToUtf32le	(	const char *	src,
		size_t	max,
		size_t *	len
	)

Convert a Latin1 string to a UTF32LE string.

Returns: pointer to new string or NULL if it fails

Parameters

src	Latin1 string to convert
max	number of characters to convert, if 0, until terminator
len	number of characters in new string, NOT including terminator

U_EMR_EXTTEXTOUTA records are "8 bit ASCII". In theory that is ASCII in an 8 bit character, but numerous applications store Latin1 in them, and some may store UTF-8 in them. Since very vew Latin1 strings are valid UTF-8 strings, call U_Utf8ToUtf32le first, and if it fails, then call this function.

char* U_Latin1ToUtf8	(	const char *	src,
		size_t	max,
		size_t *	len
	)

Convert a Latin1 string to a UTF8 string.

Returns: pointer to new string or NULL if it fails

Parameters

src	Latin1 string to convert
max	number of characters to convert, if 0, until terminator
len	number of characters in new string, NOT including terminator

WMF uses latin1, others UTF-8, all Latin1 should be able to convert to utf-8.

char* U_strdup ( const char * s )

strdup for when strict C99 compliance is enforced

Returns: duplicate string or NULL on error

Parameters

s	string to duplicate

uint16_t U_Utf16le ( const uint16_t src )

Put a single 16 bit character into UTF-16LE form.

Used in conjunction with U_Utf16leEdit(), because the character representation would otherwise be dependent on machine Endianness.

Returns: UTF16LE representation of the character.

Parameters

src	16 bit character

int U_Utf16leEdit	(	uint16_t *	src,
		uint16_t	find,
		uint16_t	replace
	)

Single character replacement in a UTF-16LE string.

Used solely for the Description field which contains embedded nulls, which makes it difficult to manipulate. Use some other character and then swap it.

Returns: number of substitutions, or -1 if src is not defined

Parameters

src	UTF16LE string to edit
find	character to replace
replace	replacestitute character

char* U_Utf16leToLatin1	(	const uint16_t *	src,
		size_t	max,
		size_t *	len
	)

Convert a UTF16LE string to a LATIN1 string.

Returns: pointer to new UTF8 string or NULL if it fails

Parameters

src	UTF16LE string to convert
max	number of characters to convert, if 0, until terminator
len	number of characters in new string, NOT including terminator

uint32_t* U_Utf16leToUtf32le	(	const uint16_t *	src,
		size_t	max,
		size_t *	len
	)

Convert a UTF16LE string to a UTF32LE string.

Returns: pointer to new string or NULL if it fails

Parameters

src	UTF16LE string to convert
max	number of characters to convert, if 0, until terminator
len	number of characters in new string, NOT including terminator

char* U_Utf16leToUtf8	(	const uint16_t *	src,
		size_t	max,
		size_t *	len
	)

Convert a UTF16LE string to a UTF8 string.

Returns: pointer to new UTF8 string or NULL if it fails

Parameters

src	UTF16LE string to convert
max	number of characters to convert, if 0, until terminator
len	number of characters in new string, NOT including terminator

uint16_t* U_Utf32leToUtf16le	(	const uint32_t *	src,
		size_t	max,
		size_t *	len
	)

Convert a UTF32LE string to a UTF16LE string.

Returns: pointer to new string or NULL if it fails

Parameters

src	wchar_t string to convert
max	number of characters to convert, if 0, until terminator
len	number of characters in new string, NOT including terminator

char* U_Utf32leToUtf8	(	const uint32_t *	src,
		size_t	max,
		size_t *	len
	)

Convert a UTF32LE string to a UTF8 string.

Returns: pointer to new string or NULL if it fails

Parameters

src	wchar_t string to convert
max	number of characters to convert, if 0, until terminator
len	number of characters in new string, NOT including terminator

char* U_Utf8ToLatin1	(	const char *	src,
		size_t	max,
		size_t *	len
	)

Convert a UTF8 string to a Latin1 string.

Returns: pointer to new string or NULL if it fails

Parameters

src	Latin1 string to convert
max	number of characters to convert, if 0, until terminator
len	number of characters in new string, NOT including terminator

WMF uses latin1, others UTF-8, only some utf-8 can be converted to latin1.

uint16_t* U_Utf8ToUtf16le	(	const char *	src,
		size_t	max,
		size_t *	len
	)

Convert a UTF-8 string to a UTF16-LE string.

Returns: pointer to new string or NULL if it fails

Parameters

src	UTF8 string to convert
max	number of characters to convert, if 0, until terminator
len	number of characters in new string, NOT including terminator

uint32_t* U_Utf8ToUtf32le	(	const char *	src,
		size_t	max,
		size_t *	len
	)

Convert a UTF8 string to a UTF32LE string.

Returns: pointer to new string or NULL if it fails

Parameters

src	UTF8 string to convert
max	number of characters to convert, if 0, until terminator
len	number of characters in new string, NOT including terminator

size_t wchar16len ( const uint16_t * src )

Find the number of (storage) characters in a 16 bit character string, not including terminator.

Parameters

src	string to examine

void wchar16show ( const uint16_t * src )

Dump a UTF16 string. Not for use in production code.

Parameters

src	string to examine

void wchar16strncpy	(	uint16_t *	dst,
		const uint16_t *	src,
		size_t	nchars
	)

Strncpy for wchar16 (UTF16).

Parameters

dst	destination (already allocated)
src	source
nchars	number of characters to copy

void wchar16strncpypad	(	uint16_t *	dst,
		const uint16_t *	src,
		size_t	nchars
	)

Fill the output string with N characters, if the input string is shorter than N, pad with nulls.

Parameters

dst	destination (already allocated)
src	source
nchars	number of characters to copy

size_t wchar32len ( const uint32_t * src )

Find the number of (storage) characters in a 32 bit character string, not including terminator.

Parameters

src	string to examine

void wchar8show ( const char * src )

Dump a UTF8 string. Not for use in production code.

Parameters

src	string to examine

void wchartshow ( const wchar_t * src )

Dump a wchar_t string. Not for use in production code.

Parameters

src	string to examine