#include "unicode.h"
#include "invariant.h"
#include <codecvt>
#include <cstdint>
#include <iomanip>
#include <locale>
#include <sstream>

Include dependency graph for unicode.cpp:

Functions
static void	utf8_append_code (unsigned int c, std::string &result)
	Appends a unicode character to a utf8-encoded string.

std::string	narrow (const wchar_t *s)

std::wstring	widen (const char *s)

std::string	narrow (const std::wstring &s)

std::wstring	widen (const std::string &s)

std::string	utf32_native_endian_to_utf8 (const std::basic_string< char32_t > &s)

std::vector< std::string >	narrow_argv (int argc, const wchar_t **argv_wide)

static void	utf16_append_code (unsigned int code, std::wstring &result)

std::wstring	utf8_to_utf16_native_endian (const std::string &in)
	Convert UTF8-encoded string to UTF-16 with architecture-native endianness.

std::u32string	utf8_to_utf32 (const std::string &utf8_str)
	Convert UTF8-encoded string to UTF-32 with architecture-native endianness.

static void	utf16_native_endian_to_java_string (const wchar_t ch, std::ostringstream &result, const std::locale &loc)
	Escapes non-printable characters, whitespace except for spaces, double quotes and backslashes.

static void	utf16_native_endian_to_java (const wchar_t ch, std::ostringstream &result, const std::locale &loc)
	Escapes non-printable characters, whitespace except for spaces, double- and single-quotes and backslashes.

std::string	utf16_native_endian_to_java (const char16_t ch)

std::string	utf16_native_endian_to_java_string (const std::wstring &in)
	Escapes non-printable characters, whitespace except for spaces, double quotes and backslashes.

std::string	utf16_native_endian_to_utf8 (const char16_t utf16_char)

std::string	utf16_native_endian_to_utf8 (const std::u16string &utf16_str)

char16_t	codepoint_hex_to_utf16_native_endian (const std::string &hex)

std::string	codepoint_hex_to_utf8 (const std::string &hex)

Function Documentation

◆ codepoint_hex_to_utf16_native_endian()

char16_t codepoint_hex_to_utf16_native_endian ( const std::string & hex )

Parameters

hex	representation of a BMP codepoint as a four-digit string (e.g. "0041" for \u0041)

Returns: encoding of the codepoint as a single UTF-16 character in architecture-native endianness encoding

Definition at line 378 of file unicode.cpp.

◆ codepoint_hex_to_utf8()

std::string codepoint_hex_to_utf8 ( const std::string & hex )

Parameters

hex	representation of a BMP codepoint as a four-digit string (e.g. "0041" for \u0041)

Returns: UTF-8 encoding of the codepoint

Definition at line 384 of file unicode.cpp.

◆ narrow() [1/2]

std::string narrow ( const std::wstring & s )

Definition at line 64 of file unicode.cpp.

◆ narrow() [2/2]

std::string narrow ( const wchar_t * s )

Definition at line 33 of file unicode.cpp.

◆ narrow_argv()

std::vector< std::string > narrow_argv	(	int	argc,
		const wchar_t **	argv_wide
	)

Definition at line 148 of file unicode.cpp.

◆ utf16_append_code()

static void utf16_append_code	(	unsigned int	code,
		std::wstring &	result
	)

static

Definition at line 162 of file unicode.cpp.

◆ utf16_native_endian_to_java() [1/2]

std::string utf16_native_endian_to_java ( const char16_t ch )

Parameters

ch	UTF-16 character in architecture-native endianness encoding

Returns: String in US-ASCII format, with \uxxxx escapes for other characters

Definition at line 335 of file unicode.cpp.

◆ utf16_native_endian_to_java() [2/2]

static void utf16_native_endian_to_java	(	const wchar_t	ch,
		std::ostringstream &	result,
		const std::locale &	loc
	)

static

Escapes non-printable characters, whitespace except for spaces, double- and single-quotes and backslashes.

This should yield a valid Java identifier.

Parameters

ch	UTF-16 character in architecture-native endianness encoding
result	stream to receive string in US-ASCII format, with \uxxxx escapes for other characters
loc	locale to check for printable characters

Definition at line 316 of file unicode.cpp.

◆ utf16_native_endian_to_java_string() [1/2]

std::string utf16_native_endian_to_java_string ( const std::wstring & in )

Escapes non-printable characters, whitespace except for spaces, double quotes and backslashes.

This should yield a valid Java string literal. Note that this specifically does not escape single quotes, as these are not required to be escaped for Java string literals.

Parameters

in	String in UTF-16 (native endianness) format

Returns: Valid Java string literal in US-ASCII format, with \uxxxx escapes for other characters

Definition at line 350 of file unicode.cpp.

◆ utf16_native_endian_to_java_string() [2/2]

static void utf16_native_endian_to_java_string	(	const wchar_t	ch,
		std::ostringstream &	result,
		const std::locale &	loc
	)

static

Escapes non-printable characters, whitespace except for spaces, double quotes and backslashes.

This should yield a valid Java string literal. Note that this specifically does not escape single quotes, as these are not required to be escaped for Java string literals.

Parameters

ch	UTF-16 character in architecture-native endianness encoding
result	stream to receive string in US-ASCII format, with \uxxxx escapes for other characters
loc	locale to check for printable characters

Definition at line 272 of file unicode.cpp.

◆ utf16_native_endian_to_utf8() [1/2]

std::string utf16_native_endian_to_utf8 ( char16_t utf16_char )

Parameters

utf16_char UTF-16 character in architecture-native endianness encoding

Returns: UTF-8 encoding of the same codepoint

Definition at line 359 of file unicode.cpp.

◆ utf16_native_endian_to_utf8() [2/2]

std::string utf16_native_endian_to_utf8 ( const std::u16string & utf16_str )

Parameters

utf16_str UTF-16 string in architecture-native endianness encoding

Returns: UTF-8 encoding of the string

Definition at line 364 of file unicode.cpp.

◆ utf32_native_endian_to_utf8()

std::string utf32_native_endian_to_utf8 ( const std::basic_string< char32_t > & s )

Parameters

s	UTF-32 encoded wide string

Returns: utf8-encoded string with the same unicode characters as the input.

Definition at line 136 of file unicode.cpp.

◆ utf8_append_code()

static void utf8_append_code	(	unsigned int	c,
		std::string &	result
	)

static

Appends a unicode character to a utf8-encoded string.

parameters: character to append, string to append to

Definition at line 110 of file unicode.cpp.

◆ utf8_to_utf16_native_endian()

std::wstring utf8_to_utf16_native_endian ( const std::string & in )

Convert UTF8-encoded string to UTF-16 with architecture-native endianness.

parameters: String in UTF-8 format

Returns: String in UTF-16 format. The encoding follows the endianness of the architecture iff swap_bytes is true.

Definition at line 191 of file unicode.cpp.

◆ utf8_to_utf32()

std::u32string utf8_to_utf32 ( const std::string & utf8_str )

Convert UTF8-encoded string to UTF-32 with architecture-native endianness.

parameters: String in UTF-8 format

Returns: String in UTF-32 format.

Definition at line 205 of file unicode.cpp.

◆ widen() [1/2]

std::wstring widen ( const char * s )

Definition at line 49 of file unicode.cpp.

◆ widen() [2/2]

std::wstring widen ( const std::string & s )

Definition at line 87 of file unicode.cpp.

Functions

Function Documentation

◆ codepoint_hex_to_utf16_native_endian()

◆ codepoint_hex_to_utf8()

◆ narrow() [1/2]

◆ narrow() [2/2]

◆ narrow_argv()

◆ utf16_append_code()

◆ utf16_native_endian_to_java() [1/2]

◆ utf16_native_endian_to_java() [2/2]

◆ utf16_native_endian_to_java_string() [1/2]

◆ utf16_native_endian_to_java_string() [2/2]

◆ utf16_native_endian_to_utf8() [1/2]

◆ utf16_native_endian_to_utf8() [2/2]

◆ utf32_native_endian_to_utf8()

◆ utf8_append_code()

◆ utf8_to_utf16_native_endian()

◆ utf8_to_utf32()

◆ widen() [1/2]

◆ widen() [2/2]