tomasz.sowa/pikotools
1
0
Fork 0
Code Issues 1 Pull Requests Releases Activity

reorganization in utf8

Browse Source 
- utf8 auxiliary functions moved to utf8_private.h file
- in utf8.h are shown only functions available for consumers
- template functions has been moved to utf8_template.h (in utf8.h are only declarations)
  utf8_template.h is included at the end of utf8.h
- functions which take std::ostream changed to template (the stream is a template argument now)
This commit is contained in:
Tomasz Sowa
2021-03-15 19:34:51 +01:00
parent effe9be0a3
commit fac3a7eb71
5 changed files with 845 additions and 606 deletions
Download Patch File Download Diff File Expand all files Collapse all files

268
utf8/utf8.h
View File

@@ -5,7 +5,7 @@
*/
/*
* Copyright (c) 2010-2018, Tomasz Sowa
* Copyright (c) 2010-2021, Tomasz Sowa
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
@@ -38,7 +38,6 @@
#ifndef headerfile_picotools_utf8_utf8
#define headerfile_picotools_utf8_utf8
#include <fstream>
#include <string>
#include "textstream/textstream.h"
@@ -46,8 +45,6 @@
namespace PT
{
/*!
UTF-8, a transformation format of ISO 10646
http://tools.ietf.org/html/rfc3629
@@ -68,6 +65,16 @@ bool UTF8_CheckRange(int c);
/*
*
*
*
* convertions from UTF-8
*
*
*
*/
/*!
converting one character from UTF-8 to an int
*/
@@ -85,25 +92,40 @@ bool UTF8ToWide(const char * utf8, std::wstring & res, bool cle
bool UTF8ToWide(const std::string & utf8, std::wstring & res, bool clear = true, int mode = 1);
bool UTF8ToWide(std::istream & utf8, std::wstring & res, bool clear = true, int mode = 1);
template<size_t stack_size, size_t heap_block_size>
bool UTF8ToWide(const char * utf8, size_t utf8_len, TextStreamBase<wchar_t, stack_size, heap_block_size> & res, bool clear = true, int mode = 1); // need to be tested
template<size_t stack_size, size_t heap_block_size>
bool UTF8ToWide(const char * utf8, TextStreamBase<wchar_t, stack_size, heap_block_size> & res, bool clear = true, int mode = 1); // need to be tested
template<size_t stack_size, size_t heap_block_size>
bool UTF8ToWide(const std::string & utf8, TextStreamBase<wchar_t, stack_size, heap_block_size> & res, bool clear = true, int mode = 1); // need to be tested
template<size_t stack_size, size_t heap_block_size>
bool UTF8ToWide(std::istream & utf8, TextStreamBase<wchar_t, stack_size, heap_block_size> & res, bool clear = true, int mode = 1); // need to be tested
/*!
converting UTF-8 string to a WTextStream stream
(need to be tested)
*/
/*
implemented as templates below
bool UTF8ToWide(const char * utf8, size_t utf8_len, WTextStream & res, bool clear = true, int mode = 1);
bool UTF8ToWide(const char * utf8, WTextStream & res, bool clear = true, int mode = 1);
bool UTF8ToWide(const std::string & utf8, WTextStream & res, bool clear = true, int mode = 1);
bool UTF8ToWide(std::istream & utf8, WTextStream & res, bool clear = true, int mode = 1);
*/
*
*
*
* convertions to UTF-8
*
*
*
*/
/*!
converting one int character to UTF-8
*/
size_t IntToUTF8(int z, char * utf8, size_t utf8_max_len);
size_t IntToUTF8(int z, std::string & utf8, bool clear = true );
size_t IntToUTF8(int z, std::ostream & utf8);
size_t IntToUTF8(int z, std::string & utf8, bool clear = true);
template<typename StreamType>
size_t IntToUTF8(int z, StreamType & utf8);
/*!
@@ -113,216 +135,32 @@ bool WideToUTF8(const wchar_t * wide_string, size_t string_len, std::string & ut
bool WideToUTF8(const wchar_t * wide_string, std::string & utf8, bool clear = true, int mode = 1);
bool WideToUTF8(const std::wstring & wide_string, std::string & utf8, bool clear = true, int mode = 1);
// implemented as a template below
//void WideToUTF8(PT::WTextStream & buffer, std::string & utf8, bool clear = true, int mode = 1);// not tested
template<typename StreamType>
bool WideToUTF8(const wchar_t * wide_string, size_t string_len, StreamType & utf8, int mode = 1);
template<typename StreamType>
bool WideToUTF8(const wchar_t * wide_string, StreamType & utf8, int mode = 1);
template<typename StreamType>
bool WideToUTF8(const std::wstring & wide_string, StreamType & utf8, int mode = 1);
bool WideToUTF8(const wchar_t * wide_string, size_t string_len, std::ostream & utf8, int mode = 1);
bool WideToUTF8(const wchar_t * wide_string, std::ostream & utf8, int mode = 1);
bool WideToUTF8(const std::wstring & wide_string, std::ostream & utf8, int mode = 1);
// implemented as a template below
//void WideToUTF8(PT::WTextStream & buffer, std::ostream & utf8, int mode = 1);// not tested
bool WideToUTF8(const wchar_t * wide_string, size_t string_len, char * utf8, size_t utf8_len, size_t & utf8_written, int mode = 1);
bool WideToUTF8(const wchar_t * wide_string, char * utf8, size_t utf8_len, size_t & utf8_written, int mode = 1);
bool WideToUTF8(const std::wstring & wide_string, char * utf8, size_t utf8_len, size_t & utf8_written, int mode = 1);
// implement void WideToUTF8(PT::WTextStream & buffer, char * utf8, size_t utf8_len, size_t & utf8_written, int mode = 1);
// implement template<typename StreamType>
bool WideToUTF8(const wchar_t * wide_string, size_t string_len, char * utf8, size_t utf8_len, int mode = 1);
bool WideToUTF8(const wchar_t * wide_string, char * utf8, size_t utf8_len, int mode = 1);
bool WideToUTF8(const std::wstring & wide_string, char * utf8, size_t utf8_len, int mode = 1);
// implement void WideToUTF8(PT::WTextStream & buffer, char * utf8, size_t utf8_len, int mode = 1);
namespace private_namespace
{
template<typename function_type>
bool UTF8ToWideGeneric(const char * utf8, size_t utf8_len, int mode, function_type convert_function)
{
int z;
size_t len;
bool correct, was_error = false;
while( utf8_len > 0 )
{
if( (unsigned char)*utf8 <= 0x7f )
{
// small optimization
len = 1;
correct = true;
z = static_cast<unsigned char>(*utf8);
}
else
{
len = UTF8ToInt(utf8, utf8_len, z, correct); // the len will be different from zero
}
if( !correct )
{
if( mode == 1 )
convert_function(0xFFFD); // U+FFFD "replacement character"
was_error = true;
}
else
{
convert_function(z);
}
utf8 += len;
utf8_len -= len;
}
return !was_error;
}
template<typename char_type, size_t stack_size, size_t heap_block_size>
void IntToWide(int c, TextStreamBase<char_type, stack_size, heap_block_size> & res)
{
if( sizeof(wchar_t)==2 && c>0xffff )
{
// UTF16 surrogate pairs
c -= 0x10000;
res << static_cast<wchar_t>(((c >> 10) & 0x3FF) + 0xD800);
res << static_cast<wchar_t>((c & 0x3FF) + 0xDC00);
}
else
{
res << static_cast<wchar_t>(c);
}
}
// not tested
// FIX ME it is not using surrogate pairs from input stream
// and mode parameter
template<typename char_type, size_t stack_size, size_t heap_block_size, typename function_type>
void WideToUTF8Generic(TextStreamBase<char_type, stack_size, heap_block_size> & buffer, int mode, function_type write_function)
{
char utf8_buffer[256];
std::size_t buffer_len = sizeof(utf8_buffer) / sizeof(char);
std::size_t utf8_sequence_max_length = 10;
std::size_t index = 0;
typename TextStreamBase<char_type, stack_size, heap_block_size>::const_iterator i = buffer.begin();
while( i != buffer.end() )
{
if( index + utf8_sequence_max_length > buffer_len )
{
write_function(utf8_buffer, index);
index = 0;
}
index += PT::IntToUTF8(*i, utf8_buffer + index, buffer_len - index);
++i;
}
if( index > 0 )
{
write_function(utf8_buffer, index);
}
}
} // namespace
// need to be tested
template<typename char_type, size_t stack_size, size_t heap_block_size>
bool UTF8ToWide(const char * utf8, size_t utf8_len, TextStreamBase<char_type, stack_size, heap_block_size> & res, bool clear = true, int mode = 1)
{
if( clear )
res.clear();
bool status = private_namespace::UTF8ToWideGeneric(utf8, utf8_len, mode, [&res](int c) {
private_namespace::IntToWide(c, res);
});
return status;
}
// need to be tested
template<typename char_type, size_t stack_size, size_t heap_block_size>
bool UTF8ToWide(const char * utf8, TextStreamBase<char_type, stack_size, heap_block_size> & res, bool clear = true, int mode = 1)
{
size_t utf8_len = 0;
while( utf8[utf8_len] != 0 )
utf8_len += 1;
return UTF8ToWide(utf8, utf8_len, res, clear, mode);
}
// need to be tested
template<typename char_type, size_t stack_size, size_t heap_block_size>
bool UTF8ToWide(const std::string & utf8, TextStreamBase<char_type, stack_size, heap_block_size> & res, bool clear = true, int mode = 1)
{
return UTF8ToWide(utf8.c_str(), utf8.size(), res, clear, mode);
}
// need to be tested
template<typename char_type, size_t stack_size, size_t heap_block_size>
bool UTF8ToWide(std::istream & utf8, TextStreamBase<char_type, stack_size, heap_block_size> & res, bool clear = true, int mode = 1)
{
int z;
bool correct, was_error = false;
if( clear )
res.clear();
while( UTF8ToInt(utf8, z, correct) > 0 )
{
if( !correct )
{
if( mode == 1 )
res << 0xFFFD; // U+FFFD "replacement character"
was_error = true;
}
else
{
private_namespace::IntToWide(z, res);
}
}
return !was_error;
}
// not tested
template<typename char_type, size_t stack_size, size_t heap_block_size>
void WideToUTF8(TextStreamBase<char_type, stack_size, heap_block_size> & buffer, std::string & utf8, bool clear = true, int mode = 1)
{
if( clear )
utf8.clear();
private_namespace::WideToUTF8Generic(buffer, mode, [&utf8](const char * utf8_buffer, std::size_t buffer_len){
utf8.append(utf8_buffer, buffer_len);
});
}
// not tested
template<typename char_type, size_t stack_size, size_t heap_block_size>
void WideToUTF8(TextStreamBase<char_type, stack_size, heap_block_size> & buffer, std::ostream & utf8, int mode = 1)
{
private_namespace::WideToUTF8Generic(buffer, mode, [&utf8](const char * utf8_buffer, std::size_t buffer_len){
utf8.write(utf8_buffer, buffer_len);
});
}
// implement template<typename StreamType>
template<size_t stack_size, size_t heap_block_size>
void WideToUTF8(TextStreamBase<wchar_t, stack_size, heap_block_size> & buffer, std::string & utf8, bool clear = true, int mode = 1); // not tested
template<size_t stack_size, size_t heap_block_size>
void WideToUTF8(TextStreamBase<wchar_t, stack_size, heap_block_size> & buffer, std::ostream & utf8, int mode = 1); // not tested
@@ -330,5 +168,7 @@ void WideToUTF8(TextStreamBase<char_type, stack_size, heap_block_size> & buffer,
} // namespace
#include "utf8/utf8_templates.h"
#endif