1
0
mirror of https://github.com/nlohmann/json.git synced 2025-07-28 12:02:00 +03:00

Merge branch 'develop' into feature/sax2

This commit is contained in:
Niels Lohmann
2018-05-06 13:29:23 +02:00
52 changed files with 582 additions and 418 deletions

View File

@ -5,6 +5,7 @@
|_____|_____|_____|_|___| https://github.com/nlohmann/json
Licensed under the MIT License <http://opensource.org/licenses/MIT>.
SPDX-License-Identifier: MIT
Copyright (c) 2013-2018 Niels Lohmann <http://nlohmann.me>.
Permission is hereby granted, free of charge, to any person obtaining a copy
@ -1688,6 +1689,176 @@ class input_buffer_adapter : public input_adapter_protocol
const char* const limit;
};
template<typename WideStringType>
class wide_string_input_adapter : public input_adapter_protocol
{
public:
wide_string_input_adapter(const WideStringType& w) : str(w) {}
std::char_traits<char>::int_type get_character() noexcept override
{
// unget_character() was called previously: return the last character
if (next_unget)
{
next_unget = false;
return last_char;
}
// check if buffer needs to be filled
if (utf8_bytes_index == utf8_bytes_filled)
{
if (sizeof(typename WideStringType::value_type) == 2)
{
fill_buffer_utf16();
}
else
{
fill_buffer_utf32();
}
assert(utf8_bytes_filled > 0);
assert(utf8_bytes_index == 0);
}
// use buffer
assert(utf8_bytes_filled > 0);
assert(utf8_bytes_index < utf8_bytes_filled);
return (last_char = utf8_bytes[utf8_bytes_index++]);
}
void unget_character() noexcept override
{
next_unget = true;
}
private:
void fill_buffer_utf16()
{
utf8_bytes_index = 0;
if (current_wchar == str.size())
{
utf8_bytes[0] = std::char_traits<char>::eof();
utf8_bytes_filled = 1;
}
else
{
// get the current character
const int wc = static_cast<int>(str[current_wchar++]);
// UTF-16 to UTF-8 encoding
if (wc < 0x80)
{
utf8_bytes[0] = wc;
utf8_bytes_filled = 1;
}
else if (wc <= 0x7FF)
{
utf8_bytes[0] = 0xC0 | ((wc >> 6));
utf8_bytes[1] = 0x80 | (wc & 0x3F);
utf8_bytes_filled = 2;
}
else if (0xD800 > wc or wc >= 0xE000)
{
utf8_bytes[0] = 0xE0 | ((wc >> 12));
utf8_bytes[1] = 0x80 | ((wc >> 6) & 0x3F);
utf8_bytes[2] = 0x80 | (wc & 0x3F);
utf8_bytes_filled = 3;
}
else
{
if (current_wchar < str.size())
{
const int wc2 = static_cast<int>(str[current_wchar++]);
const int charcode = 0x10000 + (((wc & 0x3FF) << 10) | (wc2 & 0x3FF));
utf8_bytes[0] = 0xf0 | (charcode >> 18);
utf8_bytes[1] = 0x80 | ((charcode >> 12) & 0x3F);
utf8_bytes[2] = 0x80 | ((charcode >> 6) & 0x3F);
utf8_bytes[3] = 0x80 | (charcode & 0x3F);
utf8_bytes_filled = 4;
}
else
{
// unknown character
++current_wchar;
utf8_bytes[0] = wc;
utf8_bytes_filled = 1;
}
}
}
}
void fill_buffer_utf32()
{
utf8_bytes_index = 0;
if (current_wchar == str.size())
{
utf8_bytes[0] = std::char_traits<char>::eof();
utf8_bytes_filled = 1;
}
else
{
// get the current character
const int wc = static_cast<int>(str[current_wchar++]);
// UTF-32 to UTF-8 encoding
if (wc < 0x80)
{
utf8_bytes[0] = wc;
utf8_bytes_filled = 1;
}
else if (wc <= 0x7FF)
{
utf8_bytes[0] = 0xC0 | ((wc >> 6) & 0x1F);
utf8_bytes[1] = 0x80 | (wc & 0x3F);
utf8_bytes_filled = 2;
}
else if (wc <= 0xFFFF)
{
utf8_bytes[0] = 0xE0 | ((wc >> 12) & 0x0F);
utf8_bytes[1] = 0x80 | ((wc >> 6) & 0x3F);
utf8_bytes[2] = 0x80 | (wc & 0x3F);
utf8_bytes_filled = 3;
}
else if (wc <= 0x10FFFF)
{
utf8_bytes[0] = 0xF0 | ((wc >> 18 ) & 0x07);
utf8_bytes[1] = 0x80 | ((wc >> 12) & 0x3F);
utf8_bytes[2] = 0x80 | ((wc >> 6) & 0x3F);
utf8_bytes[3] = 0x80 | (wc & 0x3F);
utf8_bytes_filled = 4;
}
else
{
// unknown character
utf8_bytes[0] = wc;
utf8_bytes_filled = 1;
}
}
}
private:
/// the wstring to process
const WideStringType& str;
/// index of the current wchar in str
std::size_t current_wchar = 0;
/// a buffer for UTF-8 bytes
std::array<std::char_traits<char>::int_type, 4> utf8_bytes = {{0, 0, 0, 0}};
/// index to the utf8_codes array for the next valid byte
std::size_t utf8_bytes_index = 0;
/// number of valid bytes in the utf8_codes array
std::size_t utf8_bytes_filled = 0;
/// the last character (returned after unget_character() is called)
std::char_traits<char>::int_type last_char = 0;
/// whether get_character() should return last_char
bool next_unget = false;
};
class input_adapter
{
public:
@ -1701,6 +1872,15 @@ class input_adapter
input_adapter(std::istream&& i)
: ia(std::make_shared<input_stream_adapter>(i)) {}
input_adapter(const std::wstring& ws)
: ia(std::make_shared<wide_string_input_adapter<std::wstring>>(ws)) {}
input_adapter(const std::u16string& ws)
: ia(std::make_shared<wide_string_input_adapter<std::u16string>>(ws)) {}
input_adapter(const std::u32string& ws)
: ia(std::make_shared<wide_string_input_adapter<std::u32string>>(ws)) {}
/// input adapter for buffer
template<typename CharT,
typename std::enable_if<
@ -13934,7 +14114,7 @@ class basic_json
@return copy of the element at key @a key or @a default_value if @a key
is not found
@throw type_error.306 if the JSON value is not an objec; in that case,
@throw type_error.306 if the JSON value is not an object; in that case,
using `value()` with a key makes no sense.
@complexity Logarithmic in the size of the container.