feedc0de/ArduinoJson
1
0
Fork 0
forked from bblanchon/ArduinoJson
Code Pull Requests Activity

Improved decoding of UTF-16 surrogate pairs (closes #1157)

Browse Source
This commit is contained in:
Kay Sievers
2020-01-02 13:14:02 +01:00
committed by Benoit Blanchon
parent 8550418875
commit 91b808381e
5 changed files with 36 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
src/ArduinoJson/Json/JsonDeserializer.hpp
View File

@ -189,6 +189,7 @@ class JsonDeserializer {
DeserializationError parseQuotedString(const char *&result) {
StringBuilder builder = _stringStorage.startString();
uint16_t surrogate1 = 0;
const char stopChar = current();
move();
@ -208,7 +209,19 @@ class JsonDeserializer {
move();
DeserializationError err = parseCodepoint(codepoint);
if (err) return err;
Utf8::encodeCodepoint(codepoint, builder);
if (codepoint >= 0xd800 && codepoint <= 0xdbff) {
if (surrogate1 > 0) return DeserializationError::InvalidInput;
surrogate1 = codepoint;
} else if (codepoint >= 0xdc00 && codepoint <= 0xdfff) {
if (surrogate1 == 0) return DeserializationError::InvalidInput;
uint32_t codepoint32 = 0x10000;
codepoint32 += static_cast<uint32_t>(surrogate1 - 0xd800) << 10;
codepoint32 += codepoint - 0xdc00;
Utf8::encodeCodepoint(codepoint32, builder);
surrogate1 = 0;
} else {
Utf8::encodeCodepoint(codepoint, builder);
}
continue;
#else
return DeserializationError::NotSupported;
@ -220,6 +233,8 @@ class JsonDeserializer {
move();
}
if (surrogate1 > 0) return DeserializationError::InvalidInput;
builder.append(c);
}

12
src/ArduinoJson/Json/Utf8.hpp
View File

@ -10,17 +10,21 @@ namespace ARDUINOJSON_NAMESPACE {
namespace Utf8 {
template <typename TStringBuilder>
inline void encodeCodepoint(uint16_t codepoint, TStringBuilder &str) {
inline void encodeCodepoint(uint32_t codepoint, TStringBuilder &str) {
if (codepoint < 0x80) {
str.append(char(codepoint));
return;
}
if (codepoint >= 0x00000800) {
if (codepoint < 0x00000800) {
str.append(char(0xc0 /*0b11000000*/ | (codepoint >> 6)));
} else if (codepoint < 0x00010000) {
str.append(char(0xe0 /*0b11100000*/ | (codepoint >> 12)));
str.append(char(((codepoint >> 6) & 0x3f /*0b00111111*/) | 0x80));
} else {
str.append(char(0xc0 /*0b11000000*/ | (codepoint >> 6)));
} else if (codepoint < 0x00110000) {
str.append(char(0xf0 /*0b11110000*/ | (codepoint >> 18)));
str.append(char(((codepoint >> 12) & 0x3f /*0b00111111*/) | 0x80));
str.append(char(((codepoint >> 6) & 0x3f /*0b00111111*/) | 0x80));
}
str.append(char((codepoint & 0x3f /*0b00111111*/) | 0x80));
}