gvsafronov/rapidjson
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Improve surrogate handling

Browse Source 
Report a single low surrogate as kParseErrorStringUnicodeSurrogateInvalid.
This commit is contained in:
Lars Klein 2020-07-04 14:34:09 +02:00
parent 88bd956d66
commit ed73d7bdb4
1 changed files with 16 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
include/rapidjson/reader.h
View File

@ -1023,15 +1023,23 @@ private:
is.Take(); is.Take();
unsigned codepoint = ParseHex4(is, escapeOffset); unsigned codepoint = ParseHex4(is, escapeOffset);
RAPIDJSON_PARSE_ERROR_EARLY_RETURN_VOID; RAPIDJSON_PARSE_ERROR_EARLY_RETURN_VOID;
if (RAPIDJSON_UNLIKELY(codepoint >= 0xD800 && codepoint <= 0xDBFF)) { if (RAPIDJSON_UNLIKELY(codepoint >= 0xD800 && codepoint <= 0xDFFF)) {
// Handle UTF-16 surrogate pair // high surrogate, check if followed by valid low surrogate
if (RAPIDJSON_UNLIKELY(!Consume(is, '\\') || !Consume(is, 'u'))) if (RAPIDJSON_LIKELY(codepoint <= 0xDBFF)) {
// Handle UTF-16 surrogate pair
if (RAPIDJSON_UNLIKELY(!Consume(is, '\\') || !Consume(is, 'u')))
RAPIDJSON_PARSE_ERROR(kParseErrorStringUnicodeSurrogateInvalid, escapeOffset);
unsigned codepoint2 = ParseHex4(is, escapeOffset);
RAPIDJSON_PARSE_ERROR_EARLY_RETURN_VOID;
if (RAPIDJSON_UNLIKELY(codepoint2 < 0xDC00 || codepoint2 > 0xDFFF))
RAPIDJSON_PARSE_ERROR(kParseErrorStringUnicodeSurrogateInvalid, escapeOffset);
codepoint = (((codepoint - 0xD800) << 10) | (codepoint2 - 0xDC00)) + 0x10000;
}
// single low surrogate
else
{
RAPIDJSON_PARSE_ERROR(kParseErrorStringUnicodeSurrogateInvalid, escapeOffset); RAPIDJSON_PARSE_ERROR(kParseErrorStringUnicodeSurrogateInvalid, escapeOffset);
unsigned codepoint2 = ParseHex4(is, escapeOffset); }
RAPIDJSON_PARSE_ERROR_EARLY_RETURN_VOID;
if (RAPIDJSON_UNLIKELY(codepoint2 < 0xDC00 || codepoint2 > 0xDFFF))
RAPIDJSON_PARSE_ERROR(kParseErrorStringUnicodeSurrogateInvalid, escapeOffset);
codepoint = (((codepoint - 0xD800) << 10) | (codepoint2 - 0xDC00)) + 0x10000;
} }
TEncoding::Encode(os, codepoint); TEncoding::Encode(os, codepoint);
} }