fix #2550 json encoding should use surrogate pairs per RFC7159

This commit is contained in:
Laytan Laats
2023-05-22 17:22:33 +02:00
parent 118ab60588
commit 5d54b710e7
5 changed files with 54 additions and 13 deletions
+10 -1
View File
@@ -2,6 +2,7 @@ package json
import "core:mem"
import "core:unicode/utf8"
import "core:unicode/utf16"
import "core:strconv"
Parser :: struct {
@@ -403,11 +404,19 @@ unquote_string :: proc(token: Token, spec: Specification, allocator := context.a
}
i += 6
// If this is a surrogate pair, decode as such by taking the next rune too.
if r >= utf8.SURROGATE_MIN && r <= utf8.SURROGATE_HIGH_MAX && len(s) > i + 2 && s[i:i+2] == "\\u" {
r2 := get_u4_rune(s[i:])
if r2 >= utf8.SURROGATE_LOW_MIN && r2 <= utf8.SURROGATE_MAX {
i += 6
r = utf16.decode_surrogate_pair(r, r2)
}
}
buf, buf_width := utf8.encode_rune(r)
copy(b[w:], buf[:buf_width])
w += buf_width
case '0':
if spec != .JSON {
b[w] = '\x00'