Have std/json reject invalid UTF-8

This commit is contained in:
Nigel Tao committed 2020-02-26 16:26:16 +11:00
1 parent 2e62fa369a
commit 550ef90114
3 files changed
+220 -32

No files matched your search

+87 -12
View File
@@ -48,7 +48,7 @@ pub func decoder.decode_tokens?(dst: base.token_writer, src: base.io_reader) {
var token_value : base.u64[..= 0xFF_FFFF]
var number_length : base.u32[..= 0x3FF]
var number_status : base.u32[..= 0x3]
var string_length : base.u32[..= 0xFFFE]
var string_length : base.u32[..= 0xFFFB]
var whitespace_length : base.u32[..= 0xFFFE]
var depth : base.u32[..= 1024]
var stack_byte : base.u32[..= (1024 / 32) - 1]
@@ -58,6 +58,7 @@ pub func decoder.decode_tokens?(dst: base.token_writer, src: base.io_reader) {
var backslash : base.u8
var char : base.u8
var class : base.u8[..= 0x0F]
var multi_byte_utf8 : base.u32
var uni4_ok : base.u8
var uni4_string : base.u64
@@ -208,10 +209,10 @@ pub func decoder.decode_tokens?(dst: base.token_writer, src: base.io_reader) {
if char == 0x00 { // Non-special ASCII.
args.src.skip32_fast!(actual: 1, worst_case: 1)
if string_length >= 0xFFFE {
if string_length >= 0xFFFB {
args.dst.write_fast_token!(
value: 0x20_0221,
length: 0xFFFF)
length: 0xFFFC)
string_length = 0
continue.string_loop
}
@@ -381,18 +382,92 @@ pub func decoder.decode_tokens?(dst: base.token_writer, src: base.io_reader) {
return "#bad backslash-escape"
} else if char <= 0x10 {
// TODO: reject invalid UTF-8.
args.src.skip32_fast!(actual: 1, worst_case: 1)
if string_length >= 0xFFFE {
args.dst.write_fast_token!(
value: 0x20_0221,
length: 0xFFFF)
} else if char == 0x03 { // 2-byte UTF-8.
if args.src.available() < 2 {
if args.src.is_closed() {
return "#bad UTF-8"
}
yield? base."$short read"
string_length = 0
char = 0
continue.string_loop
}
string_length += 1
continue
multi_byte_utf8 = args.src.peek_u16le_as_u32()
if (multi_byte_utf8 & 0xC000) == 0x8000 {
multi_byte_utf8 = (0x00_07C0 & (multi_byte_utf8 ~mod<< 6)) |
(0x00_003F & (multi_byte_utf8 >> 8))
args.src.skip32_fast!(actual: 2, worst_case: 2)
if string_length >= 0xFFF8 {
args.dst.write_fast_token!(
value: 0x20_0221,
length: (string_length + 2) as base.u64)
string_length = 0
continue.string_loop
}
string_length += 2
continue
}
} else if char == 0x04 { // 3-byte UTF-8.
if args.src.available() < 3 {
if args.src.is_closed() {
return "#bad UTF-8"
}
yield? base."$short read"
string_length = 0
char = 0
continue.string_loop
}
multi_byte_utf8 = args.src.peek_u24le_as_u32()
if (multi_byte_utf8 & 0xC0_C000) == 0x80_8000 {
multi_byte_utf8 = (0x00_F000 & (multi_byte_utf8 ~mod<< 12)) |
(0x00_0FC0 & (multi_byte_utf8 >> 2)) |
(0x00_003F & (multi_byte_utf8 >> 16))
if (0x07FF < multi_byte_utf8) and
((multi_byte_utf8 < 0xD800) or (0xDFFF < multi_byte_utf8)) {
args.src.skip32_fast!(actual: 3, worst_case: 3)
if string_length >= 0xFFF8 {
args.dst.write_fast_token!(
value: 0x20_0221,
length: (string_length + 3) as base.u64)
string_length = 0
continue.string_loop
}
string_length += 3
continue
}
}
} else if char == 0x05 { // 4-byte UTF-8.
if args.src.available() < 4 {
if args.src.is_closed() {
return "#bad UTF-8"
}
yield? base."$short read"
string_length = 0
char = 0
continue.string_loop
}
multi_byte_utf8 = args.src.peek_u32le()
if (multi_byte_utf8 & 0xC0C0_C000) == 0x8080_8000 {
multi_byte_utf8 = (0x1C_0000 & (multi_byte_utf8 ~mod<< 18)) |
(0x03_F000 & (multi_byte_utf8 ~mod<< 4)) |
(0x00_0FC0 & (multi_byte_utf8 >> 10)) |
(0x00_003F & (multi_byte_utf8 >> 24))
if (0xFFFF < multi_byte_utf8) and (multi_byte_utf8 <= 0x10_FFFF) {
args.src.skip32_fast!(actual: 4, worst_case: 4)
if string_length >= 0xFFF8 {
args.dst.write_fast_token!(
value: 0x20_0221,
length: (string_length + 4) as base.u64)
string_length = 0
continue.string_loop
}
string_length += 4
continue
}
}
}
if string_length > 0 {