1
0
mirror of https://github.com/vlang/v.git synced 2023-08-10 21:13:21 +03:00
v/vlib/encoding/utf8/utf8.v

89 lines
1.8 KiB
V
Raw Normal View History

module utf8
2020-05-26 13:50:37 +03:00
struct Utf8State {
2020-05-26 13:50:37 +03:00
mut:
index int
subindex int
2020-05-26 13:50:37 +03:00
failed bool
}
2020-05-26 13:50:37 +03:00
pub fn validate_str(str string) bool {
return validate(str.str, str.len)
}
pub fn validate(data byteptr, len int) bool {
mut state := Utf8State{}
for i := 0; i < len; i++ {
s := unsafe {data[i]}
2020-05-26 13:50:37 +03:00
if s == 0 {
break
}
state.next_state(s)
if state.failed {
return false
}
}
return !state.failed && state.subindex <= 0
}
fn (mut s Utf8State) seq(r0 bool, r1 bool, is_tail bool) bool {
if s.subindex == 0 || (s.index > 1 && s.subindex == 1) || (s.index >= 6 && s.subindex == 2) {
if (s.subindex == 0 && r0) || (s.subindex == 1 && r1) || (s.subindex == 2 && is_tail) {
s.subindex++
return true
}
2020-05-26 13:50:37 +03:00
} else {
s.failed = true
if is_tail {
s.index = 0
s.subindex = 0
s.failed = false
}
return true
}
s.index++
s.subindex = 0
return false
}
2020-05-26 13:50:37 +03:00
fn (mut s Utf8State) next_state(c byte) {
// sequence 1
if s.index == 0 {
if (c >= 0x00 + 1 && c <= 0x7F) || c == 0x00 {
return
}
s.index++
s.subindex = 0
}
is_tail := c >= 0x80 && c <= 0xBF
2020-05-26 13:50:37 +03:00
// sequence 2
if s.index == 1 && s.seq(c >= 0xC2 && c <= 0xDF, false, is_tail) {
return
}
// sequence 3
if s.index == 2 && s.seq(c == 0xE0, c >= 0xA0 && c <= 0xBF, is_tail) {
return
}
if s.index == 3 && s.seq(c >= 0xE1 && c <= 0xEC, c >= 0x80 && c <= 0xBF, is_tail) {
return
}
if s.index == 4 && s.seq(c == 0xED, c >= 0x80 && c <= 0x9F, is_tail) {
return
}
if s.index == 5 && s.seq(c >= 0xEE && c <= 0xEF, c >= 0x80 && c <= 0xBF, is_tail) {
return
}
// sequence 4
if s.index == 6 && s.seq(c == 0xF0, c >= 0x90 && c <= 0xBF, is_tail) {
return
}
if s.index == 7 && s.seq(c >= 0xF1 && c <= 0xF3, c >= 0x80 && c <= 0xBF, is_tail) {
return
}
if s.index == 8 && s.seq(c == 0xF4, c >= 0x80 && c <= 0x8F, is_tail) {
return
}
// we should never reach here
s.failed = true
2020-05-26 13:50:37 +03:00
}