Better handling of incomplete UTF-8 sequences - just emit a UTF8_INVALID if waittime is up
This commit is contained in:
parent
76147e92d4
commit
a2f3d807f8
15
termkey.c
15
termkey.c
|
@ -440,8 +440,19 @@ static termkey_result getkey_simple(termkey_t *tk, termkey_key *key, int force)
|
|||
return TERMKEY_RES_KEY;
|
||||
}
|
||||
|
||||
if(tk->buffcount < nbytes)
|
||||
return tk->waittime ? TERMKEY_RES_AGAIN : TERMKEY_RES_NONE;
|
||||
if(tk->buffcount < nbytes) {
|
||||
if(!force)
|
||||
return TERMKEY_RES_AGAIN;
|
||||
|
||||
/* There weren't enough bytes for a complete UTF-8 sequence but caller
|
||||
* demands an answer. About the best thing we can do here is eat as many
|
||||
* bytes as we have, and emit a UTF8_INVALID. If the remaining bytes
|
||||
* arrive later, they'll be invalid too.
|
||||
*/
|
||||
(*tk->method.emit_codepoint)(tk, UTF8_INVALID, key);
|
||||
(*tk->method.eat_bytes)(tk, tk->buffcount);
|
||||
return TERMKEY_RES_KEY;
|
||||
}
|
||||
|
||||
for(int b = 1; b < nbytes; b++) {
|
||||
unsigned char cb = CHARAT(b);
|
||||
|
|
Loading…
Reference in New Issue