Handle wcwidth() and mbtowc() failures in better style and drop

characters where we can't find the width (wcwidth() fails) on input, the
same as we drop invalid UTF-8. Suggested by schwarze@.
This commit is contained in:
nicm
2016-03-02 15:36:02 +00:00
parent d980d965dd
commit b8a102d26f
5 changed files with 48 additions and 21 deletions

10
input.c
View File

@@ -1960,8 +1960,14 @@ input_utf8_close(struct input_ctx *ictx)
{
struct utf8_data *ud = &ictx->utf8data;
if (utf8_append(ud, ictx->ch) != UTF8_DONE)
fatalx("UTF-8 close invalid %#x", ictx->ch);
if (utf8_append(ud, ictx->ch) != UTF8_DONE) {
/*
* An error here could be invalid UTF-8 or it could be a
* nonprintable character for which we can't get the
* width. Drop it.
*/
return (0);
}
log_debug("%s %hhu '%*s' (width %hhu)", __func__, ud->size,
(int)ud->size, ud->data, ud->width);