Reject files with invalid utf-8.

It's easier to only support files with valid utf-8.

1 files changed, 18 insertions, 6 deletions

diff --git a/src/term.rs b/src/term.rs
index e390125..831cd43 100644
--- a/src/term.rs
+++ b/src/term.rs
@@ -1,6 +1,6 @@
 use libc::{termios, winsize};
 use std::error::Error;
-use std::io::{stdout, BufRead, BufReader, Stdin, Write};
+use std::io::{stdout, BufRead, BufReader, Read, Stdin, Write};
 
 const ESC: u8 = 27;
 
@@ -83,7 +83,7 @@ pub enum Key {
     End,
     PgDn,
     PgUp,
-    Byte(u8),
+    Char(char),
 }
 
 pub struct Reader {
@@ -110,9 +110,21 @@ impl Reader {
                 return Ok(Key::Timeout);
             }
             if buf[0] != ESC {
-                let b = buf[0];
-                self.stdin.consume(1);
-                return Ok(Key::Byte(b));
+                let mut n = usize::try_from(buf[0].leading_ones())
+                    .expect("buf[0] is a u8 so it can only have 8 leading ones");
+                if n == 0 {
+                    n = 1;
+                }
+                let mut char = [0; 4];
+                self.stdin.read_exact(&mut char[..n])?;
+                let Ok(s) = std::str::from_utf8(&char) else {
+                    continue;
+                };
+                return Ok(Key::Char(
+                    s.chars()
+                        .nth(0)
+                        .expect("s should be exactly one codepoint long"),
+                ));
             }
             if buf.len() < 3 || buf[1] != b'[' {
                 // Unknown escape sequence, eat the escape and one more byte.
@@ -173,7 +185,7 @@ impl Reader {
                 }
                 "3~" => {
                     // Backspace
-                    return Ok(Key::Byte(8));
+                    return Ok(Key::Char(char::from(b'H' - b'@')));
                 }
                 _ => {
                     continue;