shell: accept unicode input

userspace/shell/src/readline.rs

@@ -9,9 +9,43 @@ enum Outcome {
     Data(usize),
 }
 
-fn readline_inner(stdin: &mut RawStdin, stdout: &mut Stdout, buffer: &mut String) -> Result<Outcome, Error> {
+struct Utf8Decoder {
+    buffer: [u8; 4],
+    len: usize,
+}
+
+impl Utf8Decoder {
+    fn push(&mut self, byte: u8) -> Option<char> {
+        self.buffer[self.len] = byte;
+        self.len += 1;
+        if let Ok(str) = std::str::from_utf8(&self.buffer[..self.len]) {
+            let ch = str.chars().next().unwrap();
+            self.len = 0;
+            Some(ch)
+        } else {
+            if self.len == 4 {
+                // Got 4 bytes and could not decode a single character
+                self.len = 0;
+                Some('\u{25A1}')
+            } else {
+                None
+            }
+        }
+    }
+}
+
+fn readline_inner(
+    stdin: &mut RawStdin,
+    stdout: &mut Stdout,
+    buffer: &mut String,
+) -> Result<Outcome, Error> {
+    let mut ch_buffer = [0; 8];
     let mut pos = 0;
     let mut ch = [0];
+    let mut decoder = Utf8Decoder {
+        buffer: [0; 4],
+        len: 0,
+    };
 
     loop {
         let len = stdin.read(&mut ch)?;
@@ -20,10 +54,13 @@ fn readline_inner(stdin: &mut RawStdin, stdout: &mut Stdout, buffer: &mut String
         }
 
         let ch = ch[0];
+        let Some(ch) = decoder.push(ch) else {
+            continue;
+        };
 
         match ch {
             // ^D
-            0x04 => {
+            c if c as u32 == 0x04 => {
                 if pos == 0 {
                     break;
                 } else {
@@ -31,10 +68,10 @@ fn readline_inner(stdin: &mut RawStdin, stdout: &mut Stdout, buffer: &mut String
                 }
             }
             // ^C
-            0x03 => return Ok(Outcome::Interrupt),
+            c if c as u32 == 0x03 => return Ok(Outcome::Interrupt),
             // TODO completion
-            b'\t' => (),
+            '\t' => (),
-            0x7F => {
+            c if c as u32 == 0x7F => {
                 if pos != 0 {
                     stdout.write_all(b"\x1B[D \x1B[D").ok();
                     stdout.flush().ok();
@@ -42,18 +79,19 @@ fn readline_inner(stdin: &mut RawStdin, stdout: &mut Stdout, buffer: &mut String
                     pos -= 1;
                 }
             }
-            ch if ch.is_ascii_graphic() || ch.is_ascii_whitespace() => {
+            ch if ch.is_whitespace() || ch.is_alphanumeric() || ch.is_ascii_graphic() => {
-                stdout.write_all(&[ch]).ok();
+                let bytes = ch.encode_utf8(&mut ch_buffer);
+                stdout.write_all(bytes.as_bytes()).ok();
                 stdout.flush().ok();
 
-                buffer.push(ch as char);
+                buffer.push(ch);
                 pos += 1;
             }
             _ => (),
         }
 
-        if ch == b'\n' || ch == b'\r' {
+        if ch == '\n' || ch == '\r' {
-            if ch == b'\r' {
+            if ch == '\r' {
                 stdout.write_all(b"\n").ok();
                 stdout.flush().ok();
             }
@@ -68,7 +106,7 @@ pub fn readline<P: Fn(&mut Stdout)>(
     stdin: &mut Stdin,
     stdout: &mut Stdout,
     buffer: &mut String,
-    prompt: P
+    prompt: P,
 ) -> Result<usize, Error> {
     let mut stdin = RawStdin::new(stdin)?;
 
@@ -80,7 +118,7 @@ pub fn readline<P: Fn(&mut Stdout)>(
             Outcome::Data(n) => break Ok(n),
             Outcome::Interrupt => {
                 stdout.write_all(b"\r\n").ok();
-            },
+            }
         }
     }
 }