Sketch: Bytestring diagnostics preservation

src/encode/bytes.rs

@@ -75,10 +75,10 @@
 }
 
 fn negative_to_bytes(bytes: &mut Vec<u8>, value: u64, bitwidth: IntegerWidth) {
     integer_to_bytes(bytes, value, bitwidth, 1);
 }
 
-fn definite_bytestring_to_bytes(bytes: &mut Vec<u8>, ByteString { data, bitwidth }: &ByteString) {
+fn definite_bytestring_to_bytes(bytes: &mut Vec<u8>, ByteString { data, bitwidth, .. }: &ByteString) {
     integer_to_bytes(bytes, data.len() as u64, *bitwidth, 2);
     bytes.extend_from_slice(data);
 }

src/encode/diag.rs

@@ -193,7 +193,7 @@ impl<'a> Context<'a> {
     }
 
     fn definite_bytestring_to_diag(&mut self, bytestring: &ByteString) {
-        match self.encoding {
+        match self.encoding.override_with(bytestring.diag_encoding) {
             Encoding::Base64Url => {
                 self.output.push_str("b64'");
                 data_encoding::BASE64URL_NOPAD.encode_append(&bytestring.data, self.output);
@@ -342,7 +342,7 @@ impl<'a> Context<'a> {
                 }
             }
             Tag::ENCODED_CBOR_SEQ => {
-                if let DataItem::ByteString(ByteString { data, bitwidth }) = value {
+                if let DataItem::ByteString(ByteString { data, bitwidth, .. }) = value {
                     let mut data = data.as_slice();
                     let mut items = Vec::new();
                     while let Ok(Some((item, len))) = crate::parse_bytes_partial(data) {
@@ -368,6 +368,8 @@ impl<'a> Context<'a> {
                         self.item_to_diag(&DataItem::ByteString(ByteString {
                             data: data.into(),
                             bitwidth: *bitwidth,
+                            // For broken CBOR, hex is a practical choice
+                            diag_encoding: Some(Encoding::Base16),
                         }));
                     }
                 } else {

src/encode/hex.rs

@@ -268,10 +268,10 @@
         };
         Line::new(hex, comment)
     })
 }
 
 fn definite_bytestring_to_hex(encoding: Option<Encoding>, bytestring: &ByteString) -> Line {
-    let ByteString { ref data, bitwidth } = *bytestring;
+    let ByteString { ref data, bitwidth, .. } = *bytestring;
 
     let mut line = length_to_hex(Some(data.len()), Some(bitwidth), 2, "bytes");
 
@@ -1137,7 +1137,7 @@
     name: &str,
     convert: impl Fn([u8; LEN]) -> String,
 ) -> Vec<Line> {
-    if let DataItem::ByteString(ByteString { data, bitwidth }) = value {
+    if let DataItem::ByteString(ByteString { data, bitwidth, .. }) = value {
         if data.len() % LEN == 0 {
             let mut line = length_to_hex(Some(data.len()), Some(*bitwidth), 2, "bytes");
             // TODO: Use slice::array_chunks when stable

src/encode/mod.rs

@@ -2,10 +2,17 @@ mod bytes;
 mod diag;
 mod hex;
 
-#[derive(Copy, Clone)]
+#[derive(Copy, Clone, Debug, PartialEq, Eq)]
 #[allow(clippy::enum_variant_names)]
 pub(crate) enum Encoding {
     Base16,
     Base64,
     Base64Url,
 }
+
+impl Encoding {
+    /// Return overrider if given, otherwise return self
+    pub(crate) fn override_with(self, overrider: Option<Self>) -> Self {
+        overrider.unwrap_or(self)
+    }
+}

src/parse/binary.rs

@@ -88,7 +88,7 @@ fn definite_bytestring(input: &[u8]) -> IResult<&[u8], ByteString> {
         .map_err(|_| Err::Error(make_error(input, ErrorKind::LengthValue)))?;
     let (input, data) = take_bytes(length)(input)?;
     let data = data.to_owned();
-    Ok((input, ByteString { data, bitwidth }))
+    Ok((input, ByteString::new(data).with_bitwidth(bitwidth)))
 }
 
 fn indefinite_bytestring(input: &[u8]) -> IResult<&[u8], DataItem> {

src/parse/diag.rs

@@ -3,6 +3,8 @@
 use std::f64;
 use std::str::FromStr;
 
+use crate::encode::Encoding;
+
 use nom::{
     branch::alt,
     bytes::complete::{escaped_transform, tag},
@@ -222,46 +224,51 @@
     )(input)
 }
 
-fn definite_bytestring(input: &str) -> IResult<&str, Vec<u8>> {
+fn definite_bytestring(input: &str) -> IResult<&str, (Vec<u8>, Option<Encoding>)> {
     wrapws(alt((
         map_res(
             preceded(
                 tag("h"),
                 delimited(tag("'"), wrapws_strings(base16_digit0), tag("'")),
             ),
-            |s| data_encoding::HEXLOWER_PERMISSIVE.decode(s.as_bytes()),
+            |s| data_encoding::HEXLOWER_PERMISSIVE.decode(s.as_bytes())
+                .map(|o| (o, Some(Encoding::Base16))),
         ),
         map_res(
             preceded(
                 tag("b32"),
                 delimited(tag("'"), wrapws_strings(base32_digit0), tag("'")),
             ),
-            |s| data_encoding::BASE32.decode(s.as_bytes()),
+            |s| data_encoding::BASE32.decode(s.as_bytes())
+                .map(|o| (o, None)), // FIXME: Could add as encoding
         ),
         map_res(
             preceded(
                 tag("h32"),
                 delimited(tag("'"), wrapws_strings(base32hex_digit0), tag("'")),
             ),
-            |s| data_encoding::BASE32HEX.decode(s.as_bytes()),
+            |s| data_encoding::BASE32HEX.decode(s.as_bytes())
+                .map(|o| (o, None)), // FIXME: Could add as encoding
         ),
         map_res(
             preceded(
                 tag("b64"),
                 delimited(tag("'"), wrapws_strings(base64url_digit0), tag("'")),
             ),
-            |s| data_encoding::BASE64URL_NOPAD.decode(s.as_bytes()),
+            |s| data_encoding::BASE64URL_NOPAD.decode(s.as_bytes())
+                .map(|o| (o, Some(Encoding::Base64))),
         ),
         map_res(
             preceded(
                 tag("b64"),
                 delimited(tag("'"), wrapws_strings(base64_digit0), tag("'")),
             ),
-            |s| data_encoding::BASE64.decode(s.as_bytes()),
+            |s| data_encoding::BASE64.decode(s.as_bytes())
+                .map(|o| (o, Some(Encoding::Base64))),
         ),
         map(
             delimited(tag("<<"), separated_list0(tag(","), data_item), tag(">>")),
-            |items| items.into_iter().flat_map(|item| item.to_bytes()).collect(),
+            |items| (items.into_iter().flat_map(|item| item.to_bytes()).collect(), None),
         ),
         map(
             delimited(
@@ -273,14 +280,15 @@
                 )),
                 tag("'"),
             ),
-            |s| s.unwrap_or_default().into_bytes(),
+            |s| (s.unwrap_or_default().into_bytes(), None), // FIXME be explicit in Encoding?
         ),
     )))(input)
 }
 
 fn concatenated_definite_bytestring(input: &str) -> IResult<&str, ByteString> {
     map(many1(definite_bytestring), |data| ByteString {
-        data: data.into_iter().flatten().collect(),
+        diag_encoding: data.get(0).and_then(|(_d, e)| e.as_ref()).copied(),
+        data: data.into_iter().map(|(d, _e)| d).flatten().collect(),
         bitwidth: IntegerWidth::Unknown,
     })(input)
 }
@@ -324,7 +332,7 @@
             definite_textstring,
             map_res(
                 many0(alt((
-                    definite_bytestring,
+                    map(definite_bytestring, |(d, _e)| d),
                     map(definite_textstring, |s| s.into_bytes()),
                 ))),
                 |rest| String::from_utf8(rest.into_iter().flatten().collect()),

src/syntax.rs

@@ -56,6 +56,9 @@ pub struct Tag(pub u64);
 pub struct Simple(pub u8);
 
 #[derive(Debug, Eq, PartialEq, Clone)]
+// FIXME: Deriving PartialEq here means different diag notations do differ; we'll have to make a
+// choice at some point whether Eq means Eq in all representations and choices (probably that's
+// indeed the right one).
 /// A string of raw bytes with no direct attached meaning.
 ///
 /// May be assigned a meaning by being enclosed in a [semantic tag](Tag).
@@ -65,9 +68,30 @@ pub struct Simple(pub u8);
 /// [RFC 2.1]: https://tools.ietf.org/html/rfc7049#section-2.1
 pub struct ByteString {
     /// The raw binary data in this byte string
-    pub data: Vec<u8>,
+    pub(crate) data: Vec<u8>,
     /// The bitwidth used for encoding the length
-    pub bitwidth: IntegerWidth,
+    pub(crate) bitwidth: IntegerWidth,
+    /// Encoding used for diagnostic notation
+    pub(crate) diag_encoding: Option<crate::encode::Encoding>,
+}
+
+impl ByteString {
+    /// Create a new ByteString
+    ///
+    /// The bitwidth of the encoding is initially unknown
+    pub fn new(data: impl Into<Vec<u8>>) -> Self {
+        let data = data.into();
+        Self {
+            data,
+            bitwidth: IntegerWidth::Unknown,
+            diag_encoding: None,
+        }
+    }
+
+    /// Builder for ByteStrings with a fixed bit width
+    pub fn with_bitwidth(self, bitwidth: IntegerWidth) -> Self {
+        Self { bitwidth, ..self }
+    }
 }
 
 #[derive(Debug, Eq, PartialEq, Clone)]