mailparsing: rebuild of binary attachments could corrupt them

We were blanket-applying the text charset conversion and it was
interpreting the binary bytes as windows-1252 content, which could cause
us to then re-encode the content as the utf8 equivalent of the 1252
interpretation of those bytes.

crates/mailparsing/src/mimepart.rs

@@ -308,12 +308,11 @@ impl<'a> MimePart<'a> {
             }
         };
 
-        let (decoded, _malformed) = info.charset.decode_without_bom_handling(&bytes);
-
         if info.is_text {
+            let (decoded, _malformed) = info.charset.decode_without_bom_handling(&bytes);
             Ok(DecodedBody::Text(decoded.to_string().into()))
         } else {
-            Ok(DecodedBody::Binary(decoded.as_bytes().to_vec()))
+            Ok(DecodedBody::Binary(bytes))
         }
     }
 
@@ -1273,4 +1272,19 @@ Ok(
             .conformance()
             .contains(MessageConformance::MISSING_COLON_VALUE));
     }
+
+    /// This is a regression test for an issue where we'd interpret the
+    /// binary bytes as default windows-1252 codepage charset, and mangle them.
+    /// The high byte is sufficient to trigger the offending code prior
+    /// to the fix
+    #[test]
+    fn rebuild_binary() {
+        let expect = &[0, 1, 2, 3, 0xbe, 4, 5];
+        let part = MimePart::new_binary("applicat/octet-stream", expect, None);
+
+        let rebuilt = part.rebuild().unwrap();
+        let body = rebuilt.body().unwrap();
+
+        assert_eq!(body, DecodedBody::Binary(expect.to_vec()));
+    }
 }

docs/changelog/main.md

@@ -100,3 +100,6 @@
   [msg:get_meta](../reference/message/get_meta.md) and
   [msg:set_meta](../reference/message/set_meta.md) now internally will ensure
   that the data or meta portion of the message is loaded from spool.
+* Rebuilding a MIME message (such as via `msg:check_fix_conformance`) that had
+  binary attachments would incorrectly re-interpret the bytes as windows-1252
+  encoded characters, damaging the attachment.