fix(format): locate the address in version-1 shared messages

A version-1 shared message reference is version, type, six reserved bytes
and then an old-style symbol table entry: link-name offset (length size),
object header address, cache type, reserved, scratch. We read the address
straight after the reserved bytes, i.e. the link-name offset, and the
committed datatype lookup failed with InvalidObjectHeaderVersion (the bytes
checked in tcompound.h5: name offset 0x10, then 0x590 = /type1). Datasets
of 1.4/1.6-era files that use a committed datatype were unreadable.

Skip the name offset. parse_shared_ref has no length size, so add
parse_shared_ref_sized and use it in every internal caller;
parse_shared_ref keeps its signature and assumes length size == offset
size. The old parse_v1_ref unit test encoded the wrong layout and now uses
the real bytes.

Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>

crates/clawhdf5-format/src/shared_message.rs

@@ -154,13 +154,29 @@ pub fn is_shared(msg_flags: u8) -> bool {
 ///
 /// When the shared flag is set on a message, the data contains a reference
 /// instead of the actual message content.
+///
+/// Assumes the file's length size equals its offset size, which only matters
+/// for version-1 references; use [`parse_shared_ref_sized`] when the
+/// superblock's length size is known.
 pub fn parse_shared_ref(data: &[u8], offset_size: u8) -> Result<SharedMessageRef, FormatError> {
+    parse_shared_ref_sized(data, offset_size, offset_size)
+}
+
+/// [`parse_shared_ref`] with the superblock's length size, which locates the
+/// object header address in a version-1 reference.
+pub fn parse_shared_ref_sized(
+    data: &[u8],
+    offset_size: u8,
+    length_size: u8,
+) -> Result<SharedMessageRef, FormatError> {
     ensure_len(data, 0, 2)?;
     let version = data[0];
     let ref_type = data[1];
 
     // Layouts (HDF5 spec IV.A.2 "Shared Message", and libhdf5's decoder):
-    //   v1: version, type, reserved(6), address     — always "committed"
+    //   v1: version, type, reserved(6), then an old-style symbol table
+    //       entry: link-name offset(length_size), object header address,
+    //       cache type(4), reserved(4), scratch(16)  — always "committed"
     //   v2: version, type, address                  — always "committed"
     //   v3: version, type, then a fractal-heap ID if type == SOHM, otherwise
     //       an address
@@ -177,7 +193,7 @@ pub fn parse_shared_ref(data: &[u8], offset_size: u8) -> Result<SharedMessageRef
         })
     };
     match version {
-        1 => address_at(2 + 6),
+        1 => address_at(2 + 6 + length_size as usize),
         2 => address_at(2),
         3 if ref_type == SHARE_TYPE_SOHM => {
             ensure_len(data, 2, FHEAP_ID_LEN)?;
@@ -434,7 +450,7 @@ pub fn message_data_with_sohm<'a>(
     if !is_shared(msg.flags) {
         return Ok(Cow::Borrowed(&msg.data));
     }
-    let shared_ref = parse_shared_ref(&msg.data, offset_size)?;
+    let shared_ref = parse_shared_ref_sized(&msg.data, offset_size, length_size)?;
     let table = if shared_ref.heap_id.is_some() {
         load_sohm_table(file_data, offset_size, length_size)?
     } else {
@@ -514,7 +530,7 @@ pub fn message_data<'a>(
     if !is_shared(msg.flags) {
         return Ok(Cow::Borrowed(&msg.data));
     }
-    let shared_ref = parse_shared_ref(&msg.data, offset_size)?;
+    let shared_ref = parse_shared_ref_sized(&msg.data, offset_size, length_size)?;
     resolve_shared_message(
         file_data,
         &shared_ref,
@@ -649,15 +665,26 @@ mod tests {
 
     #[test]
     fn parse_v1_ref() {
-        let mut data = Vec::new();
-        data.push(1); // version
-        data.push(0); // type
-        data.extend_from_slice(&[0u8; 6]); // reserved
-        data.extend_from_slice(&0x5678u64.to_le_bytes());
+        // Datatype message of `/group1/dset2` in HDF5's `tcompound.h5`
+        // (written in 2000): version 1, six reserved bytes, then an old-style
+        // symbol table entry — link-name offset 0x10, object header address
+        // 0x590 (the committed datatype `/type1`), cache type, reserved and
+        // scratch.
+        let mut data = vec![1, 0, 0, 0, 0, 0, 0, 0];
+        data.extend_from_slice(&0x10u64.to_le_bytes());
+        data.extend_from_slice(&0x590u64.to_le_bytes());
+        data.extend_from_slice(&[0; 24]);
 
-        let shared = parse_shared_ref(&data, 8).unwrap();
+        let shared = parse_shared_ref_sized(&data, 8, 8).unwrap();
         assert_eq!(shared.version, 1);
-        assert_eq!(shared.object_header_address, Some(0x5678));
+        assert_eq!(shared.object_header_address, Some(0x590));
+
+        // The name offset is a length: 4 bytes here, then an 8-byte address.
+        let mut data = vec![1, 0, 0, 0, 0, 0, 0, 0];
+        data.extend_from_slice(&0x10u32.to_le_bytes());
+        data.extend_from_slice(&0x590u64.to_le_bytes());
+        let shared = parse_shared_ref_sized(&data, 8, 4).unwrap();
+        assert_eq!(shared.object_header_address, Some(0x590));
     }
 
     #[test]