fix: expose ref metadata in bindings

Author: majin1102

java/src/main/java/org/lance/Dataset.java

@@ -1334,23 +1334,6 @@ public Map<String, String> getConfig() {
 
   private native Map<String, String> nativeGetConfig();
 
-  /**
-   * Check whether the dataset uses stable row IDs.
-   *
-   * <p>Stable row IDs remain constant when rows are moved during compaction. This reads the
-   * manifest feature flag directly rather than the user-facing config map.
-   *
-   * @return true if the dataset was created with stable row IDs enabled
-   */
-  public boolean hasStableRowIds() {
-    try (LockManager.ReadLock readLock = lockManager.acquireReadLock()) {
-      Preconditions.checkArgument(nativeDatasetHandle != 0, "Dataset is closed");
-      return nativeHasStableRowIds();
-    }
-  }
-
-  private native boolean nativeHasStableRowIds();
-
   /**
    * Get the Lance file format version of this dataset.
    *
@@ -1703,6 +1686,14 @@ public void update(String tag, Ref ref) {
       }
     }
 
+    public void updateMetadata(String tag, Optional<String> metadata) {
+      Preconditions.checkArgument(tag != null, "tag cannot be null");
+      try (LockManager.WriteLock writeLock = lockManager.acquireWriteLock()) {
+        Preconditions.checkArgument(nativeDatasetHandle != 0, "Dataset is closed");
+        nativeUpdateTagMetadata(tag, metadata);
+      }
+    }
+
     /**
      * List all tags of the dataset.
      *
@@ -1755,6 +1746,14 @@ public List<Branch> list() {
         return nativeListBranches();
       }
     }
+
+    public void updateMetadata(String branchName, Optional<String> metadata) {
+      Preconditions.checkArgument(branchName != null, "branchName cannot be null");
+      try (LockManager.WriteLock writeLock = lockManager.acquireWriteLock()) {
+        Preconditions.checkArgument(nativeDatasetHandle != 0, "Dataset is closed");
+        nativeUpdateBranchMetadata(branchName, metadata);
+      }
+    }
   }
 
   /**
@@ -1840,6 +1839,8 @@ private native MergeInsertResult nativeMergeInsert(
 
   private native void nativeUpdateTag(String tag, Ref ref);
 
+  private native void nativeUpdateTagMetadata(String tag, Optional<String> metadata);
+
   private native List<Tag> nativeListTags();
 
   private native long nativeGetVersionByTag(String tag);
@@ -1854,6 +1855,8 @@ private native Dataset nativeCreateBranch(
 
   private native List<Branch> nativeListBranches();
 
+  private native void nativeUpdateBranchMetadata(String branch, Optional<String> metadata);
+
   public Dataset shallowClone(String targetPath, Ref ref) {
     return shallowClone(targetPath, ref, null);
   }

python/python/lance/lance/__init__.pyi

@@ -346,6 +346,11 @@ class _Dataset:
         tag: str,
         reference: Optional[int | str | Tuple[Optional[str], Optional[int]]] = None,
     ): ...
+    def update_tag_metadata(
+        self,
+        tag: str,
+        metadata: Optional[str] = None,
+    ) -> None: ...
     # Branch operations
     def branches(self) -> Dict[str, Branch]: ...
     def branches_ordered(self, order: Optional[str]) -> List[Tuple[str, Branch]]: ...
@@ -357,6 +362,11 @@ class _Dataset:
         **kwargs,
     ) -> _Dataset: ...
     def delete_branch(self, branch: str) -> None: ...
+    def update_branch_metadata(
+        self,
+        branch: str,
+        metadata: Optional[str] = None,
+    ) -> None: ...
     def optimize_indices(self, **kwargs): ...
     def create_index(
         self,

python/python/tests/test_dataset.py

@@ -471,7 +471,9 @@ def test_tag(tmp_path: Path):
         ds.tags.delete("tag1")
 
     ds.tags.create("tag1", 1)
+    ds.tags.update_metadata("tag1", metadata="first tag")
     assert len(ds.tags.list()) == 1
+    assert ds.tags.list()["tag1"]["metadata"] == "first tag"
 
     with pytest.raises(ValueError):
         ds.tags.create("tag1", 1)
@@ -505,13 +507,25 @@ def test_tag(tmp_path: Path):
     ):
         ds.tags.update("tag3", 1)
 
+    ds.tags.update_metadata("tag1", metadata="updated tag")
+    ds = lance.dataset(base_dir, "tag1")
+    assert ds.version == 1
+    assert ds.tags.list()["tag1"]["metadata"] == "updated tag"
+
     ds.tags.update("tag1", 2)
     ds = lance.dataset(base_dir, "tag1")
     assert ds.version == 2
+    assert ds.tags.list()["tag1"]["metadata"] == "updated tag"
+
+    ds.tags.update_metadata("tag1", metadata=None)
+    ds = lance.dataset(base_dir, "tag1")
+    assert ds.version == 2
+    assert ds.tags.list()["tag1"]["metadata"] is None
 
     ds.tags.update("tag1", 1)
     ds = lance.dataset(base_dir, "tag1")
     assert ds.version == 1
+    assert ds.tags.list()["tag1"]["metadata"] is None
 
     version = ds.tags.get_version("tag1")
     assert version == 1
@@ -5213,6 +5227,7 @@ def test_branches(tmp_path: Path):
     ds_main = lance.write_dataset(main_table, base_dir)
 
     branch1 = ds_main.create_branch("branch1")
+    ds_main.branches.update_metadata("branch1", "branch one")
     assert branch1.version == 1
     branch1_append = pa.Table.from_pydict({"a": [7, 8], "b": [9, 10]})
     branch1 = lance.write_dataset(branch1_append, branch1, mode="append")
@@ -5262,6 +5277,7 @@ def test_branches(tmp_path: Path):
     assert isinstance(b1_meta["parent_version"], int)
     assert b1_meta["manifest_size"] > 0
     assert "create_at" in b1_meta
+    assert b1_meta["metadata"] == "branch one"
 
     try:
         ds_main.checkout_version("branch_not_exists")

python/src/dataset.rs

@@ -1682,6 +1682,7 @@ impl Dataset {
             let dict = PyDict::new(py);
             dict.set_item("version", tag_content.version)?;
             dict.set_item("manifest_size", tag_content.manifest_size)?;
+            dict.set_item("metadata", tag_content.metadata.clone())?;
 
             pylist.append((tag_name.as_str(), dict))?;
         }
@@ -1699,6 +1700,7 @@ impl Dataset {
             dict.set_item("branch", v.branch.clone())?;
             dict.set_item("version", v.version)?;
             dict.set_item("manifest_size", v.manifest_size)?;
+            dict.set_item("metadata", v.metadata.clone())?;
             pytags.set_item(k, dict.into_py_any(py)?)?;
         }
         pytags.into_py_any(py)
@@ -1754,6 +1756,16 @@ impl Dataset {
         Ok(())
     }
 
+    #[pyo3(signature = (tag, metadata=None))]
+    fn update_tag_metadata(&self, tag: String, metadata: Option<String>) -> PyResult<()> {
+        rt().block_on(
+            None,
+            self.ds.as_ref().tags().update_metadata(&tag, metadata),
+        )?
+        .infer_error()?;
+        Ok(())
+    }
+
     /// Check out the latest version of the current branch
     fn checkout_latest(&mut self) -> PyResult<()> {
         let mut new_self = self.ds.as_ref().clone();
@@ -1821,11 +1833,25 @@ impl Dataset {
             dict.set_item("parent_version", meta.parent_version)?;
             dict.set_item("create_at", meta.create_at)?;
             dict.set_item("manifest_size", meta.manifest_size)?;
+            dict.set_item("metadata", meta.metadata.clone())?;
             pybranches.set_item(name, dict.into_py_any(py)?)?;
         }
         Ok(pybranches.into())
     }
 
+    #[pyo3(signature = (branch, metadata=None))]
+    fn update_branch_metadata(&self, branch: String, metadata: Option<String>) -> PyResult<()> {
+        rt().block_on(
+            None,
+            self.ds
+                .as_ref()
+                .branches()
+                .update_metadata(&branch, metadata),
+        )?
+        .infer_error()?;
+        Ok(())
+    }
+
     /// List branches ordered by parent_version
     fn branches_ordered(
         &self,