Enforce sample ID uniqueness on append (#93)

Author: tomwhite

tests/test_append.py

@@ -476,3 +476,19 @@ def test_require_direct_copy_fails_before_mutating_when_incoming_is_not_aligned(
     root = zarr.open_group(store=store1, mode="r")
     np.testing.assert_array_equal(root["sample_id"][:], np.array(["S1", "S2"]))
     assert root["call_genotype"].shape == (2, 2, 2)
+
+
+def test_append_fails_when_duplicate_sample_id():
+    store1 = _create_minimal_append_store(
+        ["S1", "S2"],
+        _make_genotype(2, 2),
+        samples_chunk_size=2,
+    )
+    store2 = _create_minimal_append_store(
+        ["S1", "S3"],
+        _make_genotype(2, 4),
+        samples_chunk_size=2,
+    )
+
+    with pytest.raises(ValueError, match=r"Duplicate samples found: \['S1'\]"):
+        append(store1, store2)

vczstore/append.py

@@ -128,6 +128,12 @@ def append(
         )
 
         # check preconditions
+        sample_id1 = root1["sample_id"]
+        sample_id2 = root2["sample_id"]
+        common_samples = np.intersect1d(sample_id1, sample_id2)
+        if common_samples.shape[0] > 0:
+            raise ValueError(f"Duplicate samples found: {common_samples}")
+
         n_variants1 = root1["variant_contig"].shape[0]
         n_variants2 = root2["variant_contig"].shape[0]
         if n_variants1 != n_variants2:
@@ -168,9 +174,6 @@ def append(
         )
 
         # append samples
-        sample_id1 = root1["sample_id"]
-        sample_id2 = root2["sample_id"]
-
         old_num_samples = sample_id1.shape[0]
         incoming_num_samples = sample_id2.shape[0]
         new_num_samples = old_num_samples + incoming_num_samples