Always transitively merge (#9)

src/eche/cluster.py

@@ -99,10 +99,8 @@ def _contains_overlaps(self, data):
         return False
 
     def _from_sets(self, data: Iterable[Set]):
-        # check if contains overlaps
-        if self._contains_overlaps(data):
-            # merge overlapping
-            data = connected_components(data)
+        # merge overlapping
+        data = connected_components(data)
 
         for cluster_id, inner in enumerate(data):
             if not isinstance(inner, set):
@@ -158,13 +156,15 @@ def __init__(
             return
         if not isinstance(data, (dict, list)):
             raise TypeError(f"Only list or dict allowed, but got {type(data)}")
-        if isinstance(data, list):
-            self._from_sets(data)
-        elif isinstance(data, dict) and len(data) != 0:
+        if isinstance(data, dict) and len(data) != 0:
             if isinstance(next(iter(data.values())), set):
                 self._from_clusters(data)
-            else:
-                self._from_dict(data)
+                return
+            # assume binary links as key-value pairs
+            # transform to iterable sets of 2
+            data = (set(pair) for pair in data.items())
+        if isinstance(data, Iterable):
+            self._from_sets(data)
 
     def __repr__(self):
         return f"ClusterHelper(elements={self.elements!s},clusters={self.clusters!s})"

tests/test_cluster.py

@@ -105,10 +105,6 @@ def test_clusters_init():
     }
     assert {frozenset(c) for c in ch_sets.clusters.values()} == expected_clusters
 
-    # assert no selflinks
-    with pytest.raises(ValueError, match="selflinks"):
-        ClusterHelper({"1": "1"})
-
     # assert no multiple cluster memberships with cluster init
     with pytest.raises(ValueError, match="multiple membership"):
         ClusterHelper({0: {"1", "2"}, 1: {"1", "3"}})
@@ -492,3 +488,14 @@ def test_from_zipped_file(
     assert expected_prefixed_pairs == set(
         ch.pairs_in_ds_tuple(ds_tuple=_LEFT_RIGHT_NAMES)
     )
+
+
+def test_transitivity_for_all_inits():
+    gold = {0: {"a", "b", "c"}}
+    assert ClusterHelper(gold).clusters == gold
+    assert ClusterHelper({"a": "b", "b": "c"}).clusters == gold
+    assert ClusterHelper([{"a", "b"}, {"b", "c"}]).clusters == gold
+    assert ClusterHelper([{"a", "b", "c"}]).clusters == gold
+
+    # self-links should not matter anymore
+    assert ClusterHelper({"a": "a", "b": "a", "c": "b"}).clusters == gold