Issue #391 support shared PGNodes better in MultiResult

Author: soxofaan

openeo/internal/graph_building.py

@@ -451,11 +451,13 @@ def __init__(self, leaves: List[FlatGraphableMixin]):
 
     def flat_graph(self) -> Dict[str, dict]:
         result = {}
+        flattener = GraphFlattener()
         for leave in self._leaves:
-            leave_graph = leave.flat_graph()
-            existing = set(leave_graph.keys()).intersection(result.keys())
-            if existing:
-                # TODO: automatic renaming of duplicate node ids?
-                raise ValueError(f"Duplicate node ids while building multi-result process graph: {sorted(existing)}")
-            result.update(leave_graph)
+            if isinstance(leave, PGNode):
+                result = flattener.flatten(leave)
+            elif isinstance(leave, _FromNodeMixin):
+                result = flattener.flatten(leave.from_node())
+            else:
+                raise ValueError(leave)
+
         return result

openeo/rest/datacube.py

@@ -85,7 +85,9 @@ class DataCube(_ProcessGraphAbstraction):
     # TODO: set this based on back-end or user preference?
     _DEFAULT_RASTER_FORMAT = "GTiff"
 
-    def __init__(self, graph: PGNode, connection: Optional[Connection], metadata: Optional[CollectionMetadata] = None):
+    def __init__(
+        self, graph: PGNode, connection: Optional[Connection] = None, metadata: Optional[CollectionMetadata] = None
+    ):
         super().__init__(pgnode=graph, connection=connection)
         self.metadata: Optional[CollectionMetadata] = metadata
 

tests/internal/test_graphbuilding.py

@@ -14,6 +14,7 @@
     ReduceNode,
 )
 from openeo.internal.process_graph_visitor import ProcessGraphVisitException
+from openeo.rest.datacube import DataCube
 
 
 def test_pgnode_process_id():
@@ -426,7 +427,26 @@ def test_simple(self):
 
     def test_simple_duplicates(self):
         multi = MultiResult([PGNode("foo"), PGNode("foo")])
-        with pytest.raises(
-            ValueError, match=re.escape("Duplicate node ids while building multi-result process graph: ['foo1']")
-        ):
-            multi.flat_graph()
+        assert multi.flat_graph() == {
+            "foo1": {"process_id": "foo", "arguments": {}, "result": True},
+            "foo2": {"process_id": "foo", "arguments": {}, "result": True},
+        }
+
+    def test_multi_save_result_same_root(self):
+        load_collection = DataCube(PGNode("load_collection", collection_id="S2"))
+        save_a = load_collection.save_result(format="GTiff")
+        save_b = load_collection.save_result(format="NetCDF")
+        multi = MultiResult([save_a, save_b])
+        assert multi.flat_graph() == {
+            "loadcollection1": {"process_id": "load_collection", "arguments": {"collection_id": "S2"}},
+            "saveresult1": {
+                "process_id": "save_result",
+                "arguments": {"data": {"from_node": "loadcollection1"}, "format": "GTiff", "options": {}},
+                "result": True,
+            },
+            "saveresult2": {
+                "process_id": "save_result",
+                "arguments": {"data": {"from_node": "loadcollection1"}, "format": "NetCDF", "options": {}},
+                "result": True,
+            },
+        }