Move non-static data config from params into query

Author: knutwalker

graphdatascience/graph/graph_cypher_runner.py

@@ -120,6 +120,7 @@ def project(
         query_params = {"graph_name": graph_name}
 
         data_config = {}
+        data_config_is_static = True
 
         nodes = self._node_projections_spec(nodes)
         rels = self._rel_projections_spec(relationships)
@@ -152,6 +153,7 @@ def project(
 
                 data_config["sourceNodeLabels"] = "labels(source)"
                 data_config["targetNodeLabels"] = "labels(target)"
+                data_config_is_static = False
             else:
                 raise ValueError(f"Invalid value for combine_labels_with: {combine_labels_with}")
 
@@ -162,6 +164,7 @@ def project(
             else:
                 rel_var = "rel"
                 data_config["relationshipTypes"] = "type(rel)"
+                data_config_is_static = False
             match_pattern = match_pattern._replace(
                 type_filter=f"[{rel_var}:{'|'.join(spec.source_type for spec in rels)}]"
             )
@@ -179,8 +182,11 @@ def project(
         args = ["$graph_name", "source", "target"]
 
         if data_config:
-            query_params["data_config"] = data_config
-            args += ["$data_config"]
+            if data_config_is_static:
+                query_params["data_config"] = data_config
+                args += ["$data_config"]
+            else:
+                args += [self._render_map(data_config)]
 
         if config:
             query_params["config"] = config
@@ -247,6 +253,9 @@ def _rel_projection_spec(self, spec: Any, name: Optional[str] = None) -> Relatio
     def _rel_properties_spec(self, properties: dict[str, Any]) -> list[RelationshipProperty]:
         raise TypeError(f"Invalid relationship projection specification: {properties}")
 
+    def _render_map(self, mapping: dict[str, Any]) -> str:
+        return "{" + ", ".join(f"{key}: {value}" for key, value in mapping.items()) + "}"
+
     #
     # def estimate(self, *, nodes: Any, relationships: Any, **config: Any) -> "Series[Any]":
     #     pass

graphdatascience/tests/unit/test_graph_cypher.py

@@ -136,15 +136,14 @@ def test_multiple_node_labels_or(runner: CollectingQueryRunner, gds: GraphDataSc
     G, _ = gds.graph.cypher.project("g", nodes=["A", "B"], combine_labels_with="OR")
 
     assert G.name() == "g"
-    assert runner.last_params() == dict(
-        graph_name="g", data_config={"sourceNodeLabels": "labels(source)", "targetNodeLabels": "labels(target)"}
-    )
+    assert runner.last_params() == dict(graph_name="g")
 
-    assert (
-        runner.last_query()
-        == """MATCH (source)-->(target)
+    assert runner.last_query() == (
+        """MATCH (source)-->(target)
 WHERE (source:A OR source:B) AND (target:A OR target:B)
-RETURN gds.graph.project($graph_name, source, target, $data_config)"""
+RETURN gds.graph.project($graph_name, source, target, {"""
+        "sourceNodeLabels: labels(source), "
+        "targetNodeLabels: labels(target)})"
     )
 
 
@@ -153,17 +152,16 @@ def test_disconnected_nodes_multiple_node_labels_or(runner: CollectingQueryRunne
     G, _ = gds.graph.cypher.project("g", nodes=["A", "B"], combine_labels_with="OR", allow_disconnected_nodes=True)
 
     assert G.name() == "g"
-    assert runner.last_params() == dict(
-        graph_name="g", data_config={"sourceNodeLabels": "labels(source)", "targetNodeLabels": "labels(target)"}
-    )
+    assert runner.last_params() == dict(graph_name="g")
 
-    assert (
-        runner.last_query()
-        == """MATCH (source)
+    assert runner.last_query() == (
+        """MATCH (source)
 WHERE source:A OR source:B
 OPTIONAL MATCH (source)-->(target)
 WHERE target:A OR target:B
-RETURN gds.graph.project($graph_name, source, target, $data_config)"""
+RETURN gds.graph.project($graph_name, source, target, {"""
+        "sourceNodeLabels: labels(source), "
+        "targetNodeLabels: labels(target)})"
     )
 
 
@@ -207,18 +205,14 @@ def test_multiple_multi_graph(runner: CollectingQueryRunner, gds: GraphDataScien
     G, _ = gds.graph.cypher.project("g", nodes=["A", "B"], relationships=["REL1", "REL2"])
 
     assert G.name() == "g"
-    assert runner.last_params() == dict(
-        graph_name="g",
-        data_config={
-            "sourceNodeLabels": "labels(source)",
-            "targetNodeLabels": "labels(target)",
-            "relationshipTypes": "type(rel)",
-        },
-    )
+    assert runner.last_params() == dict(graph_name="g")
 
     assert (
         runner.last_query()
         == """MATCH (source)-[rel:REL1|REL2]->(target)
 WHERE (source:A OR source:B) AND (target:A OR target:B)
-RETURN gds.graph.project($graph_name, source, target, $data_config)"""
+RETURN gds.graph.project($graph_name, source, target, {"""
+        "sourceNodeLabels: labels(source), "
+        "targetNodeLabels: labels(target), "
+        "relationshipTypes: type(rel)})"
     )