update refs in exposure and semantic model definitions when downstream of a split resource

dbt_meshify/utilities/grouper.py

@@ -90,7 +90,11 @@ def _generate_resource_group(
 
         logger.info(f"Selected {len(nodes)} resources: {nodes}")
         # Check if any of the selected nodes are already in a group of a different name. If so, raise an exception.
-        nodes = set(filter(lambda x: not x.startswith("source"), nodes))
+        nodes = set(
+            filter(
+                lambda x: not x.startswith("source") and not x.startswith("semantic_model"), nodes
+            )
+        )
         for node in nodes:
             existing_group = self.project.manifest.nodes[node].config.group
 

dbt_meshify/utilities/references.py

@@ -1,11 +1,17 @@
 import re
 from pathlib import Path
-from typing import List, Optional, Union
+from typing import Any, Dict, List, Optional, Union
 
-from dbt.contracts.graph.nodes import CompiledNode
+from dbt.contracts.graph.nodes import CompiledNode, Exposure, Resource, SemanticModel
 from loguru import logger
 
-from dbt_meshify.change import ChangeSet, EntityType, FileChange, Operation
+from dbt_meshify.change import (
+    ChangeSet,
+    EntityType,
+    FileChange,
+    Operation,
+    ResourceChange,
+)
 from dbt_meshify.dbt_projects import DbtProject, DbtSubProject, PathedProject
 
 
@@ -135,32 +141,77 @@ def replace_source_with_ref__python(
 
         return new_code
 
+    def update_yml_resource_references(
+        self,
+        project_name: str,
+        upstream_resource_name: str,
+        resource: Union[Exposure, SemanticModel],
+    ) -> Dict[str, Any]:
+        new_ref = f"ref('{project_name}', '{upstream_resource_name}')"
+        if isinstance(resource, SemanticModel):
+            # we can return early, since semantic models only have one ref and no depends_on
+            return {"model": new_ref}
+        refs = resource.refs
+        ref_to_update = f"ref('{upstream_resource_name}')"
+        str_refs = []
+        for ref in refs:
+            package_clause = f"'{ref.package}', " if ref.package else ""
+            name_clause = f"'{ref.name}'"
+            version_clause = f", v={ref.version}" if ref.version else ""
+            str_ref = f"ref({package_clause}{name_clause}{version_clause})"
+            if str_ref != ref_to_update:
+                str_refs.append(str_ref)
+        str_refs.append(new_ref)
+        return {"depends_on": str_refs}
+
     def generate_reference_update(
         self,
         project_name: str,
         upstream_node: CompiledNode,
-        downstream_node: CompiledNode,
+        downstream_node: Union[Resource, CompiledNode],
         downstream_project: PathedProject,
         code: str,
-    ) -> FileChange:
+    ) -> Union[FileChange, ResourceChange]:
         """Generate FileChanges that update the references in the downstream_node's code."""
 
-        updated_code = self.ref_update_methods[downstream_node.language](
-            model_name=upstream_node.name,
-            project_name=project_name,
-            model_code=code,
-        )
+        change: Union[FileChange, ResourceChange]
+        if isinstance(downstream_node, CompiledNode):
+            updated_code = self.ref_update_methods[downstream_node.language](
+                model_name=upstream_node.name,
+                project_name=project_name,
+                model_code=code,
+            )
 
-        return FileChange(
-            operation=Operation.Update,
-            entity_type=EntityType.Code,
-            identifier=downstream_node.name,
-            path=downstream_project.resolve_file_path(downstream_node),
-            data=updated_code,
-        )
+            change = FileChange(
+                operation=Operation.Update,
+                entity_type=EntityType.Code,
+                identifier=downstream_node.name,
+                path=downstream_project.resolve_file_path(downstream_node),
+                data=updated_code,
+            )
+
+        else:
+            if isinstance(downstream_node, Exposure) or isinstance(downstream_node, SemanticModel):
+                is_exposure = isinstance(downstream_node, Exposure)
+                data = self.update_yml_resource_references(
+                    project_name=project_name,
+                    upstream_resource_name=upstream_node.name,
+                    resource=downstream_node,
+                )
+                change = ResourceChange(
+                    operation=Operation.Update,
+                    entity_type=EntityType.Exposure if is_exposure else EntityType.SemanticModel,
+                    identifier=downstream_node.name,
+                    path=downstream_project.resolve_file_path(downstream_node),
+                    data=data,
+                )
+
+        return change
 
     def update_child_refs(
-        self, resource: CompiledNode, current_change_set: Optional[ChangeSet] = None
+        self,
+        resource: CompiledNode,
+        current_change_set: Optional[ChangeSet] = None,
     ) -> ChangeSet:
         """Generate a set of FileChanges to update child references"""
 
@@ -175,33 +226,29 @@ def update_child_refs(
         else:
             compare_project = self.project.name
 
-        for model in self.project.child_map[resource.unique_id]:
-            if model in self.project.resources or model.split(".")[1] != compare_project:
-                continue
-            model_node = self.project.get_manifest_node(model)
-            if not model_node:
-                raise KeyError(f"Resource {model} not found in manifest")
-
-            # Don't process Resources missing a language attribute
-            if not hasattr(model_node, "language") or not isinstance(model_node, CompiledNode):
+        for child in self.project.child_map[resource.unique_id]:
+            if child in self.project.resources or child.split(".")[1] != compare_project:
                 continue
+            node = self.project.get_manifest_node(child)
+            if not node:
+                raise KeyError(f"Resource {child} not found in manifest")
 
             if current_change_set:
                 previous_change = get_latest_file_change(
                     changeset=current_change_set,
-                    identifier=model_node.name,
-                    path=self.project.parent_project.resolve_file_path(model_node),
+                    identifier=node.name,
+                    path=self.project.parent_project.resolve_file_path(node),
                 )
                 code = (
                     previous_change.data
                     if (previous_change and previous_change.data)
-                    else model_node.raw_code
+                    else getattr(node, "raw_code", "")
                 )
 
             change = self.generate_reference_update(
                 project_name=self.project.name,
                 upstream_node=resource,
-                downstream_node=model_node,
+                downstream_node=node,
                 code=code,
                 downstream_project=self.project.parent_project,
             )
@@ -258,8 +305,6 @@ def update_parent_refs(self, resource: CompiledNode) -> ChangeSet:
             if change.data is None:
                 raise Exception(f"Change has null data despite being provided code. {change}")
 
-            code = change.data
-
             change_set.add(change)
 
         return change_set