Split command

dbt_meshify/cli.py

@@ -11,6 +11,13 @@
     help="The path to the dbt project to operate on. Defaults to the current directory.",
 )
 
+create_path = click.option(
+    "--create-path",
+    type=click.Path(exists=True),
+    default=None,
+    help="The path to create the new dbt project. Defaults to the name argument supplied.",
+)
+
 exclude = click.option(
     "--exclude",
     "-e",

dbt_meshify/dbt_projects.py

@@ -12,8 +12,14 @@
 from dbt.contracts.project import Project
 from dbt.contracts.results import CatalogArtifact, CatalogTable
 from dbt.graph import Graph
+from dbt.node_types import NodeType
 
 from dbt_meshify.dbt import Dbt
+from dbt_meshify.storage.file_content_editors import (
+    DbtMeshConstructor,
+    filter_empty_dict_items,
+)
+from dbt_meshify.storage.file_manager import DbtFileManager
 
 logger = logging.getLogger()
 
@@ -100,7 +106,8 @@ def installed_packages(self) -> Set[str]:
                 if item.package_name:
                     _hash = hashlib.md5()
                     _hash.update(item.package_name.encode("utf-8"))
-                    project_packages.append(_hash.hexdigest())
+                    if _hash.hexdigest() != self.manifest.metadata.project_id:
+                        project_packages.append(_hash.hexdigest())
         return set(project_packages)
 
     @property
@@ -152,7 +159,19 @@ def get_catalog_entry(self, unique_id: str) -> Optional[CatalogTable]:
 
     def get_manifest_node(self, unique_id: str) -> Optional[ManifestNode]:
         """Returns the catalog entry for a model in the dbt project's catalog"""
-        return self.manifest.nodes.get(unique_id)
+        if unique_id.split(".")[0] in [
+            "model",
+            "seed",
+            "snapshot",
+            "test",
+            "analysis",
+            "snapshot",
+        ]:
+            return self.manifest.nodes.get(unique_id)
+        else:
+            pluralized = NodeType(unique_id.split('.')[0]).pluralize()
+            resources = getattr(self.manifest, pluralized)
+            return resources.get(unique_id)
 
 
 class DbtProject(BaseDbtProject):
@@ -216,10 +235,13 @@ def split(
         project_name: str,
         select: str,
         exclude: Optional[str] = None,
+        selector: Optional[str] = None,
     ) -> "DbtSubProject":
         """Create a new DbtSubProject using NodeSelection syntax."""
 
-        subproject_resources = self.select_resources(select, exclude)
+        subproject_resources = self.select_resources(
+            select=select, exclude=exclude, selector=selector, output_key="unique_id"
+        )
 
         # Construct a new project and inject the new manifest
         subproject = DbtSubProject(
@@ -242,13 +264,65 @@ class DbtSubProject(BaseDbtProject):
     def __init__(self, name: str, parent_project: DbtProject, resources: Set[str]):
         self.name = name
         self.resources = resources
-        self.parent = parent_project
+        self.parent_project = parent_project
+        self.path = parent_project.path / Path(name)
 
-        self.manifest = parent_project.manifest.deepcopy()
+        # self.manifest = parent_project.manifest.deepcopy()
+        # i am running into a bug with the core deepcopy -- checking with michelle
+        self.manifest = copy.deepcopy(parent_project.manifest)
         self.project = copy.deepcopy(parent_project.project)
         self.catalog = parent_project.catalog
+        self.custom_macros = self._get_custom_macros()
+        self.groups = self._get_indirect_groups()
+
+        self._rename_project()
 
-        super().__init__(self.manifest, self.project, self.catalog)
+        super().__init__(self.manifest, self.project, self.catalog, self.name)
+
+    def _rename_project(self) -> None:
+        """
+        edits the project yml to take any instance of the parent project name and update it to the subproject name
+        """
+        project_dict = self.project.to_dict()
+        for key in [resource.pluralize() for resource in NodeType]:
+            if self.parent_project.name in project_dict.get(key, {}).keys():
+                project_dict[key][self.name] = project_dict[key].pop(self.parent_project.name)
+        project_dict["name"] = self.name
+        self.project = Project.from_dict(project_dict)
+
+    def _get_custom_macros(self) -> Set[str]:
+        """
+        get a set of macro unique_ids for all the selected resources
+        """
+        macros_set = set()
+        for unique_id in self.resources:
+            resource = self.get_manifest_node(unique_id)
+            if not resource:
+                continue
+            macros = resource.depends_on.macros
+            project_macros = [
+                macro
+                for macro in macros
+                if hashlib.md5((macro.split(".")[1]).encode()).hexdigest()
+                == self.manifest.metadata.project_id
+            ]
+            macros_set.update(project_macros)
+        return macros_set
+
+    def _get_indirect_groups(self) -> Set[str]:
+        """
+        get a set of group unique_ids for all the selected resources
+        """
+        groups = set()
+        for unique_id in self.resources:
+            resource = self.get_manifest_node(unique_id)
+            if not resource or resource.resource_type in [NodeType.Source, NodeType.Exposure]:
+                continue
+            group = resource.group
+            if group:
+                group_unique_id = f"group.{self.parent_project.name}.{group}"
+                groups.update({group_unique_id})
+        return groups
 
     def select_resources(self, select: str, exclude: Optional[str] = None) -> Set[str]:
         """
@@ -259,7 +333,7 @@ def select_resources(self, select: str, exclude: Optional[str] = None) -> Set[st
         if exclude:
             args.extend(["--exclude", exclude])
 
-        results = self.parent.dbt.ls(self.parent.path, args)
+        results = self.parent_project.dbt.ls(self.parent_project.path, args)
 
         return set(results) - self.resources
 
@@ -276,7 +350,7 @@ def split(
         # Construct a new project and inject the new manifest
         subproject = DbtSubProject(
             name=project_name,
-            parent_project=copy.deepcopy(self.parent),
+            parent_project=copy.deepcopy(self.parent_project),
             resources=subproject_resources,
         )
 
@@ -285,13 +359,6 @@ def split(
 
         return subproject
 
-    def initialize(self, target_directory: os.PathLike):
-        """Initialize this subproject as a full dbt project at the provided `target_directory`."""
-
-        # TODO: Implement project initialization
-
-        raise NotImplementedError
-
 
 class DbtProjectHolder:
     def __init__(self) -> None:

dbt_meshify/main.py

@@ -6,7 +6,10 @@
 import yaml
 from dbt.contracts.graph.unparsed import Owner
 
+from dbt_meshify.storage.dbt_project_creator import DbtSubprojectCreator
+
 from .cli import (
+    create_path,
     exclude,
     group_yml_path,
     owner,
@@ -18,7 +21,7 @@
     selector,
 )
 from .dbt_projects import DbtProject, DbtProjectHolder, DbtSubProject
-from .storage.yaml_editors import DbtMeshModelConstructor
+from .storage.file_content_editors import DbtMeshConstructor
 
 
 # define cli group
@@ -59,40 +62,29 @@ def connect(projects_dir):
 
 
 @cli.command(name="split")
+@create_path
+@click.argument("project_name")
 @exclude
 @project_path
 @select
 @selector
-def split():
+def split(project_name, select, exclude, project_path, selector, create_path):
     """
-    !!! info
-        This command is not yet implemented
-
-    Splits dbt projects apart by adding all necessary dbt Mesh constructs based on the selection syntax.
+    Splits out a new subproject from a dbt project by adding all necessary dbt Mesh constructs to the resources based on the selected resources.
 
     """
-    path_string = input("Enter the relative path to a dbt project you'd like to split: ")
 
-    holder = DbtProjectHolder()
-
-    path = Path(path_string).expanduser().resolve()
+    path = Path(project_path).expanduser().resolve()
     project = DbtProject.from_directory(path)
-    holder.register_project(project)
-
-    while True:
-        subproject_name = input("Enter the name for your subproject ('done' to finish): ")
-        if subproject_name == "done":
-            break
-        subproject_selector = input(
-            f"Enter the selector that represents the subproject {subproject_name}: "
-        )
-
-        subproject: DbtSubProject = project.split(
-            project_name=subproject_name, select=subproject_selector
-        )
-        holder.register_project(subproject)
 
-    print(holder.project_map())
+    subproject = project.split(
+        project_name=project_name, select=select, exclude=exclude, selector=selector
+    )
+    target_directory = Path(create_path) if create_path else None
+    subproject_creator = DbtSubprojectCreator(
+        subproject=subproject, target_directory=target_directory
+    )
+    subproject_creator.initialize()
 
 
 @operation.command(name="add-contract")
@@ -117,8 +109,8 @@ def add_contract(select, exclude, project_path, selector, public_only=False):
     for model_unique_id in models:
         model_node = project.get_manifest_node(model_unique_id)
         model_catalog = project.get_catalog_entry(model_unique_id)
-        meshify_constructor = DbtMeshModelConstructor(
-            project_path=project_path, model_node=model_node, model_catalog=model_catalog
+        meshify_constructor = DbtMeshConstructor(
+            project_path=project_path, node=model_node, catalog=model_catalog
         )
         meshify_constructor.add_model_contract()
 
@@ -145,9 +137,7 @@ def add_version(select, exclude, project_path, selector, prerelease, defined_in)
     for model_unique_id in models:
         model_node = project.get_manifest_node(model_unique_id)
         if model_node.version == model_node.latest_version:
-            meshify_constructor = DbtMeshModelConstructor(
-                project_path=project_path, model_node=model_node
-            )
+            meshify_constructor = DbtMeshConstructor(project_path=project_path, node=model_node)
             meshify_constructor.add_model_version(prerelease=prerelease, defined_in=defined_in)