cognite-toolkit 0.7.30__py3-none-any.whl → 0.7.39__py3-none-any.whl

cognite_toolkit/_cdf_tk/commands/_migrate/migration_io.py

@@ -1,23 +1,43 @@
 from collections.abc import Iterable, Iterator, Mapping, Sequence
-from typing import ClassVar, cast
+from typing import ClassVar, Literal, cast
 
 from cognite.client.data_classes import Annotation
 from cognite.client.data_classes.data_modeling import EdgeId, InstanceApply, NodeId
 
 from cognite_toolkit._cdf_tk.client import ToolkitClient
-from cognite_toolkit._cdf_tk.client.data_classes.pending_instances_ids import PendingInstanceId
+from cognite_toolkit._cdf_tk.client.data_classes.legacy.pending_instances_ids import PendingInstanceId
+from cognite_toolkit._cdf_tk.client.data_classes.three_d import (
+    AssetMappingDMRequest,
+    AssetMappingResponse,
+    ThreeDModelResponse,
+)
+from cognite_toolkit._cdf_tk.commands._migrate.data_classes import ThreeDMigrationRequest
 from cognite_toolkit._cdf_tk.constants import MISSING_EXTERNAL_ID, MISSING_INSTANCE_SPACE
 from cognite_toolkit._cdf_tk.exceptions import ToolkitNotImplementedError, ToolkitValueError
 from cognite_toolkit._cdf_tk.storageio import (
     AnnotationIO,
     HierarchyIO,
     InstanceIO,
+    T_Selector,
     UploadableStorageIO,
 )
 from cognite_toolkit._cdf_tk.storageio._base import Page, UploadItem
+from cognite_toolkit._cdf_tk.storageio.selectors import (
+    ThreeDModelFilteredSelector,
+    ThreeDModelIdSelector,
+    ThreeDSelector,
+)
 from cognite_toolkit._cdf_tk.tk_warnings import MediumSeverityWarning
 from cognite_toolkit._cdf_tk.utils.collection import chunker_sequence
-from cognite_toolkit._cdf_tk.utils.http_client import HTTPClient, HTTPMessage, ItemsRequest, SuccessResponseItems
+from cognite_toolkit._cdf_tk.utils.http_client import (
+    FailedResponse,
+    HTTPClient,
+    HTTPMessage,
+    ItemsRequest,
+    SimpleBodyRequest,
+    SuccessResponseItems,
+    ToolkitAPIError,
+)
 from cognite_toolkit._cdf_tk.utils.useful_types import (
     AssetCentricKindExtended,
     AssetCentricType,
@@ -348,3 +368,200 @@ class AnnotationMigrationIO(
         selector: AssetCentricMigrationSelector | None = None,
     ) -> list[dict[str, JsonVal]]:
         raise NotImplementedError("Serializing Annotation Migrations to JSON is not supported.")
+
+
+class ThreeDMigrationIO(UploadableStorageIO[ThreeDSelector, ThreeDModelResponse, ThreeDMigrationRequest]):
+    """IO class for downloading and migrating 3D models.
+
+    Args:
+        client: The ToolkitClient to use for CDF interactions.
+        data_model_type: The type of 3D data model to download. Either "classic" or "DM".
+
+    """
+
+    KIND = "3DMigration"
+    SUPPORTED_DOWNLOAD_FORMATS = frozenset({".ndjson"})
+    SUPPORTED_COMPRESSIONS = frozenset({".gz"})
+    SUPPORTED_READ_FORMATS = frozenset({".ndjson"})
+    DOWNLOAD_LIMIT = 1000
+    CHUNK_SIZE = 1
+    UPLOAD_ENDPOINT = "/3d/migrate/models"
+    REVISION_ENDPOINT = "/3d/migrate/revisions"
+
+    def __init__(self, client: ToolkitClient, data_model_type: Literal["classic", "data modeling"] = "classic") -> None:
+        super().__init__(client)
+        self.data_model_type = data_model_type
+
+    def as_id(self, item: ThreeDModelResponse) -> str:
+        return f"{item.name}_{item.id!s}"
+
+    def _is_selected(self, item: ThreeDModelResponse, included_models: set[int] | None) -> bool:
+        return self._is_correct_type(item) and (included_models is None or item.id in included_models)
+
+    def _is_correct_type(self, item: ThreeDModelResponse) -> bool:
+        if self.data_model_type == "classic":
+            return item.space is None
+        else:
+            return item.space is not None
+
+    def stream_data(self, selector: ThreeDSelector, limit: int | None = None) -> Iterable[Page[ThreeDModelResponse]]:
+        published: bool | None = None
+        if isinstance(selector, ThreeDModelFilteredSelector):
+            published = selector.published
+        included_models: set[int] | None = None
+        if isinstance(selector, ThreeDModelIdSelector):
+            included_models = set(selector.ids)
+        cursor: str | None = None
+        total = 0
+        while True:
+            request_limit = min(self.DOWNLOAD_LIMIT, limit - total) if limit is not None else self.DOWNLOAD_LIMIT
+            response = self.client.tool.three_d.models.iterate(
+                published=published, include_revision_info=True, limit=request_limit, cursor=cursor
+            )
+            items = [item for item in response.items if self._is_selected(item, included_models)]
+            total += len(items)
+            if items:
+                yield Page(worker_id="main", items=items, next_cursor=response.next_cursor)
+            if response.next_cursor is None:
+                break
+            cursor = response.next_cursor
+
+    def count(self, selector: ThreeDSelector) -> int | None:
+        # There is no efficient way to count 3D models in CDF.
+        return None
+
+    def data_to_json_chunk(
+        self, data_chunk: Sequence[ThreeDModelResponse], selector: ThreeDSelector | None = None
+    ) -> list[dict[str, JsonVal]]:
+        raise NotImplementedError("Deserializing Annotation Migrations from JSON is not supported.")
+
+    def json_to_resource(self, item_json: dict[str, JsonVal]) -> ThreeDMigrationRequest:
+        raise NotImplementedError("Deserializing ThreeD Migrations from JSON is not supported.")
+
+    def upload_items(
+        self,
+        data_chunk: Sequence[UploadItem[ThreeDMigrationRequest]],
+        http_client: HTTPClient,
+        selector: ThreeDSelector | None = None,
+    ) -> Sequence[HTTPMessage]:
+        """Migrate 3D models by uploading them to the migrate/models endpoint."""
+        if len(data_chunk) > self.CHUNK_SIZE:
+            raise RuntimeError(f"Uploading more than {self.CHUNK_SIZE} 3D models at a time is not supported.")
+
+        results: list[HTTPMessage] = []
+        responses = http_client.request_with_retries(
+            message=ItemsRequest(
+                endpoint_url=self.client.config.create_api_url(self.UPLOAD_ENDPOINT),
+                method="POST",
+                items=list(data_chunk),
+            )
+        )
+        if (
+            failed_response := next((res for res in responses if isinstance(res, FailedResponse)), None)
+        ) and failed_response.status_code == 400:
+            raise ToolkitAPIError("3D model migration failed. You need to enable the 3D migration alpha feature flag.")
+
+        results.extend(responses)
+        success_ids = {id for res in responses if isinstance(res, SuccessResponseItems) for id in res.ids}
+        for data in data_chunk:
+            if data.source_id not in success_ids:
+                continue
+            revision = http_client.request_with_retries(
+                message=SimpleBodyRequest(
+                    endpoint_url=self.client.config.create_api_url(self.REVISION_ENDPOINT),
+                    method="POST",
+                    body_content={"items": [data.item.revision.dump(camel_case=True)]},
+                )
+            )
+            results.extend(revision.as_item_responses(data.source_id))
+        return results
+
+
+class ThreeDAssetMappingMigrationIO(UploadableStorageIO[ThreeDSelector, AssetMappingResponse, AssetMappingDMRequest]):
+    KIND = "3DMigrationAssetMapping"
+    SUPPORTED_DOWNLOAD_FORMATS = frozenset({".ndjson"})
+    SUPPORTED_COMPRESSIONS = frozenset({".gz"})
+    SUPPORTED_READ_FORMATS = frozenset({".ndjson"})
+    DOWNLOAD_LIMIT = 1000
+    CHUNK_SIZE = 100
+    UPLOAD_ENDPOINT = "/3d/models/{modelId}/revisions/{revisionId}/mappings"
+
+    def __init__(self, client: ToolkitClient, object_3D_space: str, cad_node_space: str) -> None:
+        super().__init__(client)
+        self.object_3D_space = object_3D_space
+        self.cad_node_space = cad_node_space
+        # We can only migrate asset mappings for 3D models that are already migrated to data modeling.
+        self._3D_io = ThreeDMigrationIO(client, data_model_type="data modeling")
+
+    def as_id(self, item: AssetMappingResponse) -> str:
+        return f"AssetMapping_{item.model_id!s}_{item.revision_id!s}_{item.asset_id!s}"
+
+    def stream_data(self, selector: ThreeDSelector, limit: int | None = None) -> Iterable[Page[AssetMappingResponse]]:
+        total = 0
+        for three_d_page in self._3D_io.stream_data(selector, None):
+            for model in three_d_page.items:
+                if model.last_revision_info is None or model.last_revision_info.revision_id is None:
+                    # No revisions, so no asset mappings to
+                    continue
+                cursor: str | None = None
+                while True:
+                    request_limit = (
+                        min(self.DOWNLOAD_LIMIT, limit - total) if limit is not None else self.DOWNLOAD_LIMIT
+                    )
+                    if limit is not None and total >= limit:
+                        return
+                    response = self.client.tool.three_d.asset_mappings.iterate(
+                        model_id=model.id,
+                        revision_id=model.last_revision_info.revision_id,
+                        cursor=cursor,
+                        limit=request_limit,
+                    )
+                    items = response.items
+                    total += len(items)
+                    if items:
+                        yield Page(worker_id="main", items=items, next_cursor=response.next_cursor)
+                    if response.next_cursor is None:
+                        break
+                    cursor = response.next_cursor
+
+    def count(self, selector: ThreeDSelector) -> int | None:
+        # There is no efficient way to count 3D asset mappings in CDF.
+        return None
+
+    def upload_items(
+        self,
+        data_chunk: Sequence[UploadItem[AssetMappingDMRequest]],
+        http_client: HTTPClient,
+        selector: T_Selector | None = None,
+    ) -> Sequence[HTTPMessage]:
+        """Migrate 3D asset mappings by uploading them to the migrate/asset-mappings endpoint."""
+        if not data_chunk:
+            return []
+        # Assume all items in the chunk belong to the same model and revision, they should
+        # if the .stream_data method is used for downloading.
+        first = data_chunk[0]
+        model_id = first.item.model_id
+        revision_id = first.item.revision_id
+        endpoint = self.UPLOAD_ENDPOINT.format(modelId=model_id, revisionId=revision_id)
+        responses = http_client.request_with_retries(
+            ItemsRequest(
+                endpoint_url=self.client.config.create_api_url(endpoint),
+                method="POST",
+                items=list(data_chunk),
+                extra_body_fields={
+                    "dmsContextualizationConfig": {
+                        "object3DSpace": self.object_3D_space,
+                        "cadNodeSpace": self.cad_node_space,
+                    }
+                },
+            )
+        )
+        return responses
+
+    def json_to_resource(self, item_json: dict[str, JsonVal]) -> AssetMappingDMRequest:
+        raise NotImplementedError("Deserializing 3D Asset Mappings from JSON is not supported.")
+
+    def data_to_json_chunk(
+        self, data_chunk: Sequence[AssetMappingResponse], selector: ThreeDSelector | None = None
+    ) -> list[dict[str, JsonVal]]:
+        raise NotImplementedError("Serializing 3D Asset Mappings to JSON is not supported.")

cognite_toolkit/_cdf_tk/commands/_profile.py

@@ -21,7 +21,7 @@ from rich.spinner import Spinner
 from rich.table import Table
 
 from cognite_toolkit._cdf_tk.client import ToolkitClient
-from cognite_toolkit._cdf_tk.client.data_classes.raw import RawProfileResults, RawTable
+from cognite_toolkit._cdf_tk.client.data_classes.legacy.raw import RawProfileResults, RawTable
 from cognite_toolkit._cdf_tk.constants import MAX_ROW_ITERATION_RUN_QUERY
 from cognite_toolkit._cdf_tk.exceptions import ToolkitMissingDependencyError, ToolkitThrottledError, ToolkitValueError
 from cognite_toolkit._cdf_tk.utils.aggregators import (

cognite_toolkit/_cdf_tk/commands/_purge.py

@@ -16,6 +16,7 @@ from rich.console import Console
 from rich.panel import Panel
 
 from cognite_toolkit._cdf_tk.client import ToolkitClient
+from cognite_toolkit._cdf_tk.client.data_classes.instance_api import TypedInstanceIdentifier
 from cognite_toolkit._cdf_tk.cruds import (
     AssetCRUD,
     ContainerCRUD,
@@ -59,10 +60,10 @@ from cognite_toolkit._cdf_tk.utils.aggregators import (
     TimeSeriesAggregator,
 )
 from cognite_toolkit._cdf_tk.utils.http_client import (
-    FailedRequestItems,
-    FailedResponseItems,
     HTTPClient,
     ItemsRequest,
+    ItemsRequest2,
+    ItemsSuccessResponse2,
     SuccessResponseItems,
 )
 from cognite_toolkit._cdf_tk.utils.producer_worker import ProducerWorkerExecutor
@@ -714,21 +715,18 @@ class PurgeCommand(ToolkitCommand):
             results.deleted += len(items)
             return
 
-        responses = delete_client.request_with_retries(
-            ItemsRequest(
-                delete_client.config.create_api_url("/models/instances/delete"),
+        responses = delete_client.request_items_retries(
+            ItemsRequest2(
+                endpoint_url=delete_client.config.create_api_url("/models/instances/delete"),
                 method="POST",
-                # MyPy does not understand that InstanceId.load handles dict[str, JsonVal]
-                items=[DeleteItem(item=item, as_id_fun=InstanceId.load) for item in items],  # type: ignore[arg-type]
+                items=[TypedInstanceIdentifier._load(item) for item in items],
             )
         )
         for response in responses:
-            if isinstance(response, SuccessResponseItems):
+            if isinstance(response, ItemsSuccessResponse2):
                 results.deleted += len(response.ids)
-            elif isinstance(response, FailedResponseItems | FailedRequestItems):
-                results.failed += len(response.ids)
             else:
-                results.failed += len(items)
+                results.failed += len(response.ids)
 
     @staticmethod
     def _unlink_timeseries(

cognite_toolkit/_cdf_tk/commands/build_cmd.py

@@ -14,7 +14,7 @@ from rich.progress import track
 from cognite_toolkit._cdf_tk.builders import Builder, create_builder
 from cognite_toolkit._cdf_tk.cdf_toml import CDFToml
 from cognite_toolkit._cdf_tk.client import ToolkitClient
-from cognite_toolkit._cdf_tk.client.data_classes.raw import RawDatabase
+from cognite_toolkit._cdf_tk.client.data_classes.legacy.raw import RawDatabase
 from cognite_toolkit._cdf_tk.commands._base import ToolkitCommand
 from cognite_toolkit._cdf_tk.constants import (
     _RUNNING_IN_BROWSER,

cognite_toolkit/_cdf_tk/commands/build_v2/build_cmd.py

@@ -0,0 +1,241 @@
+from pathlib import Path
+from typing import Any, Literal, TypedDict
+
+from rich import print
+from rich.panel import Panel
+
+from cognite_toolkit._cdf_tk.client import ToolkitClient
+from cognite_toolkit._cdf_tk.commands._base import ToolkitCommand
+from cognite_toolkit._cdf_tk.commands.build_cmd import BuildCommand as OldBuildCommand
+from cognite_toolkit._cdf_tk.commands.build_v2.build_input import BuildInput
+from cognite_toolkit._cdf_tk.commands.build_v2.build_issues import BuildIssue, BuildIssueList
+from cognite_toolkit._cdf_tk.data_classes import (
+    BuildConfigYAML,
+    BuildVariables,
+    BuiltModuleList,
+    ModuleDirectories,
+)
+from cognite_toolkit._cdf_tk.exceptions import ToolkitError
+from cognite_toolkit._cdf_tk.hints import verify_module_directory
+from cognite_toolkit._cdf_tk.tk_warnings import ToolkitWarning, WarningList
+from cognite_toolkit._cdf_tk.utils.file import safe_rmtree
+from cognite_toolkit._cdf_tk.validation import validate_module_selection, validate_modules_variables
+from cognite_toolkit._version import __version__
+
+
+class BuildWarnings(TypedDict):
+    warning: ToolkitWarning
+    location: list[Path]
+
+
+class BuildCommand(ToolkitCommand):
+    def __init__(self, print_warning: bool = True, skip_tracking: bool = False, silent: bool = False) -> None:
+        super().__init__(print_warning, skip_tracking, silent)
+        self.issues = BuildIssueList()
+
+    def execute(
+        self,
+        verbose: bool,
+        organization_dir: Path,
+        build_dir: Path,
+        selected: list[str | Path] | None,
+        build_env_name: str | None,
+        no_clean: bool,
+        client: ToolkitClient | None = None,
+        on_error: Literal["continue", "raise"] = "continue",
+    ) -> BuiltModuleList:
+        """
+        Build the resources into deployable artifacts in the build directory.
+        """
+
+        self.verbose = verbose
+        self.on_error = on_error
+
+        # Tracking the project and cluster for the build.
+        if client:
+            self._additional_tracking_info.project = client.config.project
+            self._additional_tracking_info.cluster = client.config.cdf_cluster
+
+        # Setting the parameters for the build.
+        input = BuildInput.load(organization_dir, build_dir, build_env_name, client, selected)
+
+        # Print the build input.
+        if self.verbose:
+            self._print_build_input(input)
+
+        # Capture warnings from module structure integrity
+        if module_selection_issues := self._validate_modules(input):
+            self.issues.extend(module_selection_issues)
+
+        # Logistics: clean and create build directory
+        if prepare_issues := self._prepare_target_directory(input, not no_clean):
+            self.issues.extend(prepare_issues)
+
+        # Compile the configuration and variables,
+        # check syntax on module and resource level
+        # for any "compilation errors and warnings"
+        built_modules, build_integrity_issues = self._build_configuration(input)
+        if build_integrity_issues:
+            self.issues.extend(build_integrity_issues)
+
+        # This is where we would add any recommendations for the user to improve the build.
+        if build_quality_issues := self._verify_build_quality(built_modules):
+            self.issues.extend(build_quality_issues)
+
+        # Finally, print warnings grouped by category/code and location.
+        self._print_or_log_warnings_by_category(self.issues)
+
+        return built_modules
+
+    def _print_build_input(self, input: BuildInput) -> None:
+        print(
+            Panel(
+                f"Building {input.organization_dir!s}:\n  - Toolkit Version '{__version__!s}'\n"
+                f"  - Environment name {input.build_env_name!r}, validation-type {input.config.environment.validation_type!r}.\n"
+                f"  - Config '{input.config.filepath!s}'",
+                expand=False,
+            )
+        )
+
+    def _prepare_target_directory(self, input: BuildInput, clean: bool = False) -> BuildIssueList:
+        """
+        Directory logistics
+        """
+        issues = BuildIssueList()
+        if input.build_dir.exists() and any(input.build_dir.iterdir()):
+            if not clean:
+                raise ToolkitError("Build directory is not empty. Run without --no-clean to remove existing files.")
+
+            if self.verbose:
+                issues.append(BuildIssue(description=f"Build directory {input.build_dir!s} is not empty. Clearing."))
+            safe_rmtree(input.build_dir)
+        input.build_dir.mkdir(parents=True, exist_ok=True)
+        return issues
+
+    def _validate_modules(self, input: BuildInput) -> BuildIssueList:
+        issues = BuildIssueList()
+        # Verify that the modules exists, are not duplicates,
+        # and at least one is selected
+        verify_module_directory(input.organization_dir, input.build_env_name)
+
+        # Validate module selection
+        user_selected_modules = input.config.environment.get_selected_modules({})
+        module_warnings = validate_module_selection(
+            modules=input.modules,
+            config=input.config,
+            packages={},
+            selected_modules=user_selected_modules,
+            organization_dir=input.organization_dir,
+        )
+        if module_warnings:
+            issues.extend(BuildIssueList.from_warning_list(module_warnings))
+
+        # Validate variables. Note: this looks for non-replaced template
+        # variables <.*?> and can be improved in the future.
+        # Keeping for reference.
+        variables_warnings = validate_modules_variables(input.variables, input.config.filepath)
+        if variables_warnings:
+            issues.extend(BuildIssueList.from_warning_list(variables_warnings))
+
+        # Track LOC of managed configuration
+        # Note: _track is not implemented yet, so we skip it for now
+        # self._track(input)
+
+        return issues
+
+    def _build_configuration(self, input: BuildInput) -> tuple[BuiltModuleList, BuildIssueList]:
+        issues = BuildIssueList()
+        # Use input.modules.selected directly (it's already a ModuleDirectories)
+        if not input.modules.selected:
+            return BuiltModuleList(), issues
+
+        # first collect variables into practical lookup
+        # TODO: parallelism is not implemented yet. I'm sure there are optimizations to be had here, but we'll focus on process parallelism since we believe loading yaml and file i/O are the biggest bottlenecks.
+
+        old_build_command = OldBuildCommand(print_warning=False, skip_tracking=False)
+        built_modules = old_build_command.build_config(
+            build_dir=input.build_dir,
+            organization_dir=input.organization_dir,
+            config=input.config,
+            packages={},
+            clean=False,
+            verbose=self.verbose,
+            client=input.client,
+            progress_bar=False,
+            on_error=self.on_error,
+        )
+        # Copy tracking info from old command to self
+        self._additional_tracking_info.package_ids.update(old_build_command._additional_tracking_info.package_ids)
+        self._additional_tracking_info.module_ids.update(old_build_command._additional_tracking_info.module_ids)
+
+        # Collect warnings from the old build command and convert to issues
+        # Always convert warnings to issues, even if the list appears empty
+        # (WarningList might have custom __bool__ behavior)
+        if old_build_command.warning_list:
+            converted_issues = BuildIssueList.from_warning_list(old_build_command.warning_list)
+            issues.extend(converted_issues)
+        return built_modules, issues
+
+    def _verify_build_quality(self, built_modules: BuiltModuleList) -> BuildIssueList:
+        issues = BuildIssueList()
+        return issues
+
+    def _write(self, input: BuildInput) -> None:
+        # Write the build to the build directory.
+        # Track lines of code built.
+        raise NotImplementedError()
+
+    def _track(self, input: BuildInput) -> None:
+        raise NotImplementedError()
+
+    def _print_or_log_warnings_by_category(self, issues: BuildIssueList) -> None:
+        pass
+
+    # Delegate to old BuildCommand for backward compatibility with tests
+    def build_modules(
+        self,
+        modules: ModuleDirectories,
+        build_dir: Path,
+        variables: BuildVariables,
+        verbose: bool = False,
+        progress_bar: bool = False,
+        on_error: Literal["continue", "raise"] = "continue",
+    ) -> BuiltModuleList:
+        """Delegate to old BuildCommand for backward compatibility."""
+        old_cmd = OldBuildCommand()
+
+        built_modules = old_cmd.build_modules(modules, build_dir, variables, verbose, progress_bar, on_error)
+        self._additional_tracking_info.package_ids.update(old_cmd._additional_tracking_info.package_ids)
+        self._additional_tracking_info.module_ids.update(old_cmd._additional_tracking_info.module_ids)
+        self.issues.extend(BuildIssueList.from_warning_list(old_cmd.warning_list or WarningList[ToolkitWarning]()))
+        return built_modules
+
+    def build_config(
+        self,
+        build_dir: Path,
+        organization_dir: Path,
+        config: BuildConfigYAML,
+        packages: dict[str, list[str]],
+        clean: bool = False,
+        verbose: bool = False,
+        client: ToolkitClient | None = None,
+        progress_bar: bool = False,
+        on_error: Literal["continue", "raise"] = "continue",
+    ) -> BuiltModuleList:
+        """Delegate to old BuildCommand for backward compatibility."""
+        old_cmd = OldBuildCommand()
+        return old_cmd.build_config(
+            build_dir, organization_dir, config, packages, clean, verbose, client, progress_bar, on_error
+        )
+
+    def _replace_variables(
+        self,
+        resource_files: list[Path],
+        variables: BuildVariables,
+        resource_name: str,
+        module_dir: Path,
+        verbose: bool = False,
+    ) -> list[Any]:
+        """Delegate to old BuildCommand for backward compatibility."""
+        old_cmd = OldBuildCommand()
+        return old_cmd._replace_variables(resource_files, variables, resource_name, module_dir, verbose)

cognite_toolkit/_cdf_tk/commands/build_v2/build_input.py

@@ -0,0 +1,85 @@
+import sys
+from functools import cached_property
+from pathlib import Path
+
+if sys.version_info >= (3, 11):
+    from typing import Self
+else:
+    from typing_extensions import Self
+
+from pydantic import BaseModel, ConfigDict
+
+from cognite_toolkit._cdf_tk.client import ToolkitClient
+from cognite_toolkit._cdf_tk.constants import DEFAULT_ENV
+from cognite_toolkit._cdf_tk.data_classes import (
+    BuildConfigYAML,
+    BuildVariables,
+    ModuleDirectories,
+)
+from cognite_toolkit._cdf_tk.tk_warnings import ToolkitWarning, WarningList
+from cognite_toolkit._cdf_tk.utils.modules import parse_user_selected_modules
+
+
+class BuildInput(BaseModel):
+    """Input to the build process."""
+
+    # need this until we turn BuildConfigYaml and ToolkitClient into Pydantic models
+    model_config = ConfigDict(frozen=True, arbitrary_types_allowed=True)
+
+    organization_dir: Path
+    build_dir: Path
+    build_env_name: str
+    config: BuildConfigYAML
+    client: ToolkitClient | None = None
+    selected: list[str | Path] | None = None
+    warnings: WarningList[ToolkitWarning] | None = None
+
+    @classmethod
+    def load(
+        cls,
+        organization_dir: Path,
+        build_dir: Path,
+        build_env_name: str | None,
+        client: ToolkitClient | None,
+        selected: list[str | Path] | None = None,
+    ) -> Self:
+        resolved_org_dir = Path.cwd() if organization_dir in {Path("."), Path("./")} else organization_dir
+        resolved_env = build_env_name or DEFAULT_ENV
+        config, warnings = cls._load_config(resolved_org_dir, resolved_env, selected)
+        return cls(
+            organization_dir=resolved_org_dir,
+            build_dir=build_dir,
+            build_env_name=resolved_env,
+            config=config,
+            client=client,
+            selected=selected,
+            warnings=warnings,
+        )
+
+    @classmethod
+    def _load_config(
+        cls, organization_dir: Path, build_env_name: str, selected: list[str | Path] | None
+    ) -> tuple[BuildConfigYAML, WarningList[ToolkitWarning]]:
+        warnings: WarningList[ToolkitWarning] = WarningList[ToolkitWarning]()
+        if (organization_dir / BuildConfigYAML.get_filename(build_env_name or DEFAULT_ENV)).exists():
+            config = BuildConfigYAML.load_from_directory(organization_dir, build_env_name or DEFAULT_ENV)
+        else:
+            # Loads the default environment
+            config = BuildConfigYAML.load_default(organization_dir)
+        if selected:
+            config.environment.selected = parse_user_selected_modules(selected, organization_dir)
+        config.set_environment_variables()
+        if environment_warning := config.validate_environment():
+            warnings.append(environment_warning)
+        return config, warnings
+
+    @cached_property
+    def modules(self) -> ModuleDirectories:
+        user_selected_modules = self.config.environment.get_selected_modules({})
+        return ModuleDirectories.load(self.organization_dir, user_selected_modules)
+
+    @cached_property
+    def variables(self) -> BuildVariables:
+        return BuildVariables.load_raw(
+            self.config.variables, self.modules.available_paths, self.modules.selected.available_paths
+        )

cognite_toolkit/_cdf_tk/commands/build_v2/build_issues.py

@@ -0,0 +1,27 @@
+import sys
+
+if sys.version_info >= (3, 11):
+    from typing import Self
+else:
+    from typing_extensions import Self
+
+from collections import UserList
+
+from pydantic import BaseModel
+
+from cognite_toolkit._cdf_tk.tk_warnings import ToolkitWarning, WarningList
+
+
+class BuildIssue(BaseModel):
+    """Issue with the build. Can have a recommendation for the user to improve the build."""
+
+    description: str
+
+
+class BuildIssueList(UserList[BuildIssue]):
+    """List of build issues."""
+
+    @classmethod
+    def from_warning_list(cls, warning_list: WarningList[ToolkitWarning]) -> Self:
+        """Create a BuildIssueList from a WarningList."""
+        return cls([BuildIssue(description=warning.get_message()) for warning in warning_list])