Make imports on dvc.repo and dvc.repo.plots lazy (#5227)

dvc/repo/__init__.py

@@ -6,12 +6,10 @@ from typing import TYPE_CHECKING
 
 from funcy import cached_property, cat
 
-from dvc.config import Config
 from dvc.exceptions import FileMissingError
 from dvc.exceptions import IsADirectoryError as DvcIsADirectoryError
 from dvc.exceptions import NotDvcRepoError, OutputNotFoundError
 from dvc.path_info import PathInfo
-from dvc.scm.base import SCMError
 from dvc.utils.fs import path_isin
 
 from .graph import build_graph, build_outs_graph, get_pipelines
@@ -94,6 +92,7 @@ class Repo:
         assert bool(scm) == bool(rev)
 
         from dvc.scm import SCM
+        from dvc.scm.base import SCMError
         from dvc.scm.git import Git
         from dvc.utils.fs import makedirs
 
@@ -131,6 +130,7 @@ class Repo:
         config=None,
     ):
         from dvc.cache import Cache
+        from dvc.config import Config
         from dvc.data_cloud import DataCloud
         from dvc.lock import LockNoop, make_lock
         from dvc.repo.live import Live

dvc/repo/fetch.py

@@ -2,7 +2,6 @@ import logging
 
 from dvc.config import NoRemoteError
 from dvc.exceptions import DownloadError
-from dvc.scm.base import CloneError
 
 from . import locked
 
@@ -81,6 +80,7 @@ def fetch(
 
 def _fetch_external(self, repo_url, repo_rev, files, jobs):
     from dvc.external_repo import external_repo
+    from dvc.scm.base import CloneError
 
     failed, downloaded = 0, 0
     cache = self.cache.local

dvc/repo/get.py

@@ -1,8 +1,6 @@
 import logging
 import os
 
-import shortuuid
-
 from dvc.exceptions import DvcException
 from dvc.utils import resolve_output
 from dvc.utils.fs import remove
@@ -19,6 +17,8 @@ class GetDVCFileError(DvcException):
 
 
 def get(url, path, out=None, rev=None):
+    import shortuuid
+
     from dvc.dvcfile import is_valid_filename
     from dvc.external_repo import external_repo
 

dvc/repo/live.py

@@ -1,14 +1,18 @@
 import contextlib
 import os
-from typing import List
+from typing import TYPE_CHECKING, List, Optional
 
 from dvc.exceptions import MetricDoesNotExistError, MetricsError
-from dvc.output import BaseOutput
-from dvc.path_info import PathInfo
-from dvc.repo import Repo
+
+if TYPE_CHECKING:
+    from dvc.output import BaseOutput
+    from dvc.path_info import PathInfo
+    from dvc.repo import Repo
 
 
-def summary_path_info(out: BaseOutput) -> PathInfo:
+def summary_path_info(out: "BaseOutput") -> Optional["PathInfo"]:
+    from dvc.output import BaseOutput
+
     assert out.live
     has_summary = True
     if isinstance(out.live, dict):
@@ -19,7 +23,7 @@ def summary_path_info(out: BaseOutput) -> PathInfo:
 
 
 class Live:
-    def __init__(self, repo: Repo):
+    def __init__(self, repo: "Repo"):
         self.repo = repo
 
     def show(self, target: str, revs: List[str] = None):

dvc/repo/plots/__init__.py

@@ -9,8 +9,6 @@ from dvc.exceptions import (
     NoMetricsFoundError,
     NoMetricsParsedError,
 )
-from dvc.repo.collect import collect
-from dvc.repo.plots.data import PlotParsingError
 from dvc.utils import relpath
 
 if TYPE_CHECKING:
@@ -95,6 +93,8 @@ class Plots:
     @staticmethod
     def render(data, revs=None, props=None, templates=None):
         """Renders plots"""
+        from dvc.repo.plots.data import PlotParsingError
+
         props = props or {}
 
         # Merge data by plot file and apply overriding props
@@ -213,6 +213,8 @@ def _collect_plots(
     rev: str = None,
     recursive: bool = False,
 ) -> Dict[str, Dict]:
+    from dvc.repo.collect import collect
+
     plots, path_infos = collect(
         repo,
         output_filter=_is_plot,

dvc/repo/run.py

@@ -3,14 +3,6 @@ from contextlib import suppress
 
 from funcy import concat, first, lfilter, without
 
-from dvc.exceptions import InvalidArgumentError
-from dvc.stage.exceptions import (
-    DuplicateStageName,
-    InvalidStageName,
-    StageFileAlreadyExistsError,
-)
-
-from ..exceptions import OutputDuplicationError
 from . import locked
 from .scm_context import scm_context
 
@@ -61,6 +53,10 @@ def _get_file_path(kwargs):
 
 def _check_stage_exists(dvcfile, stage):
     from dvc.stage import PipelineStage
+    from dvc.stage.exceptions import (
+        DuplicateStageName,
+        StageFileAlreadyExistsError,
+    )
 
     if not dvcfile.exists():
         return
@@ -80,7 +76,9 @@ def _check_stage_exists(dvcfile, stage):
 @scm_context
 def run(self, fname=None, no_exec=False, single_stage=False, **kwargs):
     from dvc.dvcfile import PIPELINE_FILE, Dvcfile
+    from dvc.exceptions import InvalidArgumentError, OutputDuplicationError
     from dvc.stage import PipelineStage, Stage, create_stage, restore_meta
+    from dvc.stage.exceptions import InvalidStageName
 
     if not kwargs.get("cmd"):
         raise InvalidArgumentError("command is not specified")

dvc/repo/stage.py

@@ -5,15 +5,12 @@ import typing
 from contextlib import suppress
 from typing import Iterable, List, NamedTuple, Optional, Set, Tuple
 
-from dvc.dvcfile import PIPELINE_FILE, Dvcfile, is_valid_filename
 from dvc.exceptions import (
     DvcException,
     NoOutputOrStageError,
     OutputNotFoundError,
 )
 from dvc.path_info import PathInfo
-from dvc.repo.graph import collect_inside_path, collect_pipeline
-from dvc.stage.exceptions import StageFileDoesNotExistError, StageNotFound
 from dvc.utils import parse_target, relpath
 
 logger = logging.getLogger(__name__)
@@ -26,6 +23,8 @@ if typing.TYPE_CHECKING:
     from dvc.stage.loader import StageLoader
     from dvc.types import OptStr
 
+PIPELINE_FILE = "dvc.yaml"
+
 
 class StageInfo(NamedTuple):
     stage: "Stage"
@@ -38,6 +37,8 @@ StageSet = Set["Stage"]
 
 
 def _collect_with_deps(stages: StageList, graph: "DiGraph") -> StageSet:
+    from dvc.repo.graph import collect_pipeline
+
     res: StageSet = set()
     for stage in stages:
         res.update(collect_pipeline(stage, graph=graph))
@@ -47,6 +48,7 @@ def _collect_with_deps(stages: StageList, graph: "DiGraph") -> StageSet:
 def _maybe_collect_from_dvc_yaml(
     loader: "StageLoad", target, with_deps: bool, **load_kwargs,
 ) -> StageIter:
+    from dvc.stage.exceptions import StageNotFound
 
     stages: StageList = []
     if loader.tree.exists(PIPELINE_FILE):
@@ -62,6 +64,8 @@ def _collect_specific_target(
     recursive: bool,
     accept_group: bool,
 ) -> Tuple[StageIter, "OptStr", "OptStr"]:
+    from dvc.dvcfile import is_valid_filename
+
     # Optimization: do not collect the graph for a specific target
     file, name = parse_target(target)
 
@@ -163,6 +167,7 @@ class StageLoad:
             glob: if true, `name` is considered as a glob, which is
                 used to filter list of stages from the given `path`.
         """
+        from dvc.dvcfile import Dvcfile
         from dvc.stage.loader import SingleStageLoader, StageLoader
 
         path = self._get_filepath(path, name)
@@ -185,6 +190,8 @@ class StageLoad:
             path: if not provided, default `dvc.yaml` is assumed.
             name: required for `dvc.yaml` files, ignored for `.dvc` files.
         """
+        from dvc.dvcfile import Dvcfile
+
         path = self._get_filepath(path, name)
         dvcfile = Dvcfile(self.repo, path)
 
@@ -251,6 +258,8 @@ class StageLoad:
             return list(graph) if graph else self.repo.stages
 
         if recursive and self.repo.tree.isdir(target):
+            from dvc.repo.graph import collect_inside_path
+
             path = os.path.abspath(target)
             return collect_inside_path(path, graph or self.graph)
 
@@ -301,6 +310,12 @@ class StageLoad:
                 except OutputNotFoundError:
                     pass
 
+            from dvc.dvcfile import is_valid_filename
+            from dvc.stage.exceptions import (
+                StageFileDoesNotExistError,
+                StageNotFound,
+            )
+
             try:
                 stages = self.collect(
                     target,
@@ -331,6 +346,8 @@ class StageLoad:
                 Otherwise, the file will be skipped, and dvc will partially
                 load the stages.
         """
+        from dvc.dvcfile import is_valid_filename
+
         stages = []
         outs = set()
         for root, dirs, files in self.tree.walk(self.repo.root_dir):