Merge pull request #3225 from pared/2647_no_cache_external_dir

remote: verify cache exists on dir checksum

dvc/exceptions.py

@@ -311,3 +311,18 @@ class PathMissingError(DvcException):
             " neither as an output nor a git-handled file."
         )
         super().__init__(msg.format(path, repo))
+
+
+class RemoteCacheRequiredError(DvcException):
+    def __init__(self, path_info):
+        super().__init__(
+            (
+                "Current operation was unsuccessful because '{}' requires "
+                "existing cache on '{}' remote. See {} for information on how "
+                "to set up remote cache."
+            ).format(
+                path_info,
+                path_info.scheme,
+                format_link("https://man.dvc.org/config#cache"),
+            )
+        )

dvc/output/base.py

@@ -6,7 +6,7 @@ from voluptuous import Any
 
 import dvc.prompt as prompt
 from dvc.cache import NamedCache
-from dvc.exceptions import CollectCacheError
+from dvc.exceptions import CollectCacheError, RemoteCacheRequiredError
 from dvc.exceptions import DvcException
 from dvc.remote.base import RemoteBASE
 
@@ -98,14 +98,9 @@ class OutputBase(object):
         self.persist = persist
         self.tags = None if self.IS_DEPENDENCY else (tags or {})
 
-        if self.use_cache and self.cache is None:
-            raise DvcException(
-                "no cache location setup for '{}' outputs.".format(
-                    self.REMOTE.scheme
-                )
-            )
-
         self.path_info = self._parse_path(remote, path)
+        if self.use_cache and self.cache is None:
+            raise RemoteCacheRequiredError(self.path_info)
 
     def _parse_path(self, remote, path):
         if remote:

dvc/remote/base.py

@@ -18,6 +18,7 @@ from dvc.exceptions import (
     DvcException,
     ConfirmRemoveError,
     DvcIgnoreInCollectedDirError,
+    RemoteCacheRequiredError,
 )
 from dvc.ignore import DvcIgnore
 from dvc.path_info import PathInfo, URLInfo
@@ -233,6 +234,9 @@ class RemoteBASE(object):
         return sorted(result, key=itemgetter(self.PARAM_RELPATH))
 
     def get_dir_checksum(self, path_info):
+        if not self.cache:
+            raise RemoteCacheRequiredError(path_info)
+
         dir_info = self._collect_dir(path_info)
         checksum, tmp_info = self._get_dir_info_checksum(dir_info)
         new_info = self.cache.checksum_to_path_info(checksum)

tests/func/test_remote.py

@@ -11,7 +11,7 @@ from dvc.exceptions import DownloadError, UploadError
 from dvc.main import main
 from dvc.path_info import PathInfo
 from dvc.remote import RemoteLOCAL, RemoteConfig
-from dvc.remote.base import RemoteBASE
+from dvc.remote.base import RemoteBASE, RemoteCacheRequiredError
 from dvc.compat import fspath
 from tests.basic_env import TestDvc
 from tests.remotes import Local
@@ -257,3 +257,14 @@ def test_modify_missing_remote(dvc):
 
     with pytest.raises(ConfigError, match=r"Unable to find remote section"):
         remote_config.modify("myremote", "gdrive_client_id", "xxx")
+
+
+def test_external_dir_resource_on_no_cache(tmp_dir, dvc, tmp_path_factory):
+    # https://github.com/iterative/dvc/issues/2647, is some situations
+    # (external dir dependency) cache is required to calculate dir md5
+    external_dir = tmp_path_factory.mktemp("external_dir")
+    (external_dir / "file").write_text("content")
+
+    dvc.cache.local = None
+    with pytest.raises(RemoteCacheRequiredError):
+        dvc.run(deps=[fspath(external_dir)])