dvc: raise when Repo.open() fails

dvc/exceptions.py

@@ -263,3 +263,10 @@ def __init__(self, etag, cached_etag):
             "ETag mismatch detected when copying file to cache! "
             "(expected: '{}', actual: '{}')".format(etag, cached_etag)
         )
+
+
+class OutputFileMissingError(DvcException):
+    def __init__(self, path):
+        super(OutputFileMissingError, self).__init__(
+            "Can't find {} neither locally nor on remote".format(path)
+        )

dvc/repo/__init__.py

@@ -8,10 +8,11 @@
     NotDvcRepoError,
     OutputNotFoundError,
     TargetNotDirectoryError,
+    OutputFileMissingError,
 )
 from dvc.ignore import DvcIgnoreFileHandler
 from dvc.path_info import PathInfo
-from dvc.utils.compat import open as _open
+from dvc.utils.compat import open as _open, fspath_py35
 from dvc.utils import relpath
 
 logger = logging.getLogger(__name__)
@@ -455,9 +456,15 @@ def open(self, path, remote=None, mode="r", encoding=None):
         if out.isdir():
             raise ValueError("Can't open a dir")
 
+        cache_file = self.cache.local.checksum_to_path_info(out.checksum)
+        cache_file = fspath_py35(cache_file)
+
         with self.state:
             cache_info = out.get_used_cache(remote=remote)
             self.cloud.pull(cache_info, remote=remote)
 
-        cache_file = self.cache.local.checksum_to_path_info(out.checksum)
-        return _open(cache_file.fspath, mode=mode, encoding=encoding)
+        # Since pull may just skip with a warning, we need to check it here
+        if not os.path.exists(cache_file):
+            raise OutputFileMissingError(relpath(path, self.root_dir))
+
+        return _open(cache_file, mode=mode, encoding=encoding)

tests/func/test_api.py

@@ -2,6 +2,7 @@
 import shutil
 
 from dvc import api
+from dvc.exceptions import OutputFileMissingError
 from dvc.main import main
 from dvc.path_info import URLInfo
 from dvc.remote.config import RemoteConfig
@@ -113,6 +114,15 @@ def test_open_external(repo_dir, dvc_repo, erepo, remote_url):
         assert fd.read() == repo_dir.FOO_CONTENTS
 
 
+def test_open_missing(erepo):
+    # Remove cache to make foo missing
+    shutil.rmtree(erepo.dvc.cache.local.cache_dir)
+
+    repo_url = "file://" + erepo.dvc.root_dir
+    with pytest.raises(OutputFileMissingError):
+        api.read(erepo.FOO, repo=repo_url)
+
+
 def _set_remote_url_and_commit(repo, remote_url):
     rconfig = RemoteConfig(repo.config)
     rconfig.modify("upstream", "url", remote_url)