tests: ssh hardlink optimization

dvc/path_info.py

@@ -7,7 +7,7 @@ import posixpath
 from funcy import cached_property
 
 from dvc.utils.compat import str, builtin_str, basestring, is_py2
-from dvc.utils.compat import pathlib, urlparse
+from dvc.utils.compat import pathlib, urlparse, urljoin
 
 
 # On Python 2.7/Windows sys.getfilesystemencoding() is set to mbcs,

dvc/remote/ssh/__init__.py

@@ -193,7 +193,9 @@ class RemoteSSH(RemoteBASE):
 
         # See dvc/remote/local/__init__.py - hardlink()
         if self.getsize(from_info) == 0:
-            self.open(to_info, "w").close()
+
+            with self.ssh(to_info) as ssh:
+                ssh.sftp.open(to_info.path, "w").close()
 
             logger.debug(
                 "Created empty file: {src} -> {dest}".format(

tests/conftest.py

@@ -177,3 +177,13 @@ def _close_pools():
 
     yield
     close_pools()
+
+
+@pytest.fixture
+def chdir_tmp(tmp_path):
+    cwd = os.getcwd()
+    try:
+        os.chdir(tmp_path)
+        yield
+    finally:
+        os.chdir(cwd)

tests/unit/remote/ssh/test_connection.py

@@ -9,15 +9,6 @@ import filecmp
 from dvc.system import System
 
 
-@pytest.fixture
-def chdir_tmp(tmp_path):
-    cwd = os.getcwd()
-    try:
-        os.chdir(tmp_path)
-        yield
-    finally:
-        os.chdir(cwd)
-
 @pytest.fixture
 def foo():
     with open("foo", "w") as fobj:

tests/unit/remote/ssh/test_ssh.py

@@ -1,6 +1,7 @@
 import getpass
 import os
 import sys
+import posixpath
 
 from unittest import TestCase
 
@@ -8,6 +9,8 @@ from mock import patch, mock_open
 import pytest
 
 from dvc.remote.ssh import RemoteSSH
+from dvc.path_info import URLInfo
+from dvc.system import System
 
 
 class TestRemoteSSH(TestCase):
@@ -182,3 +185,23 @@ def test_ssh_gss_auth(mock_file, mock_exists, config, expected_gss_auth):
     mock_exists.assert_called_with(RemoteSSH.ssh_config_filename())
     mock_file.assert_called_with(RemoteSSH.ssh_config_filename())
     assert remote.gss_auth == expected_gss_auth
+
+def test_hardlink_optimization(chdir_tmp, ssh_server):
+    config = {
+        "url": "ssh://localhost",
+        "hostname": ssh_server.test_creds["host"],
+        "port": ssh_server.test_creds["port"],
+        "user": ssh_server.test_creds["username"],
+        "keyfile": ssh_server.test_creds["key_filename"],
+    }
+    remote = RemoteSSH(None, config)
+
+    open("empty", "w").close()
+
+    def urlpath(path):
+        netloc = remote.path_info.netloc
+        path = posixpath.abspath(path)
+        return URLInfo("ssh://{}/{}".format(netloc, path))
+
+    remote.hardlink(urlpath("empty"), urlpath("bar"))
+    assert not System.is_hardlink("bar")