From b8cf72cd0db0b1077dfd122717645ea5a1935739 Mon Sep 17 00:00:00 2001
From: Saugat Pachhai <suagatchhetri@outlook.com>
Date: Mon, 2 Mar 2020 13:09:18 +0545
Subject: [PATCH] add: do not verify hardlink if file is empty

Fixes #3390
---
 dvc/remote/local.py    |  6 ++++++
 tests/func/test_add.py | 38 +++++++++++++++++++++++++++++++++++++-
 2 files changed, 43 insertions(+), 1 deletion(-)

diff --git a/dvc/remote/local.py b/dvc/remote/local.py
index 24725c526d..3e62e0628e 100644
--- a/dvc/remote/local.py
+++ b/dvc/remote/local.py
@@ -92,6 +92,12 @@ def already_cached(self, path_info):
 
         return not self.changed_cache(current_md5)
 
+    def _verify_link(self, path_info, link_type):
+        if link_type == "hardlink" and self.getsize(path_info) == 0:
+            return
+
+        super()._verify_link(path_info, link_type)
+
     def is_empty(self, path_info):
         path = path_info.fspath
 
diff --git a/tests/func/test_add.py b/tests/func/test_add.py
index 95407ee7d0..e788c176c3 100644
--- a/tests/func/test_add.py
+++ b/tests/func/test_add.py
@@ -7,7 +7,7 @@
 
 import colorama
 import pytest
-from mock import patch
+from mock import patch, call
 
 import dvc as dvc_module
 from dvc.cache import Cache
@@ -662,3 +662,39 @@ def test_not_raises_on_re_add(tmp_dir, dvc):
 
     tmp_dir.gen({"file2": "file2 content", "file": "modified file"})
     dvc.add(["file2", "file"])
+
+
+@pytest.mark.parametrize("link", ["hardlink", "symlink", "copy"])
+def test_add_empty_files(tmp_dir, dvc, link):
+    file = "foo"
+    dvc.cache.local.cache_types = [link]
+    stages = tmp_dir.dvc_gen(file, "")
+
+    assert (tmp_dir / file).exists()
+    assert (tmp_dir / (file + Stage.STAGE_FILE_SUFFIX)).exists()
+    assert os.path.exists(stages[0].outs[0].cache_path)
+
+
+@mock.patch(
+    "dvc.remote.local.RemoteLOCAL.is_hardlink",
+    side_effect=RemoteLOCAL.is_hardlink,
+)
+def test_add_optimization_for_hardlink_on_empty_files(m, tmp_dir, dvc):
+    dvc.cache.local.cache_types = ["hardlink"]
+    tmp_dir.gen({"foo": "", "bar": "", "lorem": "lorem", "ipsum": "ipsum"})
+    stages = dvc.add(["foo", "bar", "lorem", "ipsum"])
+
+    m.assert_called_once()
+    assert m.call_args != call(tmp_dir / "foo")
+    assert m.call_args != call(tmp_dir / "bar")
+
+    for stage in stages[:2]:
+        # hardlinks are not created for empty files
+        assert not System.is_hardlink(stage.outs[0].path_info)
+
+    for stage in stages[2:]:
+        assert System.is_hardlink(stage.outs[0].path_info)
+
+    for stage in stages:
+        assert os.path.exists(stage.path)
+        assert os.path.exists(stage.outs[0].cache_path)