checkout: Remove selectively on directory files when changed (#1320)

Fix: #1220

dvc/project.py

@@ -465,12 +465,10 @@ class Project(object):
 
     def checkout(self, target=None, with_deps=False):
         all_stages = self.active_stages()
+        stages = all_stages
 
         if target:
-            stages = self._collect(target,
-                                   with_deps=with_deps)
-        else:
-            stages = all_stages
+            stages = self._collect(target, with_deps=with_deps)
 
         with self.state:
             self._cleanup_unused_links(all_stages)

dvc/remote/local.py

@@ -270,7 +270,7 @@ class RemoteLOCAL(RemoteBase):
 
     def checkout(self, path_info, checksum_info):
         path = path_info['path']
-        md5 = checksum_info.get(self.PARAM_MD5, None)
+        md5 = checksum_info.get(self.PARAM_MD5)
         cache = self.get(md5)
 
         if not cache:
@@ -289,15 +289,16 @@ class RemoteLOCAL(RemoteBase):
             remove(path)
             return
 
-        if os.path.exists(path):
-            msg = u'Data \'{}\' exists. Removing before checkout.'
-            Logger.warn(msg.format(os.path.relpath(path)))
-            remove(path)
-
         msg = u'Checking out \'{}\' with cache \'{}\'.'
         Logger.info(msg.format(os.path.relpath(path), md5))
 
         if not self.is_dir_cache(cache):
+
+            if os.path.exists(path):
+                msg = u'File \'{}\' exists. Removing before checkout.'
+                Logger.warn(msg.format(os.path.relpath(path)))
+                remove(path)
+
             self.link(cache, path)
             self.state.update_link(path)
             return
@@ -319,7 +320,19 @@ class RemoteLOCAL(RemoteBase):
             m = entry[self.PARAM_MD5]
             p = os.path.join(path, relpath)
             c = self.get(m)
-            self.link(c, p)
+
+            entry_info = {
+                'scheme': path_info['scheme'],
+                self.PARAM_PATH: p,
+            }
+
+            entry_checksum_info = {
+                self.PARAM_MD5: m,
+            }
+
+            if self.changed(entry_info, entry_checksum_info):
+                remove(p)
+                self.link(c, p)
 
             if bar:
                 progress.update_target(dir_relpath, processed, dir_size)

tests/test_checkout.py

@@ -4,9 +4,11 @@ import time
 import stat
 import shutil
 import filecmp
+import collections
 
 from dvc.main import main
 from dvc.project import Project
+from dvc.system import System
 from tests.basic_env import TestDvc
 from tests.test_repro import TestRepro
 from dvc.stage import Stage
@@ -116,6 +118,20 @@ class CheckoutBase(TestDvc):
     def read_ignored(self):
         return list(map(lambda s: s.strip('\n'), open(self.GIT_IGNORE).readlines()))
 
+    def outs_info(self, stage):
+        FileInfo = collections.namedtuple('FileInfo', 'path inode')
+
+        paths = [
+            os.path.join(root, file)
+            for output in stage.outs
+            for root, _, files in os.walk(output.path)
+            for file in files
+        ]
+
+        return [
+            FileInfo(path=path, inode=System.inode(path))
+            for path in paths
+        ]
 
 class TestRemoveFilesWhenCheckout(CheckoutBase):
     def test(self):
@@ -143,6 +159,29 @@ class TestRemoveFilesWhenCheckout(CheckoutBase):
         self.assertFalse(os.path.exists(fname))
 
 
+class TestCheckoutSelectiveRemove(CheckoutBase):
+    def test(self):
+        # Use copy to test for changes in the inodes
+        ret = main(['config', 'cache.type', 'copy'])
+        self.assertEqual(ret, 0)
+
+        stages = self.dvc.add(self.DATA_DIR)
+        self.assertEqual(len(stages), 1)
+        stage = stages[0]
+        staged_files = self.outs_info(stage)
+
+        os.remove(staged_files[0].path)
+        ret = main(['checkout', stage.relpath])
+        self.assertEqual(ret, 0)
+
+        checkedout_files = self.outs_info(stage)
+
+        self.assertEqual(len(staged_files), len(checkedout_files))
+        self.assertEqual(staged_files[0].path, checkedout_files[0].path)
+        self.assertNotEqual(staged_files[0].inode, checkedout_files[0].inode)
+        self.assertEqual(staged_files[1].inode, checkedout_files[1].inode)
+
+
 class TestGitIgnoreBasic(CheckoutBase):
     def test(self):
         fname1 = 'file_1'