dvc checkout: remove untracked files

dvc/command/checkout.py

@@ -1,7 +1,38 @@
+import os
+
 from dvc.command.common.base import CmdBase
+from dvc.logger import Logger
+from dvc.system import System
 
 
 class CmdCheckout(CmdBase):
     def run(self):
+        self.remove_untracked_hardlinks()
         self.project.checkout()
         return 0
+
+    def remove_untracked_hardlinks(self):
+        untracked_cache_files = self.untracked_hardlinks_files()
+
+        for file in untracked_cache_files:
+            Logger.info(u'Remove \'{}\''.format(file))
+            os.remove(file)
+
+            dir = os.path.dirname(file)
+            if not os.listdir(dir):
+                Logger.info(u'Remove empty directory \'{}\''.format(dir))
+                os.removedirs(dir)
+        pass
+
+    def untracked_hardlinks_files(self):
+        untracked_files = set(self.project.scm.untracked_files())
+        cached_files = []
+        for cache_file in self.project.cache.all():
+            to_remove = set()
+            for file in untracked_files:
+                if System.samefile(cache_file, file):
+                    cached_files.append(file)
+                    to_remove.add(file)
+                untracked_files = untracked_files - to_remove
+
+        return cached_files

dvc/scm.py

@@ -38,6 +38,9 @@ class Base(object):
     def brancher(self, branch, new_branch):
         return BranchChanger(self, branch, new_branch)
 
+    def untracked_files(self):
+        pass
+
 
 class Git(Base):
     GITIGNORE = '.gitignore'
@@ -79,6 +82,9 @@ class Git(Base):
     def branch(self, branch):
         self.repo.git.branch(branch)
 
+    def untracked_files(self):
+        return self.repo.untracked_files
+
 
 def SCM(root_dir):
     if Git.is_repo(root_dir):