gc: take into account local cache

Signed-off-by: Ruslan Kuprieiev <[email protected]>

dvc/remote/azure.py

@@ -17,6 +17,7 @@ from dvc.logger import Logger
 from dvc.progress import progress
 from dvc.config import Config
 from dvc.remote.base import RemoteBase
+from dvc.remote.local import RemoteLOCAL
 
 
 class Callback(object):
@@ -212,6 +213,7 @@ class RemoteAzure(RemoteBase):
 
     def gc(self, checksum_infos):
         used_etags = [info[self.PARAM_ETAG] for info in checksum_infos['azure']]
+        used_etags += [info[RemoteLOCAL.PARAM_MD5] for info in checksum_infos['local']]
 
         all_blobs = self.blob_service.list_blobs(self.bucket)
 

dvc/remote/gs.py

@@ -13,6 +13,7 @@ except ImportError:
 
 from dvc.logger import Logger
 from dvc.remote.base import RemoteBase
+from dvc.remote.local import RemoteLOCAL
 from dvc.config import Config
 from dvc.progress import progress
 
@@ -190,6 +191,7 @@ class RemoteGS(RemoteBase):
 
     def gc(self, checksum_infos):
         used_etags = [info[self.PARAM_ETAG] for info in checksum_infos['gs']]
+        used_etags += [info[RemoteLOCAL.PARAM_MD5] for info in checksum_infos['local']]
 
         for etag in self._all_etags():
             if etag in used_etags:

dvc/remote/hdfs.py

@@ -11,6 +11,7 @@ except ImportError:
 
 from dvc.config import Config
 from dvc.remote.base import RemoteBase
+from dvc.remote.local import RemoteLOCAL
 from dvc.exceptions import DvcException
 from dvc.logger import Logger
 
@@ -174,6 +175,7 @@ class RemoteHDFS(RemoteBase):
 
     def gc(self, checksum_infos):
         used_checksums = [info[self.PARAM_CHECKSUM] for info in checksum_infos['hdfs']]
+        used_checksums += [info[RemoteLOCAL.PARAM_MD5] for info in checksum_infos['local']]
 
         for checksum in self._all_checksums():
             if checksum in used_checksums:

dvc/remote/s3.py

@@ -17,6 +17,7 @@ from dvc.logger import Logger
 from dvc.progress import progress
 from dvc.config import Config
 from dvc.remote.base import RemoteBase
+from dvc.remote.local import RemoteLOCAL
 
 
 class Callback(object):
@@ -230,6 +231,7 @@ class RemoteS3(RemoteBase):
 
     def gc(self, checksum_infos):
         used_etags = [info[self.PARAM_ETAG] for info in checksum_infos['s3']]
+        used_etags += [info[RemoteLOCAL.PARAM_MD5] for info in checksum_infos['local']]
 
         for etag in self._all():
             if etag in used_etags:

dvc/remote/ssh.py

@@ -13,6 +13,7 @@ from dvc.logger import Logger
 from dvc.utils import copyfile, file_md5
 from dvc.progress import progress
 from dvc.remote.base import RemoteBase
+from dvc.remote.local import RemoteLOCAL
 from dvc.config import Config
 from dvc.exceptions import DvcException
 
@@ -256,6 +257,7 @@ class RemoteSSH(RemoteBase):
 
     def gc(self, checksum_infos):
         used_md5s = [info[self.PARAM_MD5] for info in checksum_infos['ssh']]
+        used_md5s += [info[RemoteLOCAL.PARAM_MD5] for info in checksum_infos['local']]
 
         for md5 in self._all_md5s():
             if md5 in used_md5s: