Merge pull request iterative#576 from efiop/master

dvc: fix checkout/repro with stage files without md5

dvc/output.py

@@ -132,10 +132,16 @@ def __init__(self, project, path, md5=None, use_cache=True):
 
     @property
     def cache(self):
+        if not self.md5:
+            return None
+
         return self.project.cache.get(self.md5)
 
     @property
     def rel_cache(self):
+        if not self.cache:
+            return None
+
         return os.path.relpath(self.cache)
 
     def dumpd(self, cwd):
@@ -198,10 +204,10 @@ def dir_info_dict(dir_info):
         return {i['relpath']: i['md5'] for i in dir_info}
 
     def changed(self):
-        ret = True
-
         if not self.use_cache:
             ret = super(Output, self).changed()
+        elif not self.cache:
+            ret = True
         elif self.is_dir_cache(self.cache):
             ret = self._changed_dir()
         else:
@@ -280,7 +286,7 @@ def checkout(self):
             self.project.logger.debug(msg.format(self.rel_path, self.rel_cache))
             return
 
-        if not os.path.exists(self.cache):
+        if not self.cache or not os.path.exists(self.cache):
             self.project.logger.warn(u'\'{}\': cache file not found'.format(self.rel_path))
             self.remove()
             return

dvc/project.py

@@ -182,7 +182,7 @@ def _used_cache(self, target=None):
 
         for stage in stages:
             for out in stage.outs:
-                if not out.use_cache:
+                if not out.use_cache or not out.cache:
                     continue
                 cache_set |= set([out.cache])
                 if out.is_dir_cache(out.cache) and os.path.isfile(out.cache):

tests/test_checkout.py

@@ -1,4 +1,5 @@
 import os
+import yaml
 import time
 import stat
 import shutil
@@ -7,6 +8,8 @@
 from dvc.main import main
 from tests.basic_env import TestDvc
 from tests.test_repro import TestRepro
+from dvc.stage import Stage
+from dvc.output import Output
 
 
 class TestCheckout(TestRepro):
@@ -146,3 +149,17 @@ def test(self):
         main(['checkout'])
         ignored = self.read_ignored()
         self.assertIn(fname_branch, ignored)
+
+
+class TestCheckoutMissingMd5InStageFile(TestRepro):
+    def test(self):
+        with open(self.file1_stage, 'r') as fd:
+            d = yaml.load(fd)
+
+        del(d[Stage.PARAM_OUTS][0][Output.PARAM_MD5])
+        del(d[Stage.PARAM_DEPS][0][Output.PARAM_MD5])
+
+        with open(self.file1_stage, 'w') as fd:
+            yaml.dump(d, fd)
+
+        self.dvc.checkout()

tests/test_repro.py

@@ -1,4 +1,5 @@
 import os
+import yaml
 import stat
 import shutil
 import filecmp
@@ -7,6 +8,8 @@
 from dvc.command.repro import CmdRepro
 from dvc.project import ReproductionError
 from dvc.utils import file_md5
+from dvc.output import Output
+from dvc.stage import Stage
 
 from tests.basic_env import TestDvc
 
@@ -135,6 +138,21 @@ def test(self):
         self.assertEqual(len(stages), 1)
 
 
+class TestReproMissingMd5InStageFile(TestRepro):
+    def test(self):
+        with open(self.file1_stage, 'r') as fd:
+            d = yaml.load(fd)
+
+        del(d[Stage.PARAM_OUTS][0][Output.PARAM_MD5])
+        del(d[Stage.PARAM_DEPS][0][Output.PARAM_MD5])
+
+        with open(self.file1_stage, 'w') as fd:
+            yaml.dump(d, fd)
+
+        stages = self.dvc.reproduce(self.file1_stage)
+        self.assertEqual(len(stages), 1)
+
+
 class TestCmdRepro(TestRepro):
     def test(self):
         ret = main(['repro',