Merge pull request #337 from dataversioncontrol/git_file_state_file_fix

Git file output fix

dvc/command/run.py

@@ -1,6 +1,7 @@
 import os
 
 from dvc.command.common.base import CmdBase
+from dvc.data_cloud import file_md5
 from dvc.exceptions import DvcException
 from dvc.logger import Logger
 from dvc.state_file import StateFile
@@ -18,7 +19,7 @@ class CmdRun(CmdBase):
 
     def run(self):
         cmd = ' '.join(self.parsed_args.command)
-        command = StateFile(data_item=None,
+        state = StateFile(data_item=None,
                             cmd=cmd,
                             out=self.parsed_args.out,
                             out_git=self.parsed_args.out_git,
@@ -26,36 +27,43 @@ class CmdRun(CmdBase):
                             locked=self.parsed_args.lock,
                             md5=None)
 
-        self.run_command(self.settings, command)
-        return self.commit_if_needed('DVC run: {}'.format(command.cmd))
+        self.run_command(self.settings, state)
+        return self.commit_if_needed('DVC run: {}'.format(state.cmd))
 
     @staticmethod
-    def run_command(settings, command):
-        Executor.exec_cmd_only_success(command.cmd, shell=True)
+    def run_command(settings, state):
+        Executor.exec_cmd_only_success(state.cmd, shell=True)
 
-        CmdRun.apply_to_files(command.out, command, CmdRun._create_cache_and_state_files, settings)
-        CmdRun.apply_to_files(command.out_git, command, CmdRun._create_state_file, settings)
+        CmdRun.apply_to_files(state.out, state, CmdRun._create_cache_and_state_files, settings)
+        CmdRun.apply_to_files(state.out_git, state, CmdRun._create_state_file, settings)
 
     @staticmethod
-    def apply_to_files(files, command, func, settings):
-        [func(i, command, settings) for i in settings.path_factory.to_data_items(files)[0]]
+    def apply_to_files(files, state, func, settings):
+        items = settings.path_factory.to_data_items(files)[0]
+        [func(i, state, settings) for i in items]
 
     @staticmethod
-    def _create_cache_and_state_files(data_item, command, settings):
+    def _create_cache_and_state_files(data_item, state, settings):
         Logger.debug('Move output file "{}" to cache dir "{}" and create a hardlink'.format(
                      data_item.data.relative, data_item.cache_dir_abs))
         data_item.move_data_to_cache()
-        return CmdRun._create_state_file(data_item, command, settings)
+        return CmdRun._create_state_file(data_item, state, settings, cache_file_exists = True)
 
     @staticmethod
-    def _create_state_file(data_item, command, settings):
+    def _create_state_file(data_item, state, settings, cache_file_exists=False):
         Logger.debug('Create state file "{}"'.format(data_item.state.relative))
-        state_file = StateFile(data_item=data_item,
-                               cmd=command.cmd,
-                               out=command.out,
-                               out_git=command.out_git,
-                               locked=command.locked,
-                               deps=StateFile.parse_deps_state(settings, command.deps),
-                               md5=os.path.basename(data_item.cache.relative))
+
+        if cache_file_exists:
+            md5 = os.path.basename(data_item.cache.relative)
+        else:
+            md5 = file_md5(data_item.data.relative)[0]
+
+        state_file = StateFile(data_item,
+                               state.cmd,
+                               state.out,
+                               state.out_git,
+                               locked=state.locked,
+                               deps=StateFile.parse_deps_state(settings, state.deps),
+                               md5=md5)
         state_file.save()
         return state_file

dvc/git_wrapper.py

@@ -5,10 +5,9 @@ from collections import defaultdict
 from dvc import utils
 from dvc.exceptions import DvcException
 from dvc.logger import Logger
-from dvc.config import Config, ConfigI
+from dvc.config import Config
 from dvc.executor import Executor, ExecutorError
-from dvc.path.data_item import DataItemError
-from dvc.state_file import StateFile
+from dvc.path.data_item import DataItem
 from dvc.system import System
 from dvc.graph.workflow import Workflow
 from dvc.graph.commit import Commit

dvc/path/data_item.py

@@ -23,7 +23,7 @@ class DataDirError(DvcException):
 class DataItemInConfigDirError(DataDirError):
     def __init__(self, file):
         msg = 'File "{}" is in config directory'.format(file)
-        super(DataItemInStatusDirError, self).__init__(msg)
+        super(DataDirError, self).__init__(msg)
 
 
 class NotInGitDirError(DataDirError):

dvc/state_file.py

@@ -1,14 +1,10 @@
 import os
-import ntpath
-import sys
 import yaml
-import re
 
-from dvc import utils
 from dvc.exceptions import DvcException
-from dvc.system import System
 from dvc.data_cloud import file_md5
 
+
 class StateFileError(DvcException):
     def __init__(self, msg):
         DvcException.__init__(self, 'State file error: {}'.format(msg))
@@ -48,7 +44,6 @@ class StateFile(StateFileBase):
                  locked=None,
                  md5=None):
         super(StateFile, self).__init__()
-
         self.data_item = data_item
 
         self.cmd = cmd