run: Parse and quote correctly the command

Fix: #1478

dvc/command/run.py

@@ -4,26 +4,11 @@ from dvc.exceptions import DvcException
 
 
 class CmdRun(CmdBase):
-    def _joined_cmd(self):
-        if len(self.args.command) == 0:
-            return ''
-
-        if len(self.args.command) == 1:
-            return self.args.command[0]
-
-        cmd = ''
-        for chunk in self.args.command:
-            if len(chunk.split()) != 1:
-                fmt = ' "{}"'
-            else:
-                fmt = ' {}'
-            cmd += fmt.format(chunk)
-        return cmd
-
     def run(self):
         overwrite = (self.args.yes or self.args.overwrite_dvcfile)
+
         try:
-            self.project.run(cmd=self._joined_cmd(),
+            self.project.run(cmd=self._parsed_cmd(),
                              outs=self.args.outs,
                              outs_no_cache=self.args.outs_no_cache,
                              metrics_no_cache=self.args.metrics_no_cache,
@@ -39,3 +24,24 @@ class CmdRun(CmdBase):
             return 1
 
         return 0
+
+    def _parsed_cmd(self):
+        """
+        We need to take into account two cases:
+
+        - ['python code.py foo bar']: Used mainly with dvc as a library
+        - ['echo', 'foo bar']: List of arguments received from the CLI
+
+        The second case would need quoting, as it was passed through:
+                dvc run echo "foo bar"
+        """
+        if len(self.args.command) < 2:
+            return ' '.join(self.args.command)
+
+        return ' '.join(self._quote_argument(arg) for arg in self.args.command)
+
+    def _quote_argument(self, argument):
+        if ' ' not in argument or '"' in argument:
+            return argument
+
+        return '"{}"'.format(argument)

tests/test_run.py

@@ -7,6 +7,7 @@ import subprocess
 from dvc.main import main
 from dvc.utils import file_md5
 from dvc.system import System
+from dvc.stage import Stage
 from dvc.stage import StageFileBadNameError, MissingDep
 from dvc.stage import StageBadCwdError, StageFileAlreadyExistsError
 from dvc.exceptions import (OutputDuplicationError,
@@ -173,33 +174,30 @@ class TestCmdRun(TestDvc):
                     '-o', 'out',
                     '-f', 'out.dvc',
                     'python', self.CODE, self.FOO, 'out'])
+
+        stage = Stage.load(self.dvc, fname='out.dvc')
+
         self.assertEqual(ret, 0)
         self.assertTrue(os.path.isfile('out'))
         self.assertTrue(os.path.isfile('out.dvc'))
         self.assertTrue(filecmp.cmp(self.FOO, 'out', shallow=False))
+        self.assertEqual(stage.cmd, 'python code.py foo out')
+
+    def test_run_args_from_cli(self):
+        ret = main(['run', 'echo', 'foo'])
+        stage = Stage.load(self.dvc, fname='Dvcfile')
+        self.assertEqual(ret, 0)
+        self.assertEqual(stage.cmd, 'echo foo')
 
     def test_run_bad_command(self):
-        ret = main(['run',
-                    'non-existing-command'])
+        ret = main(['run', 'non-existing-command'])
         self.assertNotEqual(ret, 0)
 
     def test_run_args_with_spaces(self):
-        with open(self.CODE, 'w') as fobj:
-            fobj.write(
-                "import sys\nopen(sys.argv[1], 'w+').write(sys.argv[2])")
-
-        arg = 'arg1 arg2'
-        log = 'log'
-        ret = main(['run',
-                    'python',
-                    self.CODE,
-                    log,
-                    'arg1 arg2'])
-
+        ret = main(['run', 'echo', 'foo bar'])
+        stage = Stage.load(self.dvc, fname='Dvcfile')
         self.assertEqual(ret, 0)
-
-        with open(log, 'r') as fobj:
-            self.assertEqual(fobj.read(), arg)
+        self.assertEqual(stage.cmd, 'echo "foo bar"')
 
     @mock.patch.object(subprocess, 'Popen', side_effect=KeyboardInterrupt)
     def test_keyboard_interrupt(self, mock_popen):