diff --git a/tests/olevba/test_crypto.py b/tests/olevba/test_crypto.py
index b2dc84d..787a1ad 100644
--- a/tests/olevba/test_crypto.py
+++ b/tests/olevba/test_crypto.py
@@ -2,13 +2,11 @@
 
 import sys
 import unittest
-import os
 from os.path import join as pjoin
-from subprocess import check_output, CalledProcessError
 import json
 from collections import OrderedDict
 
-from tests.test_utils import DATA_BASE_DIR, SOURCE_BASE_DIR
+from tests.test_utils import DATA_BASE_DIR, call_and_capture
 
 from oletools import crypto
 
@@ -34,25 +32,11 @@ class OlevbaCryptoWriteProtectTest(unittest.TestCase):
     """
     def test_autostart(self):
         """Check that autostart macro is found in xls[mb] sample file."""
-        # create a PYTHONPATH environment var to prefer our olevba
-        env = os.environ
-        try:
-            env['PYTHONPATH'] = SOURCE_BASE_DIR + os.pathsep + \
-                                os.environ['PYTHONPATH']
-        except KeyError:
-            env['PYTHONPATH'] = SOURCE_BASE_DIR
-
         for suffix in 'xlsm', 'xlsb':
             example_file = pjoin(
                 DATA_BASE_DIR, 'encrypted',
                 'autostart-encrypt-standardpassword.' + suffix)
-            try:
-                output = check_output([sys.executable, '-m', 'olevba', '-j',
-                                       example_file],
-                                      universal_newlines=True, env=env)
-            except CalledProcessError as err:
-                print(err.output)
-                raise
+            output, _ = call_and_capture('olevba', args=('-j', example_file))
             data = json.loads(output, object_pairs_hook=OrderedDict)
             # debug: json.dump(data, sys.stdout, indent=4)
             self.assertEqual(len(data), 4)
diff --git a/tests/test_utils/utils.py b/tests/test_utils/utils.py
index b3b9005..8a35936 100644
--- a/tests/test_utils/utils.py
+++ b/tests/test_utils/utils.py
@@ -2,8 +2,10 @@
 
 """Utils generally useful for unittests."""
 
+import sys
 import os
 from os.path import dirname, join, abspath
+from subprocess import check_output, STDOUT, CalledProcessError
 
 
 # Base dir of project, contains subdirs "tests" and "oletools" and README.md
@@ -14,3 +16,48 @@ DATA_BASE_DIR = join(PROJECT_ROOT, 'tests', 'test-data')
 
 # Directory with source code
 SOURCE_BASE_DIR = join(PROJECT_ROOT, 'oletools')
+
+
+def call_and_capture(module, args=None, accept_nonzero_exit=False):
+    """
+    Run module as script, capturing and returning output and return code.
+
+    This is the best way to capture a module's stdout and stderr; trying to
+    modify sys.stdout/sys.stderr to StringIO-Buffers frequently causes trouble.
+
+    Only drawback sofar: stdout and stderr are merged into one (which is
+    what users see on their shell as well).
+
+    :param str module: name of module to test, e.g. `olevba`
+    :param args: arguments for module's main function
+    :param bool fail_nonzero: Raise error if command returns non-0 return code
+    :returns: ret_code, output
+    :rtype: int, str
+    """
+    # create a PYTHONPATH environment var to prefer our current code
+    env = os.environ.copy()
+    try:
+        env['PYTHONPATH'] = SOURCE_BASE_DIR + os.pathsep + \
+                            os.environ['PYTHONPATH']
+    except KeyError:
+        env['PYTHONPATH'] = SOURCE_BASE_DIR
+
+    # ensure args is a tuple
+    my_args = tuple(args) if args else ()
+
+    ret_code = -1
+    try:
+        output = check_output((sys.executable, '-m', module) + my_args,
+                              universal_newlines=True, env=env,
+                              stderr=STDOUT)
+        ret_code = 0
+
+    except CalledProcessError as err:
+        if accept_nonzero_exit:
+            ret_code = err.returncode
+            output = err.output
+        else:
+            print(err.output)
+            raise
+
+    return output, ret_code