Add a workaround around non-ascii and non-standard FQDNs (#790)

Add a wrapper around `socket.getfqdn()` which returns `invalid.hostname`
in cases where hostname, or rather FQDN contains non-ascii and
non-standard characters.

This works around the issue that non-ascii characters can't be inserted
into leapp database.

For the standard for hostnames see `hostname(7)`. The current
implementation doesn't check for length of hostname nor it's individual
segments.

Author: matejmatuska

leapp/utils/workarounds/__init__.py

@@ -1,5 +1,7 @@
 import leapp.utils.workarounds.mp
+import leapp.utils.workarounds.fqdn
 
 
 def apply_workarounds():
     leapp.utils.workarounds.mp.apply_workaround()
+    leapp.utils.workarounds.fqdn.apply_workaround()

leapp/utils/workarounds/fqdn.py

@@ -0,0 +1,18 @@
+import re
+import socket
+
+
+def decorate_getfqdn(fn):
+    def wrap(*args, **kwargs):
+        result = fn(*args, **kwargs)
+        # check whether FQDN conforms to standard, see HOSTNAME(7)
+        pattern = r"^(([a-z0-9]|[a-z0-9][a-z0-9\-]*[a-z0-9])\.)*([a-z0-9]|[a-z0-9][a-z0-9\-]*[a-z0-9])$"
+        if re.match(pattern, result, re.IGNORECASE):
+            return result
+        return 'invalid.hostname'
+
+    return wrap
+
+
+def apply_workaround():
+    setattr(socket, 'getfqdn', decorate_getfqdn(socket.getfqdn))

tests/scripts/test_workarounds_application.py

@@ -2,8 +2,10 @@
 from multiprocessing import util, Process, Manager
 
 import os
+import pytest
 
 import leapp  # noqa: F401; pylint: disable=unused-import
+from leapp.utils.workarounds import fqdn
 
 
 def test_mp_is_patched():
@@ -29,3 +31,26 @@ def child_fun(_lst):
 def test_mp_workaround_applied():
     if getattr(util, 'os', None) is None:
         assert util.Finalize.__name__ == 'FixedFinalize'
+
+
+@pytest.mark.parametrize(
+    ('input_fqdn', 'valid'),
+    [
+        ('foo.bar.com', True),
+        ('foo\xa0bar.foo.com', False),
+        ('-foo.bar', False),
+        ('foo.bar-', False),
+        ('foo.-bar.1234', False),
+        ('a1.b2.c3', True),
+        ('123.foo.bar', True),
+        ('123.f-o-o.b-a-r', True),
+    ]
+)
+def test_fqdn_is_patched(input_fqdn, valid):
+
+    def getfqdn():
+        return input_fqdn
+
+    fn = fqdn.decorate_getfqdn(getfqdn)
+    expected_fqdn = input_fqdn if valid else 'invalid.hostname'
+    assert fn() == expected_fqdn