Fix false positive non-utf symlinks reported

.pylintrc

@@ -52,7 +52,8 @@ disable=
  consider-using-with,  # on bunch spaces we cannot change that...
  duplicate-string-formatting-argument,  # TMP: will be fixed in close future
  consider-using-f-string,  # sorry, not gonna happen, still have to support py2
- use-dict-literal
+ use-dict-literal,
+ redundant-u-string-prefix  # still have py2 to support
 
 [FORMAT]
 # Maximum number of characters on a single line.

commands/upgrade/breadcrumbs.py

@@ -128,8 +128,6 @@ def _get_packages(self):
 
     def _verify_leapp_pkgs(self):
         if not os.environ.get('LEAPP_IPU_IN_PROGRESS'):
-            # NOTE(ivasilev) this can happen if LEAPP_DEVEL_TARGET_RELEASE is specified and pointing to an impossible
-            # version
             return []
         upg_path = os.environ.get('LEAPP_IPU_IN_PROGRESS').split('to')
         cmd = ['/bin/bash', '-c', 'rpm -V leapp leapp-upgrade-el{}toel{}'.format(upg_path[0], upg_path[1])]

repos/system_upgrade/common/actors/rootscanner/actor.py

@@ -1,9 +1,6 @@
-import os
-
-import six
-
 from leapp.actors import Actor
-from leapp.models import InvalidRootSubdirectory, RootDirectory, RootSubdirectory
+from leapp.libraries.actor.rootscanner import scan_dir
+from leapp.models import RootDirectory
 from leapp.tags import FactsPhaseTag, IPUWorkflowTag
 
 
@@ -19,19 +16,4 @@ class RootScanner(Actor):
     tags = (IPUWorkflowTag, FactsPhaseTag)
 
     def process(self):
-        subdirs = []
-        invalid_subdirs = []
-
-        def _create_a_subdir(subdir_cls, name, path):
-            if os.path.islink(path):
-                return subdir_cls(name=name, target=os.readlink(path))
-            return subdir_cls(name=name)
-
-        for subdir in os.listdir('/'):
-            # Note(ivasilev) non-utf encoded string will appear as byte strings
-            if isinstance(subdir, six.binary_type):
-                invalid_subdirs.append(_create_a_subdir(InvalidRootSubdirectory, subdir, os.path.join(b'/', subdir)))
-            else:
-                subdirs.append(_create_a_subdir(RootSubdirectory, subdir, os.path.join('/', subdir)))
-
-        self.produce(RootDirectory(items=subdirs, invalid_items=invalid_subdirs))
+        self.produce(scan_dir(b'/'))

repos/system_upgrade/common/actors/rootscanner/libraries/rootscanner.py

@@ -0,0 +1,34 @@
+import os
+
+import six
+
+from leapp.models import InvalidRootSubdirectory, RootDirectory, RootSubdirectory
+
+
+def scan_dir(root_dir=b'/'):
+    """
+    Scan root directory and return a RootDirectory(subdirs, invalid_subdirs) model object
+    """
+    subdirs = []
+    invalid_subdirs = []
+
+    def _create_a_subdir(subdir_cls, name, path):
+        if os.path.islink(path):
+            return subdir_cls(name=name, target=os.readlink(path))
+        return subdir_cls(name=name)
+
+    for subdir in os.listdir(root_dir):
+        # Note(ivasilev) in py3 env non-utf encoded string will appear as byte strings
+        # However in py2 env subdir will be always of str type, so verification if this is a valid utf-8 string
+        # should be done differently than formerly suggested plain six.binary_type check
+        decoded = True
+        if isinstance(subdir, six.binary_type):
+            try:
+                subdir = subdir.decode('utf-8')
+            except (AttributeError, UnicodeDecodeError):
+                decoded = False
+        if not decoded:
+            invalid_subdirs.append(_create_a_subdir(InvalidRootSubdirectory, subdir, os.path.join(b'/', subdir)))
+        else:
+            subdirs.append(_create_a_subdir(RootSubdirectory, subdir, os.path.join('/', subdir)))
+    return RootDirectory(items=subdirs, invalid_items=invalid_subdirs)

repos/system_upgrade/common/actors/rootscanner/tests/test_rootscanner.py

@@ -0,0 +1,32 @@
+# -*- coding: utf-8 -*-
+import os
+import shutil
+import tempfile
+
+import pytest
+
+from leapp.libraries.actor.rootscanner import scan_dir
+
+
+@pytest.mark.parametrize("filename,symlink,count_invalid",
+                         [(u'a_utf_file'.encode('utf-8'), u"utf8_symlink".encode('utf-8'), 0),
+                          (u'простофайл'.encode('koi8-r'), u"этонеутф8".encode('koi8-r'), 2),
+                          (u'a_utf_file'.encode('utf-8'), u"этонеутф8".encode('koi8-r'), 1)])
+def test_invalid_symlinks(filename, symlink, count_invalid):
+    # Let's create a directory with both valid utf-8 and non-utf symlinks
+    # NOTE(ivasilev) As this has to run for python2 as well can't use the nice tempfile.TemporaryDirectory way
+    tmpdirname = tempfile.mkdtemp()
+    # create the file in the temp directory
+    path_to_file = os.path.join(tmpdirname.encode('utf-8'), filename)
+    path_to_symlink = os.path.join(tmpdirname.encode('utf-8'), symlink)
+    with open(path_to_file, 'w') as f:
+        f.write('Some data here')
+    # create a symlink
+    os.symlink(path_to_file, path_to_symlink)
+    # run scan_dir
+    model = scan_dir(tmpdirname.encode('utf-8'))
+    # verify the results
+    assert len(model.items) == 2 - count_invalid
+    assert len(model.invalid_items) == count_invalid
+    # cleanup
+    shutil.rmtree(tmpdirname)