Utils: Generalize is_string from POSIX/File

is_string from POSIX/File could be used in other situations, too.
So we move it to Utils, use it from Lint/MergeFiles and replace a
custom is_binary function.

3 files changed, 22 insertions, 24 deletions

diff --git a/src/lib/Bcfg2/Client/Tools/POSIX/File.py b/src/lib/Bcfg2/Client/Tools/POSIX/File.py
index fc445e07c..1f1772d46 100644
--- a/src/lib/Bcfg2/Client/Tools/POSIX/File.py
+++ b/src/lib/Bcfg2/Client/Tools/POSIX/File.py
@@ -8,6 +8,7 @@ import tempfile
 import Bcfg2.Options
 from Bcfg2.Client.Tools.POSIX.base import POSIXTool
 from Bcfg2.Compat import unicode, b64encode, b64decode  # pylint: disable=W0622
+import Bcfg2.Utils
 
 
 class POSIXFile(POSIXTool):
@@ -17,21 +18,6 @@ class POSIXFile(POSIXTool):
     def fully_specified(self, entry):
         return entry.text is not None or entry.get('empty', 'false') == 'true'
 
-    def _is_string(self, strng, encoding):
-        """ Returns true if the string contains no ASCII control
-        characters and can be decoded from the specified encoding. """
-        for char in strng:
-            if ord(char) < 9 or ord(char) > 13 and ord(char) < 32:
-                return False
-        if not hasattr(strng, "decode"):
-            # py3k
-            return True
-        try:
-            strng.decode(encoding)
-            return True
-        except:  # pylint: disable=W0702
-            return False
-
     def _get_data(self, entry):
         """ Get a tuple of (<file data>, <is binary>) for the given entry """
         is_binary = entry.get('encoding', 'ascii') == 'base64'
@@ -181,8 +167,8 @@ class POSIXFile(POSIXTool):
                                   (entry.get("name"), sys.exc_info()[1]))
                 return False
         if not is_binary:
-            is_binary |= not self._is_string(content,
-                                             Bcfg2.Options.setup.encoding)
+            is_binary |= not Bcfg2.Utils.is_string(
+                content, Bcfg2.Options.setup.encoding)
         if is_binary:
             # don't compute diffs if the file is binary
             prompt.append('Binary file, no printable diff')
diff --git a/src/lib/Bcfg2/Server/Lint/MergeFiles.py b/src/lib/Bcfg2/Server/Lint/MergeFiles.py
index bdb97cee2..3a6251594 100644
--- a/src/lib/Bcfg2/Server/Lint/MergeFiles.py
+++ b/src/lib/Bcfg2/Server/Lint/MergeFiles.py
@@ -6,6 +6,7 @@ import copy
 from difflib import SequenceMatcher
 import Bcfg2.Server.Lint
 from Bcfg2.Server.Plugins.Cfg import CfgGenerator
+from Bcfg2.Utils import is_string
 
 
 def threshold(val):
@@ -17,12 +18,6 @@ def threshold(val):
     return rv
 
 
-def is_binary(data):
-    """ Check if a given string contains only text or binary data. """
-    text_chars = bytearray([7, 8, 9, 10, 12, 13, 27] + range(0x20, 0x100))
-    return bool(data.translate(None, text_chars))
-
-
 class MergeFiles(Bcfg2.Server.Lint.ServerPlugin):
     """ find Probes or Cfg files with multiple similar files that
     might be merged into one """
@@ -56,7 +51,8 @@ class MergeFiles(Bcfg2.Server.Lint.ServerPlugin):
         for filename, entryset in self.core.plugins['Cfg'].entries.items():
             candidates = dict([(f, e) for f, e in entryset.entries.items()
                                if (isinstance(e, CfgGenerator) and
-                                   not is_binary(e.data) and
+                                   is_string(e.data,
+                                             Bcfg2.Options.setup.encoding) and
                                    f not in ignore and
                                    not f.endswith(".crypt"))])
             similar, identical = self.get_similar(candidates)
diff --git a/src/lib/Bcfg2/Utils.py b/src/lib/Bcfg2/Utils.py
index 10057b63e..64d0d8b93 100644
--- a/src/lib/Bcfg2/Utils.py
+++ b/src/lib/Bcfg2/Utils.py
@@ -330,3 +330,19 @@ class classproperty(object):  # pylint: disable=C0103
 
     def __get__(self, instance, owner):
         return self.getter(owner)
+
+
+def is_string(strng, encoding):
+    """ Returns true if the string contains no ASCII control
+    characters and can be decoded from the specified encoding. """
+    for char in strng:
+        if ord(char) < 9 or ord(char) > 13 and ord(char) < 32:
+            return False
+    if not hasattr(strng, "decode"):
+        # py3k
+        return True
+    try:
+        strng.decode(encoding)
+        return True
+    except:  # pylint: disable=W0702
+        return False