Initial Python3 support

Create compat.py module to deal with Python2/Python3 dependencies

cybox/__init__.py

@@ -4,7 +4,8 @@
 import collections
 import inspect
 import json
-from StringIO import StringIO
+
+from .compat import StringIO, basestring, str
 
 import cybox.bindings as bindings
 import cybox.utils.idgen
@@ -131,13 +132,13 @@ def to_obj(self, return_obj=None, ns_info=None):
 
         entity_obj = self._binding_class()
 
-        vars = {}
+        members = {}
         for klass in self.__class__.__mro__:
             if klass is Entity:
                 break
-            vars.update(klass.__dict__.iteritems())
+            members.update(vars(klass))
 
-        for name, field in vars.iteritems():
+        for name, field in members.items():
             if isinstance(field, TypedField):
                 val = getattr(self, field.attr_name)
 
@@ -170,13 +171,13 @@ def to_dict(self):
             Python dict with keys set from this Entity.
         """
         entity_dict = {}
-        vars = {}
+        members = {}
         for klass in self.__class__.__mro__:
             if klass is Entity:
                 break
-            vars.update(klass.__dict__.iteritems())
+            members.update(vars(klass))
 
-        for name, field in vars.iteritems():
+        for name, field in members.items():
             if isinstance(field, TypedField):
                 val = getattr(self, field.attr_name)
 
@@ -299,7 +300,7 @@ def to_xml(self, include_namespaces=True, namespace_dict=None,
                 pretty_print=pretty
             )
 
-        s = unicode(sio.getvalue()).strip()
+        s = str(sio.getvalue()).strip()
 
         if encoding:
             return s.encode(encoding)
@@ -368,7 +369,12 @@ def _get_namespaces(self, recurse=True):
     def _get_children(self):
         #TODO: eventually everything should be in _fields, not the top level
         # of vars()
-        for k, v in vars(self).items() + self._fields.items():
+
+        members = {}
+        members.update(vars(self))
+        members.update(self._fields)
+
+        for k, v in members.items():
             if isinstance(v, Entity):
                 yield v
             elif isinstance(v, list):
@@ -411,7 +417,7 @@ def value(self):
 
     @value.setter
     def value(self, value):
-        self._value = unicode(value)
+        self._value = str(value)
 
     def to_obj(self, return_obj=None, ns_info=None):
         self._collect_ns_info(ns_info)

cybox/bindings/__init__.py

@@ -9,6 +9,8 @@
 from xml.sax import saxutils
 from lxml import etree as etree_
 
+from cybox.compat import basestring, str
+
 CDATA_START = "<![CDATA["
 CDATA_END = "]]>"
 
@@ -86,7 +88,7 @@ def gds_validate_integer_list(self, input_data, node, input_name=''):
         for value in values:
             try:
                 fvalue = float(value)
-            except (TypeError, ValueError), exp:
+            except (TypeError, ValueError) as exp:
                 raise_parse_error(node, 'Requires sequence of integers')
         return input_data
 
@@ -104,7 +106,7 @@ def gds_validate_float_list(self, input_data, node, input_name=''):
         for value in values:
             try:
                 fvalue = float(value)
-            except (TypeError, ValueError), exp:
+            except (TypeError, ValueError) as exp:
                 raise_parse_error(node, 'Requires sequence of floats')
         return input_data
 
@@ -122,7 +124,7 @@ def gds_validate_double_list(self, input_data, node, input_name=''):
         for value in values:
             try:
                 fvalue = float(value)
-            except (TypeError, ValueError), exp:
+            except (TypeError, ValueError) as exp:
                 raise_parse_error(node, 'Requires sequence of doubles')
         return input_data
 
@@ -271,17 +273,33 @@ def showIndent(lwrite, level, pretty_print=True):
             lwrite('    ' * level)
 
 
-def quote_xml(text):
+def _coerce_unicode(text):
+    # Convert `text` to Unicode string.
+
     if text is None:
-        return u''
+        text = ""
+
+    if isinstance(text, str):
+        return text
 
-    # Convert `text` to unicode string. This is mainly a catch-all for non
+    # This is mainly a catch-all for non
     # string/unicode types like bool and int.
     try:
-        text = unicode(text)
+        text = str(text)
     except UnicodeDecodeError:
         text = text.decode(ExternalEncoding)
 
+    return text
+
+
+def quote_xml(text):
+    """Format a value for display as an XML text node.
+
+    Returns:
+        Unicode string (str on Python 3, unicode on Python 2)
+    """
+    text = _coerce_unicode(text)
+
     # If it's a CDATA block, return the text as is.
     if text.startswith(CDATA_START):
         return text
@@ -292,15 +310,12 @@ def quote_xml(text):
 
 
 def quote_attrib(text):
-    if text is None:
-        return u'""'
+    """Format a value for display as an XML attribute.
 
-    # Convert `text` to unicode string. This is mainly a catch-all for non
-    # string/unicode types like bool and int.
-    try:
-        text = unicode(text)
-    except UnicodeDecodeError:
-        text = text.decode(ExternalEncoding)
+    Returns:
+        Unicode string (str on Python 3, unicode on Python 2)
+    """
+    text = _coerce_unicode(text)
 
     # Return the escaped the value of text.
     # Note: This wraps the escaped text in quotation marks.

cybox/bindings/account_object.py

@@ -4,7 +4,7 @@
 import sys
 
 from cybox.bindings import *
-import cybox_common
+from . import cybox_common
 
 
 class AuthenticationType(GeneratedsSuper):
@@ -426,7 +426,7 @@ def buildChildren(self, child_, node, nodeName_, fromsubclass_=False):
 """
 
 def usage():
-    print USAGE_TEXT
+    print(USAGE_TEXT)
     sys.exit(1)
 
 def get_root_tag(node):
@@ -472,7 +472,7 @@ def parseEtree(inFileName):
     return rootObj, rootElement
 
 def parseString(inString):
-    from StringIO import StringIO
+    from cybox.compat import StringIO
     doc = parsexml_(StringIO(inString))
     rootNode = doc.getroot()
     rootTag, rootClass = get_root_tag(rootNode)

cybox/bindings/address_object.py

@@ -4,7 +4,7 @@
 import sys
 
 from cybox.bindings import *
-import cybox_common
+from . import cybox_common
 
 
 class AddressObjectType(cybox_common.ObjectPropertiesType):
@@ -13,7 +13,7 @@ class AddressObjectType(cybox_common.ObjectPropertiesType):
     is being defined. The is_source field specifies if this is a
     "Source" addressThe is_destination field specifies if this is a
     "Destination" address"""
-    
+
     subclass = None
     superclass = cybox_common.ObjectPropertiesType
     def __init__(self, object_reference=None, Custom_Properties=None, xsi_type=None, category='ipv4-addr', is_source=None, is_destination=None, is_spoofed=None, Address_Value=None, VLAN_Name=None, VLAN_Num=None):
@@ -259,7 +259,7 @@ def buildChildren(self, child_, node, nodeName_, fromsubclass_=False):
 """
 
 def usage():
-    print USAGE_TEXT
+    print(USAGE_TEXT)
     sys.exit(1)
 
 def get_root_tag(node):
@@ -305,7 +305,7 @@ def parseEtree(inFileName):
     return rootObj, rootElement
 
 def parseString(inString):
-    from StringIO import StringIO
+    from cybox.compat import StringIO
     doc = parsexml_(StringIO(inString))
     rootNode = doc.getroot()
     rootTag, rootClass = get_root_tag(rootNode)

cybox/bindings/api_object.py

@@ -4,13 +4,13 @@
 import sys
 
 from cybox.bindings import *
-import cybox_common
+from . import cybox_common
 
 
 class APIObjectType(cybox_common.ObjectPropertiesType):
     """The APIObjectType type is intended to characterize a specific
     Application Programming Interface."""
-    
+
     subclass = None
     superclass = cybox_common.ObjectPropertiesType
     def __init__(self, object_reference=None, Custom_Properties=None, xsi_type=None, Description=None, Function_Name=None, Normalized_Function_Name=None, Platform=None, Address=None):
@@ -220,7 +220,7 @@ def buildChildren(self, child_, node, nodeName_, fromsubclass_=False):
 """
 
 def usage():
-    print USAGE_TEXT
+    print(USAGE_TEXT)
     sys.exit(1)
 
 def get_root_tag(node):
@@ -266,7 +266,7 @@ def parseEtree(inFileName):
     return rootObj, rootElement
 
 def parseString(inString):
-    from StringIO import StringIO
+    from cybox.compat import StringIO
     doc = parsexml_(StringIO(inString))
     rootNode = doc.getroot()
     rootTag, rootClass = get_root_tag(rootNode)

cybox/bindings/archive_file_object.py

@@ -4,9 +4,8 @@
 import sys
 
 from cybox.bindings import *
-import cybox_common
-
-import file_object
+from . import cybox_common
+from . import file_object
 
 
 class ArchiveFileFormatType(cybox_common.BaseObjectPropertyType):
@@ -363,7 +362,7 @@ def buildChildren(self, child_, node, nodeName_, fromsubclass_=False):
 """
 
 def usage():
-    print USAGE_TEXT
+    print(USAGE_TEXT)
     sys.exit(1)
 
 def get_root_tag(node):
@@ -409,7 +408,7 @@ def parseEtree(inFileName):
     return rootObj, rootElement
 
 def parseString(inString):
-    from StringIO import StringIO
+    from cybox.compat import StringIO
     doc = parsexml_(StringIO(inString))
     rootNode = doc.getroot()
     rootTag, rootClass = get_root_tag(rootNode)

cybox/bindings/arp_cache_object.py

@@ -4,10 +4,9 @@
 import sys
 
 from cybox.bindings import *
-import cybox_common
-
-import address_object
-import system_object
+from . import cybox_common
+from . import address_object
+from . import system_object
 
 
 class ARPCacheEntryType(GeneratedsSuper):
@@ -404,7 +403,7 @@ def buildChildren(self, child_, node, nodeName_, fromsubclass_=False):
 """
 
 def usage():
-    print USAGE_TEXT
+    print(USAGE_TEXT)
     sys.exit(1)
 
 def get_root_tag(node):
@@ -450,7 +449,7 @@ def parseEtree(inFileName):
     return rootObj, rootElement
 
 def parseString(inString):
-    from StringIO import StringIO
+    from cybox.compat import StringIO
     doc = parsexml_(StringIO(inString))
     rootNode = doc.getroot()
     rootTag, rootClass = get_root_tag(rootNode)