Fix input file encodings for Python 3

generateDS.py

@@ -225,7 +225,7 @@ _log = logging.getLogger(__name__)
 # Do not modify the following VERSION comments.
 # Used by updateversion.py.
 ##VERSION##
-VERSION = '2.29.15'
+VERSION = '2.29.16'
 ##VERSION##
 
 BaseStrTypes = six.string_types
@@ -252,7 +252,7 @@ NoDates = False
 NoVersion = False
 AlwaysExportDefault = False
 Dirpath = []
-ExternalEncoding = sys.getdefaultencoding()
+ExternalEncoding = ''
 Namespacedef = ''
 NoNameSpaceDefs = False
 CleanupNameList = [(re.compile('[-:.]'), '_')]
@@ -5727,7 +5727,11 @@ class GeneratedsSuper(object):
     @staticmethod
     def gds_encode(instring):
         if sys.version_info.major == 2:
-            return instring.encode(ExternalEncoding)
+            if ExternalEncoding:
+                encoding = ExternalEncoding
+            else:
+                encoding = 'utf-8'
+            return instring.encode(encoding)
         else:
             return instring
     @staticmethod
@@ -6908,7 +6912,7 @@ def parseAndGenerate(
     if xschemaFileName == '-':
         infile = sys.stdin
     else:
-        infile = open(xschemaFileName, 'r')
+        infile = open(xschemaFileName, 'rb')
     if SingleFileOutput:
         parser = make_parser()
         dh = XschemaHandler()
@@ -7226,7 +7230,7 @@ def main():
     processIncludes = 1
     namespacedef = ''
     NoNameSpaceDefs = False
-    ExternalEncoding = sys.getdefaultencoding()
+    ExternalEncoding = ''
     NoDates = False
     NoVersion = False
     NoQuestions = False

process_includes.py

@@ -40,7 +40,7 @@ except ImportError:
 # Do not modify the following VERSION comments.
 # Used by updateversion.py.
 ##VERSION##
-VERSION = '2.29.15'
+VERSION = '2.29.16'
 ##VERSION##
 
 CatalogDict = {}
@@ -63,7 +63,7 @@ def load_catalog(catalogpath):
     global CatalogBaseUrl
     if catalogpath:
         CatalogBaseUrl = os.path.split(catalogpath)[0]
-        catalog = etree.parse(open(catalogpath))
+        catalog = etree.parse(open(catalogpath, "rb"))
         for elements in catalog.getroot().findall(
                 "{urn:oasis:names:tc:entity:xmlns:xml:catalog}public"):
             CatalogDict[elements.get("publicId")] = elements.get("uri")
@@ -95,6 +95,7 @@ def get_all_root_file_paths(
         catalogpath=None,
         shallow=False):
     load_catalog(catalogpath)
+    # Note: infile has been opened in binary mode.
     doc1 = etree.parse(infile)
     root1 = doc1.getroot()
     rootPaths = []
@@ -209,12 +210,13 @@ def resolve_ref(node, params, options):
                     raise SchemaIOError(msg)
             else:
                 if os.path.exists(locn):
-                    infile = open(locn)
-                    unencoded_content = infile.read()
                     if sys.version_info.major == 2:
+                        infile = open(locn)
+                        unencoded_content = infile.read()
                         content = unencoded_content
                     else:
-                        content = unencoded_content.encode()
+                        infile = open(locn, 'rb')
+                        content = infile.read()
                     infile.close()
                     params.parent_url = locn
                     params.base_url = os.path.split(locn)[0]
@@ -319,6 +321,7 @@ def make_file(outFileName, options):
 
 
 def prep_schema_doc(infile, outfile, inpath, options):
+    # Note: infile has been opened in binary mode.
     doc1 = etree.parse(infile)
     root1 = doc1.getroot()
     params = Params()