Merge pull request #1258 from delocalizer/fix/base64Binary

handle encoding of base64Binary Literals

rdflib/term.py

@@ -38,7 +38,6 @@
 import warnings
 import math
 
-import base64
 import xml.dom.minidom
 
 from datetime import date, time, datetime, timedelta
@@ -53,6 +52,7 @@
     parse_duration,
     duration_isoformat,
 )
+from base64 import b64decode, b64encode
 from binascii import hexlify, unhexlify
 
 import rdflib
@@ -1435,6 +1435,7 @@ def _parseBoolean(value):
 _XSD_YEARMONTHDURATION = URIRef(_XSD_PFX + "yearMonthDuration")
 
 _OWL_RATIONAL = URIRef("http://www.w3.org/2002/07/owl#rational")
+_XSD_B64BINARY = URIRef(_XSD_PFX + "base64Binary")
 _XSD_HEXBINARY = URIRef(_XSD_PFX + "hexBinary")
 # TODO: gYearMonth, gYear, gMonthDay, gDay, gMonth
 
@@ -1559,6 +1560,8 @@ def _castPythonToLiteral(obj, datatype):
 _SpecificPythonToXSDRules = [
     ((str, _XSD_HEXBINARY), hexlify),
     ((bytes, _XSD_HEXBINARY), hexlify),
+    ((str, _XSD_B64BINARY), b64encode),
+    ((bytes, _XSD_B64BINARY), b64encode),
 ]
 
 XSDToPython = {
@@ -1593,7 +1596,7 @@ def _castPythonToLiteral(obj, datatype):
     URIRef(_XSD_PFX + "unsignedByte"): int,
     URIRef(_XSD_PFX + "float"): float,
     URIRef(_XSD_PFX + "double"): float,
-    URIRef(_XSD_PFX + "base64Binary"): lambda s: base64.b64decode(s),
+    URIRef(_XSD_PFX + "base64Binary"): b64decode,
     URIRef(_XSD_PFX + "anyURI"): None,
     _RDF_XMLLITERAL: _parseXML,
     _RDF_HTMLLITERAL: _parseHTML,

test/test_b64_binary.py

@@ -0,0 +1,29 @@
+# -*- coding: utf-8 -*-
+
+import unittest
+import base64 
+from rdflib import Literal, XSD
+
+
+class B64BinaryTestCase(unittest.TestCase):
+
+    def test_unicode(self):
+        str1 = "Test utf-8 string éàë"
+        # u b64string
+        b64_str1 = base64.b64encode(str1.encode("utf-8")).decode()
+        l1 = Literal(b64_str1, datatype=XSD.base64Binary)
+        b_str1 = l1.toPython()
+        self.assertEqual(b_str1.decode("utf-8"), str1)
+        self.assertEqual(str(l1), b64_str1)
+
+        # b b64string
+        b64_str1b = base64.b64encode(str1.encode("utf-8"))
+        l1b = Literal(b64_str1b, datatype=XSD.base64Binary)
+        b_str1b = l1b.toPython()
+        self.assertEqual(b_str1, b_str1b)
+        self.assertEqual(b_str1b.decode("utf-8"), str1)
+        self.assertEqual(str(l1b), b64_str1)
+
+
+if __name__ == "__main__":
+    unittest.main()