Merge pull request mvantellingen#1125 from da1910/features/strip_whitespace_builtin_types

Implement correct xsd:whitespace behaviour

Author: mvantellingen

src/zeep/xsd/types/builtins.py

@@ -34,6 +34,19 @@ def _wrapper(self, value):
     return _wrapper
 
 
+def treat_whitespace(behaviour):
+    def _treat_whitespace(func):
+        def _wrapper(self, value):
+            assert behaviour in ["replace", "collapse", "preserve"]
+            if behaviour == "replace":
+                return func(self, re.sub(r'[\n\r\t]', ' ', value))
+            elif behaviour == "collapse":
+                return func(self, re.sub(r'[\n\r\t ]', ' ', value).strip())
+            return func(self, value)
+        return _wrapper
+    return _treat_whitespace
+
+
 ##
 # Primitive types
 class String(BuiltinType):
@@ -58,6 +71,7 @@ class Boolean(BuiltinType):
     def xmlvalue(self, value):
         return "true" if value and value not in ("false", "0") else "false"
 
+    @treat_whitespace('collapse')
     def pythonvalue(self, value):
         """Return True if the 'true' or '1'. 'false' and '0' are legal false
         values, but we consider everything not true as false.
@@ -74,6 +88,7 @@ class Decimal(BuiltinType):
     def xmlvalue(self, value):
         return str(value)
 
+    @treat_whitespace('collapse')
     def pythonvalue(self, value):
         return _Decimal(value)
 
@@ -85,6 +100,7 @@ class Float(BuiltinType):
     def xmlvalue(self, value):
         return str(value).upper()
 
+    @treat_whitespace('collapse')
     def pythonvalue(self, value):
         return float(value)
 
@@ -97,6 +113,7 @@ class Double(BuiltinType):
     def xmlvalue(self, value):
         return str(value)
 
+    @treat_whitespace('collapse')
     def pythonvalue(self, value):
         return float(value)
 
@@ -109,6 +126,7 @@ class Duration(BuiltinType):
     def xmlvalue(self, value):
         return isodate.duration_isoformat(value)
 
+    @treat_whitespace('collapse')
     def pythonvalue(self, value):
         if value.startswith("PT-"):
             value = value.replace("PT-", "PT")
@@ -144,6 +162,7 @@ def xmlvalue(self, value):
             return isodate.isostrf.strftime(value, "%Y-%m-%dT%H:%M:%S.%f%Z")
         return isodate.isostrf.strftime(value, "%Y-%m-%dT%H:%M:%S%Z")
 
+    @treat_whitespace('collapse')
     def pythonvalue(self, value):
 
         # Determine based on the length of the value if it only contains a date
@@ -166,6 +185,7 @@ def xmlvalue(self, value):
             return isodate.isostrf.strftime(value, "%H:%M:%S.%f%Z")
         return isodate.isostrf.strftime(value, "%H:%M:%S%Z")
 
+    @treat_whitespace('collapse')
     def pythonvalue(self, value):
         return isodate.parse_time(value)
 
@@ -180,6 +200,7 @@ def xmlvalue(self, value):
             return value
         return isodate.isostrf.strftime(value, "%Y-%m-%d")
 
+    @treat_whitespace('collapse')
     def pythonvalue(self, value):
         return isodate.parse_date(value)
 
@@ -203,6 +224,7 @@ def xmlvalue(self, value):
         year, month, tzinfo = value
         return "%04d-%02d%s" % (year, month, _unparse_timezone(tzinfo))
 
+    @treat_whitespace('collapse')
     def pythonvalue(self, value):
         match = self._pattern.match(value)
         if not match:
@@ -231,6 +253,7 @@ def xmlvalue(self, value):
         year, tzinfo = value
         return "%04d%s" % (year, _unparse_timezone(tzinfo))
 
+    @treat_whitespace('collapse')
     def pythonvalue(self, value):
         match = self._pattern.match(value)
         if not match:
@@ -258,6 +281,7 @@ def xmlvalue(self, value):
         month, day, tzinfo = value
         return "--%02d-%02d%s" % (month, day, _unparse_timezone(tzinfo))
 
+    @treat_whitespace('collapse')
     def pythonvalue(self, value):
         match = self._pattern.match(value)
         if not match:
@@ -288,6 +312,7 @@ def xmlvalue(self, value):
         day, tzinfo = value
         return "---%02d%s" % (day, _unparse_timezone(tzinfo))
 
+    @treat_whitespace('collapse')
     def pythonvalue(self, value):
         match = self._pattern.match(value)
         if not match:
@@ -312,6 +337,7 @@ def xmlvalue(self, value):
         month, tzinfo = value
         return "--%d%s" % (month, _unparse_timezone(tzinfo))
 
+    @treat_whitespace('collapse')
     def pythonvalue(self, value):
         match = self._pattern.match(value)
         if not match:
@@ -352,6 +378,7 @@ class AnyURI(BuiltinType):
     def xmlvalue(self, value):
         return value
 
+    @treat_whitespace('collapse')
     def pythonvalue(self, value):
         return value
 
@@ -364,6 +391,7 @@ class QName(BuiltinType):
     def xmlvalue(self, value):
         return value
 
+    @treat_whitespace('collapse')
     def pythonvalue(self, value):
         return value
 
@@ -380,10 +408,18 @@ class Notation(BuiltinType):
 class NormalizedString(String):
     _default_qname = xsd_ns("normalizedString")
 
+    @treat_whitespace('replace')
+    def pythonvalue(self, value):
+        return value
+
 
 class Token(NormalizedString):
     _default_qname = xsd_ns("token")
 
+    @treat_whitespace('collapse')
+    def pythonvalue(self, value):
+        return value
+
 
 class Language(Token):
     _default_qname = xsd_ns("language")

tests/test_xsd_builtins.py

@@ -19,6 +19,42 @@ def test_pythonvalue(self):
         result = instance.pythonvalue("foobar")
         assert result == "foobar"
 
+        result = instance.pythonvalue(" foo\tbar\r\n ")
+        assert result == " foo\tbar\r\n "
+
+
+class TestNormalizedString:
+    def test_xmlvalue(self):
+        instance = builtins.NormalizedString()
+        result = instance.xmlvalue("foobar")
+        assert result == "foobar"
+
+    def test_pythonvalue(self):
+        instance = builtins.NormalizedString()
+        result = instance.pythonvalue("foobar")
+        assert result == "foobar"
+
+        result = instance.pythonvalue("fo\tob\rar\n")
+        assert result == "fo ob ar "
+
+
+class TestToken:
+    def test_xmlvalue(self):
+        instance = builtins.Token()
+        result = instance.xmlvalue("foobar")
+        assert result == "foobar"
+
+    def test_pythonvalue(self):
+        instance = builtins.Token()
+        result = instance.pythonvalue("foobar")
+        assert result == "foobar"
+
+        result = instance.pythonvalue("fo\tob\rar")
+        assert result == "fo ob ar"
+
+        result = instance.pythonvalue("  foobar ")
+        assert result == "foobar"
+
 
 class TestBoolean:
     def test_xmlvalue(self):
@@ -36,6 +72,7 @@ def test_pythonvalue(self):
         assert instance.pythonvalue("true") is True
         assert instance.pythonvalue("0") is False
         assert instance.pythonvalue("false") is False
+        assert instance.pythonvalue("\t \r\nfalse ") is False
 
 
 class TestDecimal:
@@ -53,6 +90,7 @@ def test_pythonvalue(self):
         assert instance.pythonvalue("10.001") == D("10.001")
         assert instance.pythonvalue("+10.001") == D("10.001")
         assert instance.pythonvalue("-10.001") == D("-10.001")
+        assert instance.pythonvalue(" \r\n 10 \t") == D("10")
 
 
 class TestFloat:
@@ -74,6 +112,7 @@ def test_pythonvalue(self):
         assert instance.pythonvalue("-0") == float(0)
         assert instance.pythonvalue("0") == float(0)
         assert instance.pythonvalue("INF") == float("inf")
+        assert instance.pythonvalue("\t \r12.78e-2\n  ") == float("0.1278")
 
 
 class TestDouble:
@@ -89,6 +128,7 @@ def test_pythonvalue(self):
         assert instance.pythonvalue("12") == float(12)
         assert instance.pythonvalue("-0") == float(0)
         assert instance.pythonvalue("0") == float(0)
+        assert instance.pythonvalue(" \r\n0 \t") == float(0)
 
 
 class TestDuration:
@@ -103,6 +143,10 @@ def test_pythonvalue(self):
         value = "P0Y1347M0D"
         assert instance.pythonvalue(value) == expected
 
+        expected = isodate.parse_duration("P0Y1347M0D")
+        value = "\r  \nP0Y1347M0D\t "
+        assert instance.pythonvalue(value) == expected
+
 
 class TestDateTime:
     def test_xmlvalue(self):
@@ -137,6 +181,9 @@ def test_pythonvalue(self):
         value = datetime.datetime(2016, 3, 4, 0, 0, 0)
         assert instance.pythonvalue("2016-03-04") == value
 
+        value = datetime.datetime(2016, 3, 4, 0, 0, 0)
+        assert instance.pythonvalue(" \r\n\t2016-03-04   ") == value
+
     def test_pythonvalue_invalid(self):
         instance = builtins.DateTime()
         with pytest.raises(ValueError):
@@ -161,6 +208,9 @@ def test_pythonvalue(self):
         value = isodate.parse_time("21:14:42.120+0200")
         assert instance.pythonvalue("21:14:42.120+0200") == value
 
+        value = datetime.time(21, 14, 42)
+        assert instance.pythonvalue("\t\r\n  21:14:42   ") == value
+
     def test_pythonvalue_invalid(self):
         instance = builtins.Time()
         with pytest.raises(ValueError):
@@ -181,6 +231,7 @@ def test_pythonvalue(self):
         assert instance.pythonvalue("2001-10-26+02:00") == datetime.date(2001, 10, 26)
         assert instance.pythonvalue("2001-10-26Z") == datetime.date(2001, 10, 26)
         assert instance.pythonvalue("2001-10-26+00:00") == datetime.date(2001, 10, 26)
+        assert instance.pythonvalue("\r\n\t 2016-03-04   ") == datetime.date(2016, 3, 4)
 
     def test_pythonvalue_invalid(self):
         instance = builtins.Date()
@@ -218,8 +269,8 @@ def test_pythonvalue(self):
 class TestgYear:
     def test_xmlvalue(self):
         instance = builtins.gYear()
-        instance.xmlvalue((2001, None)) == "2001"
-        instance.xmlvalue((2001, pytz.utc)) == "2001Z"
+        assert instance.xmlvalue((2001, None)) == "2001"
+        assert instance.xmlvalue((2001, pytz.utc)) == "2001Z"
 
     def test_pythonvalue(self):
         instance = builtins.gYear()
@@ -229,6 +280,7 @@ def test_pythonvalue(self):
         assert instance.pythonvalue("2001+00:00") == (2001, pytz.utc)
         assert instance.pythonvalue("-2001") == (-2001, None)
         assert instance.pythonvalue("-20000") == (-20000, None)
+        assert instance.pythonvalue("  \t2001+02:00\r\n ") == (2001, pytz.FixedOffset(120))
 
         with pytest.raises(builtins.ParseError):
             assert instance.pythonvalue("99")
@@ -247,6 +299,7 @@ def test_pythonvalue(self):
         assert instance.pythonvalue("--11-01-04:00") == (11, 1, pytz.FixedOffset(-240))
         assert instance.pythonvalue("--11-15") == (11, 15, None)
         assert instance.pythonvalue("--02-29") == (2, 29, None)
+        assert instance.pythonvalue("\t\r\n --05-01 ") == (5, 1, None)
 
         with pytest.raises(builtins.ParseError):
             assert instance.pythonvalue("99")
@@ -265,6 +318,7 @@ def test_pythonvalue(self):
         assert instance.pythonvalue("--11-04:00") == (11, pytz.FixedOffset(-240))
         assert instance.pythonvalue("--11") == (11, None)
         assert instance.pythonvalue("--02") == (2, None)
+        assert instance.pythonvalue("\n\t --11Z \r") == (11, pytz.utc)
 
         with pytest.raises(builtins.ParseError):
             assert instance.pythonvalue("99")
@@ -291,6 +345,7 @@ def test_pythonvalue(self):
         assert instance.pythonvalue("---01-04:00") == (1, pytz.FixedOffset(-240))
         assert instance.pythonvalue("---15") == (15, None)
         assert instance.pythonvalue("---31") == (31, None)
+        assert instance.pythonvalue("\r\n  \t---31 ") == (31, None)
         with pytest.raises(builtins.ParseError):
             assert instance.pythonvalue("99")