Proper VisibleString, IA5String, TeletexString, T61String validation

[pyderasn.git] / pyderasn.py

diff --git a/pyderasn.py b/pyderasn.py
index 54f2027f914df7db6d147dacde4d3daebf519a9e..bc5358a78cfa4e487efc99dd9db3f714bf0d20ea 100755 (executable)
--- a/pyderasn.py
+++ b/pyderasn.py
@@ -962,6 +962,14 @@ _______________
 .. autoclass:: pyderasn.PrintableString
    :members: __init__, allow_asterisk, allow_ampersand
 
+IA5String
+_________
+.. autoclass:: pyderasn.IA5String
+
+VisibleString
+_____________
+.. autoclass:: pyderasn.VisibleString
+
 UTCTime
 _______
 .. autoclass:: pyderasn.UTCTime
@@ -1193,7 +1201,7 @@ except ImportError:  # pragma: no cover
     def colored(what, *args, **kwargs):
         return what
 
-__version__ = "7.5"
+__version__ = "7.6"
 
 __all__ = (
     "agg_octet_string",
@@ -4710,27 +4718,25 @@ class CommonString(OctetString):
        :header-rows: 1
 
        * - Class
-         - Text Encoding
+         - Text Encoding, validation
        * - :py:class:`pyderasn.UTF8String`
          - utf-8
        * - :py:class:`pyderasn.NumericString`
-         - ascii
+         - proper alphabet validation
        * - :py:class:`pyderasn.PrintableString`
-         - ascii
+         - proper alphabet validation
        * - :py:class:`pyderasn.TeletexString`
-         - ascii
+         - iso-8859-1
        * - :py:class:`pyderasn.T61String`
-         - ascii
+         - iso-8859-1
        * - :py:class:`pyderasn.VideotexString`
          - iso-8859-1
        * - :py:class:`pyderasn.IA5String`
-         - ascii
+         - proper alphabet validation
        * - :py:class:`pyderasn.GraphicString`
          - iso-8859-1
-       * - :py:class:`pyderasn.VisibleString`
-         - ascii
-       * - :py:class:`pyderasn.ISO646String`
-         - ascii
+       * - :py:class:`pyderasn.VisibleString`, :py:class:`pyderasn.ISO646String`
+         - proper alphabet validation
        * - :py:class:`pyderasn.GeneralString`
          - iso-8859-1
        * - :py:class:`pyderasn.UniversalString`
@@ -4838,6 +4844,12 @@ class AllowableCharsMixin(object):
             return self._allowable_chars
         return frozenset(six_unichr(c) for c in self._allowable_chars)
 
+    def _value_sanitize(self, value):
+        value = super(AllowableCharsMixin, self)._value_sanitize(value)
+        if not frozenset(value) <= self._allowable_chars:
+            raise DecodeError("non satisfying alphabet value")
+        return value
+
 
 class NumericString(AllowableCharsMixin, CommonString):
     """Numeric string
@@ -4854,12 +4866,6 @@ class NumericString(AllowableCharsMixin, CommonString):
     asn1_type_name = "NumericString"
     _allowable_chars = frozenset(digits.encode("ascii") + b" ")
 
-    def _value_sanitize(self, value):
-        value = super(NumericString, self)._value_sanitize(value)
-        if not frozenset(value) <= self._allowable_chars:
-            raise DecodeError("non-numeric value")
-        return value
-
 
 PrintableStringState = namedtuple(
     "PrintableStringState",
@@ -4927,12 +4933,6 @@ class PrintableString(AllowableCharsMixin, CommonString):
         """
         return self._ampersand <= self._allowable_chars
 
-    def _value_sanitize(self, value):
-        value = super(PrintableString, self)._value_sanitize(value)
-        if not frozenset(value) <= self._allowable_chars:
-            raise DecodeError("non-printable value")
-        return value
-
     def __getstate__(self):
         return PrintableStringState(
             *super(PrintableString, self).__getstate__(),
@@ -4970,7 +4970,7 @@ class PrintableString(AllowableCharsMixin, CommonString):
 class TeletexString(CommonString):
     __slots__ = ()
     tag_default = tag_encode(20)
-    encoding = "ascii"
+    encoding = "iso-8859-1"
     asn1_type_name = "TeletexString"
 
 
@@ -4986,11 +4986,27 @@ class VideotexString(CommonString):
     asn1_type_name = "VideotexString"
 
 
-class IA5String(CommonString):
+class IA5String(AllowableCharsMixin, CommonString):
+    """IA5 string
+
+    Its value is properly sanitized: it is a mix of
+
+    * http://www.itscj.ipsj.or.jp/iso-ir/006.pdf (G)
+    * http://www.itscj.ipsj.or.jp/iso-ir/001.pdf (C0)
+    * DEL character (0x7F)
+
+    It is just 7-bit ASCII.
+
+    >>> IA5String().allowable_chars
+    frozenset(["NUL", ... "DEL"])
+    """
     __slots__ = ()
     tag_default = tag_encode(22)
     encoding = "ascii"
     asn1_type_name = "IA5"
+    _allowable_chars = frozenset(b"".join(
+        six_unichr(c).encode("ascii") for c in six_xrange(128)
+    ))
 
 
 LEN_YYMMDDHHMMSSZ = len("YYMMDDHHMMSSZ")
@@ -5001,11 +5017,27 @@ LEN_YYYYMMDDHHMMSSZ = len("YYYYMMDDHHMMSSZ")
 LEN_LEN_YYYYMMDDHHMMSSZ = len_encode(LEN_YYYYMMDDHHMMSSZ)
 
 
-class VisibleString(CommonString):
+class VisibleString(AllowableCharsMixin, CommonString):
+    """Visible string
+
+    Its value is properly sanitized. ASCII subset from space to tilde is
+    allowed: http://www.itscj.ipsj.or.jp/iso-ir/006.pdf
+
+    >>> VisibleString().allowable_chars
+    frozenset([" ", ... "~"])
+    """
     __slots__ = ()
     tag_default = tag_encode(26)
     encoding = "ascii"
     asn1_type_name = "VisibleString"
+    _allowable_chars = frozenset(b"".join(
+        six_unichr(c).encode("ascii") for c in six_xrange(ord(" "), ord("~") + 1)
+    ))
+
+
+class ISO646String(VisibleString):
+    __slots__ = ()
+    asn1_type_name = "ISO646String"
 
 
 UTCTimeState = namedtuple(
@@ -5425,11 +5457,6 @@ class GraphicString(CommonString):
     asn1_type_name = "GraphicString"
 
 
-class ISO646String(VisibleString):
-    __slots__ = ()
-    asn1_type_name = "ISO646String"
-
-
 class GeneralString(CommonString):
     __slots__ = ()
     tag_default = tag_encode(27)
@@ -7415,7 +7442,7 @@ def ascii_visualize(ba):
         92 2b 39 20 65 91 e6 8e  95 93 1a 58 df 02 78 ea  |.+9 e......X..x.|
                                                            ^^^^^^^^^^^^^^^^
     """
-    return "".join((chr(b) if 0x20 <= b <= 0x7E else ".") for b in ba)
+    return "".join((six_unichr(b) if 0x20 <= b <= 0x7E else ".") for b in ba)
 
 
 def hexdump(raw):