Preserve BER-related attributes during copy()

[pyderasn.git] / pyderasn.py

diff --git a/pyderasn.py b/pyderasn.py
index 4142af690d616b316ac46494dfde8857fcb25a02..1c7f3ed94603bda04df4aa52c38e8f2d293c0055 100755 (executable)
--- a/pyderasn.py
+++ b/pyderasn.py
@@ -1,7 +1,7 @@
 #!/usr/bin/env python
 # coding: utf-8
 # PyDERASN -- Python ASN.1 DER/BER codec with abstract structures
-# Copyright (C) 2017-2018 Sergey Matveev <stargrave@stargrave.org>
+# Copyright (C) 2017-2019 Sergey Matveev <stargrave@stargrave.org>
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Lesser General Public License as
@@ -384,7 +384,8 @@ constructed primitive types should be parsed successfully.
 
 * If object is encoded in BER form (not the DER one), then ``ber_encoded``
   attribute is set to True. Only ``BOOLEAN``, ``BIT STRING``, ``OCTET
-  STRING``, ``SEQUENCE``, ``SET``, ``SET OF`` can contain it.
+  STRING``, ``OBJECT IDENTIFIER``, ``SEQUENCE``, ``SET``, ``SET OF``
+  can contain it.
 * If object has an indefinite length encoding, then its ``lenindef``
   attribute is set to True. Only ``BIT STRING``, ``OCTET STRING``,
   ``SEQUENCE``, ``SET``, ``SEQUENCE OF``, ``SET OF``, ``ANY`` can
@@ -515,6 +516,7 @@ Various
 -------
 
 .. autofunction:: pyderasn.abs_decode_path
+.. autofunction:: pyderasn.colonize_hex
 .. autofunction:: pyderasn.hexenc
 .. autofunction:: pyderasn.hexdec
 .. autofunction:: pyderasn.tag_encode
@@ -1241,6 +1243,7 @@ class DecodePathDefBy(object):
 ########################################################################
 
 PP = namedtuple("PP", (
+    "obj",
     "asn1_type_name",
     "obj_name",
     "decode_path",
@@ -1266,6 +1269,7 @@ PP = namedtuple("PP", (
 
 
 def _pp(
+        obj=None,
         asn1_type_name="unknown",
         obj_name="unknown",
         decode_path=(),
@@ -1289,6 +1293,7 @@ def _pp(
         bered=False,
 ):
     return PP(
+        obj,
         asn1_type_name,
         obj_name,
         decode_path,
@@ -1317,6 +1322,12 @@ def _colourize(what, colour, with_colours, attrs=("bold",)):
     return colored(what, colour, attrs=attrs) if with_colours else what
 
 
+def colonize_hex(hexed):
+    """Separate hexadecimal string with colons
+    """
+    return ":".join(hexed[i:i + 2] for i in range(0, len(hexed), 2))
+
+
 def pp_console_row(
         pp,
         oids=None,
@@ -1387,6 +1398,15 @@ def pp_console_row(
                 value in oids
         ):
             cols.append(_colourize("(%s)" % oids[value], "green", with_colours))
+        if pp.asn1_type_name == Integer.asn1_type_name:
+            hex_repr = hex(int(pp.obj._value))[2:].upper()
+            if len(hex_repr) % 2 != 0:
+                hex_repr = "0" + hex_repr
+            cols.append(_colourize(
+                "(%s)" % colonize_hex(hex_repr),
+                "green",
+                with_colours,
+            ))
     if with_blob:
         if isinstance(pp.blob, binary_type):
             cols.append(hexenc(pp.blob))
@@ -1414,9 +1434,7 @@ def pp_console_blob(pp, decode_path_len_decrease=0):
         blob = hexenc(pp.blob).upper()
         for i in range(0, len(blob), 32):
             chunk = blob[i:i + 32]
-            yield " ".join(cols + [":".join(
-                chunk[j:j + 2] for j in range(0, len(chunk), 2)
-            )])
+            yield " ".join(cols + [colonize_hex(chunk)])
     elif isinstance(pp.blob, tuple):
         yield " ".join(cols + [", ".join(pp.blob)])
 
@@ -1551,6 +1569,9 @@ class Boolean(Obj):
         obj.offset = self.offset
         obj.llen = self.llen
         obj.vlen = self.vlen
+        obj.expl_lenindef = self.expl_lenindef
+        obj.lenindef = self.lenindef
+        obj.ber_encoded = self.ber_encoded
         return obj
 
     def __nonzero__(self):
@@ -1669,6 +1690,7 @@ class Boolean(Obj):
 
     def pps(self, decode_path=()):
         yield _pp(
+            obj=self,
             asn1_type_name=self.asn1_type_name,
             obj_name=self.__class__.__name__,
             decode_path=decode_path,
@@ -1806,6 +1828,9 @@ class Integer(Obj):
         obj.offset = self.offset
         obj.llen = self.llen
         obj.vlen = self.vlen
+        obj.expl_lenindef = self.expl_lenindef
+        obj.lenindef = self.lenindef
+        obj.ber_encoded = self.ber_encoded
         return obj
 
     def __int__(self):
@@ -1996,6 +2021,7 @@ class Integer(Obj):
 
     def pps(self, decode_path=()):
         yield _pp(
+            obj=self,
             asn1_type_name=self.asn1_type_name,
             obj_name=self.__class__.__name__,
             decode_path=decode_path,
@@ -2185,6 +2211,9 @@ class BitString(Obj):
         obj.offset = self.offset
         obj.llen = self.llen
         obj.vlen = self.vlen
+        obj.expl_lenindef = self.expl_lenindef
+        obj.lenindef = self.lenindef
+        obj.ber_encoded = self.ber_encoded
         return obj
 
     def __iter__(self):
@@ -2458,6 +2487,7 @@ class BitString(Obj):
             if len(self.specs) > 0:
                 blob = tuple(self.named)
         yield _pp(
+            obj=self,
             asn1_type_name=self.asn1_type_name,
             obj_name=self.__class__.__name__,
             decode_path=decode_path,
@@ -2596,6 +2626,9 @@ class OctetString(Obj):
         obj.offset = self.offset
         obj.llen = self.llen
         obj.vlen = self.vlen
+        obj.expl_lenindef = self.expl_lenindef
+        obj.lenindef = self.lenindef
+        obj.ber_encoded = self.ber_encoded
         return obj
 
     def __bytes__(self):
@@ -2809,6 +2842,7 @@ class OctetString(Obj):
 
     def pps(self, decode_path=()):
         yield _pp(
+            obj=self,
             asn1_type_name=self.asn1_type_name,
             obj_name=self.__class__.__name__,
             decode_path=decode_path,
@@ -2881,6 +2915,9 @@ class Null(Obj):
         obj.offset = self.offset
         obj.llen = self.llen
         obj.vlen = self.vlen
+        obj.expl_lenindef = self.expl_lenindef
+        obj.lenindef = self.lenindef
+        obj.ber_encoded = self.ber_encoded
         return obj
 
     def __eq__(self, their):
@@ -2954,6 +2991,7 @@ class Null(Obj):
 
     def pps(self, decode_path=()):
         yield _pp(
+            obj=self,
             asn1_type_name=self.asn1_type_name,
             obj_name=self.__class__.__name__,
             decode_path=decode_path,
@@ -3090,6 +3128,9 @@ class ObjectIdentifier(Obj):
         obj.offset = self.offset
         obj.llen = self.llen
         obj.vlen = self.vlen
+        obj.expl_lenindef = self.expl_lenindef
+        obj.lenindef = self.lenindef
+        obj.ber_encoded = self.ber_encoded
         return obj
 
     def __iter__(self):
@@ -3201,11 +3242,17 @@ class ObjectIdentifier(Obj):
             )
         v, tail = v[:l], v[l:]
         arcs = []
+        ber_encoded = False
         while len(v) > 0:
             i = 0
             arc = 0
             while True:
                 octet = indexbytes(v, i)
+                if i == 0 and octet == 0x80:
+                    if ctx.get("bered", False):
+                        ber_encoded = True
+                    else:
+                        raise DecodeError("non normalized arc encoding")
                 arc = (arc << 7) | (octet & 0x7F)
                 if octet & 0x80 == 0:
                     arcs.append(arc)
@@ -3237,6 +3284,8 @@ class ObjectIdentifier(Obj):
             optional=self.optional,
             _decoded=(offset, llen, l),
         )
+        if ber_encoded:
+            obj.ber_encoded = True
         return obj, tail
 
     def __repr__(self):
@@ -3244,6 +3293,7 @@ class ObjectIdentifier(Obj):
 
     def pps(self, decode_path=()):
         yield _pp(
+            obj=self,
             asn1_type_name=self.asn1_type_name,
             obj_name=self.__class__.__name__,
             decode_path=decode_path,
@@ -3261,6 +3311,7 @@ class ObjectIdentifier(Obj):
             expl_llen=self.expl_llen if self.expled else None,
             expl_vlen=self.expl_vlen if self.expled else None,
             expl_lenindef=self.expl_lenindef,
+            ber_encoded=self.ber_encoded,
             bered=self.bered,
         )
         for pp in self.pps_lenindef(decode_path):
@@ -3329,6 +3380,9 @@ class Enumerated(Integer):
         obj.offset = self.offset
         obj.llen = self.llen
         obj.vlen = self.vlen
+        obj.expl_lenindef = self.expl_lenindef
+        obj.lenindef = self.lenindef
+        obj.ber_encoded = self.ber_encoded
         return obj
 
     def __call__(
@@ -3470,6 +3524,7 @@ class CommonString(OctetString):
         if self.ready:
             value = hexenc(bytes(self)) if no_unicode else self.__unicode__()
         yield _pp(
+            obj=self,
             asn1_type_name=self.asn1_type_name,
             obj_name=self.__class__.__name__,
             decode_path=decode_path,
@@ -3702,6 +3757,7 @@ class UTCTime(CommonString):
 
     def pps(self, decode_path=()):
         yield _pp(
+            obj=self,
             asn1_type_name=self.asn1_type_name,
             obj_name=self.__class__.__name__,
             decode_path=decode_path,
@@ -3934,6 +3990,9 @@ class Choice(Obj):
         obj.offset = self.offset
         obj.llen = self.llen
         obj.vlen = self.vlen
+        obj.expl_lenindef = self.expl_lenindef
+        obj.lenindef = self.lenindef
+        obj.ber_encoded = self.ber_encoded
         value = self._value
         if value is not None:
             obj._value = (value[0], value[1].copy())
@@ -4054,6 +4113,7 @@ class Choice(Obj):
 
     def pps(self, decode_path=()):
         yield _pp(
+            obj=self,
             asn1_type_name=self.asn1_type_name,
             obj_name=self.__class__.__name__,
             decode_path=decode_path,
@@ -4173,6 +4233,9 @@ class Any(Obj):
         obj.offset = self.offset
         obj.llen = self.llen
         obj.vlen = self.vlen
+        obj.expl_lenindef = self.expl_lenindef
+        obj.lenindef = self.lenindef
+        obj.ber_encoded = self.ber_encoded
         return obj
 
     def __eq__(self, their):
@@ -4281,6 +4344,7 @@ class Any(Obj):
 
     def pps(self, decode_path=()):
         yield _pp(
+            obj=self,
             asn1_type_name=self.asn1_type_name,
             obj_name=self.__class__.__name__,
             decode_path=decode_path,
@@ -4511,6 +4575,9 @@ class Sequence(Obj):
         obj.offset = self.offset
         obj.llen = self.llen
         obj.vlen = self.vlen
+        obj.expl_lenindef = self.expl_lenindef
+        obj.lenindef = self.lenindef
+        obj.ber_encoded = self.ber_encoded
         obj._value = {k: v.copy() for k, v in self._value.items()}
         return obj
 
@@ -4779,6 +4846,7 @@ class Sequence(Obj):
 
     def pps(self, decode_path=()):
         yield _pp(
+            obj=self,
             asn1_type_name=self.asn1_type_name,
             obj_name=self.__class__.__name__,
             decode_path=decode_path,
@@ -5087,6 +5155,9 @@ class SequenceOf(Obj):
         obj.offset = self.offset
         obj.llen = self.llen
         obj.vlen = self.vlen
+        obj.expl_lenindef = self.expl_lenindef
+        obj.lenindef = self.lenindef
+        obj.ber_encoded = self.ber_encoded
         obj._value = [v.copy() for v in self._value]
         return obj
 
@@ -5284,6 +5355,7 @@ class SequenceOf(Obj):
 
     def pps(self, decode_path=()):
         yield _pp(
+            obj=self,
             asn1_type_name=self.asn1_type_name,
             obj_name=self.__class__.__name__,
             decode_path=decode_path,