Streaming of huge data support

[pyderasn.git] / pyderasn.py

diff --git a/pyderasn.py b/pyderasn.py
index fe1d4ab800624dfcb03307fe06b649f9ae259b3b..38f9fafdec12617fb68311b182dec1a5bca2eafd 100755 (executable)
--- a/pyderasn.py
+++ b/pyderasn.py
@@ -816,6 +816,7 @@ except ImportError:  # pragma: no cover
 __version__ = "7.0"
 
 __all__ = (
+    "agg_octet_string",
     "Any",
     "BitString",
     "BMPString",
@@ -3145,13 +3146,10 @@ class OctetString(Obj):
     >>> OctetString(b"hell", bounds=(4, 4))
     OCTET STRING 4 bytes 68656c6c
 
-    .. note::
-
-       Pay attention that OCTET STRING can be encoded both in primitive
-       and constructed forms. Decoder always checks constructed form tag
-       additionally to specified primitive one. If BER decoding is
-       :ref:`not enabled <bered_ctx>`, then decoder will fail, because
-       of DER restrictions.
+    Memoryviews can be used as a values. If memoryview is made on
+    mmap-ed file, then it does not take storage inside OctetString
+    itself. In CER encoding mode it will be streamed to the specified
+    writer, copying 1 KB chunks.
     """
     __slots__ = ("tag_constructed", "_bound_min", "_bound_max", "defined")
     tag_default = tag_encode(4)
@@ -3206,12 +3204,12 @@ class OctetString(Obj):
         )
 
     def _value_sanitize(self, value):
-        if value.__class__ == binary_type:
+        if value.__class__ == binary_type or value.__class__ == memoryview:
             pass
         elif issubclass(value.__class__, OctetString):
             value = value._value
         else:
-            raise InvalidValueType((self.__class__, bytes))
+            raise InvalidValueType((self.__class__, bytes, memoryview))
         if not self._bound_min <= len(value) <= self._bound_max:
             raise BoundsError(self._bound_min, len(value), self._bound_max)
         return value
@@ -3251,7 +3249,7 @@ class OctetString(Obj):
 
     def __bytes__(self):
         self._assert_ready()
-        return self._value
+        return bytes(self._value)
 
     def __eq__(self, their):
         if their.__class__ == binary_type:
@@ -3554,6 +3552,29 @@ class OctetString(Obj):
             yield pp
 
 
+def agg_octet_string(evgens, decode_path, raw, writer):
+    """Aggregate constructed string (OctetString and its derivatives)
+
+    :param evgens: iterator of generated events
+    :param decode_path: points to the string we want to decode
+    :param raw: slicebable (memoryview, bytearray, etc) with
+                the data evgens are generated one
+    :param writer: buffer.write where string is going to be saved
+    """
+    decode_path_len = len(decode_path)
+    for dp, obj, _ in evgens:
+        if dp[:decode_path_len] != decode_path:
+            continue
+        if not obj.ber_encoded:
+            write_full(writer, raw[
+                obj.offset + obj.tlen + obj.llen:
+                obj.offset + obj.tlen + obj.llen + obj.vlen -
+                (EOC_LEN if obj.expl_lenindef else 0)
+            ])
+        if len(dp) == decode_path_len:
+            break
+
+
 NullState = namedtuple("NullState", BasicState._fields, **NAMEDTUPLE_KWARGS)