]> Cypherpunks.ru repositories - pyderasn.git/blobdiff - pyderasn.py
Store OIDs as an array.array
[pyderasn.git] / pyderasn.py
index ced46ddeb0f4fde4ca6f9be75120e9d1d421e1c2..7bb1799863c454bb22bdbdc70c75411c47635f04 100755 (executable)
@@ -232,6 +232,7 @@ Currently available context options:
 * :ref:`allow_unordered_set <allow_unordered_set_ctx>`
 * :ref:`bered <bered_ctx>`
 * :ref:`defines_by_path <defines_by_path_ctx>`
 * :ref:`allow_unordered_set <allow_unordered_set_ctx>`
 * :ref:`bered <bered_ctx>`
 * :ref:`defines_by_path <defines_by_path_ctx>`
+* :ref:`evgen_mode_upto <evgen_mode_upto_ctx>`
 
 .. _pprinting:
 
 
 .. _pprinting:
 
@@ -352,7 +353,6 @@ Let's parse that output, human::
  (and its derivatives), ``SET``, ``SET OF``, ``UTCTime``, ``GeneralizedTime``
  could be BERed.
 
  (and its derivatives), ``SET``, ``SET OF``, ``UTCTime``, ``GeneralizedTime``
  could be BERed.
 
-
 .. _definedby:
 
 DEFINED BY
 .. _definedby:
 
 DEFINED BY
@@ -429,7 +429,7 @@ ______________________________
 
 Sometimes you either can not or do not want to explicitly set *defines*
 in the schema. You can dynamically apply those definitions when calling
 
 Sometimes you either can not or do not want to explicitly set *defines*
 in the schema. You can dynamically apply those definitions when calling
-``.decode()`` method.
+:py:meth:`pyderasn.Obj.decode` method.
 
 Specify ``defines_by_path`` key in the :ref:`decode context <ctx>`. Its
 value must be sequence of following tuples::
 
 Specify ``defines_by_path`` key in the :ref:`decode context <ctx>`. Its
 value must be sequence of following tuples::
@@ -493,9 +493,9 @@ useful for SEQUENCE/SET OF-s.
 BER encoding
 ------------
 
 BER encoding
 ------------
 
-By default PyDERASN accepts only DER encoded data. It always encodes to
-DER. But you can optionally enable BER decoding with setting ``bered``
-:ref:`context <ctx>` argument to True. Indefinite lengths and
+By default PyDERASN accepts only DER encoded data. By default it encodes
+to DER. But you can optionally enable BER decoding with setting
+``bered`` :ref:`context <ctx>` argument to True. Indefinite lengths and
 constructed primitive types should be parsed successfully.
 
 * If object is encoded in BER form (not the DER one), then ``ber_encoded``
 constructed primitive types should be parsed successfully.
 
 * If object is encoded in BER form (not the DER one), then ``ber_encoded``
@@ -534,6 +534,314 @@ lengths will be invalid in that case.
    This option should be used only for skipping some decode errors, just
    to see the decoded structure somehow.
 
    This option should be used only for skipping some decode errors, just
    to see the decoded structure somehow.
 
+.. _streaming:
+
+Streaming and dealing with huge structures
+------------------------------------------
+
+.. _evgen_mode:
+
+evgen mode
+__________
+
+ASN.1 structures can be huge, they can hold millions of objects inside
+(for example Certificate Revocation Lists (CRL), holding revocation
+state for every previously issued X.509 certificate). CACert.org's 8 MiB
+CRL file takes more than half a gigabyte of memory to hold the decoded
+structure.
+
+If you just simply want to check the signature over the ``tbsCertList``,
+you can create specialized schema with that field represented as
+OctetString for example::
+
+    class TBSCertListFast(Sequence):
+        schema = (
+            [...]
+            ("revokedCertificates", OctetString(
+                impl=SequenceOf.tag_default,
+                optional=True,
+            )),
+            [...]
+        )
+
+This allows you to quickly decode a few fields and check the signature
+over the ``tbsCertList`` bytes.
+
+But how can you get all certificate's serial number from it, after you
+trust that CRL after signature validation? You can use so called
+``evgen`` (event generation) mode, to catch the events/facts of some
+successful object decoding. Let's use command line capabilities::
+
+    $ python -m pyderasn --schema tests.test_crl:CertificateList --evgen revoke.crl
+         10     [1,1,   1]   . . version: Version INTEGER v2 (01) OPTIONAL
+         15     [1,1,   9]   . . . algorithm: OBJECT IDENTIFIER 1.2.840.113549.1.1.13
+         26     [0,0,   2]   . . . parameters: [UNIV 5] ANY OPTIONAL
+         13     [1,1,  13]   . . signature: AlgorithmIdentifier SEQUENCE
+         34     [1,1,   3]   . . . . . . type: AttributeType OBJECT IDENTIFIER 2.5.4.10
+         39     [0,0,   9]   . . . . . . value: [UNIV 19] AttributeValue ANY
+         32     [1,1,  14]   . . . . . 0: AttributeTypeAndValue SEQUENCE
+         30     [1,1,  16]   . . . . 0: RelativeDistinguishedName SET OF
+    [...]
+        188     [1,1,   1]   . . . . userCertificate: CertificateSerialNumber INTEGER 17 (11)
+        191     [1,1,  13]   . . . . . utcTime: UTCTime UTCTime 2003-04-01T14:25:08
+        191     [0,0,  15]   . . . . revocationDate: Time CHOICE utcTime
+        191     [1,1,  13]   . . . . . utcTime: UTCTime UTCTime 2003-04-01T14:25:08
+        186     [1,1,  18]   . . . 0: RevokedCertificate SEQUENCE
+        208     [1,1,   1]   . . . . userCertificate: CertificateSerialNumber INTEGER 20 (14)
+        211     [1,1,  13]   . . . . . utcTime: UTCTime UTCTime 2002-10-01T02:18:01
+        211     [0,0,  15]   . . . . revocationDate: Time CHOICE utcTime
+        211     [1,1,  13]   . . . . . utcTime: UTCTime UTCTime 2002-10-01T02:18:01
+        206     [1,1,  18]   . . . 1: RevokedCertificate SEQUENCE
+    [...]
+    9144992     [0,0,  15]   . . . . revocationDate: Time CHOICE utcTime
+    9144992     [1,1,  13]   . . . . . utcTime: UTCTime UTCTime 2020-02-08T07:25:06
+    9144985     [1,1,  20]   . . . 415755: RevokedCertificate SEQUENCE
+      181     [1,4,9144821]   . . revokedCertificates: RevokedCertificates SEQUENCE OF OPTIONAL
+        5     [1,4,9144997]   . tbsCertList: TBSCertList SEQUENCE
+    9145009     [1,1,   9]   . . algorithm: OBJECT IDENTIFIER 1.2.840.113549.1.1.13
+    9145020     [0,0,   2]   . . parameters: [UNIV 5] ANY OPTIONAL
+    9145007     [1,1,  13]   . signatureAlgorithm: AlgorithmIdentifier SEQUENCE
+    9145022     [1,3, 513]   . signatureValue: BIT STRING 4096 bits
+        0     [1,4,9145534]  CertificateList SEQUENCE
+
+Here we see how decoder works: it decodes SEQUENCE's tag, length, then
+decodes underlying values. It can not tell if SEQUENCE is decoded, so
+the event of the upper level SEQUENCE is the last one we see.
+``version`` field is just a single INTEGER -- it is decoded and event is
+fired immediately. Then we see that ``algorithm`` and ``parameters``
+fields are decoded and only after them the ``signature`` SEQUENCE is
+fired as a successfully decoded. There are 4 events for each revoked
+certificate entry in that CRL: ``userCertificate`` serial number,
+``utcTime`` of ``revocationDate`` CHOICE, ``RevokedCertificate`` itself
+as a one of entity in ``revokedCertificates`` SEQUENCE OF.
+
+We can do that in our ordinary Python code and understand where we are
+by looking at deterministically generated decode paths (do not forget
+about useful ``--print-decode-path`` CLI option). We must use
+:py:meth:`pyderasn.Obj.decode_evgen` method, instead of ordinary
+:py:meth:`pyderasn.Obj.decode`. It is generator yielding ``(decode_path,
+obj, tail)`` tuples::
+
+    for decode_path, obj, _ in CertificateList().decode_evgen(crl_raw):
+        if (
+            len(decode_path) == 4 and
+            decode_path[:2] == ("tbsCertList", "revokedCertificates"),
+            decode_path[3] == "userCertificate"
+        ):
+            print("serial number:", int(obj))
+
+Virtually it does not take any memory except at least needed for single
+object storage. You can easily use that mode to determine required
+object ``.offset`` and ``.*len`` to be able to decode it separately, or
+maybe verify signature upon it just by taking bytes by ``.offset`` and
+``.tlvlen``.
+
+.. _evgen_mode_upto_ctx:
+
+evgen_mode_upto
+_______________
+
+There is full ability to get any kind of data from the CRL in the
+example above. However it is not too convenient to get the whole
+``RevokedCertificate`` structure, that is pretty lightweight and one may
+do not want to disassemble it. You can use ``evgen_mode_upto``
+:ref:`ctx <ctx>` option that semantically equals to
+:ref:`defines_by_path <defines_by_path_ctx>` -- list of decode paths
+mapped to any non-None value. If specified decode path is met, then any
+subsequent objects won't be decoded in evgen mode. That allows us to
+parse the CRL above with fully assembled ``RevokedCertificate``::
+
+    for decode_path, obj, _ in CertificateList().decode_evgen(
+        crl_raw,
+        ctx={"evgen_mode_upto": (
+            (("tbsCertList", "revokedCertificates", any), True),
+        )},
+    ):
+        if (
+            len(decode_path) == 3 and
+            decode_path[:2] == ("tbsCertList", "revokedCertificates"),
+        ):
+            print("serial number:", int(obj["userCertificate"]))
+
+.. _mmap:
+
+mmap-ed file
+____________
+
+POSIX compliant systems have ``mmap`` syscall, giving ability to work
+the memory mapped file. You can deal with the file like it was an
+ordinary binary string, allowing you not to load it to the memory first.
+Also you can use them as an input for OCTET STRING, taking no Python
+memory for their storage.
+
+There is convenient :py:func:`pyderasn.file_mmaped` function that
+creates read-only memoryview on the file contents::
+
+    with open("huge", "rb") as fd:
+        raw = file_mmaped(fd)
+        obj = Something.decode(raw)
+
+.. warning::
+
+   mmap-ed files in Python2.7 does not implement buffer protocol, so
+   memoryview won't work on them.
+
+.. warning::
+
+   mmap maps the **whole** file. So it plays no role if you seek-ed it
+   before. Take the slice of the resulting memoryview with required
+   offset instead.
+
+.. note::
+
+   If you use ZFS as underlying storage, then pay attention that
+   currently most platforms does not deal good with ZFS ARC and ordinary
+   page cache used for mmaps. It can take twice the necessary size in
+   the memory: both in page cache and ZFS ARC.
+
+CER encoding
+____________
+
+We can parse any kind of data now, but how can we produce files
+streamingly, without storing their encoded representation in memory?
+SEQUENCE by default encodes in memory all its values, joins them in huge
+binary string, just to know the exact size of SEQUENCE's value for
+encoding it in TLV. DER requires you to know all exact sizes of the
+objects.
+
+You can use CER encoding mode, that slightly differs from the DER, but
+does not require exact sizes knowledge, allowing streaming encoding
+directly to some writer/buffer. Just use
+:py:meth:`pyderasn.Obj.encode_cer` method, providing the writer where
+encoded data will flow::
+
+    opener = io.open if PY2 else open
+    with opener("result", "wb") as fd:
+        obj.encode_cer(fd.write)
+
+::
+
+    buf = io.BytesIO()
+    obj.encode_cer(buf.write)
+
+If you do not want to create in-memory buffer every time, then you can
+use :py:func:`pyderasn.encode_cer` function::
+
+    data = encode_cer(obj)
+
+Remember that CER is **not valid** DER in most cases, so you **have to**
+use :ref:`bered <bered_ctx>` :ref:`ctx <ctx>` option during its
+decoding. Also currently there is **no** validation that provided CER is
+valid one -- you are sure that it has only valid BER encoding.
+
+.. warning::
+
+   SET OF values can not be streamingly encoded, because they are
+   required to be sorted byte-by-byte. Big SET OF values still will take
+   much memory. Use neither SET nor SET OF values, as modern ASN.1
+   also recommends too.
+
+Do not forget about using :ref:`mmap-ed <mmap>` memoryviews for your
+OCTET STRINGs! They will be streamingly copied from underlying file to
+the buffer using 1 KB chunks.
+
+Some structures require that some of the elements have to be forcefully
+DER encoded. For example ``SignedData`` CMS requires you to encode
+``SignedAttributes`` and X.509 certificates in DER form, allowing you to
+encode everything else in BER. You can tell any of the structures to be
+forcefully encoded in DER during CER encoding, by specifying
+``der_forced=True`` attribute::
+
+    class Certificate(Sequence):
+        schema = (...)
+        der_forced = True
+
+    class SignedAttributes(SetOf):
+        schema = Attribute()
+        bounds = (1, 32)
+        der_forced = True
+
+agg_octet_string
+________________
+
+In most cases, huge quantity of binary data is stored as OCTET STRING.
+CER encoding splits it on 1 KB chunks. BER allows splitting on various
+levels of chunks inclusion::
+
+    SOME STRING[CONSTRUCTED]
+        OCTET STRING[CONSTRUCTED]
+            OCTET STRING[PRIMITIVE]
+                DATA CHUNK
+            OCTET STRING[PRIMITIVE]
+                DATA CHUNK
+            OCTET STRING[PRIMITIVE]
+                DATA CHUNK
+        OCTET STRING[PRIMITIVE]
+            DATA CHUNK
+        OCTET STRING[CONSTRUCTED]
+            OCTET STRING[PRIMITIVE]
+                DATA CHUNK
+            OCTET STRING[PRIMITIVE]
+                DATA CHUNK
+        OCTET STRING[CONSTRUCTED]
+            OCTET STRING[CONSTRUCTED]
+                OCTET STRING[PRIMITIVE]
+                    DATA CHUNK
+
+You can not just take the offset and some ``.vlen`` of the STRING and
+treat it as the payload. If you decode it without
+:ref:`evgen mode <evgen_mode>`, then it will be automatically aggregated
+and ``bytes()`` will give the whole payload contents.
+
+You are forced to use :ref:`evgen mode <evgen_mode>` for decoding for
+small memory footprint. There is convenient
+:py:func:`pyderasn.agg_octet_string` helper for reconstructing the
+payload. Let's assume you have got BER/CER encoded ``ContentInfo`` with
+huge ``SignedData`` and ``EncapsulatedContentInfo``. Let's calculate the
+SHA512 digest of its ``eContent``::
+
+    fd = open("data.p7m", "rb")
+    raw = file_mmaped(fd)
+    ctx = {"bered": True}
+    for decode_path, obj, _ in ContentInfo().decode_evgen(raw, ctx=ctx):
+        if decode_path == ("content",):
+            content = obj
+            break
+    else:
+        raise ValueError("no content found")
+    hasher_state = sha512()
+    def hasher(data):
+        hasher_state.update(data)
+        return len(data)
+    evgens = SignedData().decode_evgen(
+        raw[content.offset:],
+        offset=content.offset,
+        ctx=ctx,
+    )
+    agg_octet_string(evgens, ("encapContentInfo", "eContent"), raw, hasher)
+    fd.close()
+    digest = hasher_state.digest()
+
+Simply replace ``hasher`` with some writeable file's ``fd.write`` to
+copy the payload (without BER/CER encoding interleaved overhead) in it.
+Virtually it won't take memory more than for keeping small structures
+and 1 KB binary chunks.
+
+SEQUENCE OF iterators
+_____________________
+
+You can use iterators as a value in :py:class:`pyderasn.SequenceOf`
+classes. The only difference with providing the full list of objects, is
+that type and bounds checking is done during encoding process. Also
+sequence's value will be emptied after encoding, forcing you to set its
+value again.
+
+This is very useful when you have to create some huge objects, like
+CRLs, with thousands and millions of entities inside. You can write the
+generator taking necessary data from the database and giving the
+``RevokedCertificate`` objects. Only binary representation of that
+objects will take memory during DER encoding.
+
 Base Obj
 --------
 .. autoclass:: pyderasn.Obj
 Base Obj
 --------
 .. autoclass:: pyderasn.Obj
@@ -550,12 +858,12 @@ _______
 Integer
 _______
 .. autoclass:: pyderasn.Integer
 Integer
 _______
 .. autoclass:: pyderasn.Integer
-   :members: __init__
+   :members: __init__, named
 
 BitString
 _________
 .. autoclass:: pyderasn.BitString
 
 BitString
 _________
 .. autoclass:: pyderasn.BitString
-   :members: __init__
+   :members: __init__, bit_len, named
 
 OctetString
 ___________
 
 OctetString
 ___________
@@ -587,7 +895,7 @@ _____________
 PrintableString
 _______________
 .. autoclass:: pyderasn.PrintableString
 PrintableString
 _______________
 .. autoclass:: pyderasn.PrintableString
-   :members: __init__
+   :members: __init__, allow_asterisk, allow_ampersand
 
 UTCTime
 _______
 
 UTCTime
 _______
@@ -597,6 +905,7 @@ _______
 GeneralizedTime
 _______________
 .. autoclass:: pyderasn.GeneralizedTime
 GeneralizedTime
 _______________
 .. autoclass:: pyderasn.GeneralizedTime
+   :members: __init__, todatetime
 
 Special types
 -------------
 
 Special types
 -------------
@@ -604,7 +913,7 @@ Special types
 Choice
 ______
 .. autoclass:: pyderasn.Choice
 Choice
 ______
 .. autoclass:: pyderasn.Choice
-   :members: __init__
+   :members: __init__, choice, value
 
 PrimitiveTypes
 ______________
 
 PrimitiveTypes
 ______________
@@ -642,7 +951,10 @@ Various
 -------
 
 .. autofunction:: pyderasn.abs_decode_path
 -------
 
 .. autofunction:: pyderasn.abs_decode_path
+.. autofunction:: pyderasn.agg_octet_string
 .. autofunction:: pyderasn.colonize_hex
 .. autofunction:: pyderasn.colonize_hex
+.. autofunction:: pyderasn.encode_cer
+.. autofunction:: pyderasn.file_mmaped
 .. autofunction:: pyderasn.hexenc
 .. autofunction:: pyderasn.hexdec
 .. autofunction:: pyderasn.tag_encode
 .. autofunction:: pyderasn.hexenc
 .. autofunction:: pyderasn.hexdec
 .. autofunction:: pyderasn.tag_encode
@@ -661,8 +973,118 @@ Various
 .. autoclass:: pyderasn.ObjNotReady
 .. autoclass:: pyderasn.InvalidValueType
 .. autoclass:: pyderasn.BoundsError
 .. autoclass:: pyderasn.ObjNotReady
 .. autoclass:: pyderasn.InvalidValueType
 .. autoclass:: pyderasn.BoundsError
+
+.. _cmdline:
+
+Command-line usage
+------------------
+
+You can decode DER/BER files using command line abilities::
+
+    $ python -m pyderasn --schema tests.test_crts:Certificate path/to/file
+
+If there is no schema for your file, then you can try parsing it without,
+but of course IMPLICIT tags will often make it impossible. But result is
+good enough for the certificate above::
+
+    $ python -m pyderasn path/to/file
+        0   [1,3,1604]  . >: SEQUENCE OF
+        4   [1,3,1453]  . . >: SEQUENCE OF
+        8   [0,0,   5]  . . . . >: [0] ANY
+                        . . . . . A0:03:02:01:02
+       13   [1,1,   3]  . . . . >: INTEGER 61595
+       18   [1,1,  13]  . . . . >: SEQUENCE OF
+       20   [1,1,   9]  . . . . . . >: OBJECT IDENTIFIER 1.2.840.113549.1.1.5
+       31   [1,1,   0]  . . . . . . >: NULL
+       33   [1,3, 274]  . . . . >: SEQUENCE OF
+       37   [1,1,  11]  . . . . . . >: SET OF
+       39   [1,1,   9]  . . . . . . . . >: SEQUENCE OF
+       41   [1,1,   3]  . . . . . . . . . . >: OBJECT IDENTIFIER 2.5.4.6
+       46   [1,1,   2]  . . . . . . . . . . >: PrintableString PrintableString ES
+    [...]
+     1409   [1,1,  50]  . . . . . . >: SEQUENCE OF
+     1411   [1,1,   8]  . . . . . . . . >: OBJECT IDENTIFIER 1.3.6.1.5.5.7.1.1
+     1421   [1,1,  38]  . . . . . . . . >: OCTET STRING 38 bytes
+                        . . . . . . . . . 30:24:30:22:06:08:2B:06:01:05:05:07:30:01:86:16
+                        . . . . . . . . . 68:74:74:70:3A:2F:2F:6F:63:73:70:2E:69:70:73:63
+                        . . . . . . . . . 61:2E:63:6F:6D:2F
+     1461   [1,1,  13]  . . >: SEQUENCE OF
+     1463   [1,1,   9]  . . . . >: OBJECT IDENTIFIER 1.2.840.113549.1.1.5
+     1474   [1,1,   0]  . . . . >: NULL
+     1476   [1,2, 129]  . . >: BIT STRING 1024 bits
+                        . . . 68:EE:79:97:97:DD:3B:EF:16:6A:06:F2:14:9A:6E:CD
+                        . . . 9E:12:F7:AA:83:10:BD:D1:7C:98:FA:C7:AE:D4:0E:2C
+    [...]
+
+Human readable OIDs
+___________________
+
+If you have got dictionaries with ObjectIdentifiers, like example one
+from ``tests/test_crts.py``::
+
+    stroid2name = {
+        "1.2.840.113549.1.1.1": "id-rsaEncryption",
+        "1.2.840.113549.1.1.5": "id-sha1WithRSAEncryption",
+        [...]
+        "2.5.4.10": "id-at-organizationName",
+        "2.5.4.11": "id-at-organizationalUnitName",
+    }
+
+then you can pass it to pretty printer to see human readable OIDs::
+
+    $ python -m pyderasn --oids tests.test_crts:stroid2name path/to/file
+    [...]
+       37   [1,1,  11]  . . . . . . >: SET OF
+       39   [1,1,   9]  . . . . . . . . >: SEQUENCE OF
+       41   [1,1,   3]  . . . . . . . . . . >: OBJECT IDENTIFIER id-at-countryName (2.5.4.6)
+       46   [1,1,   2]  . . . . . . . . . . >: PrintableString PrintableString ES
+       50   [1,1,  18]  . . . . . . >: SET OF
+       52   [1,1,  16]  . . . . . . . . >: SEQUENCE OF
+       54   [1,1,   3]  . . . . . . . . . . >: OBJECT IDENTIFIER id-at-stateOrProvinceName (2.5.4.8)
+       59   [1,1,   9]  . . . . . . . . . . >: PrintableString PrintableString Barcelona
+       70   [1,1,  18]  . . . . . . >: SET OF
+       72   [1,1,  16]  . . . . . . . . >: SEQUENCE OF
+       74   [1,1,   3]  . . . . . . . . . . >: OBJECT IDENTIFIER id-at-localityName (2.5.4.7)
+       79   [1,1,   9]  . . . . . . . . . . >: PrintableString PrintableString Barcelona
+    [...]
+
+Decode paths
+____________
+
+Each decoded element has so-called decode path: sequence of structure
+names it is passing during the decode process. Each element has its own
+unique path inside the whole ASN.1 tree. You can print it out with
+``--print-decode-path`` option::
+
+    $ python -m pyderasn --schema path.to:Certificate --print-decode-path path/to/file
+       0    [1,3,1604]  Certificate SEQUENCE []
+       4    [1,3,1453]   . tbsCertificate: TBSCertificate SEQUENCE [tbsCertificate]
+      10-2  [1,1,   1]   . . version: [0] EXPLICIT Version INTEGER v3 OPTIONAL [tbsCertificate:version]
+      13    [1,1,   3]   . . serialNumber: CertificateSerialNumber INTEGER 61595 [tbsCertificate:serialNumber]
+      18    [1,1,  13]   . . signature: AlgorithmIdentifier SEQUENCE [tbsCertificate:signature]
+      20    [1,1,   9]   . . . algorithm: OBJECT IDENTIFIER 1.2.840.113549.1.1.5 [tbsCertificate:signature:algorithm]
+      31    [0,0,   2]   . . . parameters: [UNIV 5] ANY OPTIONAL [tbsCertificate:signature:parameters]
+                         . . . . 05:00
+      33    [0,0, 278]   . . issuer: Name CHOICE rdnSequence [tbsCertificate:issuer]
+      33    [1,3, 274]   . . . rdnSequence: RDNSequence SEQUENCE OF [tbsCertificate:issuer:rdnSequence]
+      37    [1,1,  11]   . . . . 0: RelativeDistinguishedName SET OF [tbsCertificate:issuer:rdnSequence:0]
+      39    [1,1,   9]   . . . . . 0: AttributeTypeAndValue SEQUENCE [tbsCertificate:issuer:rdnSequence:0:0]
+      41    [1,1,   3]   . . . . . . type: AttributeType OBJECT IDENTIFIER 2.5.4.6 [tbsCertificate:issuer:rdnSequence:0:0:type]
+      46    [0,0,   4]   . . . . . . value: [UNIV 19] AttributeValue ANY [tbsCertificate:issuer:rdnSequence:0:0:value]
+                         . . . . . . . 13:02:45:53
+      46    [1,1,   2]   . . . . . . . DEFINED BY 2.5.4.6: CountryName PrintableString ES [tbsCertificate:issuer:rdnSequence:0:0:value:DEFINED BY 2.5.4.6]
+    [...]
+
+Now you can print only the specified tree, for example signature algorithm::
+
+    $ python -m pyderasn --schema path.to:Certificate --decode-path-only tbsCertificate:signature path/to/file
+      18    [1,1,  13]  AlgorithmIdentifier SEQUENCE
+      20    [1,1,   9]   . algorithm: OBJECT IDENTIFIER 1.2.840.113549.1.1.5
+      31    [0,0,   2]   . parameters: [UNIV 5] ANY OPTIONAL
+                         . . 05:00
 """
 
 """
 
+from array import array
 from codecs import getdecoder
 from codecs import getencoder
 from collections import namedtuple
 from codecs import getdecoder
 from codecs import getencoder
 from collections import namedtuple
@@ -670,7 +1092,11 @@ from collections import OrderedDict
 from copy import copy
 from datetime import datetime
 from datetime import timedelta
 from copy import copy
 from datetime import datetime
 from datetime import timedelta
+from io import BytesIO
 from math import ceil
 from math import ceil
+from mmap import mmap
+from mmap import PROT_READ
+from operator import attrgetter
 from string import ascii_letters
 from string import digits
 from sys import version_info
 from string import ascii_letters
 from string import digits
 from sys import version_info
@@ -698,9 +1124,10 @@ except ImportError:  # pragma: no cover
     def colored(what, *args, **kwargs):
         return what
 
     def colored(what, *args, **kwargs):
         return what
 
-__version__ = "6.3"
+__version__ = "7.0"
 
 __all__ = (
 
 __all__ = (
+    "agg_octet_string",
     "Any",
     "BitString",
     "BMPString",
     "Any",
     "BitString",
     "BMPString",
@@ -709,8 +1136,10 @@ __all__ = (
     "Choice",
     "DecodeError",
     "DecodePathDefBy",
     "Choice",
     "DecodeError",
     "DecodePathDefBy",
+    "encode_cer",
     "Enumerated",
     "ExceedingData",
     "Enumerated",
     "ExceedingData",
+    "file_mmaped",
     "GeneralizedTime",
     "GeneralString",
     "GraphicString",
     "GeneralizedTime",
     "GeneralString",
     "GraphicString",
@@ -776,8 +1205,17 @@ NAMEDTUPLE_KWARGS = {} if version_info < (3, 6) else {"module": __name__}
 SET01 = frozenset("01")
 DECIMALS = frozenset(digits)
 DECIMAL_SIGNS = ".,"
 SET01 = frozenset("01")
 DECIMALS = frozenset(digits)
 DECIMAL_SIGNS = ".,"
+NEXT_ATTR_NAME = "next" if PY2 else "__next__"
 
 
 
 
+def file_mmaped(fd):
+    """Make mmap-ed memoryview for reading from file
+
+    :param fd: file object
+    :returns: memoryview over read-only mmap-ing of the whole file
+    """
+    return memoryview(mmap(fd.fileno(), 0, prot=PROT_READ))
+
 def pureint(value):
     if not set(value) <= DECIMALS:
         raise ValueError("non-pure integer")
 def pureint(value):
     if not set(value) <= DECIMALS:
         raise ValueError("non-pure integer")
@@ -788,6 +1226,19 @@ def fractions2float(fractions_raw):
     return float("0." + fractions_raw)
 
 
     return float("0." + fractions_raw)
 
 
+def get_def_by_path(defines_by_path, sub_decode_path):
+    """Get define by decode path
+    """
+    for path, define in defines_by_path:
+        if len(path) != len(sub_decode_path):
+            continue
+        for p1, p2 in zip(path, sub_decode_path):
+            if (not p1 is any) and (p1 != p2):
+                break
+        else:
+            return define
+
+
 ########################################################################
 # Errors
 ########################################################################
 ########################################################################
 # Errors
 ########################################################################
@@ -1067,6 +1518,27 @@ def len_decode(data):
     return l, 1 + octets_num, data[1 + octets_num:]
 
 
     return l, 1 + octets_num, data[1 + octets_num:]
 
 
+LEN1K = len_encode(1000)
+
+
+def write_full(writer, data):
+    """Fully write provided data
+
+    :param writer: must comply with ``io.RawIOBase.write`` behaviour
+
+    BytesIO does not guarantee that the whole data will be written at
+    once. That function write everything provided, raising an error if
+    ``writer`` returns None.
+    """
+    data = memoryview(data)
+    written = 0
+    while written != len(data):
+        n = writer(data[written:])
+        if n is None:
+            raise ValueError("can not write to buf")
+        written += n
+
+
 ########################################################################
 # Base class
 ########################################################################
 ########################################################################
 # Base class
 ########################################################################
@@ -1077,6 +1549,22 @@ class AutoAddSlots(type):
         return type.__new__(cls, name, bases, _dict)
 
 
         return type.__new__(cls, name, bases, _dict)
 
 
+BasicState = namedtuple("BasicState", (
+    "version",
+    "tag",
+    "tag_order",
+    "expl",
+    "default",
+    "optional",
+    "offset",
+    "llen",
+    "vlen",
+    "expl_lenindef",
+    "lenindef",
+    "ber_encoded",
+), **NAMEDTUPLE_KWARGS)
+
+
 @add_metaclass(AutoAddSlots)
 class Obj(object):
     """Common ASN.1 object class
 @add_metaclass(AutoAddSlots)
 class Obj(object):
     """Common ASN.1 object class
@@ -1086,6 +1574,7 @@ class Obj(object):
     """
     __slots__ = (
         "tag",
     """
     __slots__ = (
         "tag",
+        "_tag_order",
         "_value",
         "_expl",
         "default",
         "_value",
         "_expl",
         "default",
@@ -1110,6 +1599,13 @@ class Obj(object):
         self._expl = getattr(self, "expl", None) if expl is None else expl
         if self.tag != self.tag_default and self._expl is not None:
             raise ValueError("implicit and explicit tags can not be set simultaneously")
         self._expl = getattr(self, "expl", None) if expl is None else expl
         if self.tag != self.tag_default and self._expl is not None:
             raise ValueError("implicit and explicit tags can not be set simultaneously")
+        if self.tag is None:
+            self._tag_order = None
+        else:
+            tag_class, _, tag_num = tag_decode(
+                self.tag if self._expl is None else self._expl
+            )
+            self._tag_order = (tag_class, tag_num)
         if default is not None:
             optional = True
         self.optional = optional
         if default is not None:
             optional = True
         self.optional = optional
@@ -1149,17 +1645,27 @@ class Obj(object):
     def __setstate__(self, state):
         if state.version != __version__:
             raise ValueError("data is pickled by different PyDERASN version")
     def __setstate__(self, state):
         if state.version != __version__:
             raise ValueError("data is pickled by different PyDERASN version")
-        self.tag = self.tag_default
-        self._value = None
-        self._expl = None
-        self.default = None
-        self.optional = False
-        self.offset = 0
-        self.llen = 0
-        self.vlen = 0
-        self.expl_lenindef = False
-        self.lenindef = False
-        self.ber_encoded = False
+        self.tag = state.tag
+        self._tag_order = state.tag_order
+        self._expl = state.expl
+        self.default = state.default
+        self.optional = state.optional
+        self.offset = state.offset
+        self.llen = state.llen
+        self.vlen = state.vlen
+        self.expl_lenindef = state.expl_lenindef
+        self.lenindef = state.lenindef
+        self.ber_encoded = state.ber_encoded
+
+    @property
+    def tag_order(self):
+        """Tag's (class, number) used for DER/CER sorting
+        """
+        return self._tag_order
+
+    @property
+    def tag_order_cer(self):
+        return self.tag_order
 
     @property
     def tlen(self):
 
     @property
     def tlen(self):
@@ -1191,11 +1697,11 @@ class Obj(object):
     def _encode(self):  # pragma: no cover
         raise NotImplementedError()
 
     def _encode(self):  # pragma: no cover
         raise NotImplementedError()
 
-    def _decode(self, tlv, offset, decode_path, ctx, tag_only):  # pragma: no cover
-        raise NotImplementedError()
+    def _decode(self, tlv, offset, decode_path, ctx, tag_only, evgen_mode):  # pragma: no cover
+        yield NotImplemented
 
     def encode(self):
 
     def encode(self):
-        """Encode the structure
+        """DER encode the structure
 
         :returns: DER representation
         """
 
         :returns: DER representation
         """
@@ -1204,6 +1710,26 @@ class Obj(object):
             return raw
         return b"".join((self._expl, len_encode(len(raw)), raw))
 
             return raw
         return b"".join((self._expl, len_encode(len(raw)), raw))
 
+    def encode_cer(self, writer):
+        """CER encode the structure to specified writer
+
+        :param writer: must comply with ``io.RawIOBase.write``
+                       behaviour. It takes slice to be written and
+                       returns number of bytes processed. If it returns
+                       None, then exception will be raised
+        """
+        if self._expl is not None:
+            write_full(writer, self._expl + LENINDEF)
+        if getattr(self, "der_forced", False):
+            write_full(writer, self._encode())
+        else:
+            self._encode_cer(writer)
+        if self._expl is not None:
+            write_full(writer, EOC)
+
+    def _encode_cer(self, writer):
+        write_full(writer, self._encode())
+
     def hexencode(self):
         """Do hexadecimal encoded :py:meth:`pyderasn.Obj.encode`
         """
     def hexencode(self):
         """Do hexadecimal encoded :py:meth:`pyderasn.Obj.encode`
         """
@@ -1225,32 +1751,77 @@ class Obj(object):
         :param int offset: initial data's offset
         :param bool leavemm: do we need to leave memoryview of remaining
                     data as is, or convert it to bytes otherwise
         :param int offset: initial data's offset
         :param bool leavemm: do we need to leave memoryview of remaining
                     data as is, or convert it to bytes otherwise
+        :param decode_path: current decode path (tuples of strings,
+                            possibly with DecodePathDefBy) with will be
+                            the root for all underlying objects
         :param ctx: optional :ref:`context <ctx>` governing decoding process
         :param ctx: optional :ref:`context <ctx>` governing decoding process
-        :param tag_only: decode only the tag, without length and contents
-                         (used only in Choice and Set structures, trying to
-                         determine if tag satisfies the schema)
-        :param _ctx_immutable: do we need to ``copy.copy()`` ``ctx``
-                               before using it?
+        :param bool tag_only: decode only the tag, without length and
+                              contents (used only in Choice and Set
+                              structures, trying to determine if tag satisfies
+                              the schema)
+        :param bool _ctx_immutable: do we need to ``copy.copy()`` ``ctx``
+                                    before using it?
         :returns: (Obj, remaining data)
 
         .. seealso:: :ref:`decoding`
         """
         :returns: (Obj, remaining data)
 
         .. seealso:: :ref:`decoding`
         """
+        result = next(self.decode_evgen(
+            data,
+            offset,
+            leavemm,
+            decode_path,
+            ctx,
+            tag_only,
+            _ctx_immutable,
+            _evgen_mode=False,
+        ))
+        if result is None:
+            return None
+        _, obj, tail = result
+        return obj, tail
+
+    def decode_evgen(
+            self,
+            data,
+            offset=0,
+            leavemm=False,
+            decode_path=(),
+            ctx=None,
+            tag_only=False,
+            _ctx_immutable=True,
+            _evgen_mode=True,
+    ):
+        """Decode with evgen mode on
+
+        That method is identical to :py:meth:`pyderasn.Obj.decode`, but
+        it returns the generator producing ``(decode_path, obj, tail)``
+        values. See :ref:`evgen mode <evgen_mode>`.
+        """
         if ctx is None:
             ctx = {}
         elif _ctx_immutable:
             ctx = copy(ctx)
         tlv = memoryview(data)
         if ctx is None:
             ctx = {}
         elif _ctx_immutable:
             ctx = copy(ctx)
         tlv = memoryview(data)
+        if (
+                _evgen_mode and
+                get_def_by_path(ctx.get("evgen_mode_upto", ()), decode_path) is not None
+        ):
+            _evgen_mode = False
         if self._expl is None:
         if self._expl is None:
-            result = self._decode(
-                tlv,
-                offset,
-                decode_path=decode_path,
-                ctx=ctx,
-                tag_only=tag_only,
-            )
-            if tag_only:
-                return None
-            obj, tail = result
+            for result in self._decode(
+                    tlv,
+                    offset=offset,
+                    decode_path=decode_path,
+                    ctx=ctx,
+                    tag_only=tag_only,
+                    evgen_mode=_evgen_mode,
+            ):
+                if tag_only:
+                    yield None
+                    return
+                _decode_path, obj, tail = result
+                if not _decode_path is decode_path:
+                    yield result
         else:
             try:
                 t, tlen, lv = tag_strip(tlv)
         else:
             try:
                 t, tlen, lv = tag_strip(tlv)
@@ -1279,16 +1850,20 @@ class Obj(object):
                     )
                 llen, v = 1, lv[1:]
                 offset += tlen + llen
                     )
                 llen, v = 1, lv[1:]
                 offset += tlen + llen
-                result = self._decode(
-                    v,
-                    offset=offset,
-                    decode_path=decode_path,
-                    ctx=ctx,
-                    tag_only=tag_only,
-                )
-                if tag_only:  # pragma: no cover
-                    return None
-                obj, tail = result
+                for result in self._decode(
+                        v,
+                        offset=offset,
+                        decode_path=decode_path,
+                        ctx=ctx,
+                        tag_only=tag_only,
+                        evgen_mode=_evgen_mode,
+                ):
+                    if tag_only:  # pragma: no cover
+                        yield None
+                        return
+                    _decode_path, obj, tail = result
+                    if not _decode_path is decode_path:
+                        yield result
                 eoc_expected, tail = tail[:EOC_LEN], tail[EOC_LEN:]
                 if eoc_expected.tobytes() != EOC:
                     raise DecodeError(
                 eoc_expected, tail = tail[:EOC_LEN], tail[EOC_LEN:]
                 if eoc_expected.tobytes() != EOC:
                     raise DecodeError(
@@ -1314,16 +1889,20 @@ class Obj(object):
                         decode_path=decode_path,
                         offset=offset,
                     )
                         decode_path=decode_path,
                         offset=offset,
                     )
-                result = self._decode(
-                    v,
-                    offset=offset + tlen + llen,
-                    decode_path=decode_path,
-                    ctx=ctx,
-                    tag_only=tag_only,
-                )
-                if tag_only:  # pragma: no cover
-                    return None
-                obj, tail = result
+                for result in self._decode(
+                        v,
+                        offset=offset + tlen + llen,
+                        decode_path=decode_path,
+                        ctx=ctx,
+                        tag_only=tag_only,
+                        evgen_mode=_evgen_mode,
+                ):
+                    if tag_only:  # pragma: no cover
+                        yield None
+                        return
+                    _decode_path, obj, tail = result
+                    if not _decode_path is decode_path:
+                        yield result
                 if obj.tlvlen < l and not ctx.get("allow_expl_oob", False):
                     raise DecodeError(
                         "explicit tag out-of-bound, longer than data",
                 if obj.tlvlen < l and not ctx.get("allow_expl_oob", False):
                     raise DecodeError(
                         "explicit tag out-of-bound, longer than data",
@@ -1331,7 +1910,7 @@ class Obj(object):
                         decode_path=decode_path,
                         offset=offset,
                     )
                         decode_path=decode_path,
                         offset=offset,
                     )
-        return obj, (tail if leavemm else tail.tobytes())
+        yield decode_path, obj, (tail if leavemm else tail.tobytes())
 
     def decod(self, data, offset=0, decode_path=(), ctx=None):
         """Decode the data, check that tail is empty
 
     def decod(self, data, offset=0, decode_path=(), ctx=None):
         """Decode the data, check that tail is empty
@@ -1452,6 +2031,16 @@ class Obj(object):
             )
 
 
             )
 
 
+def encode_cer(obj):
+    """Encode to CER in memory buffer
+
+    :returns bytes: memory buffer contents
+    """
+    buf = BytesIO()
+    obj.encode_cer(buf.write)
+    return buf.getvalue()
+
+
 class DecodePathDefBy(object):
     """DEFINED BY representation inside decode path
     """
 class DecodePathDefBy(object):
     """DEFINED BY representation inside decode path
     """
@@ -1690,6 +2279,7 @@ def pprint(
         with_colours=False,
         with_decode_path=False,
         decode_path_only=(),
         with_colours=False,
         with_decode_path=False,
         decode_path_only=(),
+        decode_path=(),
 ):
     """Pretty print object
 
 ):
     """Pretty print object
 
@@ -1742,27 +2332,18 @@ def pprint(
             else:
                 for row in _pprint_pps(pp):
                     yield row
             else:
                 for row in _pprint_pps(pp):
                     yield row
-    return "\n".join(_pprint_pps(obj.pps()))
+    return "\n".join(_pprint_pps(obj.pps(decode_path)))
 
 
 ########################################################################
 # ASN.1 primitive types
 ########################################################################
 
 
 
 ########################################################################
 # ASN.1 primitive types
 ########################################################################
 
-BooleanState = namedtuple("BooleanState", (
-    "version",
-    "value",
-    "tag",
-    "expl",
-    "default",
-    "optional",
-    "offset",
-    "llen",
-    "vlen",
-    "expl_lenindef",
-    "lenindef",
-    "ber_encoded",
-), **NAMEDTUPLE_KWARGS)
+BooleanState = namedtuple(
+    "BooleanState",
+    BasicState._fields + ("value",),
+    **NAMEDTUPLE_KWARGS
+)
 
 
 class Boolean(Obj):
 
 
 class Boolean(Obj):
@@ -1822,8 +2403,8 @@ class Boolean(Obj):
     def __getstate__(self):
         return BooleanState(
             __version__,
     def __getstate__(self):
         return BooleanState(
             __version__,
-            self._value,
             self.tag,
             self.tag,
+            self._tag_order,
             self._expl,
             self.default,
             self.optional,
             self._expl,
             self.default,
             self.optional,
@@ -1833,21 +2414,12 @@ class Boolean(Obj):
             self.expl_lenindef,
             self.lenindef,
             self.ber_encoded,
             self.expl_lenindef,
             self.lenindef,
             self.ber_encoded,
+            self._value,
         )
 
     def __setstate__(self, state):
         super(Boolean, self).__setstate__(state)
         self._value = state.value
         )
 
     def __setstate__(self, state):
         super(Boolean, self).__setstate__(state)
         self._value = state.value
-        self.tag = state.tag
-        self._expl = state.expl
-        self.default = state.default
-        self.optional = state.optional
-        self.offset = state.offset
-        self.llen = state.llen
-        self.vlen = state.vlen
-        self.expl_lenindef = state.expl_lenindef
-        self.lenindef = state.lenindef
-        self.ber_encoded = state.ber_encoded
 
     def __nonzero__(self):
         self._assert_ready()
 
     def __nonzero__(self):
         self._assert_ready()
@@ -1892,7 +2464,7 @@ class Boolean(Obj):
             (b"\xFF" if self._value else b"\x00"),
         ))
 
             (b"\xFF" if self._value else b"\x00"),
         ))
 
-    def _decode(self, tlv, offset, decode_path, ctx, tag_only):
+    def _decode(self, tlv, offset, decode_path, ctx, tag_only, evgen_mode):
         try:
             t, _, lv = tag_strip(tlv)
         except DecodeError as err:
         try:
             t, _, lv = tag_strip(tlv)
         except DecodeError as err:
@@ -1909,7 +2481,8 @@ class Boolean(Obj):
                 offset=offset,
             )
         if tag_only:
                 offset=offset,
             )
         if tag_only:
-            return None
+            yield None
+            return
         try:
             l, _, v = len_decode(lv)
         except DecodeError as err:
         try:
             l, _, v = len_decode(lv)
         except DecodeError as err:
@@ -1958,7 +2531,7 @@ class Boolean(Obj):
             _decoded=(offset, 1, 1),
         )
         obj.ber_encoded = ber_encoded
             _decoded=(offset, 1, 1),
         )
         obj.ber_encoded = ber_encoded
-        return obj, v[1:]
+        yield decode_path, obj, v[1:]
 
     def __repr__(self):
         return pp_console_row(next(self.pps()))
 
     def __repr__(self):
         return pp_console_row(next(self.pps()))
@@ -1990,23 +2563,11 @@ class Boolean(Obj):
             yield pp
 
 
             yield pp
 
 
-IntegerState = namedtuple("IntegerState", (
-    "version",
-    "specs",
-    "value",
-    "bound_min",
-    "bound_max",
-    "tag",
-    "expl",
-    "default",
-    "optional",
-    "offset",
-    "llen",
-    "vlen",
-    "expl_lenindef",
-    "lenindef",
-    "ber_encoded",
-), **NAMEDTUPLE_KWARGS)
+IntegerState = namedtuple(
+    "IntegerState",
+    BasicState._fields + ("specs", "value", "bound_min", "bound_max"),
+    **NAMEDTUPLE_KWARGS
+)
 
 
 class Integer(Obj):
 
 
 class Integer(Obj):
@@ -2113,11 +2674,8 @@ class Integer(Obj):
     def __getstate__(self):
         return IntegerState(
             __version__,
     def __getstate__(self):
         return IntegerState(
             __version__,
-            self.specs,
-            self._value,
-            self._bound_min,
-            self._bound_max,
             self.tag,
             self.tag,
+            self._tag_order,
             self._expl,
             self.default,
             self.optional,
             self._expl,
             self.default,
             self.optional,
@@ -2127,6 +2685,10 @@ class Integer(Obj):
             self.expl_lenindef,
             self.lenindef,
             self.ber_encoded,
             self.expl_lenindef,
             self.lenindef,
             self.ber_encoded,
+            self.specs,
+            self._value,
+            self._bound_min,
+            self._bound_max,
         )
 
     def __setstate__(self, state):
         )
 
     def __setstate__(self, state):
@@ -2135,16 +2697,6 @@ class Integer(Obj):
         self._value = state.value
         self._bound_min = state.bound_min
         self._bound_max = state.bound_max
         self._value = state.value
         self._bound_min = state.bound_min
         self._bound_max = state.bound_max
-        self.tag = state.tag
-        self._expl = state.expl
-        self.default = state.default
-        self.optional = state.optional
-        self.offset = state.offset
-        self.llen = state.llen
-        self.vlen = state.vlen
-        self.expl_lenindef = state.expl_lenindef
-        self.lenindef = state.lenindef
-        self.ber_encoded = state.ber_encoded
 
     def __int__(self):
         self._assert_ready()
 
     def __int__(self):
         self._assert_ready()
@@ -2174,6 +2726,8 @@ class Integer(Obj):
 
     @property
     def named(self):
 
     @property
     def named(self):
+        """Return named representation (if exists) of the value
+        """
         for name, value in iteritems(self.specs):
             if value == self._value:
                 return name
         for name, value in iteritems(self.specs):
             if value == self._value:
                 return name
@@ -2240,7 +2794,7 @@ class Integer(Obj):
                     break
         return b"".join((self.tag, len_encode(len(octets)), octets))
 
                     break
         return b"".join((self.tag, len_encode(len(octets)), octets))
 
-    def _decode(self, tlv, offset, decode_path, ctx, tag_only):
+    def _decode(self, tlv, offset, decode_path, ctx, tag_only, evgen_mode):
         try:
             t, _, lv = tag_strip(tlv)
         except DecodeError as err:
         try:
             t, _, lv = tag_strip(tlv)
         except DecodeError as err:
@@ -2257,7 +2811,8 @@ class Integer(Obj):
                 offset=offset,
             )
         if tag_only:
                 offset=offset,
             )
         if tag_only:
-            return None
+            yield None
+            return
         try:
             l, llen, v = len_decode(lv)
         except DecodeError as err:
         try:
             l, llen, v = len_decode(lv)
         except DecodeError as err:
@@ -2328,7 +2883,7 @@ class Integer(Obj):
                 decode_path=decode_path,
                 offset=offset,
             )
                 decode_path=decode_path,
                 offset=offset,
             )
-        return obj, tail
+        yield decode_path, obj, tail
 
     def __repr__(self):
         return pp_console_row(next(self.pps()))
 
     def __repr__(self):
         return pp_console_row(next(self.pps()))
@@ -2359,23 +2914,11 @@ class Integer(Obj):
             yield pp
 
 
             yield pp
 
 
-BitStringState = namedtuple("BitStringState", (
-    "version",
-    "specs",
-    "value",
-    "tag",
-    "expl",
-    "default",
-    "optional",
-    "offset",
-    "llen",
-    "vlen",
-    "expl_lenindef",
-    "lenindef",
-    "ber_encoded",
-    "tag_constructed",
-    "defined",
-), **NAMEDTUPLE_KWARGS)
+BitStringState = namedtuple(
+    "BitStringState",
+    BasicState._fields + ("specs", "value", "tag_constructed", "defined"),
+    **NAMEDTUPLE_KWARGS
+)
 
 
 class BitString(Obj):
 
 
 class BitString(Obj):
@@ -2533,9 +3076,8 @@ class BitString(Obj):
     def __getstate__(self):
         return BitStringState(
             __version__,
     def __getstate__(self):
         return BitStringState(
             __version__,
-            self.specs,
-            self._value,
             self.tag,
             self.tag,
+            self._tag_order,
             self._expl,
             self.default,
             self.optional,
             self._expl,
             self.default,
             self.optional,
@@ -2545,6 +3087,8 @@ class BitString(Obj):
             self.expl_lenindef,
             self.lenindef,
             self.ber_encoded,
             self.expl_lenindef,
             self.lenindef,
             self.ber_encoded,
+            self.specs,
+            self._value,
             self.tag_constructed,
             self.defined,
         )
             self.tag_constructed,
             self.defined,
         )
@@ -2553,16 +3097,6 @@ class BitString(Obj):
         super(BitString, self).__setstate__(state)
         self.specs = state.specs
         self._value = state.value
         super(BitString, self).__setstate__(state)
         self.specs = state.specs
         self._value = state.value
-        self.tag = state.tag
-        self._expl = state.expl
-        self.default = state.default
-        self.optional = state.optional
-        self.offset = state.offset
-        self.llen = state.llen
-        self.vlen = state.vlen
-        self.expl_lenindef = state.expl_lenindef
-        self.lenindef = state.lenindef
-        self.ber_encoded = state.ber_encoded
         self.tag_constructed = state.tag_constructed
         self.defined = state.defined
 
         self.tag_constructed = state.tag_constructed
         self.defined = state.defined
 
@@ -2573,6 +3107,8 @@ class BitString(Obj):
 
     @property
     def bit_len(self):
 
     @property
     def bit_len(self):
+        """Returns number of bits in the string
+        """
         self._assert_ready()
         return self._value[0]
 
         self._assert_ready()
         return self._value[0]
 
@@ -2593,6 +3129,10 @@ class BitString(Obj):
 
     @property
     def named(self):
 
     @property
     def named(self):
+        """Named representation (if exists) of the bits
+
+        :returns: [str(name), ...]
+        """
         return [name for name, bit in iteritems(self.specs) if self[bit]]
 
     def __call__(
         return [name for name, bit in iteritems(self.specs) if self[bit]]
 
     def __call__(
@@ -2638,7 +3178,31 @@ class BitString(Obj):
             octets,
         ))
 
             octets,
         ))
 
-    def _decode(self, tlv, offset, decode_path, ctx, tag_only):
+    def _encode_cer(self, writer):
+        bit_len, octets = self._value
+        if len(octets) + 1 <= 1000:
+            write_full(writer, self._encode())
+            return
+        write_full(writer, self.tag_constructed)
+        write_full(writer, LENINDEF)
+        for offset in six_xrange(0, (len(octets) // 999) * 999, 999):
+            write_full(writer, b"".join((
+                BitString.tag_default,
+                LEN1K,
+                int2byte(0),
+                octets[offset:offset + 999],
+            )))
+        tail = octets[offset+999:]
+        if len(tail) > 0:
+            tail = int2byte((8 - bit_len % 8) % 8) + tail
+            write_full(writer, b"".join((
+                BitString.tag_default,
+                len_encode(len(tail)),
+                tail,
+            )))
+        write_full(writer, EOC)
+
+    def _decode(self, tlv, offset, decode_path, ctx, tag_only, evgen_mode):
         try:
             t, tlen, lv = tag_strip(tlv)
         except DecodeError as err:
         try:
             t, tlen, lv = tag_strip(tlv)
         except DecodeError as err:
@@ -2650,7 +3214,8 @@ class BitString(Obj):
             )
         if t == self.tag:
             if tag_only:  # pragma: no cover
             )
         if t == self.tag:
             if tag_only:  # pragma: no cover
-                return None
+                yield None
+                return
             try:
                 l, llen, v = len_decode(lv)
             except DecodeError as err:
             try:
                 l, llen, v = len_decode(lv)
             except DecodeError as err:
@@ -2697,8 +3262,9 @@ class BitString(Obj):
                     offset=offset,
                 )
             v, tail = v[:l], v[l:]
                     offset=offset,
                 )
             v, tail = v[:l], v[l:]
+            bit_len = (len(v) - 1) * 8 - pad_size
             obj = self.__class__(
             obj = self.__class__(
-                value=((len(v) - 1) * 8 - pad_size, v[1:].tobytes()),
+                value=None if evgen_mode else (bit_len, v[1:].tobytes()),
                 impl=self.tag,
                 expl=self._expl,
                 default=self.default,
                 impl=self.tag,
                 expl=self._expl,
                 default=self.default,
@@ -2706,7 +3272,10 @@ class BitString(Obj):
                 _specs=self.specs,
                 _decoded=(offset, llen, l),
             )
                 _specs=self.specs,
                 _decoded=(offset, llen, l),
             )
-            return obj, tail
+            if evgen_mode:
+                obj._value = (bit_len, None)
+            yield decode_path, obj, tail
+            return
         if t != self.tag_constructed:
             raise TagMismatch(
                 klass=self.__class__,
         if t != self.tag_constructed:
             raise TagMismatch(
                 klass=self.__class__,
@@ -2721,7 +3290,8 @@ class BitString(Obj):
                 offset=offset,
             )
         if tag_only:  # pragma: no cover
                 offset=offset,
             )
         if tag_only:  # pragma: no cover
-            return None
+            yield None
+            return
         lenindef = False
         try:
             l, llen, v = len_decode(lv)
         lenindef = False
         try:
             l, llen, v = len_decode(lv)
@@ -2768,14 +3338,26 @@ class BitString(Obj):
                     )
             sub_decode_path = decode_path + (str(len(chunks)),)
             try:
                     )
             sub_decode_path = decode_path + (str(len(chunks)),)
             try:
-                chunk, v_tail = BitString().decode(
-                    v,
-                    offset=sub_offset,
-                    decode_path=sub_decode_path,
-                    leavemm=True,
-                    ctx=ctx,
-                    _ctx_immutable=False,
-                )
+                if evgen_mode:
+                    for _decode_path, chunk, v_tail in BitString().decode_evgen(
+                            v,
+                            offset=sub_offset,
+                            decode_path=sub_decode_path,
+                            leavemm=True,
+                            ctx=ctx,
+                            _ctx_immutable=False,
+                    ):
+                        yield _decode_path, chunk, v_tail
+                else:
+                    _, chunk, v_tail = next(BitString().decode_evgen(
+                        v,
+                        offset=sub_offset,
+                        decode_path=sub_decode_path,
+                        leavemm=True,
+                        ctx=ctx,
+                        _ctx_immutable=False,
+                        _evgen_mode=False,
+                    ))
             except TagMismatch:
                 raise DecodeError(
                     "expected BitString encoded chunk",
             except TagMismatch:
                 raise DecodeError(
                     "expected BitString encoded chunk",
@@ -2804,13 +3386,15 @@ class BitString(Obj):
                     decode_path=decode_path + (str(chunk_i),),
                     offset=chunk.offset,
                 )
                     decode_path=decode_path + (str(chunk_i),),
                     offset=chunk.offset,
                 )
-            values.append(bytes(chunk))
+            if not evgen_mode:
+                values.append(bytes(chunk))
             bit_len += chunk.bit_len
         chunk_last = chunks[-1]
             bit_len += chunk.bit_len
         chunk_last = chunks[-1]
-        values.append(bytes(chunk_last))
+        if not evgen_mode:
+            values.append(bytes(chunk_last))
         bit_len += chunk_last.bit_len
         obj = self.__class__(
         bit_len += chunk_last.bit_len
         obj = self.__class__(
-            value=(bit_len, b"".join(values)),
+            value=None if evgen_mode else (bit_len, b"".join(values)),
             impl=self.tag,
             expl=self._expl,
             default=self.default,
             impl=self.tag,
             expl=self._expl,
             default=self.default,
@@ -2818,9 +3402,11 @@ class BitString(Obj):
             _specs=self.specs,
             _decoded=(offset, llen, vlen + (EOC_LEN if lenindef else 0)),
         )
             _specs=self.specs,
             _decoded=(offset, llen, vlen + (EOC_LEN if lenindef else 0)),
         )
+        if evgen_mode:
+            obj._value = (bit_len, None)
         obj.lenindef = lenindef
         obj.ber_encoded = True
         obj.lenindef = lenindef
         obj.ber_encoded = True
-        return obj, (v[EOC_LEN:] if lenindef else v)
+        yield decode_path, obj, (v[EOC_LEN:] if lenindef else v)
 
     def __repr__(self):
         return pp_console_row(next(self.pps()))
 
     def __repr__(self):
         return pp_console_row(next(self.pps()))
@@ -2831,7 +3417,7 @@ class BitString(Obj):
         if self.ready:
             bit_len, blob = self._value
             value = "%d bits" % bit_len
         if self.ready:
             bit_len, blob = self._value
             value = "%d bits" % bit_len
-            if len(self.specs) > 0:
+            if len(self.specs) > 0 and blob is not None:
                 blob = tuple(self.named)
         yield _pp(
             obj=self,
                 blob = tuple(self.named)
         yield _pp(
             obj=self,
@@ -2866,24 +3452,17 @@ class BitString(Obj):
             yield pp
 
 
             yield pp
 
 
-OctetStringState = namedtuple("OctetStringState", (
-    "version",
-    "value",
-    "bound_min",
-    "bound_max",
-    "tag",
-    "expl",
-    "default",
-    "optional",
-    "offset",
-    "llen",
-    "vlen",
-    "expl_lenindef",
-    "lenindef",
-    "ber_encoded",
-    "tag_constructed",
-    "defined",
-), **NAMEDTUPLE_KWARGS)
+OctetStringState = namedtuple(
+    "OctetStringState",
+    BasicState._fields + (
+        "value",
+        "bound_min",
+        "bound_max",
+        "tag_constructed",
+        "defined",
+    ),
+    **NAMEDTUPLE_KWARGS
+)
 
 
 class OctetString(Obj):
 
 
 class OctetString(Obj):
@@ -2902,17 +3481,15 @@ class OctetString(Obj):
     >>> OctetString(b"hell", bounds=(4, 4))
     OCTET STRING 4 bytes 68656c6c
 
     >>> 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)
     asn1_type_name = "OCTET STRING"
     """
     __slots__ = ("tag_constructed", "_bound_min", "_bound_max", "defined")
     tag_default = tag_encode(4)
     asn1_type_name = "OCTET STRING"
+    evgen_mode_skip_value = True
 
     def __init__(
             self,
 
     def __init__(
             self,
@@ -2962,12 +3539,12 @@ class OctetString(Obj):
         )
 
     def _value_sanitize(self, value):
         )
 
     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:
             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
         if not self._bound_min <= len(value) <= self._bound_max:
             raise BoundsError(self._bound_min, len(value), self._bound_max)
         return value
@@ -2979,10 +3556,8 @@ class OctetString(Obj):
     def __getstate__(self):
         return OctetStringState(
             __version__,
     def __getstate__(self):
         return OctetStringState(
             __version__,
-            self._value,
-            self._bound_min,
-            self._bound_max,
             self.tag,
             self.tag,
+            self._tag_order,
             self._expl,
             self.default,
             self.optional,
             self._expl,
             self.default,
             self.optional,
@@ -2992,6 +3567,9 @@ class OctetString(Obj):
             self.expl_lenindef,
             self.lenindef,
             self.ber_encoded,
             self.expl_lenindef,
             self.lenindef,
             self.ber_encoded,
+            self._value,
+            self._bound_min,
+            self._bound_max,
             self.tag_constructed,
             self.defined,
         )
             self.tag_constructed,
             self.defined,
         )
@@ -3001,22 +3579,12 @@ class OctetString(Obj):
         self._value = state.value
         self._bound_min = state.bound_min
         self._bound_max = state.bound_max
         self._value = state.value
         self._bound_min = state.bound_min
         self._bound_max = state.bound_max
-        self.tag = state.tag
-        self._expl = state.expl
-        self.default = state.default
-        self.optional = state.optional
-        self.offset = state.offset
-        self.llen = state.llen
-        self.vlen = state.vlen
-        self.expl_lenindef = state.expl_lenindef
-        self.lenindef = state.lenindef
-        self.ber_encoded = state.ber_encoded
         self.tag_constructed = state.tag_constructed
         self.defined = state.defined
 
     def __bytes__(self):
         self._assert_ready()
         self.tag_constructed = state.tag_constructed
         self.defined = state.defined
 
     def __bytes__(self):
         self._assert_ready()
-        return self._value
+        return bytes(self._value)
 
     def __eq__(self, their):
         if their.__class__ == binary_type:
 
     def __eq__(self, their):
         if their.__class__ == binary_type:
@@ -3061,7 +3629,29 @@ class OctetString(Obj):
             self._value,
         ))
 
             self._value,
         ))
 
-    def _decode(self, tlv, offset, decode_path, ctx, tag_only):
+    def _encode_cer(self, writer):
+        octets = self._value
+        if len(octets) <= 1000:
+            write_full(writer, self._encode())
+            return
+        write_full(writer, self.tag_constructed)
+        write_full(writer, LENINDEF)
+        for offset in six_xrange(0, (len(octets) // 1000) * 1000, 1000):
+            write_full(writer, b"".join((
+                OctetString.tag_default,
+                LEN1K,
+                octets[offset:offset + 1000],
+            )))
+        tail = octets[offset+1000:]
+        if len(tail) > 0:
+            write_full(writer, b"".join((
+                OctetString.tag_default,
+                len_encode(len(tail)),
+                tail,
+            )))
+        write_full(writer, EOC)
+
+    def _decode(self, tlv, offset, decode_path, ctx, tag_only, evgen_mode):
         try:
             t, tlen, lv = tag_strip(tlv)
         except DecodeError as err:
         try:
             t, tlen, lv = tag_strip(tlv)
         except DecodeError as err:
@@ -3073,7 +3663,8 @@ class OctetString(Obj):
             )
         if t == self.tag:
             if tag_only:
             )
         if t == self.tag:
             if tag_only:
-                return None
+                yield None
+                return
             try:
                 l, llen, v = len_decode(lv)
             except DecodeError as err:
             try:
                 l, llen, v = len_decode(lv)
             except DecodeError as err:
@@ -3091,9 +3682,19 @@ class OctetString(Obj):
                     offset=offset,
                 )
             v, tail = v[:l], v[l:]
                     offset=offset,
                 )
             v, tail = v[:l], v[l:]
+            if evgen_mode and not self._bound_min <= len(v) <= self._bound_max:
+                raise DecodeError(
+                    msg=str(BoundsError(self._bound_min, len(v), self._bound_max)),
+                    klass=self.__class__,
+                    decode_path=decode_path,
+                    offset=offset,
+                )
             try:
                 obj = self.__class__(
             try:
                 obj = self.__class__(
-                    value=v.tobytes(),
+                    value=(
+                        None if (evgen_mode and self.evgen_mode_skip_value)
+                        else v.tobytes()
+                    ),
                     bounds=(self._bound_min, self._bound_max),
                     impl=self.tag,
                     expl=self._expl,
                     bounds=(self._bound_min, self._bound_max),
                     impl=self.tag,
                     expl=self._expl,
@@ -3116,7 +3717,8 @@ class OctetString(Obj):
                     decode_path=decode_path,
                     offset=offset,
                 )
                     decode_path=decode_path,
                     offset=offset,
                 )
-            return obj, tail
+            yield decode_path, obj, tail
+            return
         if t != self.tag_constructed:
             raise TagMismatch(
                 klass=self.__class__,
         if t != self.tag_constructed:
             raise TagMismatch(
                 klass=self.__class__,
@@ -3131,7 +3733,8 @@ class OctetString(Obj):
                 offset=offset,
             )
         if tag_only:
                 offset=offset,
             )
         if tag_only:
-            return None
+            yield None
+            return
         lenindef = False
         try:
             l, llen, v = len_decode(lv)
         lenindef = False
         try:
             l, llen, v = len_decode(lv)
@@ -3153,8 +3756,10 @@ class OctetString(Obj):
                 offset=offset,
             )
         chunks = []
                 offset=offset,
             )
         chunks = []
+        chunks_count = 0
         sub_offset = offset + tlen + llen
         vlen = 0
         sub_offset = offset + tlen + llen
         vlen = 0
+        payload_len = 0
         while True:
             if lenindef:
                 if v[:EOC_LEN].tobytes() == EOC:
         while True:
             if lenindef:
                 if v[:EOC_LEN].tobytes() == EOC:
@@ -3169,16 +3774,33 @@ class OctetString(Obj):
                         decode_path=decode_path + (str(len(chunks) - 1),),
                         offset=chunks[-1].offset,
                     )
                         decode_path=decode_path + (str(len(chunks) - 1),),
                         offset=chunks[-1].offset,
                     )
-            sub_decode_path = decode_path + (str(len(chunks)),)
             try:
             try:
-                chunk, v_tail = OctetString().decode(
-                    v,
-                    offset=sub_offset,
-                    decode_path=sub_decode_path,
-                    leavemm=True,
-                    ctx=ctx,
-                    _ctx_immutable=False,
-                )
+                if evgen_mode:
+                    sub_decode_path = decode_path + (str(chunks_count),)
+                    for _decode_path, chunk, v_tail in OctetString().decode_evgen(
+                            v,
+                            offset=sub_offset,
+                            decode_path=sub_decode_path,
+                            leavemm=True,
+                            ctx=ctx,
+                            _ctx_immutable=False,
+                    ):
+                        yield _decode_path, chunk, v_tail
+                        if not chunk.ber_encoded:
+                            payload_len += chunk.vlen
+                    chunks_count += 1
+                else:
+                    sub_decode_path = decode_path + (str(len(chunks)),)
+                    _, chunk, v_tail = next(OctetString().decode_evgen(
+                        v,
+                        offset=sub_offset,
+                        decode_path=sub_decode_path,
+                        leavemm=True,
+                        ctx=ctx,
+                        _ctx_immutable=False,
+                        _evgen_mode=False,
+                    ))
+                    chunks.append(chunk)
             except TagMismatch:
                 raise DecodeError(
                     "expected OctetString encoded chunk",
             except TagMismatch:
                 raise DecodeError(
                     "expected OctetString encoded chunk",
@@ -3186,13 +3808,22 @@ class OctetString(Obj):
                     decode_path=sub_decode_path,
                     offset=sub_offset,
                 )
                     decode_path=sub_decode_path,
                     offset=sub_offset,
                 )
-            chunks.append(chunk)
             sub_offset += chunk.tlvlen
             vlen += chunk.tlvlen
             v = v_tail
             sub_offset += chunk.tlvlen
             vlen += chunk.tlvlen
             v = v_tail
+        if evgen_mode and not self._bound_min <= payload_len <= self._bound_max:
+            raise DecodeError(
+                msg=str(BoundsError(self._bound_min, payload_len, self._bound_max)),
+                klass=self.__class__,
+                decode_path=decode_path,
+                offset=offset,
+            )
         try:
             obj = self.__class__(
         try:
             obj = self.__class__(
-                value=b"".join(bytes(chunk) for chunk in chunks),
+                value=(
+                    None if evgen_mode else
+                    b"".join(bytes(chunk) for chunk in chunks)
+                ),
                 bounds=(self._bound_min, self._bound_max),
                 impl=self.tag,
                 expl=self._expl,
                 bounds=(self._bound_min, self._bound_max),
                 impl=self.tag,
                 expl=self._expl,
@@ -3217,7 +3848,7 @@ class OctetString(Obj):
             )
         obj.lenindef = lenindef
         obj.ber_encoded = True
             )
         obj.lenindef = lenindef
         obj.ber_encoded = True
-        return obj, (v[EOC_LEN:] if lenindef else v)
+        yield decode_path, obj, (v[EOC_LEN:] if lenindef else v)
 
     def __repr__(self):
         return pp_console_row(next(self.pps()))
 
     def __repr__(self):
         return pp_console_row(next(self.pps()))
@@ -3256,19 +3887,32 @@ class OctetString(Obj):
             yield pp
 
 
             yield pp
 
 
-NullState = namedtuple("NullState", (
-    "version",
-    "tag",
-    "expl",
-    "default",
-    "optional",
-    "offset",
-    "llen",
-    "vlen",
-    "expl_lenindef",
-    "lenindef",
-    "ber_encoded",
-), **NAMEDTUPLE_KWARGS)
+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 on
+    :param writer: buffer.write where string is going to be saved
+    :param writer: where string is going to be saved. Must comply
+                   with ``io.RawIOBase.write`` behaviour
+    """
+    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)
 
 
 class Null(Obj):
 
 
 class Null(Obj):
@@ -3307,6 +3951,7 @@ class Null(Obj):
         return NullState(
             __version__,
             self.tag,
         return NullState(
             __version__,
             self.tag,
+            self._tag_order,
             self._expl,
             self.default,
             self.optional,
             self._expl,
             self.default,
             self.optional,
@@ -3318,19 +3963,6 @@ class Null(Obj):
             self.ber_encoded,
         )
 
             self.ber_encoded,
         )
 
-    def __setstate__(self, state):
-        super(Null, self).__setstate__(state)
-        self.tag = state.tag
-        self._expl = state.expl
-        self.default = state.default
-        self.optional = state.optional
-        self.offset = state.offset
-        self.llen = state.llen
-        self.vlen = state.vlen
-        self.expl_lenindef = state.expl_lenindef
-        self.lenindef = state.lenindef
-        self.ber_encoded = state.ber_encoded
-
     def __eq__(self, their):
         if not issubclass(their.__class__, Null):
             return False
     def __eq__(self, their):
         if not issubclass(their.__class__, Null):
             return False
@@ -3355,7 +3987,7 @@ class Null(Obj):
     def _encode(self):
         return self.tag + len_encode(0)
 
     def _encode(self):
         return self.tag + len_encode(0)
 
-    def _decode(self, tlv, offset, decode_path, ctx, tag_only):
+    def _decode(self, tlv, offset, decode_path, ctx, tag_only, evgen_mode):
         try:
             t, _, lv = tag_strip(tlv)
         except DecodeError as err:
         try:
             t, _, lv = tag_strip(tlv)
         except DecodeError as err:
@@ -3372,7 +4004,8 @@ class Null(Obj):
                 offset=offset,
             )
         if tag_only:  # pragma: no cover
                 offset=offset,
             )
         if tag_only:  # pragma: no cover
-            return None
+            yield None
+            return
         try:
             l, _, v = len_decode(lv)
         except DecodeError as err:
         try:
             l, _, v = len_decode(lv)
         except DecodeError as err:
@@ -3395,7 +4028,7 @@ class Null(Obj):
             optional=self.optional,
             _decoded=(offset, 1, 0),
         )
             optional=self.optional,
             _decoded=(offset, 1, 0),
         )
-        return obj, v
+        yield decode_path, obj, v
 
     def __repr__(self):
         return pp_console_row(next(self.pps()))
 
     def __repr__(self):
         return pp_console_row(next(self.pps()))
@@ -3424,21 +4057,11 @@ class Null(Obj):
             yield pp
 
 
             yield pp
 
 
-ObjectIdentifierState = namedtuple("ObjectIdentifierState", (
-    "version",
-    "value",
-    "tag",
-    "expl",
-    "default",
-    "optional",
-    "offset",
-    "llen",
-    "vlen",
-    "expl_lenindef",
-    "lenindef",
-    "ber_encoded",
-    "defines",
-), **NAMEDTUPLE_KWARGS)
+ObjectIdentifierState = namedtuple(
+    "ObjectIdentifierState",
+    BasicState._fields + ("value", "defines"),
+    **NAMEDTUPLE_KWARGS
+)
 
 
 class ObjectIdentifier(Obj):
 
 
 class ObjectIdentifier(Obj):
@@ -3508,7 +4131,7 @@ class ObjectIdentifier(Obj):
 
     def __add__(self, their):
         if their.__class__ == tuple:
 
     def __add__(self, their):
         if their.__class__ == tuple:
-            return self.__class__(self._value + their)
+            return self.__class__(self._value + array("L", their))
         if isinstance(their, self.__class__):
             return self.__class__(self._value + their._value)
         raise InvalidValueType((self.__class__, tuple))
         if isinstance(their, self.__class__):
             return self.__class__(self._value + their._value)
         raise InvalidValueType((self.__class__, tuple))
@@ -3518,10 +4141,15 @@ class ObjectIdentifier(Obj):
             return value._value
         if isinstance(value, string_types):
             try:
             return value._value
         if isinstance(value, string_types):
             try:
-                value = tuple(pureint(arc) for arc in value.split("."))
+                value = array("L", (pureint(arc) for arc in value.split(".")))
             except ValueError:
                 raise InvalidOID("unacceptable arcs values")
         if value.__class__ == tuple:
             except ValueError:
                 raise InvalidOID("unacceptable arcs values")
         if value.__class__ == tuple:
+            try:
+                value = array("L", value)
+            except OverflowError as err:
+                raise InvalidOID(repr(err))
+        if value.__class__ is array:
             if len(value) < 2:
                 raise InvalidOID("less than 2 arcs")
             first_arc = value[0]
             if len(value) < 2:
                 raise InvalidOID("less than 2 arcs")
             first_arc = value[0]
@@ -3544,8 +4172,8 @@ class ObjectIdentifier(Obj):
     def __getstate__(self):
         return ObjectIdentifierState(
             __version__,
     def __getstate__(self):
         return ObjectIdentifierState(
             __version__,
-            self._value,
             self.tag,
             self.tag,
+            self._tag_order,
             self._expl,
             self.default,
             self.optional,
             self._expl,
             self.default,
             self.optional,
@@ -3555,22 +4183,13 @@ class ObjectIdentifier(Obj):
             self.expl_lenindef,
             self.lenindef,
             self.ber_encoded,
             self.expl_lenindef,
             self.lenindef,
             self.ber_encoded,
+            self._value,
             self.defines,
         )
 
     def __setstate__(self, state):
         super(ObjectIdentifier, self).__setstate__(state)
         self._value = state.value
             self.defines,
         )
 
     def __setstate__(self, state):
         super(ObjectIdentifier, self).__setstate__(state)
         self._value = state.value
-        self.tag = state.tag
-        self._expl = state.expl
-        self.default = state.default
-        self.optional = state.optional
-        self.offset = state.offset
-        self.llen = state.llen
-        self.vlen = state.vlen
-        self.expl_lenindef = state.expl_lenindef
-        self.lenindef = state.lenindef
-        self.ber_encoded = state.ber_encoded
         self.defines = state.defines
 
     def __iter__(self):
         self.defines = state.defines
 
     def __iter__(self):
@@ -3590,7 +4209,7 @@ class ObjectIdentifier(Obj):
 
     def __eq__(self, their):
         if their.__class__ == tuple:
 
     def __eq__(self, their):
         if their.__class__ == tuple:
-            return self._value == their
+            return self._value == array("L", their)
         if not issubclass(their.__class__, ObjectIdentifier):
             return False
         return (
         if not issubclass(their.__class__, ObjectIdentifier):
             return False
         return (
@@ -3639,7 +4258,7 @@ class ObjectIdentifier(Obj):
         v = b"".join(octets)
         return b"".join((self.tag, len_encode(len(v)), v))
 
         v = b"".join(octets)
         return b"".join((self.tag, len_encode(len(v)), v))
 
-    def _decode(self, tlv, offset, decode_path, ctx, tag_only):
+    def _decode(self, tlv, offset, decode_path, ctx, tag_only, evgen_mode):
         try:
             t, _, lv = tag_strip(tlv)
         except DecodeError as err:
         try:
             t, _, lv = tag_strip(tlv)
         except DecodeError as err:
@@ -3656,7 +4275,8 @@ class ObjectIdentifier(Obj):
                 offset=offset,
             )
         if tag_only:  # pragma: no cover
                 offset=offset,
             )
         if tag_only:  # pragma: no cover
-            return None
+            yield None
+            return
         try:
             l, llen, v = len_decode(lv)
         except DecodeError as err:
         try:
             l, llen, v = len_decode(lv)
         except DecodeError as err:
@@ -3681,7 +4301,7 @@ class ObjectIdentifier(Obj):
                 offset=offset,
             )
         v, tail = v[:l], v[l:]
                 offset=offset,
             )
         v, tail = v[:l], v[l:]
-        arcs = []
+        arcs = array("L")
         ber_encoded = False
         while len(v) > 0:
             i = 0
         ber_encoded = False
         while len(v) > 0:
             i = 0
@@ -3692,10 +4312,23 @@ class ObjectIdentifier(Obj):
                     if ctx.get("bered", False):
                         ber_encoded = True
                     else:
                     if ctx.get("bered", False):
                         ber_encoded = True
                     else:
-                        raise DecodeError("non normalized arc encoding")
+                        raise DecodeError(
+                            "non normalized arc encoding",
+                            klass=self.__class__,
+                            decode_path=decode_path,
+                            offset=offset,
+                        )
                 arc = (arc << 7) | (octet & 0x7F)
                 if octet & 0x80 == 0:
                 arc = (arc << 7) | (octet & 0x7F)
                 if octet & 0x80 == 0:
-                    arcs.append(arc)
+                    try:
+                        arcs.append(arc)
+                    except OverflowError:
+                        raise DecodeError(
+                            "too huge value for local unsigned long",
+                            klass=self.__class__,
+                            decode_path=decode_path,
+                            offset=offset,
+                        )
                     v = v[i + 1:]
                     break
                 i += 1
                     v = v[i + 1:]
                     break
                 i += 1
@@ -3717,7 +4350,7 @@ class ObjectIdentifier(Obj):
             first_arc = 2
             second_arc -= 80
         obj = self.__class__(
             first_arc = 2
             second_arc -= 80
         obj = self.__class__(
-            value=tuple([first_arc, second_arc] + arcs[1:]),
+            value=array("L", (first_arc, second_arc)) + arcs[1:],
             impl=self.tag,
             expl=self._expl,
             default=self.default,
             impl=self.tag,
             expl=self._expl,
             default=self.default,
@@ -3726,7 +4359,7 @@ class ObjectIdentifier(Obj):
         )
         if ber_encoded:
             obj.ber_encoded = True
         )
         if ber_encoded:
             obj.ber_encoded = True
-        return obj, tail
+        yield decode_path, obj, tail
 
     def __repr__(self):
         return pp_console_row(next(self.pps()))
 
     def __repr__(self):
         return pp_console_row(next(self.pps()))
@@ -4210,6 +4843,7 @@ class UTCTime(VisibleString):
     tag_default = tag_encode(23)
     encoding = "ascii"
     asn1_type_name = "UTCTime"
     tag_default = tag_encode(23)
     encoding = "ascii"
     asn1_type_name = "UTCTime"
+    evgen_mode_skip_value = False
 
     def __init__(
             self,
 
     def __init__(
             self,
@@ -4380,6 +5014,9 @@ class UTCTime(VisibleString):
         value = self._encode_time()
         return b"".join((self.tag, len_encode(len(value)), value))
 
         value = self._encode_time()
         return b"".join((self.tag, len_encode(len(value)), value))
 
+    def _encode_cer(self, writer):
+        write_full(writer, self._encode())
+
     def todatetime(self):
         return self._value
 
     def todatetime(self):
         return self._value
 
@@ -4575,21 +5212,11 @@ class BMPString(CommonString):
     asn1_type_name = "BMPString"
 
 
     asn1_type_name = "BMPString"
 
 
-ChoiceState = namedtuple("ChoiceState", (
-    "version",
-    "specs",
-    "value",
-    "tag",
-    "expl",
-    "default",
-    "optional",
-    "offset",
-    "llen",
-    "vlen",
-    "expl_lenindef",
-    "lenindef",
-    "ber_encoded",
-), **NAMEDTUPLE_KWARGS)
+ChoiceState = namedtuple(
+    "ChoiceState",
+    BasicState._fields + ("specs", "value",),
+    **NAMEDTUPLE_KWARGS
+)
 
 
 class Choice(Obj):
 
 
 class Choice(Obj):
@@ -4666,6 +5293,9 @@ class Choice(Obj):
             self.default = default_obj
             if value is None:
                 self._value = copy(default_obj._value)
             self.default = default_obj
             if value is None:
                 self._value = copy(default_obj._value)
+        if self._expl is not None:
+            tag_class, _, tag_num = tag_decode(self._expl)
+            self._tag_order = (tag_class, tag_num)
 
     def _value_sanitize(self, value):
         if (value.__class__ == tuple) and len(value) == 2:
 
     def _value_sanitize(self, value):
         if (value.__class__ == tuple) and len(value) == 2:
@@ -4694,9 +5324,8 @@ class Choice(Obj):
     def __getstate__(self):
         return ChoiceState(
             __version__,
     def __getstate__(self):
         return ChoiceState(
             __version__,
-            self.specs,
-            copy(self._value),
             self.tag,
             self.tag,
+            self._tag_order,
             self._expl,
             self.default,
             self.optional,
             self._expl,
             self.default,
             self.optional,
@@ -4706,21 +5335,14 @@ class Choice(Obj):
             self.expl_lenindef,
             self.lenindef,
             self.ber_encoded,
             self.expl_lenindef,
             self.lenindef,
             self.ber_encoded,
+            self.specs,
+            copy(self._value),
         )
 
     def __setstate__(self, state):
         super(Choice, self).__setstate__(state)
         self.specs = state.specs
         self._value = state.value
         )
 
     def __setstate__(self, state):
         super(Choice, self).__setstate__(state)
         self.specs = state.specs
         self._value = state.value
-        self._expl = state.expl
-        self.default = state.default
-        self.optional = state.optional
-        self.offset = state.offset
-        self.llen = state.llen
-        self.vlen = state.vlen
-        self.expl_lenindef = state.expl_lenindef
-        self.lenindef = state.lenindef
-        self.ber_encoded = state.ber_encoded
 
     def __eq__(self, their):
         if (their.__class__ == tuple) and len(their) == 2:
 
     def __eq__(self, their):
         if (their.__class__ == tuple) and len(their) == 2:
@@ -4749,14 +5371,27 @@ class Choice(Obj):
 
     @property
     def choice(self):
 
     @property
     def choice(self):
+        """Name of the choice
+        """
         self._assert_ready()
         return self._value[0]
 
     @property
     def value(self):
         self._assert_ready()
         return self._value[0]
 
     @property
     def value(self):
+        """Value of underlying choice
+        """
         self._assert_ready()
         return self._value[1]
 
         self._assert_ready()
         return self._value[1]
 
+    @property
+    def tag_order(self):
+        self._assert_ready()
+        return self._value[1].tag_order if self._tag_order is None else self._tag_order
+
+    @property
+    def tag_order_cer(self):
+        return min(v.tag_order_cer for v in itervalues(self.specs))
+
     def __getitem__(self, key):
         if key not in self.specs:
             raise ObjUnknown(key)
     def __getitem__(self, key):
         if key not in self.specs:
             raise ObjUnknown(key)
@@ -4787,7 +5422,11 @@ class Choice(Obj):
         self._assert_ready()
         return self._value[1].encode()
 
         self._assert_ready()
         return self._value[1].encode()
 
-    def _decode(self, tlv, offset, decode_path, ctx, tag_only):
+    def _encode_cer(self, writer):
+        self._assert_ready()
+        self._value[1].encode_cer(writer)
+
+    def _decode(self, tlv, offset, decode_path, ctx, tag_only, evgen_mode):
         for choice, spec in iteritems(self.specs):
             sub_decode_path = decode_path + (choice,)
             try:
         for choice, spec in iteritems(self.specs):
             sub_decode_path = decode_path + (choice,)
             try:
@@ -4810,15 +5449,28 @@ class Choice(Obj):
                 offset=offset,
             )
         if tag_only:  # pragma: no cover
                 offset=offset,
             )
         if tag_only:  # pragma: no cover
-            return None
-        value, tail = spec.decode(
-            tlv,
-            offset=offset,
-            leavemm=True,
-            decode_path=sub_decode_path,
-            ctx=ctx,
-            _ctx_immutable=False,
-        )
+            yield None
+            return
+        if evgen_mode:
+            for _decode_path, value, tail in spec.decode_evgen(
+                    tlv,
+                    offset=offset,
+                    leavemm=True,
+                    decode_path=sub_decode_path,
+                    ctx=ctx,
+                    _ctx_immutable=False,
+            ):
+                yield _decode_path, value, tail
+        else:
+            _, value, tail = next(spec.decode_evgen(
+                tlv,
+                offset=offset,
+                leavemm=True,
+                decode_path=sub_decode_path,
+                ctx=ctx,
+                _ctx_immutable=False,
+                _evgen_mode=False,
+            ))
         obj = self.__class__(
             schema=self.specs,
             expl=self._expl,
         obj = self.__class__(
             schema=self.specs,
             expl=self._expl,
@@ -4827,7 +5479,7 @@ class Choice(Obj):
             _decoded=(offset, 0, value.fulllen),
         )
         obj._value = (choice, value)
             _decoded=(offset, 0, value.fulllen),
         )
         obj._value = (choice, value)
-        return obj, tail
+        yield decode_path, obj, tail
 
     def __repr__(self):
         value = pp_console_row(next(self.pps()))
 
     def __repr__(self):
         value = pp_console_row(next(self.pps()))
@@ -4894,27 +5546,18 @@ class PrimitiveTypes(Choice):
     ))
 
 
     ))
 
 
-AnyState = namedtuple("AnyState", (
-    "version",
-    "value",
-    "tag",
-    "expl",
-    "optional",
-    "offset",
-    "llen",
-    "vlen",
-    "expl_lenindef",
-    "lenindef",
-    "ber_encoded",
-    "defined",
-), **NAMEDTUPLE_KWARGS)
+AnyState = namedtuple(
+    "AnyState",
+    BasicState._fields + ("value", "defined"),
+    **NAMEDTUPLE_KWARGS
+)
 
 
 class Any(Obj):
     """``ANY`` special type
 
     >>> Any(Integer(-123))
 
 
 class Any(Obj):
     """``ANY`` special type
 
     >>> Any(Integer(-123))
-    ANY 020185
+    ANY INTEGER -123 (0X:7B)
     >>> a = Any(OctetString(b"hello world").encode())
     ANY 040b68656c6c6f20776f726c64
     >>> hexenc(bytes(a))
     >>> a = Any(OctetString(b"hello world").encode())
     ANY 040b68656c6c6f20776f726c64
     >>> hexenc(bytes(a))
@@ -4934,28 +5577,47 @@ class Any(Obj):
         """
         :param value: set the value. Either any kind of pyderasn's
                       **ready** object, or bytes. Pay attention that
         """
         :param value: set the value. Either any kind of pyderasn's
                       **ready** object, or bytes. Pay attention that
-                      **no** validation is performed is raw binary value
-                      is valid TLV
+                      **no** validation is performed if raw binary value
+                      is valid TLV, except just tag decoding
         :param bytes expl: override default tag with ``EXPLICIT`` one
         :param bool optional: is object ``OPTIONAL`` in sequence
         """
         super(Any, self).__init__(None, expl, None, optional, _decoded)
         :param bytes expl: override default tag with ``EXPLICIT`` one
         :param bool optional: is object ``OPTIONAL`` in sequence
         """
         super(Any, self).__init__(None, expl, None, optional, _decoded)
-        self._value = None if value is None else self._value_sanitize(value)
+        if value is None:
+            self._value = None
+        else:
+            value = self._value_sanitize(value)
+            self._value = value
+            if self._expl is None:
+                if value.__class__ == binary_type:
+                    tag_class, _, tag_num = tag_decode(tag_strip(value)[0])
+                else:
+                    tag_class, tag_num = value.tag_order
+            else:
+                tag_class, _, tag_num = tag_decode(self._expl)
+            self._tag_order = (tag_class, tag_num)
         self.defined = None
 
     def _value_sanitize(self, value):
         if value.__class__ == binary_type:
         self.defined = None
 
     def _value_sanitize(self, value):
         if value.__class__ == binary_type:
+            if len(value) == 0:
+                raise ValueError("Any value can not be empty")
             return value
         if isinstance(value, self.__class__):
             return value._value
             return value
         if isinstance(value, self.__class__):
             return value._value
-        if isinstance(value, Obj):
-            return value.encode()
-        raise InvalidValueType((self.__class__, Obj, binary_type))
+        if not isinstance(value, Obj):
+            raise InvalidValueType((self.__class__, Obj, binary_type))
+        return value
 
     @property
     def ready(self):
         return self._value is not None
 
 
     @property
     def ready(self):
         return self._value is not None
 
+    @property
+    def tag_order(self):
+        self._assert_ready()
+        return self._tag_order
+
     @property
     def bered(self):
         if self.expl_lenindef or self.lenindef:
     @property
     def bered(self):
         if self.expl_lenindef or self.lenindef:
@@ -4967,9 +5629,10 @@ class Any(Obj):
     def __getstate__(self):
         return AnyState(
             __version__,
     def __getstate__(self):
         return AnyState(
             __version__,
-            self._value,
             self.tag,
             self.tag,
+            self._tag_order,
             self._expl,
             self._expl,
+            None,
             self.optional,
             self.offset,
             self.llen,
             self.optional,
             self.offset,
             self.llen,
@@ -4977,28 +5640,24 @@ class Any(Obj):
             self.expl_lenindef,
             self.lenindef,
             self.ber_encoded,
             self.expl_lenindef,
             self.lenindef,
             self.ber_encoded,
+            self._value,
             self.defined,
         )
 
     def __setstate__(self, state):
         super(Any, self).__setstate__(state)
         self._value = state.value
             self.defined,
         )
 
     def __setstate__(self, state):
         super(Any, self).__setstate__(state)
         self._value = state.value
-        self.tag = state.tag
-        self._expl = state.expl
-        self.optional = state.optional
-        self.offset = state.offset
-        self.llen = state.llen
-        self.vlen = state.vlen
-        self.expl_lenindef = state.expl_lenindef
-        self.lenindef = state.lenindef
-        self.ber_encoded = state.ber_encoded
         self.defined = state.defined
 
     def __eq__(self, their):
         if their.__class__ == binary_type:
         self.defined = state.defined
 
     def __eq__(self, their):
         if their.__class__ == binary_type:
-            return self._value == their
+            if self._value.__class__ == binary_type:
+                return self._value == their
+            return self._value.encode() == their
         if issubclass(their.__class__, Any):
         if issubclass(their.__class__, Any):
-            return self._value == their._value
+            if self.ready and their.ready:
+                return bytes(self) == bytes(their)
+            return self.ready == their.ready
         return False
 
     def __call__(
         return False
 
     def __call__(
@@ -5015,7 +5674,10 @@ class Any(Obj):
 
     def __bytes__(self):
         self._assert_ready()
 
     def __bytes__(self):
         self._assert_ready()
-        return self._value
+        value = self._value
+        if value.__class__ == binary_type:
+            return value
+        return self._value.encode()
 
     @property
     def tlen(self):
 
     @property
     def tlen(self):
@@ -5023,9 +5685,20 @@ class Any(Obj):
 
     def _encode(self):
         self._assert_ready()
 
     def _encode(self):
         self._assert_ready()
-        return self._value
+        value = self._value
+        if value.__class__ == binary_type:
+            return value
+        return value.encode()
+
+    def _encode_cer(self, writer):
+        self._assert_ready()
+        value = self._value
+        if value.__class__ == binary_type:
+            write_full(writer, value)
+        else:
+            value.encode_cer(writer)
 
 
-    def _decode(self, tlv, offset, decode_path, ctx, tag_only):
+    def _decode(self, tlv, offset, decode_path, ctx, tag_only, evgen_mode):
         try:
             t, tlen, lv = tag_strip(tlv)
         except DecodeError as err:
         try:
             t, tlen, lv = tag_strip(tlv)
         except DecodeError as err:
@@ -5062,14 +5735,15 @@ class Any(Obj):
                 chunk_i += 1
             tlvlen = tlen + llen + vlen + EOC_LEN
             obj = self.__class__(
                 chunk_i += 1
             tlvlen = tlen + llen + vlen + EOC_LEN
             obj = self.__class__(
-                value=tlv[:tlvlen].tobytes(),
+                value=None if evgen_mode else tlv[:tlvlen].tobytes(),
                 expl=self._expl,
                 optional=self.optional,
                 _decoded=(offset, 0, tlvlen),
             )
             obj.lenindef = True
             obj.tag = t.tobytes()
                 expl=self._expl,
                 optional=self.optional,
                 _decoded=(offset, 0, tlvlen),
             )
             obj.lenindef = True
             obj.tag = t.tobytes()
-            return obj, v[EOC_LEN:]
+            yield decode_path, obj, v[EOC_LEN:]
+            return
         except DecodeError as err:
             raise err.__class__(
                 msg=err.msg,
         except DecodeError as err:
             raise err.__class__(
                 msg=err.msg,
@@ -5087,24 +5761,32 @@ class Any(Obj):
         tlvlen = tlen + llen + l
         v, tail = tlv[:tlvlen], v[l:]
         obj = self.__class__(
         tlvlen = tlen + llen + l
         v, tail = tlv[:tlvlen], v[l:]
         obj = self.__class__(
-            value=v.tobytes(),
+            value=None if evgen_mode else v.tobytes(),
             expl=self._expl,
             optional=self.optional,
             _decoded=(offset, 0, tlvlen),
         )
         obj.tag = t.tobytes()
             expl=self._expl,
             optional=self.optional,
             _decoded=(offset, 0, tlvlen),
         )
         obj.tag = t.tobytes()
-        return obj, tail
+        yield decode_path, obj, tail
 
     def __repr__(self):
         return pp_console_row(next(self.pps()))
 
     def pps(self, decode_path=()):
 
     def __repr__(self):
         return pp_console_row(next(self.pps()))
 
     def pps(self, decode_path=()):
+        value = self._value
+        if value is None:
+            pass
+        elif value.__class__ == binary_type:
+            value = None
+        else:
+            value = repr(value)
         yield _pp(
             obj=self,
             asn1_type_name=self.asn1_type_name,
             obj_name=self.__class__.__name__,
             decode_path=decode_path,
         yield _pp(
             obj=self,
             asn1_type_name=self.asn1_type_name,
             obj_name=self.__class__.__name__,
             decode_path=decode_path,
-            blob=self._value if self.ready else None,
+            value=value,
+            blob=self._value if self._value.__class__ == binary_type else None,
             optional=self.optional,
             default=self == self.default,
             impl=None if self.tag == self.tag_default else tag_decode(self.tag),
             optional=self.optional,
             default=self == self.default,
             impl=None if self.tag == self.tag_default else tag_decode(self.tag),
@@ -5172,21 +5854,11 @@ def abs_decode_path(decode_path, rel_path):
     return decode_path + rel_path
 
 
     return decode_path + rel_path
 
 
-SequenceState = namedtuple("SequenceState", (
-    "version",
-    "specs",
-    "value",
-    "tag",
-    "expl",
-    "default",
-    "optional",
-    "offset",
-    "llen",
-    "vlen",
-    "expl_lenindef",
-    "lenindef",
-    "ber_encoded",
-), **NAMEDTUPLE_KWARGS)
+SequenceState = namedtuple(
+    "SequenceState",
+    BasicState._fields + ("specs", "value",),
+    **NAMEDTUPLE_KWARGS
+)
 
 
 class Sequence(Obj):
 
 
 class Sequence(Obj):
@@ -5277,6 +5949,12 @@ class Sequence(Obj):
     defaulted values existence validation by setting
     ``"allow_default_values": True`` :ref:`context <ctx>` option.
 
     defaulted values existence validation by setting
     ``"allow_default_values": True`` :ref:`context <ctx>` option.
 
+    .. warning::
+
+       Check for default value existence is not performed in
+       ``evgen_mode``, because previously decoded values are not stored
+       in memory, to be able to compare them.
+
     Two sequences are equal if they have equal specification (schema),
     implicit/explicit tagging and the same values.
     """
     Two sequences are equal if they have equal specification (schema),
     implicit/explicit tagging and the same values.
     """
@@ -5341,9 +6019,8 @@ class Sequence(Obj):
     def __getstate__(self):
         return SequenceState(
             __version__,
     def __getstate__(self):
         return SequenceState(
             __version__,
-            self.specs,
-            {k: copy(v) for k, v in iteritems(self._value)},
             self.tag,
             self.tag,
+            self._tag_order,
             self._expl,
             self.default,
             self.optional,
             self._expl,
             self.default,
             self.optional,
@@ -5353,22 +6030,14 @@ class Sequence(Obj):
             self.expl_lenindef,
             self.lenindef,
             self.ber_encoded,
             self.expl_lenindef,
             self.lenindef,
             self.ber_encoded,
+            self.specs,
+            {k: copy(v) for k, v in iteritems(self._value)},
         )
 
     def __setstate__(self, state):
         super(Sequence, self).__setstate__(state)
         self.specs = state.specs
         self._value = state.value
         )
 
     def __setstate__(self, state):
         super(Sequence, self).__setstate__(state)
         self.specs = state.specs
         self._value = state.value
-        self.tag = state.tag
-        self._expl = state.expl
-        self.default = state.default
-        self.optional = state.optional
-        self.offset = state.offset
-        self.llen = state.llen
-        self.vlen = state.vlen
-        self.expl_lenindef = state.expl_lenindef
-        self.lenindef = state.lenindef
-        self.ber_encoded = state.ber_encoded
 
     def __eq__(self, their):
         if not isinstance(their, self.__class__):
 
     def __eq__(self, their):
         if not isinstance(their, self.__class__):
@@ -5439,7 +6108,13 @@ class Sequence(Obj):
         v = b"".join(v.encode() for v in self._values_for_encoding())
         return b"".join((self.tag, len_encode(len(v)), v))
 
         v = b"".join(v.encode() for v in self._values_for_encoding())
         return b"".join((self.tag, len_encode(len(v)), v))
 
-    def _decode(self, tlv, offset, decode_path, ctx, tag_only):
+    def _encode_cer(self, writer):
+        write_full(writer, self.tag + LENINDEF)
+        for v in self._values_for_encoding():
+            v.encode_cer(writer)
+        write_full(writer, EOC)
+
+    def _decode(self, tlv, offset, decode_path, ctx, tag_only, evgen_mode):
         try:
             t, tlen, lv = tag_strip(tlv)
         except DecodeError as err:
         try:
             t, tlen, lv = tag_strip(tlv)
         except DecodeError as err:
@@ -5456,7 +6131,8 @@ class Sequence(Obj):
                 offset=offset,
             )
         if tag_only:  # pragma: no cover
                 offset=offset,
             )
         if tag_only:  # pragma: no cover
-            return None
+            yield None
+            return
         lenindef = False
         ctx_bered = ctx.get("bered", False)
         try:
         lenindef = False
         ctx_bered = ctx.get("bered", False)
         try:
@@ -5500,21 +6176,33 @@ class Sequence(Obj):
                 continue
             sub_decode_path = decode_path + (name,)
             try:
                 continue
             sub_decode_path = decode_path + (name,)
             try:
-                value, v_tail = spec.decode(
-                    v,
-                    sub_offset,
-                    leavemm=True,
-                    decode_path=sub_decode_path,
-                    ctx=ctx,
-                    _ctx_immutable=False,
-                )
+                if evgen_mode:
+                    for _decode_path, value, v_tail in spec.decode_evgen(
+                            v,
+                            sub_offset,
+                            leavemm=True,
+                            decode_path=sub_decode_path,
+                            ctx=ctx,
+                            _ctx_immutable=False,
+                    ):
+                        yield _decode_path, value, v_tail
+                else:
+                    _, value, v_tail = next(spec.decode_evgen(
+                        v,
+                        sub_offset,
+                        leavemm=True,
+                        decode_path=sub_decode_path,
+                        ctx=ctx,
+                        _ctx_immutable=False,
+                        _evgen_mode=False,
+                    ))
             except TagMismatch as err:
                 if (len(err.decode_path) == len(decode_path) + 1) and spec.optional:
                     continue
                 raise
 
             defined = get_def_by_path(ctx.get("_defines", ()), sub_decode_path)
             except TagMismatch as err:
                 if (len(err.decode_path) == len(decode_path) + 1) and spec.optional:
                     continue
                 raise
 
             defined = get_def_by_path(ctx.get("_defines", ()), sub_decode_path)
-            if defined is not None:
+            if not evgen_mode and defined is not None:
                 defined_by, defined_spec = defined
                 if issubclass(value.__class__, SequenceOf):
                     for i, _value in enumerate(value):
                 defined_by, defined_spec = defined
                 if issubclass(value.__class__, SequenceOf):
                     for i, _value in enumerate(value):
@@ -5566,31 +6254,32 @@ class Sequence(Obj):
             vlen += value_len
             sub_offset += value_len
             v = v_tail
             vlen += value_len
             sub_offset += value_len
             v = v_tail
-            if spec.default is not None and value == spec.default:
-                if ctx_bered or ctx_allow_default_values:
-                    ber_encoded = True
-                else:
-                    raise DecodeError(
-                        "DEFAULT value met",
-                        klass=self.__class__,
-                        decode_path=sub_decode_path,
-                        offset=sub_offset,
-                    )
-            values[name] = value
-
-            spec_defines = getattr(spec, "defines", ())
-            if len(spec_defines) == 0:
-                defines_by_path = ctx.get("defines_by_path", ())
-                if len(defines_by_path) > 0:
-                    spec_defines = get_def_by_path(defines_by_path, sub_decode_path)
-            if spec_defines is not None and len(spec_defines) > 0:
-                for rel_path, schema in spec_defines:
-                    defined = schema.get(value, None)
-                    if defined is not None:
-                        ctx.setdefault("_defines", []).append((
-                            abs_decode_path(sub_decode_path[:-1], rel_path),
-                            (value, defined),
-                        ))
+            if not evgen_mode:
+                if spec.default is not None and value == spec.default:
+                    # This will not work in evgen_mode
+                    if ctx_bered or ctx_allow_default_values:
+                        ber_encoded = True
+                    else:
+                        raise DecodeError(
+                            "DEFAULT value met",
+                            klass=self.__class__,
+                            decode_path=sub_decode_path,
+                            offset=sub_offset,
+                        )
+                values[name] = value
+                spec_defines = getattr(spec, "defines", ())
+                if len(spec_defines) == 0:
+                    defines_by_path = ctx.get("defines_by_path", ())
+                    if len(defines_by_path) > 0:
+                        spec_defines = get_def_by_path(defines_by_path, sub_decode_path)
+                if spec_defines is not None and len(spec_defines) > 0:
+                    for rel_path, schema in spec_defines:
+                        defined = schema.get(value, None)
+                        if defined is not None:
+                            ctx.setdefault("_defines", []).append((
+                                abs_decode_path(sub_decode_path[:-1], rel_path),
+                                (value, defined),
+                            ))
         if lenindef:
             if v[:EOC_LEN].tobytes() != EOC:
                 raise DecodeError(
         if lenindef:
             if v[:EOC_LEN].tobytes() != EOC:
                 raise DecodeError(
@@ -5619,7 +6308,7 @@ class Sequence(Obj):
         obj._value = values
         obj.lenindef = lenindef
         obj.ber_encoded = ber_encoded
         obj._value = values
         obj.lenindef = lenindef
         obj.ber_encoded = ber_encoded
-        return obj, tail
+        yield decode_path, obj, tail
 
     def __repr__(self):
         value = pp_console_row(next(self.pps()))
 
     def __repr__(self):
         value = pp_console_row(next(self.pps()))
@@ -5681,15 +6370,22 @@ class Set(Sequence):
     asn1_type_name = "SET"
 
     def _encode(self):
     asn1_type_name = "SET"
 
     def _encode(self):
-        raws = [v.encode() for v in self._values_for_encoding()]
-        raws.sort()
-        v = b"".join(raws)
+        v = b"".join(value.encode() for value in sorted(
+            self._values_for_encoding(),
+            key=attrgetter("tag_order"),
+        ))
         return b"".join((self.tag, len_encode(len(v)), v))
 
         return b"".join((self.tag, len_encode(len(v)), v))
 
-    def _specs_items(self):
-        return iteritems(self.specs)
+    def _encode_cer(self, writer):
+        write_full(writer, self.tag + LENINDEF)
+        for v in sorted(
+                self._values_for_encoding(),
+                key=attrgetter("tag_order_cer"),
+        ):
+            v.encode_cer(writer)
+        write_full(writer, EOC)
 
 
-    def _decode(self, tlv, offset, decode_path, ctx, tag_only):
+    def _decode(self, tlv, offset, decode_path, ctx, tag_only, evgen_mode):
         try:
             t, tlen, lv = tag_strip(tlv)
         except DecodeError as err:
         try:
             t, tlen, lv = tag_strip(tlv)
         except DecodeError as err:
@@ -5706,7 +6402,8 @@ class Set(Sequence):
                 offset=offset,
             )
         if tag_only:
                 offset=offset,
             )
         if tag_only:
-            return None
+            yield None
+            return
         lenindef = False
         ctx_bered = ctx.get("bered", False)
         try:
         lenindef = False
         ctx_bered = ctx.get("bered", False)
         try:
@@ -5742,12 +6439,13 @@ class Set(Sequence):
         ber_encoded = False
         ctx_allow_default_values = ctx.get("allow_default_values", False)
         ctx_allow_unordered_set = ctx.get("allow_unordered_set", False)
         ber_encoded = False
         ctx_allow_default_values = ctx.get("allow_default_values", False)
         ctx_allow_unordered_set = ctx.get("allow_unordered_set", False)
-        value_prev = memoryview(v[:0])
+        tag_order_prev = (0, 0)
+        _specs_items = copy(self.specs)
 
         while len(v) > 0:
             if lenindef and v[:EOC_LEN].tobytes() == EOC:
                 break
 
         while len(v) > 0:
             if lenindef and v[:EOC_LEN].tobytes() == EOC:
                 break
-            for name, spec in self._specs_items():
+            for name, spec in iteritems(_specs_items):
                 sub_decode_path = decode_path + (name,)
                 try:
                     spec.decode(
                 sub_decode_path = decode_path + (name,)
                 try:
                     spec.decode(
@@ -5768,16 +6466,29 @@ class Set(Sequence):
                     decode_path=decode_path,
                     offset=offset,
                 )
                     decode_path=decode_path,
                     offset=offset,
                 )
-            value, v_tail = spec.decode(
-                v,
-                sub_offset,
-                leavemm=True,
-                decode_path=sub_decode_path,
-                ctx=ctx,
-                _ctx_immutable=False,
-            )
+            if evgen_mode:
+                for _decode_path, value, v_tail in spec.decode_evgen(
+                        v,
+                        sub_offset,
+                        leavemm=True,
+                        decode_path=sub_decode_path,
+                        ctx=ctx,
+                        _ctx_immutable=False,
+                ):
+                    yield _decode_path, value, v_tail
+            else:
+                _, value, v_tail = next(spec.decode_evgen(
+                    v,
+                    sub_offset,
+                    leavemm=True,
+                    decode_path=sub_decode_path,
+                    ctx=ctx,
+                    _ctx_immutable=False,
+                    _evgen_mode=False,
+                ))
+            value_tag_order = value.tag_order
             value_len = value.fulllen
             value_len = value.fulllen
-            if value_prev.tobytes() > v[:value_len].tobytes():
+            if tag_order_prev >= value_tag_order:
                 if ctx_bered or ctx_allow_unordered_set:
                     ber_encoded = True
                 else:
                 if ctx_bered or ctx_allow_unordered_set:
                     ber_encoded = True
                 else:
@@ -5799,10 +6510,12 @@ class Set(Sequence):
                     offset=sub_offset,
                 )
             values[name] = value
                     offset=sub_offset,
                 )
             values[name] = value
-            value_prev = v[:value_len]
+            del _specs_items[name]
+            tag_order_prev = value_tag_order
             sub_offset += value_len
             vlen += value_len
             v = v_tail
             sub_offset += value_len
             vlen += value_len
             v = v_tail
+
         obj = self.__class__(
             schema=self.specs,
             impl=self.tag,
         obj = self.__class__(
             schema=self.specs,
             impl=self.tag,
@@ -5821,7 +6534,6 @@ class Set(Sequence):
                 )
             tail = v[EOC_LEN:]
             obj.lenindef = True
                 )
             tail = v[EOC_LEN:]
             obj.lenindef = True
-        obj._value = values
         for name, spec in iteritems(self.specs):
             if name not in values and not spec.optional:
                 raise DecodeError(
         for name, spec in iteritems(self.specs):
             if name not in values and not spec.optional:
                 raise DecodeError(
@@ -5830,27 +6542,17 @@ class Set(Sequence):
                     decode_path=decode_path,
                     offset=offset,
                 )
                     decode_path=decode_path,
                     offset=offset,
                 )
+        if not evgen_mode:
+            obj._value = values
         obj.ber_encoded = ber_encoded
         obj.ber_encoded = ber_encoded
-        return obj, tail
+        yield decode_path, obj, tail
 
 
 
 
-SequenceOfState = namedtuple("SequenceOfState", (
-    "version",
-    "spec",
-    "value",
-    "bound_min",
-    "bound_max",
-    "tag",
-    "expl",
-    "default",
-    "optional",
-    "offset",
-    "llen",
-    "vlen",
-    "expl_lenindef",
-    "lenindef",
-    "ber_encoded",
-), **NAMEDTUPLE_KWARGS)
+SequenceOfState = namedtuple(
+    "SequenceOfState",
+    BasicState._fields + ("spec", "value", "bound_min", "bound_max"),
+    **NAMEDTUPLE_KWARGS
+)
 
 
 class SequenceOf(Obj):
 
 
 class SequenceOf(Obj):
@@ -5879,9 +6581,21 @@ class SequenceOf(Obj):
     >>> ints
     Ints SEQUENCE OF[INTEGER 123, INTEGER 345]
 
     >>> ints
     Ints SEQUENCE OF[INTEGER 123, INTEGER 345]
 
-    Also you can initialize sequence with preinitialized values:
+    You can initialize sequence with preinitialized values:
 
     >>> ints = Ints([Integer(123), Integer(234)])
 
     >>> ints = Ints([Integer(123), Integer(234)])
+
+    Also you can use iterator as a value:
+
+    >>> ints = Ints(iter(Integer(i) for i in range(1000000)))
+
+    And it won't be iterated until encoding process. Pay attention that
+    bounds and required schema checks are done only during the encoding
+    process in that case! After encode was called, then value is zeroed
+    back to empty list and you have to set it again. That mode is useful
+    mainly with CER encoding mode, where all objects from the iterable
+    will be streamed to the buffer, without copying all of them to
+    memory first.
     """
     __slots__ = ("spec", "_bound_min", "_bound_max")
     tag_default = tag_encode(form=TagFormConstructed, num=16)
     """
     __slots__ = ("spec", "_bound_min", "_bound_max")
     tag_default = tag_encode(form=TagFormConstructed, num=16)
@@ -5925,21 +6639,31 @@ class SequenceOf(Obj):
                 self._value = copy(default_obj._value)
 
     def _value_sanitize(self, value):
                 self._value = copy(default_obj._value)
 
     def _value_sanitize(self, value):
+        iterator = False
         if issubclass(value.__class__, SequenceOf):
             value = value._value
         if issubclass(value.__class__, SequenceOf):
             value = value._value
+        elif hasattr(value, NEXT_ATTR_NAME):
+            iterator = True
+            value = value
         elif hasattr(value, "__iter__"):
             value = list(value)
         else:
         elif hasattr(value, "__iter__"):
             value = list(value)
         else:
-            raise InvalidValueType((self.__class__, iter))
-        if not self._bound_min <= len(value) <= self._bound_max:
-            raise BoundsError(self._bound_min, len(value), self._bound_max)
-        for v in value:
-            if not isinstance(v, self.spec.__class__):
-                raise InvalidValueType((self.spec.__class__,))
+            raise InvalidValueType((self.__class__, iter, "iterator"))
+        if not iterator:
+            if not self._bound_min <= len(value) <= self._bound_max:
+                raise BoundsError(self._bound_min, len(value), self._bound_max)
+            class_expected = self.spec.__class__
+            for v in value:
+                if not isinstance(v, class_expected):
+                    raise InvalidValueType((class_expected,))
         return value
 
     @property
     def ready(self):
         return value
 
     @property
     def ready(self):
+        if hasattr(self._value, NEXT_ATTR_NAME):
+            return True
+        if self._bound_min > 0 and len(self._value) == 0:
+            return False
         return all(v.ready for v in self._value)
 
     @property
         return all(v.ready for v in self._value)
 
     @property
@@ -5949,13 +6673,12 @@ class SequenceOf(Obj):
         return any(v.bered for v in self._value)
 
     def __getstate__(self):
         return any(v.bered for v in self._value)
 
     def __getstate__(self):
+        if hasattr(self._value, NEXT_ATTR_NAME):
+            raise ValueError("can not pickle SequenceOf with iterator")
         return SequenceOfState(
             __version__,
         return SequenceOfState(
             __version__,
-            self.spec,
-            [copy(v) for v in self._value],
-            self._bound_min,
-            self._bound_max,
             self.tag,
             self.tag,
+            self._tag_order,
             self._expl,
             self.default,
             self.optional,
             self._expl,
             self.default,
             self.optional,
@@ -5965,6 +6688,10 @@ class SequenceOf(Obj):
             self.expl_lenindef,
             self.lenindef,
             self.ber_encoded,
             self.expl_lenindef,
             self.lenindef,
             self.ber_encoded,
+            self.spec,
+            [copy(v) for v in self._value],
+            self._bound_min,
+            self._bound_max,
         )
 
     def __setstate__(self, state):
         )
 
     def __setstate__(self, state):
@@ -5973,16 +6700,6 @@ class SequenceOf(Obj):
         self._value = state.value
         self._bound_min = state.bound_min
         self._bound_max = state.bound_max
         self._value = state.value
         self._bound_min = state.bound_min
         self._bound_max = state.bound_max
-        self.tag = state.tag
-        self._expl = state.expl
-        self.default = state.default
-        self.optional = state.optional
-        self.offset = state.offset
-        self.llen = state.llen
-        self.vlen = state.vlen
-        self.expl_lenindef = state.expl_lenindef
-        self.lenindef = state.lenindef
-        self.ber_encoded = state.ber_encoded
 
     def __eq__(self, their):
         if isinstance(their, self.__class__):
 
     def __eq__(self, their):
         if isinstance(their, self.__class__):
@@ -6033,11 +6750,9 @@ class SequenceOf(Obj):
         self._value.append(value)
 
     def __iter__(self):
         self._value.append(value)
 
     def __iter__(self):
-        self._assert_ready()
         return iter(self._value)
 
     def __len__(self):
         return iter(self._value)
 
     def __len__(self):
-        self._assert_ready()
         return len(self._value)
 
     def __setitem__(self, key, value):
         return len(self._value)
 
     def __setitem__(self, key, value):
@@ -6052,10 +6767,54 @@ class SequenceOf(Obj):
         return iter(self._value)
 
     def _encode(self):
         return iter(self._value)
 
     def _encode(self):
-        v = b"".join(v.encode() for v in self._values_for_encoding())
-        return b"".join((self.tag, len_encode(len(v)), v))
+        iterator = hasattr(self._value, NEXT_ATTR_NAME)
+        if iterator:
+            values = []
+            values_append = values.append
+            class_expected = self.spec.__class__
+            values_for_encoding = self._values_for_encoding()
+            self._value = []
+            for v in values_for_encoding:
+                if not isinstance(v, class_expected):
+                    raise InvalidValueType((class_expected,))
+                values_append(v.encode())
+            if not self._bound_min <= len(values) <= self._bound_max:
+                raise BoundsError(self._bound_min, len(values), self._bound_max)
+            value = b"".join(values)
+        else:
+            value = b"".join(v.encode() for v in self._values_for_encoding())
+        return b"".join((self.tag, len_encode(len(value)), value))
 
 
-    def _decode(self, tlv, offset, decode_path, ctx, tag_only, ordering_check=False):
+    def _encode_cer(self, writer):
+        write_full(writer, self.tag + LENINDEF)
+        iterator = hasattr(self._value, NEXT_ATTR_NAME)
+        if iterator:
+            class_expected = self.spec.__class__
+            values_count = 0
+            values_for_encoding = self._values_for_encoding()
+            self._value = []
+            for v in values_for_encoding:
+                if not isinstance(v, class_expected):
+                    raise InvalidValueType((class_expected,))
+                v.encode_cer(writer)
+                values_count += 1
+            if not self._bound_min <= values_count <= self._bound_max:
+                raise BoundsError(self._bound_min, values_count, self._bound_max)
+        else:
+            for v in self._values_for_encoding():
+                v.encode_cer(writer)
+        write_full(writer, EOC)
+
+    def _decode(
+            self,
+            tlv,
+            offset,
+            decode_path,
+            ctx,
+            tag_only,
+            evgen_mode,
+            ordering_check=False,
+    ):
         try:
             t, tlen, lv = tag_strip(tlv)
         except DecodeError as err:
         try:
             t, tlen, lv = tag_strip(tlv)
         except DecodeError as err:
@@ -6072,7 +6831,8 @@ class SequenceOf(Obj):
                 offset=offset,
             )
         if tag_only:
                 offset=offset,
             )
         if tag_only:
-            return None
+            yield None
+            return
         lenindef = False
         ctx_bered = ctx.get("bered", False)
         try:
         lenindef = False
         ctx_bered = ctx.get("bered", False)
         try:
@@ -6106,6 +6866,7 @@ class SequenceOf(Obj):
         vlen = 0
         sub_offset = offset + tlen + llen
         _value = []
         vlen = 0
         sub_offset = offset + tlen + llen
         _value = []
+        _value_count = 0
         ctx_allow_unordered_set = ctx.get("allow_unordered_set", False)
         value_prev = memoryview(v[:0])
         ber_encoded = False
         ctx_allow_unordered_set = ctx.get("allow_unordered_set", False)
         value_prev = memoryview(v[:0])
         ber_encoded = False
@@ -6113,15 +6874,27 @@ class SequenceOf(Obj):
         while len(v) > 0:
             if lenindef and v[:EOC_LEN].tobytes() == EOC:
                 break
         while len(v) > 0:
             if lenindef and v[:EOC_LEN].tobytes() == EOC:
                 break
-            sub_decode_path = decode_path + (str(len(_value)),)
-            value, v_tail = spec.decode(
-                v,
-                sub_offset,
-                leavemm=True,
-                decode_path=sub_decode_path,
-                ctx=ctx,
-                _ctx_immutable=False,
-            )
+            sub_decode_path = decode_path + (str(_value_count),)
+            if evgen_mode:
+                for _decode_path, value, v_tail in spec.decode_evgen(
+                        v,
+                        sub_offset,
+                        leavemm=True,
+                        decode_path=sub_decode_path,
+                        ctx=ctx,
+                        _ctx_immutable=False,
+                ):
+                    yield _decode_path, value, v_tail
+            else:
+                _, value, v_tail = next(spec.decode_evgen(
+                    v,
+                    sub_offset,
+                    leavemm=True,
+                    decode_path=sub_decode_path,
+                    ctx=ctx,
+                    _ctx_immutable=False,
+                    _evgen_mode=False,
+                ))
             value_len = value.fulllen
             if ordering_check:
                 if value_prev.tobytes() > v[:value_len].tobytes():
             value_len = value.fulllen
             if ordering_check:
                 if value_prev.tobytes() > v[:value_len].tobytes():
@@ -6135,13 +6908,22 @@ class SequenceOf(Obj):
                             offset=sub_offset,
                         )
                 value_prev = v[:value_len]
                             offset=sub_offset,
                         )
                 value_prev = v[:value_len]
-            _value.append(value)
+            _value_count += 1
+            if not evgen_mode:
+                _value.append(value)
             sub_offset += value_len
             vlen += value_len
             v = v_tail
             sub_offset += value_len
             vlen += value_len
             v = v_tail
+        if evgen_mode and not self._bound_min <= _value_count <= self._bound_max:
+            raise DecodeError(
+                msg=str(BoundsError(self._bound_min, _value_count, self._bound_max)),
+                klass=self.__class__,
+                decode_path=decode_path,
+                offset=offset,
+            )
         try:
             obj = self.__class__(
         try:
             obj = self.__class__(
-                value=_value,
+                value=None if evgen_mode else _value,
                 schema=spec,
                 bounds=(self._bound_min, self._bound_max),
                 impl=self.tag,
                 schema=spec,
                 bounds=(self._bound_min, self._bound_max),
                 impl=self.tag,
@@ -6168,7 +6950,7 @@ class SequenceOf(Obj):
             obj.lenindef = True
             tail = v[EOC_LEN:]
         obj.ber_encoded = ber_encoded
             obj.lenindef = True
             tail = v[EOC_LEN:]
         obj.ber_encoded = ber_encoded
-        return obj, tail
+        yield decode_path, obj, tail
 
     def __repr__(self):
         return "%s[%s]" % (
 
     def __repr__(self):
         return "%s[%s]" % (
@@ -6214,19 +6996,32 @@ class SetOf(SequenceOf):
     tag_default = tag_encode(form=TagFormConstructed, num=17)
     asn1_type_name = "SET OF"
 
     tag_default = tag_encode(form=TagFormConstructed, num=17)
     asn1_type_name = "SET OF"
 
+    def _value_sanitize(self, value):
+        value = super(SetOf, self)._value_sanitize(value)
+        if hasattr(value, NEXT_ATTR_NAME):
+            raise ValueError(
+                "SetOf does not support iterator values, as no sense in them"
+            )
+        return value
+
     def _encode(self):
     def _encode(self):
-        raws = [v.encode() for v in self._values_for_encoding()]
-        raws.sort()
-        v = b"".join(raws)
+        v = b"".join(sorted(v.encode() for v in self._values_for_encoding()))
         return b"".join((self.tag, len_encode(len(v)), v))
 
         return b"".join((self.tag, len_encode(len(v)), v))
 
-    def _decode(self, tlv, offset, decode_path, ctx, tag_only):
+    def _encode_cer(self, writer):
+        write_full(writer, self.tag + LENINDEF)
+        for v in sorted(encode_cer(v) for v in self._values_for_encoding()):
+            write_full(writer, v)
+        write_full(writer, EOC)
+
+    def _decode(self, tlv, offset, decode_path, ctx, tag_only, evgen_mode):
         return super(SetOf, self)._decode(
             tlv,
             offset,
             decode_path,
             ctx,
             tag_only,
         return super(SetOf, self)._decode(
             tlv,
             offset,
             decode_path,
             ctx,
             tag_only,
+            evgen_mode,
             ordering_check=True,
         )
 
             ordering_check=True,
         )
 
@@ -6348,14 +7143,22 @@ def main():  # pragma: no cover
         help="Allow explicit tag out-of-bound",
     )
     parser.add_argument(
         help="Allow explicit tag out-of-bound",
     )
     parser.add_argument(
-        "DERFile",
+        "--evgen",
+        action="store_true",
+        help="Turn on event generation mode",
+    )
+    parser.add_argument(
+        "RAWFile",
         type=argparse.FileType("rb"),
         type=argparse.FileType("rb"),
-        help="Path to DER file you want to decode",
+        help="Path to BER/CER/DER file you want to decode",
     )
     args = parser.parse_args()
     )
     args = parser.parse_args()
-    args.DERFile.seek(args.skip)
-    der = memoryview(args.DERFile.read())
-    args.DERFile.close()
+    if PY2:
+        args.RAWFile.seek(args.skip)
+        raw = memoryview(args.RAWFile.read())
+        args.RAWFile.close()
+    else:
+        raw = file_mmaped(args.RAWFile)[args.skip:]
     oid_maps = (
         [obj_by_path(_path) for _path in (args.oids or "").split(",")]
         if args.oids else ()
     oid_maps = (
         [obj_by_path(_path) for _path in (args.oids or "").split(",")]
         if args.oids else ()
@@ -6372,10 +7175,9 @@ def main():  # pragma: no cover
     }
     if args.defines_by_path is not None:
         ctx["defines_by_path"] = obj_by_path(args.defines_by_path)
     }
     if args.defines_by_path is not None:
         ctx["defines_by_path"] = obj_by_path(args.defines_by_path)
-    obj, tail = schema().decode(der, ctx=ctx)
     from os import environ
     from os import environ
-    print(pprinter(
-        obj,
+    pprinter = partial(
+        pprinter,
         oid_maps=oid_maps,
         with_colours=environ.get("NO_COLOR") is None,
         with_decode_path=args.print_decode_path,
         oid_maps=oid_maps,
         with_colours=environ.get("NO_COLOR") is None,
         with_decode_path=args.print_decode_path,
@@ -6383,7 +7185,13 @@ def main():  # pragma: no cover
             () if args.decode_path_only is None else
             tuple(args.decode_path_only.split(":"))
         ),
             () if args.decode_path_only is None else
             tuple(args.decode_path_only.split(":"))
         ),
-    ))
+    )
+    if args.evgen:
+        for decode_path, obj, tail in schema().decode_evgen(raw, ctx=ctx):
+            print(pprinter(obj, decode_path=decode_path))
+    else:
+        obj, tail = schema().decode(raw, ctx=ctx)
+        print(pprinter(obj))
     if tail != b"":
         print("\nTrailing data: %s" % hexenc(tail))
 
     if tail != b"":
         print("\nTrailing data: %s" % hexenc(tail))