nrbf

Python script to parse Microsoft NRBF serialized streams into records or stdout.

Background

The .NET Remoting: Binary Format Data Structure defines a set of structures that represent object graph or method invocation information as an octet stream.

While working on a malware implant the NRBF format was encountered which was a little bit annoying to parse using some poor Python file parsing, and proved not giving back the data in the format that was easier to parse using existing Python implementations. This tool was thus created to have some reference code to parse the Microsoft NRBF format, while also allowing extendability and usability with the support for flow.record.

Installation

Installing can be done using pip:

python -m pip install .

CLI Usage

The script can be used straight from the command line, or by using the API. Since the script supports the output in the form of flow.record, rdump can be used to format the output if necessary. For rdump usage examples you can check out the rdump documentation.

Simply print the records in the given hex stream to stdout:

$ nrbf-extract --hex 000000000000000000010000000000000016110800001210416464726573732072656365697665640B stdout
<nrbf/serializationheader type='SerializationHeaderRecord' RootId=0 HeaderId=0 MajorVersion=1 MinorVersion=0>
<nrbf/binarymethodreturn type='BinaryMethodReturn' MessageEnum='ReturnValueInline|NoContext|NoArgs' ReturnValue='Address received' CallContext=None Args=None>

Print the records in the given file to stdout:

nrbf-extract --file nrbf_addr_received_buffer.bin stdout
<nrbf/serializationheader type='SerializationHeaderRecord' RootId=0 HeaderId=0 MajorVersion=1 MinorVersion=0>
<nrbf/binarymethodreturn type='BinaryMethodReturn' MessageEnum='ReturnValueInline|NoContext|NoArgs' ReturnValue='Address received' CallContext=None Args=None>

Output BinaryObjectString records from the specified hex stream using rdump:

$ nrbf-extract --hex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records | rdump -s "'BinaryObjectString' in r.type"
[reading from stdin]
<nrbf/binaryobjectstring type='BinaryObjectString' ObjectId=4 Value='One Microsoft Way'>
<nrbf/binaryobjectstring type='BinaryObjectString' ObjectId=5 Value='Redmond'>
<nrbf/binaryobjectstring type='BinaryObjectString' ObjectId=6 Value='WA'>
<nrbf/binaryobjectstring type='BinaryObjectString' ObjectId=7 Value='98054'>

API Usage

The script was made to be easy in use when parsing NRBF streams as flow.record well.

from io import BytesIO

from nrbf.nrbf import NRBF

stream = BytesIO(bytes.fromhex("000000000000000000010000000000000016110800001210416464726573732072656365697665640B"))
nrbf = NRBF(stream=stream)
for record in nrbf.records():
    # Do something with the record

Or getting the raw header and record as cstruct objects for further parsing:

from io import BytesIO

from nrbf.nrbf import NRBF

stream = BytesIO(bytes.fromhex("000000000000000000010000000000000016110800001210416464726573732072656365697665640B"))
nrbf = NRBF(stream=stream)
for header, record in nrbf.parse():
    # Do something with the header and record

TODO