DC3 Malware Configuration Parser (DC3-MWCP) is a framework for parsing configuration information from malware. The information extracted from malware includes items such as addresses, passwords, filenames, and mutex names. A parser module is usually created per malware family. DC3-MWCP is designed to help ensure consistency in parser function and output, ease parser development, and facilitate parser sharing. DC3-MWCP supports both analyst directed analysis and large-scale automated execution, utilizing either the native python API, a REST API, or a provided command line tool. DC3-MWCP is authored by the Defense Cyber Crime Center (DC3).
- Install
- Builtin Parsers
- Dragodis Support
- DC3-Kordesii Support
- Usage
- Schema
- STIX Output
- YARA Matching
- Helper Utilities
- Parser Development
- Parser Components
- Parser Installation
- Parser Testing
- Python Style Guide
- Construct Tutorial
- Style Guide
- Testing
> pip install mwcp
Alternatively you can clone this repo and install locally.
> git clone https://github.com/Defense-Cyber-Crime-Center/DC3-MWCP.git
> pip install ./DC3-MWCP
For a development mode use the -e
flag to install in editable mode:
> git clone https://github.com/Defense-Cyber-Crime-Center/DC3-MWCP.git
> pip install -e ./DC3-MWCP
DC3-MWCP includes a handful of builtin parsers to get you started. These can be used as-is, subclassed, or included in your own parser groups.
To view the available parsers:
$ mwcp list
Parsers are installed under the dc3
source name. To include them in a group simply add them with
the dc3:
prefix.
SuperMalware:
description: SuperMalware component
author: acme
parsers:
- dc3:Archive.Zip
- .Dropper
- .Implant
- dc3:Decoy
DC3-MWCP optionally supports Dragodis
if it is installed. This allows you to obtain a disassembler agnostic interface for parsing
the file's disassembly from the mwcp.FileObject
object with the .disassembly()
function.
You can install Dragodis along with DC3-MWCP by adding [dragodis]
to your appropriate install command:
pip install mwcp[dragodis]
pip install ./DC3-MWCP[dragodis]
pip install -e ./DC3-MWCP[dragodis]
After installation make sure to follow Dragodis's installation instructions to setup a backend disassembler.
It is recommended to also install Rugosa for emulation and regex/yara matching capabilities using Dragodis.
DC3-MWCP optionally supports DC3-Kordesii
if it is installed. This will allow you to run any DC3-Kordesii decoder from the
mwcp.FileObject
object with the run_kordesii_decoder
function.
You can install DC3-Kordesii along with DC3-MWCP by adding [kordesii]
to your appropriate install command:
pip install mwcp[kordesii]
pip install ./DC3-MWCP[kordesii]
pip install -e ./DC3-MWCP[kordesii]
DC3-MWCP is designed to allow easy development and use of malware config parsers. DC3-MWCP is also designed to ensure that these parsers are scalable and that DC3-MWCP can be integrated in other systems.
Most automated processing systems will use a condition, such as a yara signature match, to trigger execution of an DC3-MWCP parser.
There are 3 options for integration of DC3-MWCP:
- CLI:
mwcp
- REST API:
mwcp serve
- Python API
DC3-MWCP also includes a utility for test case generation and execution.
DC3-MWCP can be used directly from the command line using the mwcp
command.
> mwcp parse foo ./README.md
----- File: README.md -----
Field Value
------------ ----------------------------------------------------------------
Parser foo
File Path README.md
Description Foo
Architecture
MD5 b21df2332fe87c0fae95bdda00b5a3c0
SHA1 8841a1fff55687ccddc587935b62667173b14bcd
SHA256 0097c13a3541a440d64155a7f4443d76597409e0f40ce3ae67f73f51f59f1930
Compile Time
Tags
---- Socket ----
Tags Address Network Protocol
------ --------- ------------------
127.0.0.1 tcp
---- URL ----
Tags Url Address Network Protocol Application Protocol
------ ---------------- --------- ------------------ ----------------------
http://127.0.0.1 127.0.0.1 tcp http
---- Residual Files ----
Tags Filename Description MD5 Arch Compile Time
------ ----------------- ------------------- -------------------------------- ------ --------------
fooconfigtest.txt example output file 5eb63bbbe01eeed093cb22bb8f5acdc3
---- Logs ----
[+] File README.md identified as Foo.
[+] size of inputfile is 15560 bytes
[+] README.md dispatched residual file: fooconfigtest.txt
[+] File fooconfigtest.txt described as example output file
[+] operating on inputfile README.md
----- File Tree -----
<README.md (b21df2332fe87c0fae95bdda00b5a3c0) : Foo>
└── <fooconfigtest.txt (5eb63bbbe01eeed093cb22bb8f5acdc3) : example output file>
see mwcp parse -h
for full set of options
DC3-MWCP can be used as a web service. The web service provides a web application as well as a REST API for some commonly used functions:
/run_parser/<parser>
-- executes a parser on uploaded file/run_parser
-- executes a parser on uploaded file using YARA matching to determine the parser./descriptions
-- provides list of available parsers/schema.json
-- provides the schema for report output
To use, first start the server by running:
> mwcp serve
Then you can either use an HTTP client to create REST requests.
The REST API for /run_parser
will accept a number of request parameters for customizing the processing and output results.
data
-- The input file data.legacy
-- If this argument is set toTrue
, the legacy JSON schema output will be produced. Defaults to the new schema.- NOTE: Legacy output will eventually be removed in a 4.0 release.
output
-- Sets the output format for parsing results.json
-- JSON format (this is the default)zip
-- Generates a ZIP file containing results and extracted residual files.stix
-- STIX 2.1 JSON format
include_logs
-- Whether to include logs in the report. Defaults to True.no_file_data
-- If this argument is set toTrue
, binary data for extracted residual files won't be included in the report.recursive
- Whether to recursively process unidentified files with YARA matched parsers. (YARA_REPO must be setup for this option to be active.)external_strings
-- Whether to create external string reports for reported decoded strings found in each file. Defaults to False.- These reports will be returned in the same manner as residual files.
- When enabled, the strings in the main report will be removed.
param
/parameter
-- Provides external parameters which will be injected into theknowledge_base
before parsing starts.- Values should be a key/value pair split by a
:
. (e.g.param="key_name:secret"
) - This can be provided multiple times for multiple parameters.
- Values should be a key/value pair split by a
> curl --form [email protected] http://localhost:8080/run_parser/foo
import requests
req = requests.post("http://localhost:8080/run_parser/foo", files={'data': open("README.md", 'rb')})
req.json()
The default parsing results will be in JSON format following the standardized schema.
[
{
"type": "report",
"tags": [],
"input_file": {
"type": "input_file",
"tags": [],
"name": "README.md",
"description": "Foo",
"md5": "80a3d9b88c956c960d1fea265db0882e",
"sha1": "994aa37fd26dd88272b8e661631eec8a5f425920",
"sha256": "3bef8d5dc4cd94c0ee92c9b6d7ee47a4794e550d287ee1affde84c2b7bcdf3cb",
"architecture": null,
"compile_time": null,
"file_path": "README.md",
"data": null
},
"parser": "foo",
"errors": [],
"logs": [
"[+] File README.md identified as Foo.",
"[+] size of inputfile is 15887 bytes",
"[+] README.md dispatched residual file: fooconfigtest.txt",
"[+] File fooconfigtest.txt described as example output file",
"[+] operating on inputfile README.md"
],
"metadata": [
{
"type": "url",
"tags": [],
"url": "http://127.0.0.1",
"socket": {
"type": "socket",
"tags": [],
"address": "127.0.0.1",
"port": null,
"network_protocol": "tcp",
"c2": null,
"listen": null
},
"path": null,
"query": "",
"application_protocol": "http",
"credential": null
},
{
"type": "socket",
"tags": [],
"address": "127.0.0.1",
"port": null,
"network_protocol": "tcp",
"c2": null,
"listen": null
},
{
"type": "residual_file",
"tags": [],
"name": "fooconfigtest.txt",
"description": "example output file",
"md5": "5eb63bbbe01eeed093cb22bb8f5acdc3",
"sha1": "2aae6c35c94fcfb415dbe95f408b9ce91ee846ed",
"sha256": "b94d27b9934d3e08a52e52d7da7dabfac484efe37a5380ee9088f7ace2efcde9",
"architecture": null,
"compile_time": null,
"file_path": "README.md_mwcp_output\\5eb63_fooconfigtest.txt",
"data": null
}
]
}
]
A simple HTML interface is also available at the same address. By default this
is http://localhost:8080/
. Individual samples can be submitted and results
saved as JSON, plain text, or ZIP archives.
DC3-MWCP can be run directly from Python.
#!/usr/bin/env python
"""
Simple example to demonstrate use of the API provided by DC3-MWCP framework.
"""
# first, import mwcp
import mwcp
# register the builtin MWCP parsers and any other parser packages installed on the system
mwcp.register_entry_points()
# register a directory containing parsers
mwcp.register_parser_directory(r'C:\my_parsers')
# view all available parsers
print(mwcp.get_parser_descriptions(config_only=False))
# Call the run() function to generate a mwcp.Report object.
report = mwcp.run("FooParser", file_path=r"C:\input.exe")
# Run on provided data buffer.
report = mwcp.run("FooParser", data=b"lorem ipsum")
# Let MWCP determine parser(s) to run based on YARA match results by excluding the parser.
# (YARA_REPO should be setup with `mwcp config` or passed in with "yara_repo" argument)
report = mwcp.run(file_path=r"C:\input.exe")
report = mwcp.run(data=b"lorem ipsum")
# Provide external knowledge by supplying a knowledge_base dictionary.
report = mwcp.run("FooParser", file_path=r"C:\input.exe", knowledge_base={"key": "secret"})
# Display report results in a variety of formats:
print(report.as_dict())
print(report.as_json())
print(report.as_text())
print(report.as_markdown())
print(report.as_html())
print(report.as_csv())
print(report.as_dataframe()) # Pandas dataframe object.
print(report.as_stix()) # STIX 2.1 JSON formatted text.
# To get the legacy format use the following:
print(report.as_dict_legacy())
print(report.as_json_legacy())
# You can also programmatically view results of report:
from mwcp import metadata
# display errors that may occur
for log in report.errors:
print(log)
# display data about original input file
print(report.input_file)
# get all url's using ftp protocol or has a query
for url in report.get(metadata.URL):
if url.application_protocol == "ftp" or url.query:
print(url.url)
# get residual files
for residual_file in report.get(metadata.File):
print(residual_file.name)
print(residual_file.description)
print(residual_file.md5)
print(repr(residual_file.data))
# iterate through all metadata elements
for element in report:
print(element)
DC3-MWCP uses a configuration file which is located within the user's
profile directory. (%APPDATA%\Local\mwcp\config.yml
for Windows or ~/.config/mwcp/config.yml
for Linux)
This configuration file is used to manage configurable parameters, such as the location of the malware repository used for testing or the default parser source.
To configure this file, run mwcp config
to open up the file in your default text
editor.
An alternative configuration file can also be temporarily set using the --config
parameter.
> mwcp --config='new_config.yml' test Foo
Individual configuration parameters can be overwritten on the command line using the respective parameter.
DC3-MWCP uses Python's builtin in logging
module to log all messages.
By default, logging is configured using the log_config.yml configuration
file. Which is currently set to log all messages to the console and error messages to %LOCALAPPDATA%/mwcp/errors.log
.
You can provide your own custom log configuration file by adding the path
to the configuration parameter LOG_CONFIG_PATH
.
(Please see Python's documentation for more information on how to write your own configuration file.)
You may also use the --verbose
or --debug
flags to adjust the logging level when using the mwcp
tool.
One of the major goals of DC3-MWCP is to standardize output for malware configuration parsers, making the data from one parser comparable with that of other parsers. This is achieved by establishing a schema of standardized metadata elements that represent the common malware configuration items seen across malware families.
A formal JSON Schema can be found at schema.json, by calling mwcp schema
in the command line, or programmatically by calling mwcp.schema()
.
This schema is versioned the same as DC3-MWCP. A change in the version may not necessarily
reflect a change in the actual schema. However, any major or minor changes to the schema will
be reflected in an appropriate change to the version and will be noted in the changelog.
Please ensure you pin DC3-MWCP appropriately.
It is acknowledged that a set of generic elements will often not be adequate to capture the nuances of individual malware families. To ensure that malware family specific attributes are appropriately captured in parser output, the schema includes an "Other" element which supports arbitrary key-value pairs. The keys and values are arbitrary to permit flexibility in describing the peculiarities of individual malware families. Information not captured in the abstract standardized elements is captured through this mechanism.
The use of tags is encouraged to provide additional context for the configuration items. For example, if a specific url is used to download a second stage component, a tag of "download" could be added to the reported URL element. Alternatively, if the URL is used for a proxy, a tag of "proxy" could be included. There is no standard on what tags are available or when they should be included. This should be determined by your organization.
It is possible to extend the schema to include your own custom metadata elements.
This can be accomplished by creating a class that inherits from mwcp.metadata.Metadata
.
This class must be decorated with attr using the custom configuration mwcp.metadata.config
.
NOTE: The class name must be unique from other metadata elements.
from typing import List
import attr
import mwcp
from mwcp import metadata
@attr.s(**metadata.config)
class MyCustom(metadata.Metadata):
"""
This is my custom metadata item.
"""
field_a: str
field_b: int
field_c: List[str] = attr.ib(factory=list)
item = MyCustom(field_a="hello", field_b=42, field_c=["a", "b"])
print(item)
print(item.as_dict())
# Custom items can be included in the report like normal.
# MWCP will automatically format and display the custom element in the report.
report = mwcp.Report()
with report:
report.add(item)
print(report.as_text())
MyCustom(tags=set(), field_a='hello', field_b=42, field_c=['a', 'b'])
{'type': 'my_custom', 'tags': [], 'field_a': 'hello', 'field_b': 42, 'field_c': ['a', 'b']}
---- My Custom ----
Tags Field A Field B Field C
------ --------- --------- ----------
hello 42 a, b
Please note, that extending the schema will obviously cause the schema.json file to be incorrect.
To regenerate the schema to also include the custom element run mwcp.schema()
afterwards.
import json
import mwcp
with open("schema.json", "w") as fo:
json.dump(mwcp.schema(id="https://acme.org/0.1/schema.json"), fo, indent=4)
MWCP can generate a STIX 2.1 JSON output that is suitable for integration into many systems that support the STIX standard. This output format makes use of three SCO extensions and one property extension in addition to the currently defined STIX objects order to accurately convey MWCP's scan results.
Some tools may not support these extensions yet which can result in the following data being omitted when ingesting MWCP's STIX output. The following provides a list of STIX objects and extensions are used and what MWCP classes these are associated with:
- artifact (SCO)
- File -- only used if the original binary is requested
- crypto-currency-address (SCO Extension)
- CryptoAddress
- directory (SCO)
- File
- Path
- Service
- domain-name (SCO)
- Socket
- URL
- email-address (SCO)
- EmailAddress
- file (SCO)
- File
- Path
- Service
- ipv4-address (SCO)
- Socket
- URL
- ipv6-address (SCO)
- Socket
- URL
- malware-analysis (SDO)
- MWCP's scan results are tied together via a malware-analysis object showing the input object and the outputs
- mutex (SCO)
- Mutex
- network-traffic (SCO)
- Socket
- URL
- note (SDO)
- Boolean and Integer values for Other. These are added to the description of the Note.
- Descriptions and other narrative text tied to SCOs
- Tags for SCOs excluding files
- observed-string (SCO Extension)
- DecodedString
- MissionID
- Other
- Pipe
- User Agent
- UUID
- process (SCO)
- Command
- Service
- relationship (SRO)
- DecodedString
- URL
- RSA Private Key (Property Extension for x509-certificate)
- RSAPrivateKey
- symmetric-encryption (SCO)
- EncryptionKey
- user-account (SCO)
- Credential
- url (SCO)
- URL
- x509-certificate (SCO)
- RSAPrivateKey
- RSAPublicKey
- SSLCertSHA1
- windows-registry-key (SCO)
- Registry2
MWCP includes a runner that can use YARA match results to determine which parser(s) to run on a given file.
This will be used whenever you use -
instead of specifying a parser on the command line,
when a parser isn't specified in mwcp.run()
, or when a parser isn't specified in a server request.
$ mwcp parse - input.exe
$ curl --form [email protected] http://localhost:8080/run_parser
import mwcp
mwcp.register_entry_points()
report = mwcp.run(data=b"file data")
As well, YARA matching will be recursively used on unidentified residual files.
If you want to disable this, either set --no-recursive
on the command line or set recursive=False
on mwcp.run()
.
To enable YARA matching you'll need to specify a directory containing YARA signatures which use the mwcp
meta field to map a signature to a comma delimited list of parsers. Parsers can be specified in the same
way as on the command line or Python API. That is, parser group names, .
notation for specific parser components,
and the use of :
for specifying a parser source are all valid.
Any signatures that don't have the mwcp
meta field will be ignored.
rule SuperMalware {
meta:
mwcp = "SuperMalware"
...
}
To setup a YARA repo, set the YARA_REPO
field to point to a directory containing YARA signatures (subdirectories allowed)
in the configuration file that appears when you call mwcp config
.
If you have upgraded from an older version of MWCP, you may need to first backup and remove the original configuration file and
then run mwcp config
again to have MWCP recreate the file.
Alternatively, the yara repo can be specified in the command line with --yara-repo
. But the former method
is necessary to use YARA matching with the server.
Recursive YARA matching for unidentified files can be done when creating test cases. Simply include the --recursive
flag
when adding a new or updating an existing test case.
$ mwcp test SuperMalware --add="C:\input.exe" --recursive
$ mwcp test SuperMalware -u --recursive
Once a test case is created with recursion turned on, anybody running your test case must also have a YARA repo setup with the same YARA signatures so the test will pass for them. For this reason, it is recommended to turn on recursion for a test only if the parser's full functionality depends on it.
MWCP comes with a few helper utilities (located in mwcp.utils
) that may become useful for parsing malware files.
pefileutils
- Provides helper functions for common routines done with thepefile
library. (obtaining or checking for exports, imports, resources, sections, etc.)elffileutils
- Provides helper functions for common routines done with theelftools
library. Provides a consistent interface similar topefileutils
.custombase64
- Provides functions for base64 encoding/decoding data with a custom alphabet.construct
- Provides extended functionality to the construct library and brings back some lost features from version 2.8 into 2.9.- This library has replaced the
enstructured
library originally found in the resources directory. - Please follow this tutorial for migrating from
enstructured
toconstruct
.
- This library has replaced the
pecon
- PE file reconstruction utility.- Please see docstring in pecon.py for more information.
poshdeob
- An experimental powershell deobfuscator utility used to statically deobfuscate code and extract strings.