dnsmule

DNSMule

Package for rule based dependency scanning and service fingerprinting via DNS.

This package provides utilities for writing and evaluating verbose and easy to read rule definitions in YAML-format. There are two builtin rule formats with more available as plugins.

Installation

pip install dnsmule[full] dnsmule_plugins[full]

This will install everything available for DNSMule. You can also choose to install components as necessary.

For installing from the repo you can use:

pip install -e .

Overview

The DNSMule tool is composed in the following way:

• DNSMule
  • Backend
  • Rules
    • Rule
  • Plugins

Examples

Check out the examples in the examples folder. They should get you up and running quickly.

YAML Configuration

Summary

The tool configuration is done through one or multiple rule configuration files. The file structure is defined in the schema file. In addition to some builtin rule types, it is possible to create new types by registering handlers or rules programmatically.

Rules support registration per DNS record type and priority for controlling invocation order.

version: '0.0.1'
rules:
  - name: o365
    priority: 10
    type: dns.regex
    record: txt
    config:
      pattern: '^MS=ms'
      identification: MICROSOFT::O365
  - name: ses
    type: dns.regex
    record: txt
    config:
      pattern: '^amazonses:'
      identification: AMAZON::SES

Example

from dnsmule.definitions import Record, Result
from dnsmule.rules import Rules, Rule

rules: Rules

...


@rules.add.A[10]
def my_scan(record: Record) -> Result:
    from dnsmule.logger import get_logger
    get_logger().info('Address %s', record)
    return record.tag('MY::SCAN')


@rules.register('my.rule')
def create_my_rule(**arguments) -> Rule:
    ...

Here the add is used to directly register a new rule into the ruleset with a given priority. The register call creates a new handler for rules of type my.rule. Any future my.rule creations from YAML or code would be routed to this factory function.

See the examples folder for more examples and how to use rules in code.

Plugins and Backends

Plugins

It is possible to register plugins using the YAML file:

plugins:
  - name: dnsmule_plugins.CertCheckPlugin
    config:
      callback: false

These are required to extend the dnsmule.plugins.Plugin class. Plugins are evaluated and initialized before rules. Any rules requiring a plugin should list their plugin in this block. Plugins are only initialized once and if a plugin already exists in the receiving DNSMule instance it will be ignored.

Backends

It is possible to define a single backend in a YAML file:

backend:
  name: 'dnspython.DNSPythonBackend'
  config:
    timeout: 5.5
    resolver: 8.8.8.8

The backend must extend the dnsmule.backends.Backend class. This declaration is ignored if this is not used in DNSMule.load or DNSMule(file=file).

Editor Support

Type Hints and JSON Schema (IntelliJ IDEA, PyCharm, etc.)

It is possible to register the schema file as a custom JSON schema in IntelliJ editors. This will give access to typehints and schema validation inside rule files and is especially nice for dynamic rule definitions as you get full editor features for python inside the snippets.

• settings...
• Languages & Frameworks > Schemas and DTDs > JSON Schema Mappings
• Add a new mapping with the schema file and specified file or pattern

This is configured in the schema using the custom intellij language injection tag:

x-intellij-language-injection:
  language: Python
  prefix: |+0
    code hints go here
    check the schema for more info
  suffix: ''

Currently, this supports dns.regex pattern regex language injection and dns.dynamic rule code language injection. Type hints and quick documentation are available.

Builtin Rules

Regex rules

Regex rules can be defined with either one pattern or multiple patterns. An example is in the following snippet:

rules:
  - name: test
    type: dns.regex
    record: txt
    config:
      pattern: '^.*\.hello_world\.'
      identification: HELLO::WORLD
      flags:
        - UNICODE
        - DOTALL
  - name: generic_verification
    type: dns.regex
    record: txt
    priority: 10
    description: Generic Site Regex Collection
    config:
      patterns:
        - '^(.+)(?:-(?:site|domain))?-verification='
        - '^(.+)(?:site|domain)verification'
        - '^(.+)_verify_'
        - '^(\w+)-code:'
      group: 1

The full definition and additional info is available from the schema file, examples, and code.

Dynamic Rules

Dynamic rules are defined as code snippets with one or two methods

An init method that is invoked once after creation

def init() -> None:
    add_rule(...)

A process function that is invoked once for each record

def process(record: Record) -> Result:
    add_rule(...)
    return record.result()

Both of these functions have access to the following rule creation method:

def add_rule(
        record_type: Union[str, int, RRType],
        rule_type: str,
        name: str,
        *,
        priority: int = 0,
        **options,
) -> None:
    """
    :param record_type: Valid DNS record type as text, int, or type
    :param rule_type:   Valid rule type factory e.g. dns.regex
    :param name:        Name of the created rule
    :param priority:    Priority for the created rule, default 0
    :param options:     Any additional options for the rule factory
    """

The only globals passed to these methods are:

• __builtins__
• RRType, Record, Result, Domain, Tag, Config
  • The Config contains the config property passed to the rule from YAML
• add_rule
• Any additional globals created by the code itself

When the code is exec'd the result is inspected for:

• init function without parameters
• process function with a single parameter

Some notes:

• The init function is invoked exactly once.
• The process function is invoked exactly once for every single Record.
• Any rules created from the init method will be invoked for every suitable record.
• Any rules created from the process method will be invoked for suitable records found after creation.
• Creating DynamicRules from init or process is considered undefined behaviour and care should be taken
  • The user should call init manually and include fail-safes for only calling it once
  • The add_rule callback might not be available so you need to pass it manually to the rule

Other

Example server

The repo has a Dockerfile for easily running the tool using an example server in Docker:

$ ./build-image
$ ./run-server

Notice

This package is under active development.

Additional

• RnD Scripts under scripts
• Example server under server
• Examples for coding under examples