dnspyjson: Seamlessly Serialize dnspython2.x Answers Into JSON or Native Python Objects

This package provides a json.JSONEncoder subclass for the standard cPython json package that allows a dnspython>=2 response to be easily converted into a JSON string, native Python objects, or a JSON file. It also includes a public interfaces so you don't need to interact directly with the JSON encoder or even with dnspython (optional)

NOTE: See the usage examples below; you can not simply pass the JSON encoder class that is implemented in this package directly to json.dump using cls=

Requirements

• Python >= 3.5 (Python2 will not work, Python < 3.5 is not tested; note that PEP-484 type hints are used in this code)
• dnspython >= 2 (dnspython < 2 will likely not work)

Supported Answer Types

There aren't any known types that are lacking support at this time, though there are some enhancements to be made (see TODO section)

Installing

Install master/latest stable:

$ pip install 'git+https://github.com/mzpqnxow/dnspyjson@master#egg=dnspyjson'

Install a specific release by tag:

$ pip install 'git+https://github.com/mzpqnxow/[email protected]#egg=dnspyjson'

You can also use this syntax in requirements.txt

Using In A Project As A setuptools Dependency

Master/latest stable:

install_requires = dnspyjson @ git+https://github.com/mzpqnxow/dnspyjson@master#egg=dnspyjson  # Master (latest stable)

Pin a specific release by tag:

install_requires = dnspyjson @ git+https://github.com/mzpqnxow/[email protected]#egg=dnspyjson  # Pin to 0.0.8

Known Working Query Types

Confirmed working against all sorts of different Answer types, including but not limited to:

• A
• AAAA
• CERT
• SRV
• NS
• SOA
• PTR
• SPF
• DNSKEY
• MX
• RRSIG
• TXT

Example Pretty Print Output

{
  "qname": "walmart.com.",
  "rdtype": 6,
  "rdclass": 1,
  "nameserver": "127.0.0.1",
  "port": 53,
  "canonical_name": "walmart.com.",
  "rrset": [
    {
      "rname": "hostmaster.walmart.com.",
      "minimum": 300,
      "mname": "pdns1.ultradns.net.",
      "refresh": 3600,
      "retry": 3600,
      "expire": 604800,
      "serial": 2016113072,
      "rdclass": 1,
      "rdtype": "SOA"
    }
  ],
  "expiration": 1602519121.0416613
}

Usage

If you already have an dns.resolver.Answer object from your own direct use of dnspython, you can use:

from dnspyjson import dns_answer_to_json
...
answer = resolver.resolve(qname, qtype)  # Using dnspython directly
json_string = dns_answer_to_json(answer)  # Returns a Python str containing JSON
native_object = dns_answer_to_json(answer, as_native_object=True)  # A native Python object (list of dict)
dns_answer_to_json(answer, to_file='response.out)  # Ouput the JSON to a file

If you don't want to use dnspython directly, you can use the json_dns_query() wrapper:

from dnspyjson import json_dns_query
...
qname = 'test.com'
qtype = 'A'
resolvers = ['1.1.1.1', '8.8.8.8']
result = json_dns_query('test.com', 'TXT', ['4.2.2.2', '8.8.8.8'])

All of the kwargs values supported by dns_answer_to_json() are accepted and passed along by json_dns_query():

result = json_dns_query('test.com', 'TXT', ['4.2.2.2', '8.8.8.8'], as_native_object=True)

Finally, you can use the CLI app, json-resolve (or see the source in dnspyjson/app.py) by specifying a specific record type, or using the special type "ALL" which will query all of the known record types in succession

$ json-resolve test.com TXT | jq '.' -c
Nameservers
===========
- 1.1.1.1
- 8.8.8.8
{"qname":"test.com.","rdtype":16,"rdclass":1,"nameserver":"1.1.1.1","port":53,"chaining_result":{"canonical_name":"test.com.","answer":[{"ttl":3501,"rdclass":"IN","rdtype":"TXT","strings":["google-site-verification=kW9t2V_S7WjOX57zq0tP8Ae_WJhRwUcZoqpdEkvuXJk"]}],"minimum_ttl":3501,"cnames":[]},"canonical_name":"test.com.","rrset":[{"ttl":3501,"rdclass":"IN","rdtype":"TXT","strings":["google-site-verification=kW9t2V_S7WjOX57zq0tP8Ae_WJhRwUcZoqpdEkvuXJk"]}],"expiration":1612744578.296427,"isoformat_expiration":"2021-02-08T00:36:18","request_timestamp":"2021-02-07T23:37:57.296469"}

The CLI app is very primitive and is only meant as a test driver. If you would like to use it for very specialized tasks/queries, you will need to modify it. Contributing it back via a Pull Request would be nice of you

Use-Case(s)

The use-case for development of this was for a simple web application that performed comprehensive DNS interrogation for a domain (e.g. MX, TXT, A, AAAA, ...) and returned them in a pretty printed way in the browser that a user could click into to expand and collapse with Javascript. If you want to do a very, very large amount of queries, you should use a specialized tool like massdns- this is meant as a replacement for tools like dig, host and nslookup, or for use as a backend for a web service/application

Example Use-Cases

• Storage of JSON directly into a PostgreSQL JSONB column, to sidestep the need to create a complex schema capable of properly storing all different types of answers
• Render JSON as a response in a web application that performs DNS lookups as part of its functionality
• Conversion into pandas.DataFrame objects, where large-scale analysis/processing can be done efficiently
• Conversion into native Python objects, if that's more comfortable for you (that's a pretty lame one)

Example Applications: PostgreSQL: Creating indexes and querying JSONB (structured JSON) columns in PostgreSQL >= 9.2

PostgreSQL supported JSON columns for quite some time, but there wasn't much value as the data wasn't indexable. The addition of JSONB to PostgreSQL 9.2 was a game-changer as it allowed JSON objects to be easily stored as well as efficiently indexed and searched as if the keys were predefined columns. You can refer here for some information on the difference between JSON and JSONB in PostgreSQL

What follows are a few very brief example of how you can use JSONB columns in PostgreSQL to store and then search on DNS data. For comprehensive reference covering the syntax for accessing deeply nested, how to operate on arrays/lists and anything else you can think of, see the PostgreSQL 'JSONB' documentation here

Create a very basic table with a JSONB column

Create a table with an auto-incrementing ID, a creation timestamp, and a JSONB column to store the JSON data in an indexable way:

CREATE TABLE dns_answer (
    id BIGSERIAL PRIMARY KEY,
    queried_ts TIMESTAMPTZ NOT NULL DEFAULT NOW(),
    answer JSONB);

Create a few simple indexes on 'JSONB' columns

Create an index on the creation timestamp and one across both the creation timestamp and the canonical name from the answer, to optimize queries using those columns

CREATE INDEX dns_answer_data_queried_ts_idx ON dns_answer (queried_ts);
CREATE INDEX dns_answer_data_queried_ts_canonical_name_idx ON dns_answer (queried_ts, (data->>'canonical_name'));

You can create alternate index types such as GiST and GIN indexes on specific keys within the 'JSONB' column as well. Dealing with the 'JSONB' column is much like dealing with a table.

Make some simple queries on 'JSONB' columns using indexes

SELECT * FROM dns_answer WHERE data->>'canonical_name' = 'somesite.com';
SELECT data->>'canonical_name', data->>'rrset' FROM dns_answer WHERE queried_ts

For more advanced queries, see PostgreSQL documentation here

TODO (Maybe)

• Intelligent handling of certain integer and floating-point values; no compelling use-case for the moment, though maybe at some point it will be useful
  • Detect epoch-times and (optionally) convert them to human-readable date format
  • Detect ttl values and (optionally) convert them from seconds to minutes, hours, days
• Support other dnspython data-types (e.g. QueryMessage)
  • I don't have any use-case for this currently
• Expand the scope to store responses in PostgreSQL or some none SQL datastore
  • I don't have a use-case for this, and it would be a significant change in scope, from an afternoon project to a week-long project, which I don't have the time for right now