The National Institute of Standards and Technology (NIST) maintain and distribute a set of known file hashes called the National Software Reference Library (NSRL) Reference Data Sets (RDS). This data set has several applications, including malware analysis and computer forensics.

This library has everything you need to:

• Ingest NIST NSRL RDS formatted CSV files into an Elasticsearch instance, including from the ISO images distributed by NIST,
• A conventient Python library for accessing the data from Elasticsearch,
• A HTTP microservice which lets you expose the indexed data without unsafely exposing your Elasticsearch instance,
• Another convenient Python library for accessing the data from the HTTP service (which exposes the same API as the the Elasticsearch client), and
• A Command Line Interface (CLI) for ingesting the NIST NSRL RDS data which can use either of the previously described clients.

After installation, it may be desired to configure the HTTP server and the CLI tool. To do so, run the `configure` subcommand. This will prompt for configuration options along with a description of what they do. For testing purposes, the defaults should be fine.

Once configured, the CLI tool can be used to run ingest NSRL data, run the HTTP server and query the data set.

Examples of each of these are provided below.

$ nsrlsearch configure
Interface for web server to listen on (default: localhost)? 
Port for web server to listen on (default: 8080)?
.
. (edited out for brevity)
.
New configuration written to /home/user/.nsrlsearch.cfg.

$ nsrlsearch ingest RDS_ios.iso
Inserting mfg info... done! Put 80456 in 7.739879s
Inserting os info... done! Put 913 in 0.157383s
.
. (edited out for brevity)
.
File ingest done! Put 14390472 in 2481.755694s

$ nsrlsearch web
Server's client will connect to Elasticsearch with config:
  create_indices: true
  eskwargs: {'hosts': ['localhost:9200']}
  index_base: nsrl
  connection_check: true
Server's client passed Elasticsearch connection check.
Starting server (gunicorn)...
Bottle v0.12.13 server starting up (using GunicornServer(workers=8))...
Listening on http://localhost:8080/
Hit Ctrl-C to quit.

$ nsrlsearch query aaaaaaaa bbbbbbbb cccccccc
aaaaaaaa: False
bbbbbbbb: False
cccccccc: False

The `nsrlsearch command and all its subcommands have help available via the -h/--help` options.

Typically, when using this library from Python the main goal will be to either ingest or query hashes from the Elasticsearch server. This is achieved by the client classes of :pynsrlsearch.client.EsClient (which works directly with Elasticsearch) and :pynsrlsearch.client.HttpClient (which works directly with the HTTP server).

Once instantiated, these two clients expose the same interface for all querying and indexing operations.

In [1]: from elasticsearch import Elasticsearch

In [2]: from nsrlsearch.client import EsClient, HttpClient

In [3]: # Create a Elasticsearch client for nsrlsearch's EsClient to use.

In [4]: es = Elasticsearch(hosts=["localhost:9200"])

In [5]: # Create nsrlsearch's EsClient to query the the Elasticsearch server.

In [6]: client = EsClient(es=es)

In [7]: client.get_digest_exists("00000000") # can be crc32, sha1, md5
Out[7]: False

In [8]: # Similarly, the HttpClient can be used to talk to the Http server (if it is running)

In[9]: client = HttpClient(uri="http://localhost:8080")

In[10]: client.get_digest_exists("00000000") # can be crc32, sha1, md5
Out[10]: False

There is much that could be done here, including:

• a properly RESTful HTTP interface on the microservice
• better support for alternate sets of data
• support/testing against Elasticsearch 6.x

Contributions on the above or other useful features will be appreciated.

Version 0.0.2.60 - Initial release:

• ingest and query NSRL directly into Elasticsearch
• ingest and query NSRL data through and intermediary HTTP microservice
• clients for direct communication with Elasticsearch and the intermediary HTTP microservice expose identical interfaces for both querying and ingest
• support ingest direct from ISO image
• tested against NSRL RDS 2.60 on Elasticsearch 5.6 (Python 2 and 3)

The nsrlsearch library and package is licensed under APLv2.

The National Institute of Standards and Technology for their ongoing work building and curating the NSRL data set.