Python-Iocextract - Advanced Indicator Of Compromise (IOC) Extractor

Advanced Indicator of Compromise (IOC) extractor.


Overview
This library extracts URLs, IP addresses, MD5/SHA hashes, email addresses, and YARA rules from text corpora. It includes some encoded and "defanged" IOCs in the output, and optionally decodes/refangs them.


The Problem
It is common practice for malware analysts or endpoint software to "defang" IOCs such as URLs and IP addresses, in order to prevent accidental exposure to live malicious content. Being able to extract and aggregate these IOCs is often valuable for analysts. Unfortunately, existing "IOC extraction" tools often pass right by them, as they are not caught by standard regex.
For example, the simple defanging technique of surrounding periods with brackets:

127[.]0[.]0[.]1

Existing tools that use a simple IP address regex will ignore this IOC entirely.


The Solution
By combining specially crafted regex with some custom postprocessing, we are able to both detect and deobfuscate "defanged" IOCs. This saves time and effort for the analyst, who might otherwise have to manually find and convert IOCs into machine-readable format.


A Simple Use Case
Many Twitter users post C2s or other valuable IOC information with defanged URLs. For example, this tweet from @InQuest:

Recommended reading and great work from @unit42_intel:
https://researchcenter.paloaltonetworks.com/2018/02/unit42-sofacy-attacks-multiple-government-entities/ ...
InQuest customers have had detection for threats delivered from hotfixmsupload[.]com
since 6/3/2017 and cdnverify[.]net since 2/1/18.

If we run this through the extractor, we can easily pull out the URLs:

https://researchcenter.paloaltonetworks.com/2018/02/unit42-sofacy-attacks-multiple-government-entities/
hotfixmsupload[.]com
cdnverify[.]net

Passing in refang=True at extraction time would remove the obfuscation, but since these are real IOCs, let's leave them defanged in our documentation. :)


Installation
You may need to install the Python development headers in order to install the regex dependency. On Ubuntu/Debian-based systems, try:

sudo apt-get install python-dev

Then install iocextract from pip:

pip install iocextract

If you have problems installing on Windows, try installing regex directly by downloading the appropriate wheel from PyPI and running e.g.:

pip install regex-2018.06.21-cp27-none-win_amd64.whl

Usage
Try extracting some defanged URLs:

>>> content = """
... I really love example[.]com!
... All the bots are on hxxp://example.com/bad/url these days.
... C2: tcp://example[.]com:8989/bad
... """
>>> import iocextract
>>> for url in iocextract.extract_urls(content):
...     print url
...
hxxp://example.com/bad/url
tcp://example[.]com:8989/bad
example[.]com
tcp://example[.]com:8989/bad

Note that some URLs may show up twice if they are caught by multiple regexes.
If you want, you can also "refang", or remove common obfuscation methods from IOCs:

>>> for url in iocextract.extract_urls(content, refang=True):
...     print url
...
http://example.com/bad/url
http://example.com:8989/bad
http://example.com
http://example.com:8989/bad

You can even extract and decode hex-encoded and base64-encoded URLs:

>>> content = '612062756e6368206f6620776f72647320687474703a2f2f6578616d706c652e636f6d2f70617468206d6f726520776f726473'
>>> for url in iocextract.extract_urls(content):
...     print url
...
687474703a2f2f6578616d706c652e636f6d2f70617468
>>> for url in iocextract.extract_urls(content, refang=True):
...     print url
...
http://example.com/path

All extract_* functions in this library return iterators, not lists. The benefit of this behavior is that iocextract can process extremely large inputs, with a very low overhead. However, if for some reason you need to iterate over the IOCs more than once, you will have to save the results as a list:

>>> list(iocextract.extract_urls(content))
['hxxp://example.com/bad/url', 'tcp://example[.]com:8989/bad', 'example[.]com', 'tcp://example[.]com:8989/bad']

A command-line tool is also included:

$ iocextract -h
usage: iocextract [-h] [--input INPUT] [--output OUTPUT] [--extract-emails]
              [--extract-ips] [--extract-ipv4s] [--extract-ipv6s]
              [--extract-urls] [--extract-yara-rules] [--extract-hashes]
              [--custom-regex REGEX_FILE] [--refang] [--strip-urls]
              [--wide]

Advanced Indicator of Compromise (IOC) extractor. If no arguments are
specified, the default behavior is to extract all IOCs.

optional arguments:
  -h, --help            show this help message and exit
  --input INPUT         default: stdin
  --output OUTPUT       default: stdout
  --extract-emails
  --extract-ips
  --extract-ipv4s
  --extract-ipv6s
  --extract-urls
  --extract-yara-rules
  --extract-hashes
  --custom-regex REGEX_FILE
                        file with custom regex strings, one per line, with one
capture group each
  --refang              default: no
  --strip-urls          remove possible garbage from the end of urls. default:
                        no
  --wide                preprocess input to allow wide-encoded character
                        matches. default: no

Only URLs, emails, and IPv4 addresses can be "refanged".


Should I Use iocextract?
Are you...
Extracting possibly-defanged IOCs from plain text, like the contents of tweets or blog posts?
Yes! This is exactly what iocextract was designed for, and where it performs best. Want to go a step farther and automate extraction and storage? Check out ThreatIngestor.
Extracting URLs that have been hex or base64 encoded?
Yes, but the CLI might not give you the best results. Try writing a Python script and calling iocextract.extract_encoded_urls directly.
Note that you will most likely end up with extra garbage at the end of URLs.
Extracting IOCs that have not been defanged, from HTML/XML/RTF?
Maybe, but you should consider using the --strip-urls CLI flag (or the strip=True parameter in the library), and you may still get some extra garbage in your output.
If you're extracting from HTML, consider using something like Beautiful Soup to first isolate the text content, and then pass that to iocextract, like this.
Extracting IOCs that have not been defanged, from binary data like executables, or very large inputs?
Probably not. The regex in iocextract is designed to be flexible to catch defanged IOCs, so it performs significantly worse than a solution that is designed to catch only standard IOCs.
Consider using something like Cacador instead.