README.md

challenge

challenge for TruSTAR

Install

$ virtualenv -p python3 venv 
$ source venv/bin/activate
$ pip3 install -r requirements.txt

How to run the tests

$ source venv/bin/activate 
$ pip3 install -r test-requirements.txt
$ pytest tests/

Usage

Using the example of the challenge (give it a few seconds, the lib I've used for navigating through github has to clone the repo to a temp directory):

$ python3 cli.py --repo 'https://github.com/mitre/cti' --folder 'enterprise-attack/attack-pattern' --props "id" "objects[0].name" "objects[0].kill_chain_phases"

alternative usage

we could build a python package instead, by doing

$ rm -rf dist build */*.egg-info *.egg-info # optional unnecessary the first time
$ python3 setup_package.py clean bdist_wheel
$ pip3 install dist/apptim_cli-0.0.1-py2.py3-none-any.whl

After that, it can be used in other projects as follows:

from challenge.challenge import GitExtractor
ge = GitExtractor(repo='https://github.com/mitre/cti',
                  folder='enterprise-attack/attack-pattern',
                  props=["id", "objects[0].name", "objects[0].kill_chain_phases"])
for file, result in ge.files():
    print(f"For the file {file} extracted:")
    print(result)

Question 5 "how would you change it to make it less error prone"

The basic issue is that the list of properties doesn't give much information about the chain of properties, as stated on the challenge, it is just a string, so we need to interpret it (spliting it, finding hints to see if we need to access an array and so on) alternatively we could use a markup language (json, xml, yaml) to define a schema:

this

["id", "objects[0].name", "objects[0].kill_chain_phases"]

becomes this:

{
  "properties": {
    "id": {
      "type": "string"
    },
    "objects": {
      "type": "array",
      "items": {
        "0": [
          {
            "properties": {
              "kill_chain_phases": {
                "type": "array",
                "index": 0
              },
              "name": {
                "type": "string",
                "index": 0
              }
            }
          }
        ]
      }
    }
  }
}

Pros: all the info and the data types are set on the json, no need to infer anything.

Cons: its obviously more verbose, and sadly, we don't know from where does the property list comes from, so maybe its something out of our control.