convey 1.0

CSV swiss knife brought by CSIRT.cz. Convenable way to process large files that might freeze your spreadsheet processor.

Convey

A tool for information conveying – CSV swiss knife brought by CSIRT.cz. Convenable way to process large files that might freeze your spreadsheet processor.

It takes any CSV (any delimiter, header or whatever) and perform one or more actions:

1. Pick or delete columns (if only some columns are needed)
2. Add a column (computes one field from another – see below)
3. Unique filter (no value duplicates)
4. Value filter (only rows with a specific values are preserved)
5. Split by a column (produce separate files instead of single file; these can then be sent by generic SMTP or through OTRS)
6. Change CSV dialect (change delimiter or quoting character)

Installation and first run

Launch as a package:

# (optional) setup virtual environment
python3 -m venv venv
. venv/bin/activate
(venv) $ ... # continue below

# download from GitHub
pip3 install git+https://github.com/CZ-NIC/convey.git  # without root use may want to use --user

# launch
python3 -m convey [filename]  # program start

Parameter [filename] is the path to CSV source file. If not present, script asks for it.

OR launch from a directory

# download from GitHub
git clone [e-mail]:CZ-NIC/convey.git
cd convey
pip3 install -r requirements.txt  --user

# launch
./convey.py [filename]

Dependencies and troubleshooting

• You'll be asked to install dialog library at the first run if not already present in the system.
• If something is missing on your system, maybe you may find help in this command: sudo apt install python3-pip git python3-tk && pip3 install setuptools

Customisation

• A file config.ini is automatically created in user config folder. This file may be edited for further customisation.
• If config.ini is present at working directory, that one is used over the one in the user config folder.
• Launch convey with --help flag to see further options.

Computable fields

• abusemail – got abuse e-mail contact from whois
• ans – got from whois
• base64 – encode/decode
• country – country code from whois
• csirt-contact – e-mail addres corresponding with country code, taken from your personal contacts_foreign CSV in the format country,abusemail. Path to this file has to be specified in config.ini » contacts_foreign
• custom – you specify method in a custom .py file that receives the field and generates the value for you, see below
• hostname – domain from url
• incident-contact – if the IP comes from local country (specified in config.ini » local_country) the field gets abusemail, otherwise we get country. When splitting by this field, convey is subsequently able to send the splitted files to local abuse and foreign csirt contacts
• ip – translated from url
• netname – got from whois
• prefix – got from whois

Custom field example

If you wish to compute a custom field, you'll be dialogued for a path of a python file and desired method that should be used. The contents of the file can be as simple as this:

def any_method(value):
    # do something
    return "modified :)"

You may as well hard code custom fields in the config.ini by providing paths to the entrypoint Python files delimited by a comma: custom_fields_modules = /tmp/myfile.py, /tmp/anotherfile.py. All the public methods in the defined files will become custom fields!

Handsome feature if you're willing to use the Shodan API as our partner or to do anything else.

Usecase

We are using the tool to automate incident handling tasks. The input is any CSV we receive from partners; there is at least one column with IP addresses or URLs. We fetch whois information and produce a set of CSV grouped by country AND/OR abusemail related to IPs. These CSVs are then sent by our OTRS via HTTP from within the tool.

It is able to parse Apache log format files as well. It can bear ##.##.##.##.port format for ip address. If there is invalid lines, they will come to dedicated file to be reprocessed again. It connects to all whois servers I know.

I've tried a file with 3,6* 10^6 lines (300 MB). First 600 000 took around 6 minutes to analyze, the rest around two minutes. It asked 369× arin server and 709× ripe server. There were only 960 of unique IPs in 702 different IP prefixes.