CT election runner

Python library and command line tool to pull Connecticut's election data in real time.

About

Connecticut's election night reporting system has an undocumented API that I reverse engineered and used as a data journalist starting in 2015 to build live results portals (code) for The Connecticut Mirror, where I used to work. Update: These links have died since I left the Mirror, it looks like.

I've always intended to make this code public once I had time to clean it up. I have never had time to really do that, and I doubt I ever will, so I'm just releasing it as-is.

This repo isn't all that clean, it's not well documented, and it lacks test coverage and even error handling. However, I felt that putting this out there in its current state would be better than waiting until I have a lot of time to spend on this, which will probably never happen. And I really just needed to get it off my plate and stop kicking myself about not getting to it every time an election season comes and goes.

CLI tool and library

This package contains a command line tool "ctrunner" and a Python library for downloading data from the election night reporting system and to convert the raw data to more useful spreadsheets.

Install

pip install -e git+https:/jakekara/ct-election-runner.git#egg=ct-election-runner
ctrunner --help

This will output:

usage: ctrunner - Connecticut election data runner [-h] {archive,parse,download} ...

positional arguments:
  {archive,parse,download}
                        commands

optional arguments:
  -h, --help            show this help message and exit

CLI cookbook

Dump all of the data

Probably the most useful command is archive, which will download all of the election data the directory archive. I've added an archived-data directory to this repo so you can just look at the data if that's all you're interested in.

mkdir archive
ctrunner archive --dest archive

Inside the archived-data directory, there's a directory for each election with the latestd data in the raw.json and results.csv file, plus a version directory for each version. Once election data is closed, new versions stop coming in.

Get list of elections

This pipes the ctrunner downloader and the ctrunner parser together to download the data in raw JSON format then convert it to a spreadsheet:

ctrunner download --election-list | ctrunner parse --data-type=election-list

This will output:

ID,Name,DefaultElection
54,11/03/2020 -- Presidential Election,Y
55,08/11/2020 -- Democratic Presidential Preference Primary,N
57,08/11/2020 -- Republican Presidential Preference Primary,N
58,08/11/2020 -- Republican Primary,N
59,08/11/2020 -- Democratic Primary,N
50,01/21/2020 -- Special Election 151st Assembly,N
49,01/14/2020 -- Special Election 48th & 132nd Assembly,N
36,11/05/2019 -- November 2019 Municipal Election,N
42,09/10/2019 -- Democratic Primary,N
43,09/10/2019 -- Republican Primary,N
39,05/07/2019 -- Special Election 130th Assembly District,N
38,05/06/2019 -- May 2019 Municipal Elections,N
37,04/16/2019 -- Special Election 19th Assembly District,N
35,"02/26/2019 -- Special Election 3rd, 5th & 6th Senatorial Districts and 39th & ",N
31,11/06/2018 -- November 2018 State Election,N
33,08/14/2018 -- August 2018 Democratic Primary,N
34,08/14/2018 -- August 2018 Republican Primary,N
18,11/07/2017 -- November 2017 Municipal Election,N
24,04/25/2017 -- Special Election Assembly Districts 7th & 68th,N
1,11/08/2016 -- November Presidential Election,N

Download an election and convert it to CSV

Looking at the list above, we can pick an election ID and download it, then parse it on the fly to a CSV. Let's choose election 1:

ctrunner  download --election 1 | ctrunner parse --data-type=election-data > election-1.csv

Archive an election

Download the latest version of an election into a given archive directory.

While the download and parse subcommands use standard input and standard out, the archive subcommand automates a common use case where you'll want to download to results into a single output folder containing many elections, and keep past versions in a subfolder. Keeping past versions was especially important on election night because I wanted to be prepared to roll back in case a bad version came through the API. It was also useful to go back and use past versions of the data to replay how quickly the results of the election came in. That was particularly important for news coverage when this system came out because it took several days for all towns to put their results in when participation was not mandatory.

ctrunner  archive --dest examples --election 1

Check out the examples folder in this repo to see the data that the archive subcommand generates.

Library

You can learn more about the library functions by reading the cli tool code. Here are the basics.

The downloader is for downloadin raw JSON files from the API. The parser is for converting these JSON files to CSVs. The JSON files contain a lot more data than the CSVs, so you might want to create your own tools based around the raw JSON files.

ctrunner.downloader

• get_election_list - get a list of all elections in the election system
• get_data_for_election - get data for a specific election given an ID corresponding to IDs from the get_election_list return data

There are plenty of other functions in that module, but these will generally suffice.

ctrunner.parser

• convert_list(election_list_data) - convert the results from ctrunner.downloader.get_election_list to a CSV
• convert_election_data(election_data) - convert the results from ctrunner.download.get_data_for_election to a CSV.

To see these in action, check out the CLI tool. The subcommands documented in the help message are in the cli/commands folder, and those are the scripts that make use of the library.