# Trump vocabulary

**NOTE:** this was written in a few minutes without bothering with clean and robust
code.

This code goes through the tweets of Donald Trump and produces a ranked list of words
used.

The result (not much updated, though) can be found
[here](https://tobast.fr/files/trumprank.txt).

## Methodology

A word is considered to be a contiguous sequence of letters and quotes (`'`)
only. Words that have less than four occurrences are removed (considered
irrelevant — probably some random name).

## Install

Clone this reopsitory with submodules: `git clone --recurse-submodules`

Alternatively, if you already cloned the repo, you can run

```bash
git submodule update --init --depth 1
```

## Get a shell

You can explore the data in a shell by using `count_words.py` as an init script for
your favorite shell, eg.

```bash
ipython -i count_words.py
```

The following will be available to you as variables:

* `tweets`: the list of all tweets ever,
* `occur`: python dictionary of occurrences of words in Trump's tweets
* `ranked`: ranked list of occurrences of words in Trump's tweets

## Generating the list

Simply run

```bash
python ./generate_list.py [OUTPUT_FILE]
```

If you omit `OUTPUT_FILE`, the list will be generated to `trumprank.txt`.