Andrew Huynh
and ramblings.
Email me!
Follow me on Twitter!
Check me out on LinkedIn.

Adventures in Rust: CLI

2018 August 5 - San Francisco | 1031 words


My second project in Rust is a little more practical than my first (which you can read here). This project involves creating a command-line utility that is able to interact with the Bear Writer application, an OS X app that I use for taking notes, writing blogs, and generally keeping my life organized in a single, cloud-synced place.

Bear is a truly wonderful app that lacks a CLI so when I’m in the terminal it’s requires tediously switching back and forth between two screens. Honestly, it’s not that big of a deal, but why not use it as an excuse to build a little utility that I can use.

Plus I thought up a fantastic name for the utility, cub for Command-line Utility for Bear.

You can check out the code for this project here.

What this will cover

Here are the things I cover in this post in case you want to skip around:

Setting up a CLI

After checking out a couple ways of extracting arguments from the command line, I settled on clap mainly due to the ability to configure the parser using a yaml file. This keeps my rust code focused on the implementation logic of the utility and moves most of the configuration boilerplate into a completely separate non-rust file.

For a real life example of how this works, below is the configuration I use to split the arguments and subcommands with their accompanying flags/options:

name: cub
version: "0.1.0"
author: Andrew Huynh 
about: Command-line Utility for Bear.
  - db:
      help: Bear data file to pull data from.
      short: d
      global: true
      takes_value: true
  - ls:
      about: List notes.
        - all:
            short: a
            help: Show *all* notes.
            conflicts_with: limit
  - show:
      about: Show a single note.
        - NOTE:
            help: Note ID
            required: true
            index: 1

This then enables me to initialize the parser in two very simple lines as seen below:

extern crate clap;
use clap::App;

fn main() {
    let yaml = load_yaml!("cli.yml");
    let matches = App::from_yaml(yaml).get_matches(); stuff...

In addition to setting up the subcommands and flags, clap also generates useful documentation that shows up when you run --help.

CUB example

Speaking to the Bear

Great, now that we have a way to take subcommands and arguments, lets look into how we can list out notes and run some basic queries. The Bear app stores the notes and associated metadata in a sqlite database, I’m assuming due to the use of Apple’s CoreData frameworks.

To keep things safe, this first version will only do read only actions. With a little poking around I was able to determine that the local storage for note data is under:

$HOME/Library/Containers/net.shinyfrog.bear/Data/Documents/Application Data

Reverse Engineering the Data Format

Luckily, if you know your way around sqlite, it’s trivial to look through how the application stores its data. Let's start with determining how notes and the associated tags are stored. We can list out all the tables in a database file using the .tables command.

> sqlite3 database.sqlite
sqlite> .tables

Here we can see some potentially useful tables to rifle through. ZSFNOTE looks the most promising and we can use the .schema command to determine the structure of the table.

sqlite> .schema ZSFNOTE

NOTE: I left out some columns to make the schema more readable

Perfect! This has the entire note text data as well as some useful flags for that particular note that we can use to apply filters to our CLI. Looking at the schema of ZSFNOTETAG and Z_5TAGS we also have the connection between notes and tags.

Running Queries

To connect to the database, I’m using the rusqlite framework. Setting up rusqlite is very straightforward. Here’s a small example where I connect to the database and run a simple select query on the notes table just to really hammer home how simple it is.

use rusqlite::{ Connection };
use libcub::note::{ Note };

fn connect_to_db(datafile: &str) -> Connection {
    return Connection::open(datafile).unwrap();

fn main() {
    let conn = connect_to_db();
    let mut stmt = conn.prepare("SELECT * FROM ZSFNOTE").unwrap();
    let notes = stmt.query_map(&[&val], |row| {
    for note in notes {
        println!("{:?}", note);

And thats all there is to it!

To support things like filtering by different flags. I set up a rust enum to represent each filter and modify the query accordingly. For example, when I added the filters for archived and trashed notes, this is how the query string was modified:

for filter in filters {
    match filter {
        NoteStatus::ARCHIVED => {
            filter_sql.push("ZARCHIVED = 1")
        NoteStatus::TRASHED => {
            filter_sql.push("ZTRASHED = 1")
        NoteStatus::NORMAL => {
            filter_sql.push("(ZARCHIVED = 0 AND ZTRASHED = 0)")

Each filter is converted into its equivalent SQL syntax and then added to a vector of filters. This vector is then later on joined to the query string using a string format.

format!("{} WHERE {}", query, filter_sql.join(" OR "));

Learnings and Up Next

This was a short weekend(iso) project that was a lot of fun to work on. Here are some random learnings that I ran into while working on this: