I'm a lazy programmer. I love building stuff, but I hate repeating myself. Anytime I start a new article for this site, I do the following:

• navigate to my site's directory in terminal
• open up VS Code
• create a new file
• set the title
• set the slug
• set the date
• set its published flag to false

By no means is that a lot of work, but when I have an idea for something to write, I want to start writing it immediately. So, like any self-respecting lazy programmer, I put together a process to remove as much friction as possible. And, to be honest, I used this as an example to learn more about Alfred workflows and bash scripting.

So, What's the Goal?

I use Alfred all the time, so I thought this would be the perfect interface for my new process. I wanted to simply open the Alfred prompt and type art <some article title> and have it perform the steps listed above.

The Implementation

Gatsby Post Manager has a posts new command that I could use to create the actual file. The command itself looks like this:

$ gpm posts new path/to/posts "hello world"

This command automatically creates the file and assigns it the title, slug, and date, but I still need to pass it the path to the posts. This path never changes for me, so I shouldn't need to type it every time.

My first thought was to create an Alfred workflow that passes that command to terminal. That would work, but it has two downsides:

• Alfred will open a new terminal window anytime I invoke the command
• I still need to open VS Code to actually write the article, which means I need to pipe the output of gpm to code

The better approach is to write a bash script that invokes gpm and then opens VS Code.

The Bash Script

#!/bin/bash
# Alfred sometimes has trouble finding Node; the below fixes that
export PATH=$PATH:/usr/local/bin
# Require a single argument for the article name
if [ $# -ne 1 ]; then
  echo 1>&2 "Usage: $0 article-name"
  exit 3
fi
# Create the article file
# Uses Gatsby Post Manager (gpm): https://github.com/steven-mercatante/gatsby-post-manager
RESULT=`/usr/local/bin/gpm posts new /Users/steve/sites/stevemerc.com/content/posts "$1"`
# Split $RESULT into an array
# $RESULT normally looks like: "/path/to/posts/post.md successfully created", but we only want the filename
# This is the delimiter that $RESULT will be split on
IFS=' '
# Creates $ARTICLE_ARR array from $RESULT string
read -ra ARTICLE_ARR <<< $RESULT
# Grab just the filename from the success or error message
ARTICLE=${ARTICLE_ARR[0]}
# Open up VS Code on the proper project, then open up the newly created article
/usr/local/bin/code /Users/steve/sites/stevemerc.com && /usr/local/bin/code $ARTICLE

Here's a link to the repo.

I've done my best to document what the script does. If you have any questions, tweet at me.

Now that we've got our script that orchestrates everything, we need to wire it up to Alfred.

Creating the Alfred Workflow

I'm using Alfred v4, so your setup may differ a bit from mine. But the gist is:

• open Alfred's preferences
• navigate to the Workflows section
• click the + button near the bottom and select Templates > Essentials > Keyword to Script

Name your workflow whatever you want and click the Create button. In my case, I called it "Create Article". This is used for organization purposes.

You should then see your newly created workflow and two nodes:

Double click that Keyword node and you'll be prompted to define the command name and title.

Next, double click the Run Script node and you'll be prompted to either write the bash script right there, or you can select External Script (<- do that).

As soon as you enter a path to an existing bash script in the Script File: input, its contents will appear below.

Everything should be set up now, and you can type art hello world in the Alfred prompt to create your new article.

This is what the contents of the newly created hello-world.md file looks like:

---
title: Hello World
path: 'hello-world'
tags: []
published: false
date: '2019-09-24'
---
Add your content here

Debugging

I had some trouble getting my bash script and Alfred to work together. Something to do with paths... who knows. Luckily, Alfred lets us debug our workflows.

Once you click that little bug icon, it opens a console window that'll print out the steps Alfred takes, as well as any output or errors from your bash script. Getting this process to work without the debugger would've definitely taken longer.

Wrapping Up

I don't expect many people to want to solve the same problem that I just solved, but, knowing how to wire up command line tools, bash scripts, and Alfred to orchestrate common and repetitive tasks is a valuable skill.