Hacky scripts

As a CS student, I see a lot of people around me doing courses onlineto learn to code. Don’t get me wrong – it probably works for some.Everyone learns differently. But that’s only going to get you so far.Great you know the syntax, you can solve some competitive programmingproblems, but that’s not quite enough, is it? The actual learning comesfrom applying it in solving actual problems – not made up ones.(inb4 some seething CP bro comes at me)

Now, what’s an actual problem? Some might define it as real worldproblems that people out there face, and solving it probably requiresbuilding a product. This is what you see in hackathons, generally.

If you ask me, however, I like to define it as problems that you yourselfface. This could be anything. Heck, it might not even be a “problem”. Itcould just be an itch that you want to scratch. And this is where hacky scripts come in. Unclear? Let me illustrate with a fewexamples.

Now playing status in my bar

If you weren’t aware already – I rice my desktop. A lot. And a part ofthis cohesive experience I try to create involves a status bar up at thetop of my screen, showing the time, date, volume and battery statuses etc.

So here’s the “problem”. I wanted to have my currently playing song(Spotify), show up on my bar. How did I approach this? A few ideaspopped up in my head:

• Send playerctl’s STDOUT into my bar
• Write a Python script to query Spotify’s API
• Write a Python/shell script to query Last.fm’s API

The first approach bombed instantly. playerctl didn’t recognize mySpotify client and whined about some dbus issues to top it off.I spent a while in that rabbit hole but eventually gave up.

My next avenue was the Spotify Web API. One look at the docs andI realize that I’ll have to make more than one request to fetch theartist and track details. Nope, I need this to work fast.

Last resort – Last.fm’s API. Spolier alert, this worked. Also, arguablythe best choice, since it shows the track status regardless of wherethe music is being played. Here’s the script in its entirety:

#!/usr/bin/env bash
# now playing
# requires the last.fm API key

source ~/.lastfm # `export API_KEY="<key>"`
fg="$(xres color15)"
light="$(xres color8)"

USER="icyphox"
URL="http://ws.audioscrobbler.com/2.0/?method=user.getrecenttracks"
URL+="&user=$USER&api_key=$API_KEY&format=json&limit=1&nowplaying=true"
NOTPLAYING=" " # I like to have it show nothing
RES=$(curl -s $URL)
NOWPLAYING=$(jq '.recenttracks.track[0]."@attr".nowplaying' <<< "$RES" | tr -d '"')

if [["$NOWPLAYING" = "true"]]
then
    TRACK=$(jq '.recenttracks.track[0].name' <<< "$RES" | tr -d '"')
    ARTIST=$(jq '.recenttracks.track[0].artist."#text"' <<< "$RES" | tr -d '"')
    echo -ne "%{F$light}$TRACK %{F$fg}by $ARTIST"
else
    echo -ne "$NOTPLAYING"
fi

The source command is used to fetch the API key which I store at~/.lastfm. The fg and light variables can be ignored, they’re onlyfor coloring output on my bar. The rest is fairly trivial and justinvolves JSON parsing with jq.That’s it! It’s so small, but I learnt a ton. For those curious, here’swhat it looks like running:

Update latest post on the index page

This pertains to this very blog that you’re reading. I wanted a quickway to update the “latest post” section in the home page and theblog listing, with a link to the latest post. This would requireediting the Markdown sourceof both pages.

This was a veryinteresting challenge to me, primarily because it requires in-placeediting of the file, not just appending. Sure, I could’ve come up withsome sed one-liner, but that didn’t seem very fun. Also I hateregexes. Did a lot of research (read: Googling) on in-place editing offiles in Python, sorting lists of files by modification time etc. andthis is what I ended up on, ultimately:

#!/usr/bin/env python3

from markdown2 import markdown_path
import os
import fileinput
import sys

# change our cwd
os.chdir("bin")

blog = "../pages/blog/"

# get the most recently created file
def getrecent(path):
    files = [path + f for f in os.listdir(blog) if f not in ["_index.md", "feed.xml"]]
    files.sort(key=os.path.getmtime, reverse=True)
    return files[0]

# adding an entry to the markdown table
def update_index(s):
    path = "../pages/_index.md"
    with open(path, "r") as f:
        md = f.readlines()
    ruler = md.index("| --- | --: |\n")
    md[ruler + 1] = s + "\n"

    with open(path, "w") as f:
        f.writelines(md)

# editing the md source in-place
def update_blog(s):
    path = "../pages/blog/_index.md"
    s = s + "\n"
    for l in fileinput.FileInput(path, inplace=1):
        if "--:" in l:
            l = l.replace(l, l + s)
        print(l, end=""),

# fetch title and date
meta = markdown_path(getrecent(blog), extras=["metadata"]).metadata
fname = os.path.basename(os.path.splitext(getrecent(blog))[0])
url = "/blog/" + fname
line = f"| [{meta['title']}]({url}) | `{meta['date']}` |"

update_index(line)
update_blog(line)

I’m going to skip explaining this one out, but in essence, it’s onemassive hack. And in the end, that’s my point exactly. It’s veryhacky, but the sheer amount I learnt by writing this ~50line script can’t be taught anywhere.

This was partially howvite was born. It was originallyintended to be a script to build my site, but grew into a full-blownPython package. I could’ve just used an off-the-shelf static site generatorgiven that there are so many of them, butI chose to write one myself.

And that just about sums up what I wanted to say. The best and most funway to learn to code – write hacky scripts. You heard it here.