Making mlseo pip installable
I'm using mlseo as a pip installable module to get the process of making packages into my fingers. I'm creating a package called mlseo which will include my latest techniques for pulling data from Google Analytics and Search Console. I'm setting up the settings.ini file and extracting the packages, and I'm following the nbdev instructions to install and create a Github personal access token. Come follow my journey and learn how to set up your own mlseo package!
Creating mlseo: My Journey to Setting Up a Pip Installable Module
By Michael Levin
Wednesday, February 16, 2022
Well, so another day. I’m going to use mlseo as the pip installable module name that will take the place of pipulate in my heart and mind for awhile, because I don’t want to go breaking things on people.
Do it all the nbdev way. I have to get the making of packages into my fingers. Think! Your publishing and sharing of the fruits of your labor is no harder than git committing and pushing this to the mikelevinseo repo in Github. Thanks to Jekyll’s magic publishing system, this pushes out. And, okay, think!
1, 2, 3… 1? If doing things the nbdev way, follow the nbdev instructions. Recycle the existing mlseo private repo that I have. I can keep files in location so long as there’s not naming convention collisions. I can bill this as doing a revision into nbdev. Documentation!
Step #1: Starting point? Oh yeah, it’s https://nbdev.fast.ai/
Okay, so visualize the result. I want to have a package called mlseo. It should have all my latest techniques, especially focused around pulling data out of Google Analytics, Search Console, performing “SERPs” and various other data pulling, transforming and saving options. The idea is for it to become a sort of Mike’s Best-of Python for SEO tricks.
Okay, so make sure nbdev is installed and at it’s latest version. That’s important because I’m experiencing things breaking and being fixed in the nbdev_build_lib process among other things. Break and fix bleeding edge with nbdev, for sure.
It’s also worth noting that all the package creation and extraction work is happening on the WSL2-side of my Windows laptop. So that (Linux terminal) is where I am doing:
pip install -U nbdev
Okay, try:
nbdev_new
Ugh, it’s a private repo. I had to create a Github personal access token. Okay, yes from terminal I had to type:
export GITHUB_TOKEN=[Github provided token]
nbdev_new
Executing: git config --local include.path ../.gitconfig
Success: hooks are installed and repo's .gitconfig is now trusted
Wow, I didn’t think it would be that easy. Okay, this is Jeremy Howard we’re talking about here. Get ready to have a meme infect your mind. Be ready and open to it because his preaching most suits this choir. I’m ready to jump on the nbdev bandwagon, then soon fast.ai thereafter. But first…
Look at the mlseo repo. Do ls -la command. Do git status command. Figure out what nbdev_new added to the folder. Oh yeah, settings.ini is the big one. That is what needs to be setup correctly before any sort of releases, and perhaps even before any sort of build. There’s now a folder also named mlseo which is where the packages will be “extracted” to, but I already have an mlseo.py file, so this will be a little awkward for awhile. Good news is that unlike Pipulate which may be deployed out there, mlseo’s been a private repo, so I won’t piss anyone off. I also won’t be too quickly sharing, even though yes I am documenting it here where it’s public. Perhaps I’ll throw it public in time, but for now… next step?
Oh yes! This is where things start moving over to Jupyter! I’m starting to appreciate the selection of the name Jupiter and changing of the i to a y. It’s omage to Python. I should really talk about that in a video. Not only that, but it’s subversive also in that it’s going to be working more and more like a large body with lots of mass gobbling up everything around it.
That was actually a hurdle overcome, nbdev_new’ing into an already existing private repo. Check!