{"id":802,"date":"2025-10-21T00:03:36","date_gmt":"2025-10-21T07:03:36","guid":{"rendered":"https:\/\/konukoii.com\/blog\/?p=802"},"modified":"2025-10-21T17:06:22","modified_gmt":"2025-10-22T00:06:22","slug":"5-min-tutorial-building-pypi-packages","status":"publish","type":"post","link":"https:\/\/konukoii.com\/blog\/2025\/10\/21\/5-min-tutorial-building-pypi-packages\/","title":{"rendered":"5-Min Tutorial: Building PyPI Packages"},"content":{"rendered":"Reading Time: <\/span> 4<\/span> minutes<\/span><\/span>\n

A few days ago, I was coding casually while re-watching some episodes of Bob\u2019s Burgers. Inspired by the show, and with a vague memory of an old SpongeBob web-game I use to play as a kid, I decided to create a simple burger-flipping game. While the game itself isn\u2019t particularly significant (I may end up sharing it later), what is important is that I needed a lot of images. Since I have Google\u2019s Gemini, I started using it. I would sketch a basic idea on paper and then prompt Gemini to re-draw it in a different style. As you can imagine, this process became quite tedious.<\/p>\n\n\n\n

Instead of continuing with the Gemini-based approach, I decided to build my own tool called Imagenation<\/code>. Initially, it was a command-line tool, but I soon realized that I would likely need to reuse it in other projects. Therefore, I decided to learn how to create my own package and eventually upload it to PyPI. As a Python enthusiast, this was a long-standing curiosity of mine.<\/p>\n\n\n\n

At a high-level, the tool needed to:<\/p>\n\n\n\n

    \n
  1. Generate images from text prompts (Text-to-Image).<\/li>\n\n\n\n
  2. Modify existing images with text prompts (Text+Image-to-Image).<\/li>\n\n\n\n
  3. Runs as a simple CLI tool that can chew through a CSV or JSON file of prompts.<\/li>\n\n\n\n
  4. Works as an importable Python library for other projects.<\/li>\n<\/ol>\n\n\n\n

    Git Repo: https:\/\/github.com\/pmsosa\/Imagenation<\/a><\/strong><\/p>\n\n\n\n

    PyPI: https:\/\/pypi.org\/project\/imagenation<\/a><\/strong><\/p>\n\n\n\n

    Leveraging Google's Imagen<\/h2>\n\n\n\n

    First things first: just talking to the API. Google's generative AI tools<\/a> are incredibly powerful, and getting a single image generated in Python is refreshingly simple.<\/p>\n\n\n\n

    You set up your API key, initialize the model, and essentially just\u2026 ask for what you want.<\/p>\n\n\n\n

    First, go pick up your API key from: https:\/\/aistudio.google.com\/api-keys<\/a><\/p>\n\n\n

    \nimport google.generativeai as genai\n\n# Make sure your GOOGLE_API_KEY is an env variable\ngenai.configure(api_key=os.environ["GOOGLE_API_KEY"])\n\nmodel = genai.GenerativeModel('gemini-pro-vision') # Or your model of choice\n\nresponse = model.generate_content(\n    [\n        "A beautiful sunset over mountains",\n        PIL.Image.open("my-image.png") # Optional input image\n    ]\n)\n\n# ...then you have to process the response, get the bytes, save to a file...\n<\/pre><\/div>\n\n\n
    The first time this works, it feels like magic. But the second<\/em> time you run it\u2014and the third, fourth, fifth, and sixth time, all within a minute\u2014you hit the real boss: Rate Limiting<\/strong>.<\/p>\n\n\n\n
    The free tier for the API (at the time of writing) allows about 5 requests per minute. If you send that sixth request, you get a 429<\/code> error, and your script comes to a crashing halt.<\/p>\n\n\n\n
    I added a --delay<\/code> parameter that defaults to 12 seconds (just enough for 5 requests\/min), but you can set it to 0.2<\/code> if you're on a paid tier (300 requests\/min) or 15.0+<\/code> if you're running a massive batch and want to be extra careful.<\/p>\n\n\n\n
    From Script to Package<\/h2>\n\n\n\n
    The project could have ended there. I had a imagenation.py<\/code> script that worked for me. But I've been doing this long enough to know that \"future me\" is going to want to use this again, and he's not going to want to hunt for a stray .py<\/code> file. He's going to want to pip install<\/code> it.<\/p>\n\n\n\n
    This was the part I was most excited about: turning this pile of code into a real, distributable PyPI package.<\/p>\n\n\n\n
    It turns out, with modern Python tooling, this is way more accessible than I thought. The process was basically:<\/p>\n\n\n\n
      \n
    1. Restructure:<\/strong> Stop putting everything in one file. I split the logic into:\n