[Side Project] Paris Plouf: The MVP is almost ready

After several coding sessions, I am excited to announce that the MVP (Minimum Viable Product) for Paris Plouf is now operational. As a reminder, the goal is simple: find the most accessible swimming pool in Paris based on your GPS location and real-time opening hours.

How does it work?

Currently, the API receives your GPS coordinates and your departure time. It then returns the top 5 establishments by applying strict filters to guarantee a real swimming session:

• The "45-Minute Contract": The pool must be open long enough to allow for at least 45 minutes of actual swimming.
• Real-Time Calculation: I don’t just look at the closing time. The system subtracts travel time, 30 minutes for the mandatory pool evacuation, and a 10-minute buffer for the locker rooms.
• Anticipation: If a pool is closed at the time of the request but opens by the time you arrive, it is included in the results.

To keep the data accurate, a CRON job runs every night. It queries the City of Paris Open Data to refresh the schedules, as precision is paramount even if they remain relatively stable.

The Data Challenge: Open Data isn't enough

While developing, I noticed that the official City of Paris website sometimes displays real-time attendance and, more importantly, exceptional closures (maintenance, competitions) that don't always appear in the Open Data. Even worse: some pools are completely missing from the public API.

I have therefore implemented a scraping script to enrich my database. My next steps on this front are:

• Retrieving schedules for pools missing from the Open Data.
• Parsing temporary closure alerts to prevent users from showing up to a closed door.

The Dilemma: Proximity vs. Session Quality

A dilemma quickly surfaced during testing: should I prioritize the closest pool or the one that stays open the longest? Finding a universal "perfect ratio" proved impossible. Consequently, I added an optional parameter to the API allowing users to choose between two modes:

• Best Journey: Priority is given to travel speed (on foot or via public transport).
• Best Session: Priority is given to the widest available time slot for swimming, even if it means a slightly longer commute.

Pricing and the "Super-Pool" Bias

Not all Parisian pools are equal when it comes to the wallet. While most follow the standard municipal rate, some (such as Suzanne Berlioux in Châtelet or those on houseboats) are more expensive due to extended hours or premium infrastructure.

The problem? A central, late-opening pool like Châtelet would systematically rank first. To fix this bias, I need to identify "special rate" pools and implement business logic to penalize or filter these establishments.

Infrastructure & Deployment

Regarding hosting, I’ve decided to run the API and the Telegram bot on my personal server. This allows me to keep costs at zero while sticking to my original constraint: building a lightweight, resource-efficient application.

To-Do List for the next phase:

1. Optimize web parsing for non-Open Data pools and closures.

2. Index pricing to identify "Premium" entries.

3. Implement penalty logic based on price.

4. Dockerize the stack and deploy the Telegram bot.

Stay tuned for the next update!