search

covh / cov19de

401 districts of Germany, covid19 prevalence and incidence visualized and sorted
https://covh.github.io/cov19de
GNU General Public License v3.0
2 stars 2 forks source link

Project Ending? #5

Open donnergottt opened 2 years ago

donnergottt commented 2 years ago

The data hasn't been updated since 04.09.2021. Is there something broken or won't it be discontinued since change of data basis? (incidence vs. hospitalised people)

covh commented 2 years ago

Yes. Broken. Needed human intervention.

Let me explain.

The system is running amazingly on "autopilot" actually, and for weeks or months without any human intervention. It is updating itself once daily on the basis of the newest RiskLayer CSV. Then the whole 'cov19de' site is regenerated, and afterwards you can see (usually yesterday's) newest data plotted and tabled.

However, there were circa 10 cases now, in which Risklayer did something unexpected to their CSV. e.g. changed the format of their file, or put the letter O instead of a zero, etc. It's quite interesting in how many creative ways this can go wrong. Have a look at this code, to get an idea: The function repairData() now contains and calls most of the fixes. I tried to generalize as much as possible, so that if a similar mistake appears a few months later, my script would fix it by itself, without getting stuck this time. However, RiskLayer kept on finding new innovative ways to break my CSV data importer, lol.

Then I even built a CSV testing tool: cov19de.herokuapp.com (patience, might need 30 seconds to wake up), to quickly see if the is a CSV problem, and perhaps even guess what kind of problem it might be. Use this first when the next problem appears.

Another data source is RiskLayer's googlesheet spreadsheet. Whenever they dropped or renamed columns there, I needed to adapt the importer, see these different versions here. That fix was actually also among the recent 3 problems:

This weekend now, January 2022, I finally went through fixing three different problems that had appeared since September. And 'http://tiny.cc/cov19de' is working fine again. For now. Until the next CSV or googlesheet change by RiskLayer would break it again. Which I might not even notice any time soon, because I have long moved on.

And here comes the bigger news now:

redirect

  1. Not gonna update this project regularly anymore. Forgive me please. BUT:
  2. Please use the superb, improved fork by @jalsti, he has been keeping it updated:

jalsti.github.io/cov19de --> Deutschland.html

So Long, and Thanks for All the Fish