jxeeno / aust-govt-covid19-stats Goto Github PK

Hi jxeeno,

Seems data is missing for 6-Oct-2021.
Is it possible to fill in this gap?

It turns out that that the QLik Sense (the product which powers the graphs on the page) has a remarkably well-documented API which provides data a lot more accurately than the current scraping process using Puppeteer.

Advantages of this approach

• Decimals will no longer be rounded. For example, the positive test percentage for Australia yesterday was 0.0014373561076329474, but the scraped data is rounded to 0.1
• Scraping is much faster now - my implementation takes ~12 seconds compared to the current ~1 minute.
• The API provides a modification datetime as ISO8601, so the regex that looks for a date in the footnote can be now avoided:

Unfortunately we still need to hardcode the ID of each graph as graph titles aren't included for some reason, but these don't change often, and will still be stored in the raw data.

I'm happy to open a PR for this sometime in the next week if you'd like me to. You can see my basic implementation here.

Hi @jxeeno,

Just letting you know the scraper seems to be down at the moment. From a quick look locally it seems that the QLik server is unexpectedly saying that a few graphs in the page HTML don't exist on the server. Specifically, the right-hand row of the KPI grid is returning a null response from the API.

The current code currently doesn't check for an undefined graph handle, which then causes it to send an invalid API request, bringing down the script once it receives an error response. I'll open a PR next year... 😆