How does the parser deal with multiple countries?

[corneliusroemer]

Ever since the CWA supports multiple countries, I'm wondering how you deal with that fact?

I'm very curious how the setup is:

• Are there separate servers for different countries? Or are they all bunched together?
• Can you tell from the keys which country they came from, or is this impossible?

Obviously this has implications for the interpretation of key numbers. I somehow assume that the dashboards have been able to filter to Germany etc. but it'd be reassuring if you could confirm how this works. Thanks!

[mh-]

The parser doesn't download the DK file, you have to download it yourself and call the parser with the file name.

Now you could download yesterday's DK key set from https://svc90.main.px.t-online.de/version/v1/diagnosis-keys/country/DE/date/2020-12-22 or from
https://svc90.main.px.t-online.de/version/v1/diagnosis-keys/country/EUR/date/2020-12-22

WAC currently uses the latter.

The EUR set is slightly larger, but not significantly larger, than the DE set. So I don't think that you only find German DKs in the DE set.

Would be interesting to learn more about the difference between these two sets.

[corneliusroemer]

Ah great, thanks. Yeah, sorry for being sloppy regarding the parser's role.

So there are different countries, and EUR presumably includes all participating countries, whereas DE is just Germany.

The dashboards presumably use just DE but that needs to be checked one by one. What was the default? DE, and EUR was introduced later?

[mh-]

See https://covidapps.slack.com/archives/C0194ML0MLN/p1608741919207700
The DE file has 51684 keys, the EUR file has 52589 keys.
So I don't think that EUR includes all participating countries, whereas DE is just Germany. I think the difference is too small. But I could be wrong.

[tkerby]

The UK set now includes a few countries and is available from the endpoints in the code here: https://github.com/russss/nhs-covid19-app-data Note that it doesn’t save the zip files without some minor changes. I’d be interested to compare files to see who has the same keys and understand which countries are sharing data now.

…

On 23 Dec 2020, at 17:10, Michael Huebler ***@***.***> wrote:  See https://covidapps.slack.com/archives/C0194ML0MLN/p1608741919207700 The DE file has 51684 keys, the EUR file has 52589 keys. So I don't think that EUR includes all participating countries, whereas DE is just Germany. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.

[mh-]

@tkerby The German distribution at https://svc90.main.px.t-online.de/version/v1/diagnosis-keys/country/EUR/date/2020-12-22
seems to contain keys from Germany, Denmark, Ireland, Italy, Croatia, Latvia, Netherlands, Poland, Spain, Cyprus
according to the message displayed in the CWA app.

[stefanb]

Hi!

Recently Slovenia joined the EUR exchange and foreign keys caused a significant bump in our charts on 2021-02-10:

...hiding the interesting local details ☹️

On that day the number of keys grew significantly more in the EUR export (17232), compared to the SI export (4456), meaning that only a subset of EUR keys is included in the SI export.

It would be very interesting to allow filtering out those keys coming from EUR exchange, but am not sure this is possible. Is it?

I presume the same problem applies for other smaller countries (compared to Germany, at least).

[mh-]

I don't think there's a way to filter the keys by country of origin, unless your country's server distributes different packages for that purpose. The data does not contain a "country of origin" field.

[stefanb]

National health service distributes packages for SI and EUR region. SI is subset of EUR, they both include keys from Slovenia, so there's no way to just substract EUR keys from the ones in SI 😞
I also asked them if they could provide the local ones before adding the EUR ones and that seems impossible.

Maybe we could do something by substracting all other countries' packages...