acdh-oeaw / rdfproxy

GNU General Public License v3.0
2 stars 0 forks source link

fix(tests): Target DBPedia instead of Wikidata for remote tests #69

Closed lu-pl closed 2 months ago

lu-pl commented 2 months ago

As of recently, Wikidata apparently rejects the default SPARQLWrapper user agent, so remote tests targeting Wikidata fail unless the agent parameter is explicitly set. See https://github.com/RDFLib/sparqlwrapper/issues/237.

All affected remote tests use toy VALUES queries, so targeting DBPedia (or any other triplestore) resolves the issue for now.

sennierer commented 2 months ago

We are running daily and weekly updates of local typesense indexes from wikidata using SPARQLWrapper in Prefect. The queries returned fine this morning (and the morning before), We got 403s on 2024-08-21 but not before or after. So I guess our IPs got on a black list (maybe due to some of your tests?) around that day, but now everything is back to normal.

lu-pl commented 2 months ago

We are running daily and weekly updates of local typesense indexes from wikidata using SPARQLWrapper in Prefect. The queries returned fine this morning (and the morning before), We got 403s on 2024-08-21 but not before or after. So I guess our IPs got on a black list (maybe due to some of your tests?) around that day, but now everything is back to normal.

Interesting, thanks for the info!

I think the SPARQLWrapper default user agent got blocked, SPARQLWrapper didn't produce 403s with an explicitly set fake agent.

Anyways, I should mock requests for tests, see #72.