Blocked by publication of opendatasoft-explore-api library.

Rebased

case

Refactor to a cleaner implementation, move sparql mod to services

To avoid match/let ... = ...() in the middle of a HOF chain, I've prepared helper methods to read the inner content of an enum:

Fix SPARQL client, refactor

Works for me, but depends of D2731.

Move more report code to commands::wikidate::report mod

Skip first CSV line

This version is ready and can correctly update Wikidata. If we can solve the 403 issue when connecting to SPARQL endpoint, it's ready to commit. Meanwhile, to test this code, you can run the query and save the result (in XML format) at tmp/wikidata-query-result.xml

Fix CRLF issues

Clean comment and debug code

Don't hardcore wikidata table name twice

Rebased

Merged.

fantoir2db is now fantoir-datasource

Remove test limit

A lot of code to handle exceptions, mainly because there are
several formats used to represent the FANTOIR code.

I've regenerated /usr/share/ca-certificates and /etc/ssl/certs based on a "recent" Debian ca-certificates package against Debian Sid, version 20211016. Works like a charm now.

Container still runs on Jessie, we need to deploy new @DorianWinty work on recent Debian PHP 8 container, with correct certificates.

Disabled Sentry, so we can have a correct log of the stacktrace, and updated description accordingly.

"The notification lifecycle: how a notification is provided and consumed" schema

Extract run_multiple_queries method to the db module. We can't easily move the create_table function, as it needs schema-dependant replacement.

Switch to subcommands pattern (like git), so we can extend it

Creation date can be unknown, add code_fantoir to help cross data with other db, fix last line record parsing

Draft implementation: rFANTDB59128e4bfd41 (needs to add all FANTOIR field).