Fix insertion of access tokens, when swapping renderer

[birkskyum]

Going from e.g. MapTiler to OpenLayers and back will lose the maptlier key.

This code finds the urls in the style that has "{key}" and insert the correct API keys

Fixes the error reported here, cc @nyurik

• Fixes Switching to OpenLayers and back to MapLibre hides all styles #874 (comment)

Related to:

• MapTiler Access Token does not get added to URLs #869

Launch Checklist

• Briefly describe the changes in this PR.
• Link to related issues.
• Include before/after visuals or gifs if this PR includes visual changes.
• Write tests for all new functionality.
• Add an entry to CHANGELOG.md under the ## main section.

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 64.37%. Comparing base (af01346) to head (42ba9dc).
Report is 57 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1021      +/-   ##
==========================================
+ Coverage   59.84%   64.37%   +4.53%     
==========================================
  Files         104      104              
  Lines        3011     5825    +2814     
  Branches      680     1724    +1044     
==========================================
+ Hits         1802     3750    +1948     
- Misses       1209     2073     +864     
- Partials        0        2       +2     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[HarelM]

Is there a test we can add to make sure this behavior is kept in the future?

[birkskyum]

I'm not sure how to test this through dom/cypress. It's quite different from the other tests, as it's about network calls and visual output on canvas.

Cypress clearly is capable of both (network and snapshots), I just have to study it.

• https://docs.cypress.io/app/guides/network-requests
• https://docs.cypress.io/app/guides/screenshots-and-videos

[HarelM]

You can use Cypress to intercept network calls I believe.
Also you can check what is stored in the lical storage, or is the local storage content "correct"?

[birkskyum]

@HarelM , figured out the test now. It checks the requests made after the render change has the access token instead of just {key}

[HarelM]

The test needs some cleaning up, otherwise this looks good, thanks!

[HarelM]

Firefox and cypress do not play together well, this is not the fire time the tests are passing with chrome but not with firefox, maybe we should drop support for firefox and hope it works well, IDK...

[birkskyum]

@HarelM , I've removed firefox in this pr too.

[HarelM]

Great! Thanks!

[birkskyum]

Not sure why it's still showing an "E2E tests using firefox"

[birkskyum]

@HarelM , is this PR approved?

[HarelM]

It was required in the setting of the branch protection, I've removed it and fixed it now.

[HarelM]

@birkskyum can you take a look at the added cypress test here?
I think it causes some CI instability and prevents automatic dependencies updates...