Fix highlighting in hover tooltips and start testing the hover feature

[krassowski]

References

A step towards #297

Code changes

• Fixed missing new lines in markdown generation which were causing the code in the tooltips to be rendered without syntax highlighting
• Added setting for hover tooltip debouncer.
• Removed unneeded margin/padding on the bottom and right side of the tooltip
• Added className option to the custom tooltip subclass
• Added tests for hover in notebooks; they check that the syntax highlighting is correctly applied
• Fix the underline
• Improve multi-element hover display (deprecated API)

User-facing changes

before	after

Final:

Backwards-incompatible changes

None

Chores

• linted
• tested
• documented
• changelog entry

[krassowski]

Actually, we used to get these nice underlines indicating whether hover is available or not. This seems to be broken in 2.0.

• css is in place
• hover marker is getting set... theoretically...

[krassowski]

Underlines are now fixed; the only remaining problem is with tests. Everything works well locally but the tests are slow and flaky beyond reason.

TODO:

• try action chains
• if action chains fail, fallback to testing delay-based trigger only (which needs to be implemented first!)
• update changelog (either 2.0.7 or 2.1)
• record new gif, ideally showing when ctrl is being preseed
• do not trigger if not over a token (but after it) - could accelerate the overall responsiveness in large documents

[krassowski]

@bollwyvl, how would you approach the action chains issue? There is https://github.com/hmalphettes/robotframework-selenium2library-extensions but it was not updated in ages nor released. We could try to extend SeleniumLibrary https://github.com/robotframework/SeleniumLibrary/blob/master/docs/extending/extending.rst (lot's of reading it seems, no time at the moment..)

[bollwyvl]

Sorry, didn't have a chance to dig into this yet, nor have I been able to find my old code. I'd probably make a quick and dirty library that used raw selenium stuff, e.g. hover_with_modifier(selector, modifier=None) and hide the action chain stuff in there. We should not take that library you mention as a dependency, it's caused me issues in the past. Speaking if freaky dependencies: I have also been looking into: https://github.com/MarketSquare/robotframework-browser ...which uses the nodejs-based playwright, which theoretically supports more browsers, though with less control, with more rando binary blobs. Anyhow, its "Hover" keyword already accepts modifiers: *https://marketsquare.github.io/robotframework-browser/Browser.html#Hover <https://marketsquare.github.io/robotframework-browser/Browser.html#Hover>* I'll be trying it out on some greenfield stuff, and if it's useful, looking into packaging it, yadda yadda.