Implement MultilineRawStringIndentation

[BraisGabin]

This implements one of the proposed rules in #4705

To be more precise it implements this one:

When using trimIndent() or trimMargin() check that the left-most indentation is indented x characters from the enclosing element

This rule was a complex one. Format is hard >_<.

[schalkms]

@BraisGabin did you mix up the changes required for this new rule with another branch? In particular, this PR seems to include too much.

[BraisGabin]

Yes, I had one extra commit (the squash strategy doesn't work too good with chained PRs).

Probably this PR could be reviewed commit by commit. There are two prior refactors that I needed: One to be able to tests this PR properly and other to reduce the duplicated code.

I can split this PR if you feel it will make it easier to review.

[marschwar]

I will finish tomorrow. Just a few comments now ;)

[marschwar]

Do we actually want to report one violation for each line that is incorrectly indented? Consider a code block like in detekt tests. If you get the indentation wrong, it is most likely wrong for entire block. Would it not be enough to mark the multi line string as a violation as a whole? That most likely would make the rule much easier to implement.

Specifically I do not like the fact that

val a = """
Hello world!
How are you?
""".trimIndent()

produces 2 violations, while

val a = """
        Hello world!
        How are you?
""".trimIndent()

produces one

[BraisGabin]

Yes... this was a difficult decission. I was thinking about a long raw string with only one/few lines with the wrong identation. In that case I wanted to point out exactly which line was the wrong one. The problem is that if all of them are wrong I end up raising all of them

The second case was easy because if it happens I know that all the lines are wrong so I just raise the error once.

---

I feel that this is a tradeoff:

• If I raise only one the cases where all of them are wrong will be easier,
• But if only some of them are wrong this approuch would be better.

🤷 I'm fine with both of them to be honest.

[marschwar]

I see. I think the code could have been easier to write and to maintain if it were just one violation per multi line string, but I see your point.

[BraisGabin]

I'm merging it like this. We can always revisit this once we get the users feedback.

[BraisGabin]

@marschwar Thank you so much for the this great review :). I fix some of your comments and answer others.

I'm not against the changes that you point our in the comments were I answered I just wanted to share the why behind that code/decission. If you still think that it would be beneficial to change that part of the code I will change it :)

[codecov]

Codecov Report

Merging #5058 (e4a2bf5) into main (e4a2bf5) will not change coverage.
The diff coverage is n/a.

❗ Current head e4a2bf5 differs from pull request most recent head ab9ab07. Consider uploading reports for the commit ab9ab07 to get more accurate results

@@            Coverage Diff            @@
##               main    #5058   +/-   ##
=========================================
  Coverage     84.91%   84.91%           
  Complexity     3614     3614           
=========================================
  Files           502      502           
  Lines         11887    11887           
  Branches       2237     2237           
=========================================
  Hits          10094    10094           
  Misses          690      690           
  Partials       1103     1103           

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update e4a2bf5...ab9ab07. Read the comment docs.

[marschwar]

I'm not against the changes that you point our in the comments were I answered I just wanted to share the why behind that code/decission. If you still think that it would be beneficial to change that part of the code I will change it :)

Same here. I do not feel strongly enough about them but wanted to discuss them with you.