feat(forge fmt): Adds tab support as indent char in fmt

[mathewmeconry]

Motivation

I want to start to contribute and the issue #10754 was marked as a good first issue.

Solution

Adds support for different styles in fmt config. This allows the user to configure whether tabs or spaces should be used (I don't want to start the age-old discussion here).
tab_width will be ignored by fmt if the style is set to tab.
If no style is given in foundry.toml space will be used as default.

Coverting block comments from space to tab results in issues where the previous spaces will not be ignored. I couldn't figure out a way to solve this properly but I believe it is a non-issue because most teams won't switch from spaces to tabs that often and fixing some comments is fine from my point of view.

Additional to the tab support, a bug in the parsing in InlineConfig for forgefmt: disable-line handling was found and fixed. The previous code doesn't handle CRLF line endings properly and returns the wrong loc for the disabled line (offset by 1 byte).

PR Checklist

• Added Tests
• Added Documentation
• Breaking changes

[grandizzy]

thank you! looks like it breaks tests for win

[mathewmeconry]

thank you! looks like it breaks tests for win

Thanks for running the tests. Fixing this will take a while for me because there seems to be a bug with newlines.
You can reproduce the same issue by formatting the original.sol in Repros with tab_width set to anything else than 4 and the file formatted with CRLF line endings.

With CRLF it still adds indents (tab_width set to 8) :

        function test4() public {
                // forgefmt: disable-next-line
                  assembly{
                        sstore(1, 1)
                        sstore(2, 2)
                        sstore(3,    3)// forgefmt: disable-line
                        sstore(4, 4)
        }// forgefmt: disable-line
                if (condition) execute(); // comment7
        }

While with LF line endings it doesn't (tab_width set to 8):

        function test4() public {
                // forgefmt: disable-next-line
                  assembly{
                        sstore(1, 1)
                        sstore(2, 2)
            sstore(3,    3)// forgefmt: disable-line
                        sstore(4, 4)
        }// forgefmt: disable-line
                if (condition) execute(); // comment7
        }

I assume there is a bug in handling //forgefmt: disable-line when the line ending is set to CRLF.
Do you have any hints where this bug could hide?

[mathewmeconry]

@grandizzy I think I found and fixed the issue with the last commit. Tests succeed on my fork.
Please run the tests on the pull request to see if this resolves the issue.

[mathewmeconry]

@grandizzy I think I fixed the bug, but other tests are now failing in parts that I haven't touched. Could this be a race condition or timeouts that happen in tests?

[grandizzy]

@grandizzy I think I fixed the bug, but other tests are now failing in parts that I haven't touched. Could this be a race condition or timeouts that happen in tests?

hey @mathewmeconry yeah, we're debugging some unexpected failures in CI, will get back to the pr once figured out. Thank you