Code indenter for Erlang.

• Uniformize indentation across multiple code bases.
• Maintains user decisions related to formatting.
• Dirt simple (someone else is working on a proper formatter).
• Should clean up trailing whitespaces
• Should be self-contained. Dependency on OTP is ok.

As a side-effect of keeping things stupid-simple, edent has one indentation style. There are currently no plans or desire to make it customizable.

If you see problems, then please create a PR containing a test with what the indentation should look like. Keep in mind that if your change is subjective in nature it will likely be denied. If that angers you, you can always fork and modify to suite your personal predilections.

The only dependency required to run this is OTP; any relatively recent version should be fine.

$ ./edent
usage: edent [-v] <input>...
       edent [-v] -o <output> <input>

The first form will overwrite each input file with their indented versions. This will be commonly used like this:

$ find . -name '*.erl' | xargs ./edent

The second form is mostly useful for testing and debugging where you want to not overwrite the orignal file. It's currently used for the test suite of edent.

Note that pre-commit hook is also available for git and expects edent to exist at the root of your repo.

Finally, the -v option exists for debugging edent and outputs internal debugging information.

edent is a pretty simple escript so nothing too fancy is required to get up and running.

To run tests:

$ ./tests.sh
$ ./tests.sh <test>

The general idea of the tests, is that the files in the tests folder describe the expected behaviour of edent. The test harness will run edent on a file and diff the result against the original. A difference indicates a bug.

Otherwise, any changes should ideally be covered by a test because it's already pretty hard to keep track of how everything should line up so tests are the path to sanity.

• erl_scan eats the \n character after a . character BUT NOT if there's any form of whitespace in between the two characters.

  1> erl_scan:string("a.\nb", 0, [return]).
  {ok,[{atom,0,a},{dot,0},{atom,1,b}],1}
  
  2> erl_scan:string("a. \nb", 0, [return]).
  {ok,[{atom,0,a},{dot,0},{white_space,0,"\n"},{atom,1,b}],1}

  Because of this behaviour, edent inserts an \n after each . so if you have trailing whitespaces then it's possible that edent will insert extra end-of-lines in your file. This might be fixable upstream, and I'll open an issue at some point.

• There's a few known pieces of syntax that edent can't handle. These are documented in the hell.erl test cases. Hopefully these should be relatively rare in the wild.

• edent is quite slow at the moment and I consider that a bug.