Option delimiter

[phlptp]

This PR addresses an issue I came across working on the Validators. The problem was with vectors and Validators if you wanted a vector with a range of say [0, 20] There are options for Validating individual elements but not after the conversion if you also wanted to use a custom delimiter. Furthermore if you wanted to require 3 elements, that had to be 3 elements before the delimiter was applied.

What is comes down to is the delimiter was being applied too late, and had to be applied in a bunch of places if you wanted to use it.

The solution was to move it as part of the Option class. Any option can now have a delimiter, and the delimiter is applied when the results are loaded into an option. This can allow options that require multiple arguments to accept them as a comma separated list or something along those lines and have that meet the requirement.

One twist on the add_complex is you could add ->delimiter('+') and it would accept argument such as "3+2j" as a single command line argument since the + on +2j is redundant. a negative number would still need to be written like 4+-2j for that to work.

My use case I wanted to able to specify a bunch of flags like --flags=flag1,flag2,flag3,.. and have them checked against a hash table. That would only work if the delimiter was applied before the Validator.

[phlptp]

@henryiii if you are ok with a few breaking changes regarding the delimiter argument with some of the options, I think you could eliminate a couple of the overloads for options that have the delimiter argument in them. Since there is now a delimiter(char) function on the Option.

[codecov]

Codecov Report

Merging #240 into master will not change coverage.
The diff coverage is 100%.

@@          Coverage Diff          @@
##           master   #240   +/-   ##
=====================================
  Coverage     100%   100%           
=====================================
  Files          12     12           
  Lines        2212   2212           
=====================================
  Hits         2212   2212

Impacted Files	Coverage Δ
include/CLI/Option.hpp	100% <100%> (ø)	⬆️
include/CLI/App.hpp	100% <100%> (ø)	⬆️

---

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 7254ea7...a3cf2d8. Read the comment docs.

[henryiii]

@henryiii if you are ok with a few breaking changes regarding the delimiter argument with some of the options, I think you could eliminate a couple of the overloads for options that have the delimiter argument in them. Since there is now a delimiter(char) function on the Option.

I think that's better. And API that is not in a released version of CLI11 is fine to change! I strongly prefer Option methods, since they are easy to read, rather than positional arguments that do not describe what they do.

[henryiii]

Let me know when you are ready for final review & merge. Do you want to drop the extra argument on add_option in this PR or should we do it later?

[henryiii]

Please update the readme as well (optionally the changelog too, but that's completely optional, since it can cause conflicts when merging multiple PRs, and is pretty easy to do later).

[phlptp]

I think I will let you do the changelog, that is probably worth making sure the style is consistent. I updated the readMe. Let me know if I need to add the defaulted version as a separate overload for vectors. If so it might be desirable to have a test that actually needs that and wouldn't compile without the overload.

[henryiii]

If so it might be desirable to have a test that actually needs that and wouldn't compile without the overload.

That would be great, especially since I want to try adding a function to make the overload unneeded. But that's after this PR gets merged.

Edit: Added Issue #241.

[henryiii]

Thanks!

[rafiw]

Great work

[henryiii]

What about 3 + 2j and "3 + 2j"? Along with - not working by itself, I think this wouldn't work very well. (this still is fantastic, I expect using ; and : as separators might be popular)