fix: ensure SemVer instance passed to inc are not modified

[louis-bompart]

inc can be used three ways:

1. As a an option on the CLI : npx semver 1.0.0 -i major
2. As a method of the SemVer class: semverInstance.inc('major')
3. As a function: inc('1.0.0', 'major')

This PR does not impact the behaviour of the first two and focuses only on the last one.

The underlying issue is described in #425, but tl;dr inc should be a pure function (I think, from its documentation on NPM), and here it has a side-effect.

I remove this side-effect by ensuring that the SemVer instance created inside the inc function is created from a string, which removes the mutation vector.

References

Depends on #426 - just to fix the tests, that's why 563d75e is polluting the changeset
Fixes #425

[lukekarrys]

Thanks for this PR @louis-bompart! I think this is the right direction for inc as well as any other "pure" functions. Would you be willing to update this PR with this change made across all those functions? I also think we could export semver/functions/clone that would clone an instance and use that internally.

And I apologize for letting this sit so long, it now has some conficts and shouldn't need any dep updates as those have all been updated.

[lukekarrys]

See previous comment

[louis-bompart]

I also think we could export semver/functions/clone that would clone an instance and use that internally.

That's my two cents, so take it as it is, but I think that a 'clone' function would be a bit out of place:
As a user, I'd be confused about what 'cloning' a semver is doing, especially when the other functions are quite so 'math-oriented'.

I think adding a factory function to the SemVer class would maybe be more in place.

So instead of doing:

import {clone, SemVer} from 'semver';
const verA = new SemVer('1.0.0');
const verB = clone(verA);

One would do this:

import {SemVer} from 'semver';
const verA = new SemVer('1.0.0');
const verB = SemVer.clone(verA);

Another solution would be to do nothing and use structuredClone internally. This would require to enforce Node >=17.x tho.

[louis-bompart]

Would you be willing to update this PR with this change made across all those functions?

✔️. As far as I could tell, only parse was returning an actual SemVer object from a SemVer object.

[lukekarrys]

Thanks @louis-bompart! This looks great!

As far as I could tell, only parse was returning an actual SemVer object from a SemVer object.

Thanks for looking into this, not sure why I thought there would be more. And sorry for the back and forth but I now think that this should only apply to inc since that is designed to change the state of a version.

I can see people using parse as a safe way to handle a variable that might be a string or a semver instance and I don't want to break the strict equality that returning the instance provides.

> s = new semver.SemVer('1.0.0')
SemVer {
  options: {},
  loose: false,
  includePrerelease: false,
  raw: '1.0.0',
  major: 1,
  minor: 0,
  patch: 0,
  prerelease: [],
  build: [],
  version: '1.0.0'
}
> s === semver.parse(s)
true
> s === semver.parse('1.0.0')
false

[lukekarrys]

Can you remove the commit updating parse? After that I think this is good to go!