Implement string-inflect-final-position

[johannes-mueller]

Closes #42

This introduces string-inflect-final-position as proposed in #42. Along with this the behavior for (use-region-p) has been improved. When a region is active the inflection operation now does the following:

• The operation is performed on all symbols in the region.
• The region remains active after the inflection.
• As the region remains active the -cycle- operations now make sense.

[akicho8]

Thank you for the pull request.

I set string-inflection-final-position to 'remain and have been using it for a bit.
However, I noticed an issue when editing a Ruby file. For example:

FOO_BAR
       ^---- cursor position

After calling string-inflection-ruby-style-cycle, the result is:

FooBar

^---- cursor position

The cursor ends up here, which interrupts continuous transformations.

The correct behavior should be:

FooBar
      ^---- cursor position

or

FooBar
     ^---- cursor position

This is the only part that concerns me.

[akicho8]

I am trying to convert a full file path with underscore notation to a class name.
However, when I execute string-inflection-ruby-style-cycle with the cursor positioned as shown below, nothing happens.

path/to/foo_bar.rb
        ^---- cursor position

I am running this within org-mode. It works correctly in ruby-mode.

[johannes-mueller]

FOO_BAR
       ^---- cursor position

After calling string-inflection-ruby-style-cycle, the result is:

FooBar

^---- cursor position

The cursor ends up here, which interrupts continuous transformations.

Thanks for the catch, nice find. The problem is that if the cursor is at the end of the inflected symbol and after the inflection the inflected string is shortened remaining means actually that we end up one char too far.

8ad5b9d fixes this.

[johannes-mueller]

I am trying to convert a full file path with underscore notation to a class name. However, when I execute string-inflection-ruby-style-cycle with the cursor positioned as shown below, nothing happens.

path/to/foo_bar.rb
        ^---- cursor position

I am running this within org-mode. It works correctly in ruby-mode.

The problem here is that (symbol-at-point) finds path/to/foo_bar in org-mode and foo_bar in ruby-mode. In the org-mode case path/to/foo_bar neither matches string-inflection-underscore-p nor string-inflection-upcase-p, so the default (string-inflection-underscore is applied)

Thanks for the catch. I will look for a solution.

[akicho8]

FOO_BAR
       ^---- cursor position

After calling string-inflection-ruby-style-cycle, the result is:

FooBar

^---- cursor position

The cursor ends up here, which interrupts continuous transformations.

Thanks for the catch, nice find. The problem is that if the cursor is at the end of the inflected symbol and after the inflection the inflected string is shortened remaining means actually that we end up one char too far.

8ad5b9d fixes this.

Thank you for the fix. It's working smoothly now.

[johannes-mueller]

path/to/foo_bar.rb
        ^---- cursor position

This turns out to be way trickier. We could adjust the string-inflect-*-p functions so that they still match. I have briefly looked into this. While this seems doable, it would significantly complicate the regexes and sometimes even two matches would be required. That would probably introduce a whole bunch of new edge cases.

Moreover, in such a complex symbol there always rises the question, which kind of case "wins". Consider for example

some_path/to/MyProject/foo-bar.txt

Is it undescore, camel case or kebab case? And this can be taken further for programming modes when there is a symbol named someFunction_to_do_someThing.

I don't think we should introduce all that much complexity for these kind of edge cases, but I'd like to suggest a possible feature to work around that.

We could use a region not only to extend the operation space to multiple symbols but also to restrict it to a substring of a symbol

someFunction_to_do_someThing
^----------^
  region

would cycle to some_function_to_do_someThing, wheras

someFunction_to_do_someThing
             ^---^
             region

would cycle to someFunction_ToDo_someThing.

That would also work around the issue of path/to/foo_bar.rb as one could simple mark the foo_bar part.

[akicho8]

path/to/foo_bar.rb
        ^---- cursor position

This turns out to be way trickier. We could adjust the string-inflect-*-p functions so that they still match. I have briefly looked into this. While this seems doable, it would significantly complicate the regexes and sometimes even two matches would be required. That would probably introduce a whole bunch of new edge cases.

Moreover, in such a complex symbol there always rises the question, which kind of case "wins". Consider for example

some_path/to/MyProject/foo-bar.txt

Is it undescore, camel case or kebab case? And this can be taken further for programming modes when there is a symbol named someFunction_to_do_someThing.

I don't think we should introduce all that much complexity for these kind of edge cases, but I'd like to suggest a possible feature to work around that.

We could use a region not only to extend the operation space to multiple symbols but also to restrict it to a substring of a symbol

someFunction_to_do_someThing
^----------^
  region

would cycle to some_function_to_do_someThing, wheras

someFunction_to_do_someThing
             ^---^
             region

would cycle to someFunction_ToDo_someThing.

That would also work around the issue of path/to/foo_bar.rb as one could simple mark the foo_bar part.

While the range selection might be slightly cumbersome, it also offers the advantage of partial conversion. Therefore, let's proceed with this method for selecting the range.

[johannes-mueller]

Implemented the partial conversion behavior in 2612de8.

[akicho8]

Thank you for the implementation.
The selected portions are being correctly converted.