Feat: add composable mappings, e.g. use a separate key to jump between snippet tabstops

[smjonas]

• add "actions" users can use to compose their own mappings
• update readme

[smjonas]

@mathjiajia could you please test this PR and confirm that everything works as expected? :)

[smjonas]

I noticed an annoying behaviour in the mappings, will fix it

[smjonas]

I'll wait for @mathjiajia before merging, then I will squash

[mathjiajia]

Dear @smjonas, thanks so much.
I tested with this PR (with the 'alternative' mapping) a bit and it works well.

BTW, if you want to add some explanation in README, below may be better?
Since for those who does not Tab to jump forward, they will not use S-Tab to jump back.
How do you think?

["<Tab>"] = cmp.mapping(
          function(fallback)
            cmp_ultisnips_mappings.expand_or_jump_forwards(fallback)
            -- Alternatively, if you prefer a separate key for jumping between snippet tabstops:
            -- cmp_ultisnips_mappings.expand(fallback)
          end,
          {
            "i", "s",
            -- add this line when using cmp-cmdline:
            -- "c",
          }
        ),
["<S-Tab>"] = cmp.mapping(
          function(fallback)
            cmp_ultisnips_mappings.jump_backwards(fallback)
            -- Alternatively, if you prefer a separate key for jumping between snippet tabstops:
            -- if cmp.visible() then
            --   cmp.select_prev_item()
            -- else
            --   fallback()
            -- end
          end,
          {
            "i", "s",
            -- add this line when using cmp-cmdline:
            -- "c",
          }
        ),

[smjonas]

Thanks for the testing it! I just had another idea how to handle this problem: we could
make the mappings "composable": what I mean is we can provide the basic functions expand, jump-forwards, jump_backwards, select_next, select_prev. We can still keep the defaults as they are now in the readme. Then users can define which of them they want to use in a specific order by passing a table. And we can still provide a default example in the readme. What do you think?

[quangnguyen30192]

Sounds good to me 💪

[smjonas]

Unrelated: do you guys still see =UltiSnips#ExpandSnippet() in the command line when pressing tab? That was supposed to be fixed already but I just saw it again...

[quangnguyen30192]

it still persists yeah

[smjonas]

But it only happens when nothing is selected, right? Like at the beginning of a line?

Edit: it seems to occur when cmp calls fallback(). I am not sure if it can be fixed. At least it happens less often now.

Another thing I noticed: are you able to jump when in select mode in a mapping? I thought it should work with UltiSnipsRemoveSelectModeMappings.

[mathjiajia]

Thanks to both of you.
The "composable" mapping idea is great.

BTW, =UltiSnips#ExpandSnippet() disappears in my computer since this PR #42.
Maybe in my usage, fallback() occurs rarely.

[smjonas]

Btw I didnt't forget about this PR, just thought it was fun to work on treesitter integration for a bit ;) (see #50)

[quangnguyen30192]

Looks pretty good

[smjonas]

Thanks