Codecov Report

Merging #8497 (8ae8f8a) into master (22f85d0) will decrease coverage by 0.07%.
The diff coverage is 100.00%.

❗ Current head 8ae8f8a differs from pull request most recent head 92c5c2b. Consider uploading reports for the commit 92c5c2b to get more accurate results

@@            Coverage Diff             @@
##           master    #8497      +/-   ##
==========================================
- Coverage   90.02%   89.94%   -0.08%     
==========================================
  Files         149      145       -4     
  Lines      168044   167043    -1001     
==========================================
- Hits       151278   150254    -1024     
- Misses      16766    16789      +23     

Flags with carried forward coverage won't be shown. Click here to find out more.

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 22f85d0...92c5c2b. Read the comment docs.

I think it's better to keep the dot. If we would leave it out, it's too
easy to have an invalid pattern go unnoticed.

Okay, I'll leave it as is. However, please note, \%l, \%c and \%v is currently not an error, except for ("E486: Pattern not found").

Thanks for noticing the other two things. I think I have fixed that now, will push a new version in a minute (after rebasing)

Thank you very much for the feature; it's extremely useful.

I don't know whether it's possible to implement, but \%<=.c would have been also useful. Without, we sometimes have to write this (which is a bit harder to read):

\%(\%<.c\|\%.c\)

Here is a real example where it's necessary. Suppose you want a mapping to insert the text under the cursor and between backticks on the command-line:

cnoremap <C-R>` <C-R><C-R>=<SID>InsertCodespan()<CR>
def InsertCodespan(): string
    var pat: string = '\%(\%<.c\|\%.c\)`\zs[^`]\+\ze`\%>.c'
    return getline('.')
        ->matchstr(pat)
        ->substitute('^:', '', '')
enddef

Working test:

vim9script
'# some comment with a `codespan` we want to insert on the command-line'->setline(1)
search('codespan')

cnoremap <C-R>` <C-R><C-R>=<SID>InsertCodespan()<CR>
def InsertCodespan(): string
    var pat: string = '\%(\%<.c\|\%.c\)`\zs[^`]\+\ze`\%>.c'
    return getline('.')
        ->matchstr(pat)
        ->substitute('^:', '', '')
enddef

feedkeys(":\<C-R>`")

# expected: the command-line contains "codespan"
# actual:   the command-line contains "codespan"

But notice this regex in the first line of the function:

\%(\%<.c\|\%.c\)`\zs[^`]\+\ze`\%>.c

And in particular, notice the start:

\%(\%<.c\|\%.c\)`\zs[^`]\+\ze`\%>.c
^--------------^

We can't just write \%<.c, because the mapping would fail if we're on the first backtick, which might seem unexpected to the user. Failing test:

vim9script
'# some comment with a `codespan` we want to insert on the command-line'->setline(1)
search('`codespan')

cnoremap <C-R>` <C-R><C-R>=<SID>InsertCodespan()<CR>
def InsertCodespan(): string
    var pat: string = '\%<.c`\zs[^`]\+\ze`\%>.c'
    return getline('.')
        ->matchstr(pat)
        ->substitute('^:', '', '')
enddef

feedkeys(":\<C-R>`")

# expected: the command-line contains "codespan"
# actual:   the command-line is empty

With \%<=.c, we could simplify the regex in the working code like this:

\%<=.c`\zs[^`]\+\ze`\%>.c
^----^

With regards to \%.v, it doesn't work when we're on a multicell characters (like a tab):

vim9script
'the quick brown fox jumps over the lazy dog'->setline(1)
substitute/ /\t/g
search('jumps')
normal! h
feedkeys('/\%.v' .. "\<CR>")

E486: Pattern not found: \%.v

I guess that's because – internally – \%.v must be translated into \%123v where 123 is the index of the cell under the cursor. And on a multicell character, the cell under the cursor is usually the last one (unless we've tweaked 'virtualedit'). But \%123v does not want the index of the last cell; it wants the index of the first cell.

When \%.v is translated into \%123v (again, internally), could Vim make sure to use the index of the first cell, rather than the index of the last one?

When %.v is translated into %123v (again, internally), could Vim make sure to use the index of the first cell, rather than the index of the last one?

Otherwise, we have to write this (which is much harder to read):

var vcol: number = virtcol([line('.'), col('.') - 1]) + 1
var pat: string = '\%' .. vcol .. 'v'

Example:

vim9script
'the quick brown fox jumps over the lazy dog'->setline(1)
substitute/ /\t/g
search('jumps')
normal! h
var vcol: number = virtcol([line('.'), col('.') - 1]) + 1
var pat: string = '\%' .. vcol .. 'v'
feedkeys('/' .. pat .. "\<CR>")