The idea is to have a map that allows me to fold blocks of code enclosed in curly braces.
nnoremap zff 0f{zf%
This works as expected but only on the opening brace.
The following version seems to work as long as { and } are on the same line. Should they be on different lines though the cursor only jumps to the opening { and no fold is created.
nnoremap zff 0f{f}zf%
edit:
The problem appears to be that once there is an error in one of the motion commands the map breaks.
:silent! prevents ex commands from breaking a map in case of an error. Is there an equivalent for motion commands?
Your first mapping should work, unless you have a filetype detection or syntax problem that prevents proper pair-matching.
Your second mapping can't work because f (and FtT) only works on the current line. Alternatives:
f{v/}<CR>zf
f{zf/}<CR>
But you can use zf with text-objects:
zfi{
zfat
zfip
so… why bother with a mapping?
You're right that an error in a command sequence breaks the sequence, and this is usually right. If you want to continue even in case of errors, just execute the commands separately through :normal!. This can be chained in a single command-line via :execute:
nnoremap zff :exe 'normal! 0f{' | exe 'normal! f}' | exe 'normal! zf%'
Related
The following command appears to invoke the desired function.
:execute "normal \<Plug>VimwikiAddHeaderLevel<CR>"
However, putting it inside a mapping seems to cause problems.
:nmap <buffer> = execute "normal \<Plug>VimwikiAddHeaderLevel<CR>"
Here's the output when I type =.
E114: Missing quote: "normal \<Plug>VimwikiAddHeaderLevel
E15: Invalid expression: "normal \<Plug>VimwikiAddHeaderLevel
Is there some special syntax that would allow me to perform this mapping?
You don't enter command-line mode (for :execute) from the normal mode mapping; the : is missing.
The <Plug> and <CR> are already evaluated by the mapping; the double quotes do not protect them. The <CR> submits the (incomplete, as Vim hasn't seen the trailing ") command-line, and that causes the E114.
After escaping < as <lt>, you still need another <CR> to conclude the :execute command.
:nmap <buffer> = :execute "normal \<lt>Plug>VimwikiAddHeaderLevel\<lt>CR>"<CR>
As I've noted in your other question, you need a :for loop if you really want to work around the plugin's failure to accept a count. Though it would be possible to do all that inline in the mapping's right-hand side, separating the loop into a separate :function is highly recommended, exactly to avoid such escaping issues. Inside a function, the plugin invocation is a simple
:execute "normal \<Plug>VimwikiAddHeaderLevel"
I have my .vimrc in a different path, that I source from my main ~/.vimrc (so I can share same settings across Windows, bash on Windows, etc).
I'm trying to write something in the .vimrc in question, that would make a hotkey for editing said .vimrc, without hard coding the path.
What I currently have is this:
let g:vimrc_path = expand('<sfile>')
:map <Leader>v exec(":e " + g:vimrc_path + "<CR>")
But this doesn't seem to do anything. I've verified that g:vimrc_path is the right value, and that the <Leader>v ends up being called by subbing in echo messages, but I'm not wrapping my head around why the variable I'm trying to define doesn't get expanded correctly.
String concatenation is done with ., not with +, which performs coercion into numbers and addition. But :execute takes multiple arguments (which it space-separates), so you don't actually need this here.
You should use :noremap; it makes the mapping immune to remapping and recursion.
Also, I doubt you need visual and operator-pending modes (:help map-modes), so define this just for normal mode.
:exec[ute is an Ex command, so for a normal-mode mapping, you need to first enter command-line mode. So :exec 'edit' instead of exec ':edit'.
Also, this is not a function (though Vim 8 now also has execute()), so the parentheses are superfluous.
The <silent> avoids the printing of the whole command (you'll notice the loading of the vimrc file, anyway); it's optional.
The fnameescape() ensures that pathological path names are also handled; probably not necessary here.
let g:vimrc_path = expand('<sfile>')
nnoremap <silent> <Leader>v :execute 'edit' fnameescape(g:vimrc_path)<CR>
Alternative
As the script path is static, you can move the variable interpolation from runtime (mapping execution) to mapping definition, and get rid of the variable:
execute 'nnoremap <Leader>v :edit' fnameescape(expand('<sfile>')) . '<CR>'
Strings in vimscript are concatenated with ., not with +. For example:
:echo "Hello"." world!"
will echo
Hello world!
If you were to type
:echo "Hello" + " world!"
vim would echo
0
This is because the + operator is only for numbers, so vim attempts to cast these strings to numbers. If you were to run
:echo "3" + "1"
vim would echo "4".
So basically, you just need to change
:map <Leader>v exec(":e " + g:vimrc_path + "<CR>")
to
:map <Leader>v exec(":e ".g:vimrc_path."<CR>")
Another problem you might have not seen is that "<CR>" evaluates to the literal text "<CR>", so it only messes up your function. If you want a literal carriage return, you would need a backslash. However, you definitely do not want to do this! Seriously, try it out and see.
You can see the issue. It looks for a file that has a literal carriage return at the end of the filename! There is a very simple fix though. Remove the "\<cr>" completely. Since :exec runs ex commands by default, the carriage return (and the colon too for that matter) are unnecessary.
Also, as a nitpick,
The parenthesis are not needed for the "exec" function, and
Use nnoremap instead to avoid recursive mappings.
Taking all of this into consideration, I would simplify it to
:nnoremap <Leader>v :exec "e ".g:vimrc_path<cr>
There is a important part of vim mapping:
:map <C-j>r f{^
I noticed, that when the { character is not found in the line, then the rest of the mapping is not executed.
Is there a way, how to force the mapping to continue even though the search character is not found? (In this case execute the return to the beginning of the file)
Yes, Vim aborts on an error in a mapping. You can use :silent! normal! ... to continue regardless of errors. With a sequence of :normal! commands, you can even check (e.g. whether the cursor moved) and react to it. Sketch:
:map <C-j>r :exe 'normal! maf{'<Bar>if getpos("'a") == getpos(".")<Bar>echo "no move"<Bar>endif<CR>
Note that this doesn't scale well. You're better of moving the commands into a :function.
Also, you should probably use :noremap; it makes the mapping immune to remapping and recursion.
The literal answer to this question would be:
:map <C-j>r :silent! exe "normal f{^"<CR>
What do you need to properly jump to a matched search result?
To reproduce, make a macro with a search in it after you've run vim -u NONE to ensure there's no vimrc interfering. You'll need to make a file with at least 2 lines and put the cursor on the line without the text TEST_TEXT.
map x :norm gg/TEST_TEXT^MIthis
My intention is that when I press x, it goes to the top of the file, looks for TEST_TEXT and then puts this at the start of the line that matches the search. The ^M is a literal newline, achieved with the CtrlQ+Enter keypress. What's happening instead is either nothing happens, or the text gets entered on the same line as when I called the macro.
If I just run the :norm gg/TEST_TEXT^MIthis command without mapping it to a key, the command executes successfully.
I had an initially longer command involving a separate file and the tcomment plugin, but I've gotten it narrowed down to this.
What is the correct sequence of keys to pull this off once I've mapped it to a key?
The problem is that the ^M concludes the :normal Ex command, so your search command is aborted instead of executed. The Ithis is then executed outside of :normal.
In fact, you don't need :normal here at all. And, it's easier and more readable to use the special key notation with mappings:
:map x gg/TEST_TEXT<CR>Ithis
If you really wanted to use :normal, you'd have to wrap this in :execute, like this:
:map x :exe "norm gg/TEST_TEXT\<lt>CR>Ithis"<CR>
Bonus tips
You should use :noremap; it makes the mapping immune to remapping and recursion.
Better restrict the mapping to normal mode, as in its current form, it won't behave as expected in visual and operator-pending mode: :nnoremap
This clobbers the last search pattern and its highlighting. Use of lower-level functions like search() is recommended instead.
There are many ways of doing this however this is my preferred method:
nnoremap x :0/TEST_TEXT/norm! Itest<esc>
Explanation:
:{range}norm! {cmd} - execute normal commands, {cmd}, on a range of lines,{range}.
! on :normal means the commands will not be remapped.
The range 0/TEST_TEXT start before the first line and then finds the first matching line.
I have a few issues with your current mapping:
You are not specifying noremap. You usually want to use noremap
It would be best to specifiy a mode like normal mode, e.g. nnoremap
It is usually best to use <cr> notation with mappings
You are using :normal when your command is already in normal mode but not using any of the ex command features, e.g. a range.
For more help see:
:h :map
:h :norm
:h range
try this mapping:
nnoremap x gg/TEST_TEXT<cr>Ithis<esc>
note that, if you map x on this operation, you lost the original x feature.
I found the following execute command to be useful in vim:
:execute "normal! mqA;\<esc>`q"
it goes to normal mode, then it makes a mark "q" at cursor position,
goes to end of the line and adds a semicolon ";", then it goes to
normal mode again and returns to the original cursor position.
How could I map all this command to a key called "scc"?
I have tried:
imap scc <Esc>:execute "normal! mqA;\<esc>`q"<CR>
however it didn't work. Thanks.
You don't need :execute - normal, you can simply do (and use the "noremap" form):
inoremap scc <Esc>mqA;<Esc>`q
But your map keys are not very good chosen(?), try e.g. <F3> for function key 3.
Others have dealt with the core problem but I should had that, instead of creating an alphabetical mark, you could use a "context mark":
inoremap <something> <Esc>m`A;<Esc>``
Using an alphabetical mark is not wrong, mind you, but I think they are more useful elsewhere.