Nvim :help pages, generated from source using the tree-sitter-vimdoc parser.

Key mapping, abbreviations and user-defined commands.
This subject is introduced in sections 05.3, 24.7 and 40.1 of the user manual.
Key mapping is used to change the meaning of typed keys. The most common use is to define a sequence of commands for a function key. Example:
:map <F2> a<C-R>=strftime("%c")<CR><Esc>
This appends the current date and time after the cursor (in <> notation <>).
There are commands to enter new mappings, remove mappings and list mappings. See map-overview for the various forms of "map" and their relationships with modes.
{lhs} means left-hand-side {lhs} {rhs} means right-hand-side {rhs}
:map {lhs} {rhs} mapmode-nvo :map :nm[ap] {lhs} {rhs} mapmode-n :nm :nmap :vm[ap] {lhs} {rhs} mapmode-v :vm :vmap :xm[ap] {lhs} {rhs} mapmode-x :xm :xmap :smap {lhs} {rhs} mapmode-s :smap :om[ap] {lhs} {rhs} mapmode-o :om :omap :map! {lhs} {rhs} mapmode-ic :map! :im[ap] {lhs} {rhs} mapmode-i :im :imap :lm[ap] {lhs} {rhs} mapmode-l :lm :lma :lmap :cm[ap] {lhs} {rhs} mapmode-c :cm :cmap :tma[p] {lhs} {rhs} mapmode-t :tma :tmap Map the key sequence {lhs} to {rhs} for the modes where the map command applies. The result, including {rhs}, is then further scanned for mappings. This allows for nested and recursive use of mappings. Note: Trailing spaces are included in the {rhs}, because space is a valid Normal mode command. See map-trailing-white.
:nore :norem :no[remap] {lhs} {rhs} mapmode-nvo :no :noremap :nor :nn[oremap] {lhs} {rhs} mapmode-n :nn :nnoremap :vn[oremap] {lhs} {rhs} mapmode-v :vn :vnoremap :xn[oremap] {lhs} {rhs} mapmode-x :xn :xnoremap :snor[emap] {lhs} {rhs} mapmode-s :snor :snore :snoremap :ono[remap] {lhs} {rhs} mapmode-o :ono :onoremap :no[remap]! {lhs} {rhs} mapmode-ic :no! :noremap! :ino[remap] {lhs} {rhs} mapmode-i :ino :inor :inoremap :ln[oremap] {lhs} {rhs} mapmode-l :ln :lnoremap :cno[remap] {lhs} {rhs} mapmode-c :cno :cnor :cnoremap :tno[remap] {lhs} {rhs} mapmode-t :tno :tnoremap Map the key sequence {lhs} to {rhs} for the modes where the map command applies. Disallow mapping of {rhs}, to avoid nested and recursive mappings. Often used to redefine a command. Note: When <Plug> appears in the {rhs} this part is always applied even if remapping is disallowed.
:unm[ap] {lhs} mapmode-nvo :unm :unmap :nun[map] {lhs} mapmode-n :nun :nunmap :vu[nmap] {lhs} mapmode-v :vu :vunmap :xu[nmap] {lhs} mapmode-x :xu :xunmap :sunm[ap] {lhs} mapmode-s :sunm :sunmap :ou[nmap] {lhs} mapmode-o :ou :ounmap :unm[ap]! {lhs} mapmode-ic :unm! :unmap! :iu[nmap] {lhs} mapmode-i :iu :iunmap :lu[nmap] {lhs} mapmode-l :lu :lunmap :cu[nmap] {lhs} mapmode-c :cu :cun :cunmap :tunma[p] {lhs} mapmode-t :tunma :tunmap Remove the mapping of {lhs} for the modes where the map command applies. The mapping may remain defined for other modes where it applies. It also works when {lhs} matches the {rhs} of a mapping. This is for when an abbreviation applied. Note: Trailing spaces are included in the {lhs}. See map-trailing-white.
:mapc[lear] mapmode-nvo :mapc :mapclear :nmapc[lear] mapmode-n :nmapc :nmapclear :vmapc[lear] mapmode-v :vmapc :vmapclear :xmapc[lear] mapmode-x :xmapc :xmapclear :smapc[lear] mapmode-s :smapc :smapclear :omapc[lear] mapmode-o :omapc :omapclear :mapc[lear]! mapmode-ic :mapc! :mapclear! :imapc[lear] mapmode-i :imapc :imapclear :lmapc[lear] mapmode-l :lmapc :lmapclear :cmapc[lear] mapmode-c :cmapc :cmapclear :tmapc[lear] mapmode-t :tmapc :tmapclear Remove ALL mappings for the modes where the map command applies. Use the <buffer> argument to remove buffer-local mappings :map-<buffer> Warning: This also removes the default-mappings.
:map mapmode-nvo :nm[ap] mapmode-n :vm[ap] mapmode-v :xm[ap] mapmode-x :sm[ap] mapmode-s :om[ap] mapmode-o :map! mapmode-ic :im[ap] mapmode-i :lm[ap] mapmode-l :cm[ap] mapmode-c :tma[p] mapmode-t List all key mappings for the modes where the map command applies. Note that ":map" and ":map!" are used most often, because they include the other modes.
:map {lhs} mapmode-nvo :map_l :nm[ap] {lhs} mapmode-n :nmap_l :vm[ap] {lhs} mapmode-v :vmap_l :xm[ap] {lhs} mapmode-x :xmap_l :sm[ap] {lhs} mapmode-s :smap_l :om[ap] {lhs} mapmode-o :omap_l :map! {lhs} mapmode-ic :map_l! :im[ap] {lhs} mapmode-i :imap_l :lm[ap] {lhs} mapmode-l :lmap_l :cm[ap] {lhs} mapmode-c :cmap_l :tma[p] {lhs} mapmode-t :tmap_l List the key mappings for the key sequences starting with {lhs} in the modes where the map command applies.
These commands are used to map a key or key sequence to a string of characters. You can use this to put command sequences under function keys, translate one key into another, etc. See :mkexrc for how to save and restore the current mappings.
map-ambiguous When two mappings start with the same sequence of characters, they are ambiguous. Example:
:imap aa foo
:imap aaa bar
When Vim has read "aa", it will need to get another character to be able to decide if "aa" or "aaa" should be mapped. This means that after typing "aa" that mapping won't get expanded yet, Vim is waiting for another character. If you type a space, then "foo" will get inserted, plus the space. If you type "a", then "bar" will get inserted.
map-trailing-white This unmap command does NOT work:
:map @@ foo
:unmap @@ | print
Because it tries to unmap "@@ ", including the white space before the command separator "|". Other examples with trailing white space:
unmap @@ 
unmap @@     " comment
An error will be issued, which is very hard to identify, because the ending whitespace character in unmap @@ is not visible.
A generic solution is to put the command separator "|" right after the mapped keys. After that white space and a comment may follow:
unmap @@|    " comment

1.2 SPECIAL ARGUMENTS :map-arguments

"<buffer>", "<nowait>", "<silent>", "<script>", "<expr>" and "<unique>" can be used in any order. They must appear right after the command, before any other arguments.
:map-local :map-<buffer> :map-buffer E224 E225 If the first argument to one of these commands is "<buffer>" the mapping will be effective in the current buffer only. Example:
:map <buffer>  ,w  /[.,;]<CR>
Then you can map ",w" to something else in another buffer:
:map <buffer>  ,w  /[#&!]<CR>
The local buffer mappings are used before the global ones. See <nowait> below to make a short local mapping not taking effect when a longer global one exists. The "<buffer>" argument can also be used to clear mappings:
:unmap <buffer> ,w
:mapclear <buffer>
Local mappings are also cleared when a buffer is deleted, but not when it is unloaded. Just like local option values. Also see map-precedence.
:map-<nowait> :map-nowait When defining a buffer-local mapping for "," there may be a global mapping that starts with ",". Then you need to type another character for Vim to know whether to use the "," mapping or the longer one. To avoid this add the <nowait> argument. Then the mapping will be used when it matches, Vim does not wait for more characters to be typed. However, if the characters were already typed they are used. Note that this works when the <nowait> mapping fully matches and is found before any partial matches. This works when:
There is only one matching buffer-local mapping, since these are always found before global mappings.
There is another buffer-local mapping that partly matches, but it is defined earlier (last defined mapping is found first).
:map-<silent> :map-silent To define a mapping which will not be echoed on the command line, add "<silent>" as the first argument. Example:
:map <silent> ,h /Header<CR>
The search string will not be echoed when using this mapping. Messages from the executed command are still given though. To shut them up too, add a ":silent" in the executed command:
:map <silent> ,h :exe ":silent normal /Header\r"<CR>
Note that the effect of a command might also be silenced, e.g., when the mapping selects another entry for command line completion it won't be displayed. Prompts will still be given, e.g., for inputdialog(). Using "<silent>" for an abbreviation is possible, but will cause redrawing of the command line to fail.
:map-<script> :map-script If the first argument to one of these commands is "<script>" and it is used to define a new mapping or abbreviation, the mapping will only remap characters in the {rhs} using mappings that were defined local to a script, starting with "<SID>". This can be used to avoid that mappings from outside a script interfere (e.g., when CTRL-V is remapped in mswin.vim), but do use other mappings defined in the script. Note: ":map <script>" and ":noremap <script>" do the same thing. The "<script>" overrules the command name. Using ":noremap <script>" is preferred, because it's clearer that remapping is (mostly) disabled.
:map-<unique> :map-unique E226 E227 If the first argument to one of these commands is "<unique>" and it is used to define a new mapping or abbreviation, the command will fail if the mapping or abbreviation already exists. Example:
:map <unique> ,w  /[#&!]<CR>
When defining a local mapping, there will also be a check if a global map already exists which is equal. Example of what will fail:
:map ,w  /[#&!]<CR>
:map <buffer> <unique> ,w  /[.,;]<CR>
If you want to map a key and then have it do what it was originally mapped to, have a look at maparg().
:map-<expr> :map-expression If the first argument to one of these commands is "<expr>" and it is used to define a new mapping or abbreviation, the argument is an expression. The expression is evaluated to obtain the {rhs} that is used. Example:
:inoremap <expr> . <SID>InsertDot()
The result of the s:InsertDot() function will be inserted. It could check the text before the cursor and start omni completion when some condition is met. Using a script-local function is preferred, to avoid polluting the global namespace. Use <SID> in the RHS so that the script that the mapping was defined in can be found.
For abbreviations v:char is set to the character that was typed to trigger the abbreviation. You can use this to decide how to expand the {lhs}. You should not either insert or change the v:char.
In case you want the mapping to not do anything, you can have the expression evaluate to an empty string. If something changed that requires Vim to go through the main loop (e.g. to update the display), return "\<Ignore>". This is similar to "nothing" but makes Vim return from the loop that waits for input.
Keep in mind that the expression may be evaluated when looking for typeahead, before the previous command has been executed. For example:
func StoreColumn()
  let g:column = col('.')
  return 'x'
nnoremap <expr> x StoreColumn()
nmap ! f!x
You will notice that g:column has the value from before executing "f!", because "x" is evaluated before "f!" is executed. This can be solved by inserting <Ignore> before the character that is expression-mapped:
nmap ! f!<Ignore>x
Be very careful about side effects! The expression is evaluated while obtaining characters, you may very well make the command dysfunctional. Therefore the following is blocked for <expr> mappings:
Changing the buffer text textlock.
Editing another buffer.
The :normal command.
Moving the cursor is allowed, but it is restored afterwards.
If the cmdline is changed, the old text and cursor position are restored. If you want the mapping to do any of these let the returned characters do that, or use a <Cmd> mapping instead.
You can use getchar(), it consumes typeahead if there is any. E.g., if you have these mappings:
inoremap <expr> <C-L> nr2char(getchar())
inoremap <expr> <C-L>x "foo"
If you now type CTRL-L nothing happens yet, Vim needs the next character to decide what mapping to use. If you type 'x' the second mapping is used and "foo" is inserted. If you type any other key the first mapping is used, getchar() gets the typed key and returns it.
Here is an example that inserts a list number that increases:
let counter = 0
inoremap <expr> <C-L> ListItem()
inoremap <expr> <C-R> ListReset()
func ListItem()
  let g:counter += 1
  return g:counter .. '. '
func ListReset()
  let g:counter = 0
  return ''
CTRL-L inserts the next number, CTRL-R resets the count. CTRL-R returns an empty string, so that nothing is inserted.
Note that using 0x80 as a single byte before other text does not work, it will be seen as a special key.
<Cmd> :map-cmd The <Cmd> pseudokey begins a "command mapping", which executes the command directly without changing modes. Where you might use ":...<CR>" in the {rhs} of a mapping, you can instead use "<Cmd>...<CR>". Example:
noremap x <Cmd>echo mode(1)<CR>
This is more flexible than :<C-U> in Visual and Operator-pending mode, or <C-O>: in Insert mode, because the commands are executed directly in the current mode, instead of always going to Normal mode. Visual mode is preserved, so tricks with gv are not needed. Commands can be invoked directly in Command-line mode (which would otherwise require timer hacks). Example of using <Cmd> halfway Insert mode:
nnoremap <F3> aText <Cmd>echo mode(1)<CR> Added<Esc>
Unlike <expr> mappings, there are no special restrictions on the <Cmd> command: it is executed as if an (unrestricted) autocommand was invoked or an async event was processed.
Because <Cmd> avoids mode-changes (unlike ":") it does not trigger CmdlineEnter and CmdlineLeave events. This helps performance.
For the same reason, keycodes like <C-R><C-W> are interpreted as plain, unmapped keys.
The command is not echo'ed, no need for <silent>.
The {rhs} is not subject to abbreviations nor to other mappings, even if the mapping is recursive.
In Visual mode you can use line('v') and col('v') to get one end of the Visual area, the cursor is at the other end.
E1255 E1136 <Cmd> commands must terminate, that is, they must be followed by <CR> in the {rhs} of the mapping definition. Command-line mode is never entered.
There are seven sets of mappings
For Normal mode: When typing commands.
For Visual mode: When typing commands while the Visual area is highlighted.
For Select mode: like Visual mode but typing text replaces the selection.
For Operator-pending mode: When an operator is pending (after "d", "y", "c", etc.). See below: omap-info.
For Insert mode. These are also used in Replace mode.
For Command-line mode: When entering a ":" or "/" command.
For Terminal mode: When typing in a :terminal buffer.
Special case: While typing a count for a command in Normal mode, mapping zero is disabled. This makes it possible to map zero without making it impossible to type a count with a zero.
map-overview map-modes Overview of which map command works in which mode. More details below. :map :noremap :unmap Normal, Visual, Select, Operator-pending :nmap :nnoremap :nunmap Normal :vmap :vnoremap :vunmap Visual and Select :smap :snoremap :sunmap Select :xmap :xnoremap :xunmap Visual :omap :onoremap :ounmap Operator-pending :map! :noremap! :unmap! Insert and Command-line :imap :inoremap :iunmap Insert :lmap :lnoremap :lunmap Insert, Command-line, Lang-Arg :cmap :cnoremap :cunmap Command-line :tmap :tnoremap :tunmap Terminal
Same information in a table: map-table [nore]map | yes | - | - | yes | yes | yes | - | - | n[nore]map | yes | - | - | - | - | - | - | - | [nore]map! | - | yes | yes | - | - | - | - | - | i[nore]map | - | yes | - | - | - | - | - | - | c[nore]map | - | - | yes | - | - | - | - | - | v[nore]map | - | - | - | yes | yes | - | - | - | x[nore]map | - | - | - | yes | - | - | - | - | s[nore]map | - | - | - | - | yes | - | - | - | o[nore]map | - | - | - | - | - | yes | - | - | t[nore]map | - | - | - | - | - | - | yes | - | l[nore]map | - | yes | yes | - | - | - | - | yes |
:map :noremap :unmap :mapclear yes yes yes :nmap :nnoremap :nunmap :nmapclear yes - - :vmap :vnoremap :vunmap :vmapclear - yes - :omap :onoremap :ounmap :omapclear - - yes
:nunmap can also be used outside of a monastery. mapmode-x mapmode-s Some commands work both in Visual and Select mode, some in only one. Note that quite often "Visual" is mentioned where both Visual and Select mode apply. Select-mode-mapping NOTE: Mapping a printable character in Select mode may confuse the user. It's better to explicitly use :xmap and :smap for printable characters. Or use :sunmap after defining the mapping.
:vmap :vnoremap :vunmap :vmapclear yes yes :xmap :xnoremap :xunmap :xmapclear yes - :smap :snoremap :sunmap :smapclear - yes
mapmode-ic mapmode-i mapmode-c mapmode-l Some commands work both in Insert mode and Command-line mode, some not:
:map! :noremap! :unmap! :mapclear! yes yes - :imap :inoremap :iunmap :imapclear yes - - :cmap :cnoremap :cunmap :cmapclear - yes - :lmap :lnoremap :lunmap :lmapclear yes* yes* yes*
* If 'iminsert' is 1, see language-mapping below.
The original Vi did not have separate mappings for Normal/Visual/Operator-pending mode and for Insert/Command-line mode. Therefore the ":map" and ":map!" commands enter and display mappings for several modes. In Vim you can use the ":nmap", ":vmap", ":omap", ":cmap" and ":imap" commands to enter mappings for each mode separately.
omap-info Operator-pending mappings can be used to define a movement command that can be used with any operator. Simple example:
:omap { w
makes "y{" work like "yw" and "d{" like "dw".
To ignore the starting cursor position and select different text, you can have the omap start Visual mode to select the text to be operated upon. Example that operates on a function name in the current line:
onoremap <silent> F :<C-U>normal! 0f(hviw<CR>
The CTRL-U (<C-U>) is used to remove the range that Vim may insert. The Normal mode commands find the first '(' character and select the first word before it. That usually is the function name.
To enter a mapping for Normal and Visual mode, but not Operator-pending mode, first define it for all three modes, then unmap it for Operator-pending mode:
:map    xx something-difficult
:ounmap xx
Likewise for a mapping for Visual and Operator-pending mode or Normal and Operator-pending mode.
language-mapping ":lmap" defines a mapping that applies to:
Insert mode
Command-line mode
when entering a search pattern
the argument of the commands that accept a text character, such as "r" and "f"
for the input() line Generally: Whenever a character is to be typed that is part of the text in the buffer, not a Vim command character. "Lang-Arg" isn't really another mode, it's just used here for this situation. The simplest way to load a set of related language mappings is by using the 'keymap' option. See 45.5. In Insert mode and in Command-line mode the mappings can be disabled with the CTRL-^ command i_CTRL-^ c_CTRL-^. These commands change the value of the 'iminsert' option. When starting to enter a normal command line (not a search pattern) the mappings are disabled until a CTRL-^ is typed. The state last used is remembered for Insert mode and Search patterns separately. The state for Insert mode is also used when typing a character as an argument to command like "f" or "t". Language mappings will never be applied to already mapped characters. They are only used for typed characters. This assumes that the language mapping was already done when typing the mapping. Correspondingly, language mappings are applied when recording macros, rather than when applying them.
When listing mappings the characters in the first two columns are:
<Space> Normal, Visual, Select and Operator-pending n Normal v Visual and Select s Select x Visual o Operator-pending ! Insert and Command-line i Insert l ":lmap" mappings for Insert, Command-line and Lang-Arg c Command-line t Terminal-Job
Just before the {rhs} a special character can appear: * indicates that it is not remappable & indicates that only script-local mappings are remappable @ indicates a buffer-local mapping
Everything from the first non-blank after {lhs} up to the end of the line (or '|') is considered to be part of {rhs}. This allows the {rhs} to end with a space.
Note: When using mappings for Visual mode, you can use the "'<" mark, which is the start of the last selected Visual area in the current buffer '<.
The :filter command can be used to select what mappings to list. The pattern is matched against the {lhs} and {rhs} in the raw form. If a description was added using nvim_set_keymap() or nvim_buf_set_keymap() then the pattern is also matched against it.
:map-verbose When 'verbose' is non-zero, listing a key map will also display where it was last defined. Example:
:verbose map <C-W>*
n  <C-W>*      * <C-W><C-S>*
        Last set from ~/.config/nvim/init.vim
See :verbose-cmd for more information.
To map a function key, use the internal code for it. To enter such a mapping type CTRL-K and then hit the function key, or use the form "<F2>", "<F10>", "<Up>", "<S-Down>", "<S-F7>", etc. (see table of keys key-notation, all keys from <Up> can be used).

1.6 SPECIAL CHARACTERS :map-special-chars

map_backslash map-backslash Note that only CTRL-V is mentioned here as a special character for mappings and abbreviations. When 'cpoptions' does not contain 'B', a backslash can also be used like CTRL-V. The <> notation can be fully used then <>. But you cannot use "<C-V>" like CTRL-V to escape the special meaning of what follows.
To map a backslash, or use a backslash literally in the {rhs}, the special sequence "<Bslash>" can be used. This avoids the need to double backslashes when using nested mappings.
map_CTRL-C map-CTRL-C Using CTRL-C in the {lhs} is possible, but it will only work when Vim is waiting for a key, not when Vim is busy with something. When Vim is busy CTRL-C interrupts/breaks the command. When using the GUI version on MS-Windows CTRL-C can be mapped to allow a Copy command to the clipboard. Use CTRL-Break to interrupt Vim.
map_space_in_lhs map-space_in_lhs To include a space in {lhs} precede it with a CTRL-V (type two CTRL-Vs for each space). map_space_in_rhs map-space_in_rhs If you want a {rhs} that starts with a space, use "<Space>". To be fully Vi compatible (but unreadable) don't use the <> notation, precede {rhs} with a single CTRL-V (you have to type CTRL-V two times). map_empty_rhs map-empty-rhs You can create an empty {rhs} by typing nothing after a single CTRL-V (you have to type CTRL-V two times). Unfortunately, you cannot do this in a vimrc file. <Nop> An easier way to get a mapping that doesn't produce anything, is to use "<Nop>" for the {rhs}. For example, to disable function key 8:
:map  <F8>  <Nop>
:map! <F8>  <Nop>
map-multibyte It is possible to map multibyte characters, but only the whole character. You cannot map the first byte only. This was done to prevent problems in this scenario:
:set encoding=latin1
:imap <M-C> foo
:set encoding=utf-8
The mapping for <M-C> is defined with the latin1 encoding, resulting in a 0xc3 byte. If you type the character á (0xe1 <M-a>) in UTF-8 encoding this is the two bytes 0xc3 0xa1. You don't want the 0xc3 byte to be mapped then or otherwise it would be impossible to type the á character.
<Leader> mapleader To define a mapping which uses the "g:mapleader" variable, the special string "<Leader>" can be used. It is replaced with the string value of "g:mapleader". If "g:mapleader" is not set or empty, a backslash is used instead. Example:
map <Leader>A  oanother line<Esc>
Works like:
map \A  oanother line<Esc>
But after:
let mapleader = ","
It works like:
map ,A  oanother line<Esc>
Note that the value of "g:mapleader" is used at the moment the mapping is defined. Changing "g:mapleader" after that has no effect for already defined mappings.
<LocalLeader> maplocalleader <LocalLeader> is just like <Leader>, except that it uses "maplocalleader" instead of "mapleader". <LocalLeader> is to be used for mappings which are local to a buffer. Example:
:map <buffer> <LocalLeader>A  oanother line<Esc>
In a global plugin <Leader> should be used and in a filetype plugin <LocalLeader>. "mapleader" and "maplocalleader" can be equal. Although, if you make them different, there is a smaller chance of mappings from global plugins to clash with mappings for filetype plugins. For example, you could keep "mapleader" at the default backslash, and set "maplocalleader" to an underscore.
map-<SID> In a script the special key name "<SID>" can be used to define a mapping that's local to the script. See <SID> for details.
<Plug> The special key name "<Plug>" can be used for an internal mapping, which is not to be matched with any key sequence. This is useful in plugins using-<Plug>.
<MouseMove> The special key name "<MouseMove>" can be used to handle mouse movement. It needs to be enabled with 'mousemoveevent'. The getmousepos() function can be used to obtain the mouse position.
<Char> <Char-> To map a character by its decimal, octal or hexadecimal number the <Char> construct can be used: <Char-123> character 123 <Char-033> character 27 <Char-0x7f> character 127 <S-Char-114> character 114 ('r') shifted ('R') This is useful to specify a (multibyte) character in a 'keymap' file. Upper and lowercase differences are ignored.
map-comments It is not possible to put a comment after these commands, because the " character is considered to be part of the {lhs} or {rhs}. However, one can use |", since this starts a new, empty command with a comment.
map_bar map-bar Since the '|' character is used to separate a map command from the next command, you will have to do something special to include a '|' in {rhs}. There are three methods: <Bar> always :map _l :!ls <Bar> more^M \| 'b' is not in 'cpoptions' :map _l :!ls \| more^M ^V| always :map _l :!ls ^V| more^M
(here ^V stands for CTRL-V; to get one CTRL-V you have to type it twice; you cannot use the <> notation "<C-V>" here).
All three work when you use the default setting for 'cpoptions'.
When 'b' is present in 'cpoptions', "\|" will be recognized as a mapping ending in a '\' and then another command. This is Vi compatible, but illogical when compared to other commands.
map_return map-return When you have a mapping that contains an Ex command, you need to put a line terminator after it to have it executed. The use of <CR> is recommended for this (see <>). Example:
:map  _ls  :!ls -l %:S<CR>:echo "the end"<CR>
To avoid mapping of the characters you type in insert or Command-line mode, type a CTRL-V first. map-error Note that when an error is encountered (that causes an error message or might cause a beep) the rest of the mapping is not executed. This is Vi-compatible.
Note that the second character (argument) of the commands @zZtTfF[]rm'`"v and CTRL-X is not mapped. This was done to be able to use all the named registers and marks, even when the command with the same name has been mapped.
If you are going to map something, you will need to choose which key(s) to use for the {lhs}. You will have to avoid keys that are used for Vim commands, otherwise you would not be able to use those commands anymore. Here are a few suggestions:
Function keys <F2>, <F3>, etc.. Also the shifted function keys <S-F1>, <S-F2>, etc. Note that <F1> is already used for the help command.
Meta-keys (with the ALT key pressed). Depending on your keyboard accented characters may be used as well. :map-alt-keys
Use the '_' or ',' character and then any other character. The "_" and "," commands do exist in Vim (see _ and ,), but you probably never use them.
Use a key that is a synonym for another command. For example: CTRL-P and CTRL-N. Use an extra character to allow more mappings.
The key defined by <Leader> and one or more other keys. This is especially useful in scripts. mapleader
See the file "index" for keys that are not used and thus can be mapped without losing any builtin function. You can also use ":help {key}^D" to find out if a key is used for some command. ({key} is the specific key you want to find out about, ^D is CTRL-D).
A few examples (as you type them: for "<CR>" you type four characters).
:map <F3>  o#include
:map <M-g> /foo<CR>cwbar<Esc>
:map _x    d/END/e<CR>
:map! qq   quadrillion questions
Multiplying a count
When you type a count before triggering a mapping, it's like the count was typed before the {lhs}. For example, with this mapping:
:map <F4>  3w
Typing 2<F4> will result in "23w". Thus not moving 2 * 3 words but 23 words. If you want to multiply counts use the expression register:
:map <F4>  @='3w'<CR>
The part between quotes is the expression being executed. @=
Vim will compare what you type with the start of a mapped sequence. If there is an incomplete match, it will get more characters until there either is a complete match or until there is no match at all. Example: If you map! "qq", the first 'q' will not appear on the screen until you type another character. This is because Vim cannot know if the next character will be a 'q' or not. If the 'timeout' option is on (which is the default) Vim will only wait for one second (or as long as specified with the 'timeoutlen' option). After that it assumes that the 'q' is to be interpreted as such. If you type slowly, or your system is slow, reset the 'timeout' option. Then you might want to set the 'ttimeout' option.
map-precedence Buffer-local mappings (defined using :map-<buffer>) take precedence over global mappings. When a buffer-local mapping is the same as a global mapping, Vim will use the buffer-local mapping. In addition, Vim will use a complete mapping immediately if it was defined with <nowait>, even if a longer mapping has the same prefix. For example, given the following two mappings:
:map <buffer> <nowait> \a   :echo "Local \a"<CR>
:map                   \abc :echo "Global \abc"<CR>
When typing \a the buffer-local mapping will be used immediately. Vim will not wait for more characters to see if the user might be typing \abc.
map-keys-fails There are situations where key codes might not be recognized:
Vim can only read part of the key code. Mostly this is only the first character. This happens on some Unix versions in an xterm.
The key code is after character(s) that are mapped. E.g., "<F1><F1>" or "g<F1>".
The result is that the key code is not recognized in this situation, and the mapping fails. There are two actions needed to avoid this problem:
Remove the 'K' flag from 'cpoptions'. This will make Vim wait for the rest of the characters of the function key.
When using <F1> to <F4> the actual key code generated may correspond to <xF1> to <xF4>. There are mappings from <xF1> to <F1>, <xF2> to <F2>, etc., but these are not recognized after another half a mapping. Make sure the key codes for <F1> to <F4> are correct:
:set <F1>=<type CTRL-V><type F1>
Type the <F1> as four characters. The part after the "=" must be done with the actual keys, not the literal text. Another solution is to use the actual key code in the mapping for the second special key:
:map <F1><Esc>OP :echo "yes"<CR>
Don't type a real <Esc>, Vim will recognize the key code and replace it with <F1> anyway.
recursive_mapping If you include the {lhs} in the {rhs} you have a recursive mapping. When {lhs} is typed, it will be replaced with {rhs}. When the {lhs} which is included in {rhs} is encountered it will be replaced with {rhs}, and so on. This makes it possible to repeat a command an infinite number of times. The only problem is that the only way to stop this is by causing an error. The macros to solve a maze uses this, look there for an example. There is one exception: If the {rhs} starts with {lhs}, the first character is not mapped again (this is Vi compatible). For example:
:map ab abcd
will execute the "a" command and insert "bcd" in the text. The "ab" in the {rhs} will not be mapped again.
If you want to exchange the meaning of two keys you should use the :noremap command. For example:
:noremap k j
:noremap j k
This will exchange the cursor up and down commands.
With the normal :map command mapping takes place until the text is found not to be a part of a {lhs}. For example, if you use:
:map x y
:map y x
Vim will replace x with y, and then y with x, etc. When this has happened 'maxmapdepth' times (default 1000), Vim will give the error message "recursive mapping".
:map-undo If you include an undo command inside a mapped sequence, this will bring the text back in the state before executing the macro. This is compatible with the original Vi, as long as there is only one undo command in the mapped sequence (having two undo commands in a mapped sequence did not make sense in the original Vi, you would get back the text before the first undo).
For a readable mapping command the <A-k> form can be used. Note that <A-k> and <A-K> are different, the latter will use an upper case letter. Actually, <A-K> and <A-S-K> are the same. Instead of "A" you can use "M". If you have an actual Meta modifier key, please see :map-meta-keys.
In the GUI Nvim handles the ALT key itself, thus mapping keys with ALT should always work. But in a terminal Nvim gets a sequence of bytes and has to figure out whether ALT was pressed. Terminals may use ESC to indicate that ALT was pressed. If ESC is followed by a {key} within 'ttimeoutlen' milliseconds, the ESC is interpreted as: <ALT-{key}> otherwise it is interpreted as two key presses: <ESC> {key}
Mapping keys with the Meta modifier works very similar to using the Alt key. What key on your keyboard produces the Meta modifier depends on your keyboard and configuration.
Note that mapping <M-a> actually is for using the Alt key. That can be confusing! It cannot be changed, it would not be backwards compatible.
For the Meta modifier the "T" character is used. For example, to map Meta-b in Insert mode:
:imap <T-b> terrible
1.12 MAPPING SUPER-KEYS or COMMAND-KEYS :map-super-keys :map-cmd-key
The Super / Command modifier is available if the terminal or GUI supports it. The character "D" is used for the Super / Command modifier.
For example, to map Command-b in Insert mode:
:imap <D-b> barritone

1.13 MAPPING AN OPERATOR :map-operator

An operator is used before a {motion} command. To define your own operator you must create a mapping that first sets the 'operatorfunc' option and then invoke the g@ operator. After the user types the {motion} command the specified function will be called.
g@ E774 E775 g@{motion} Call the function set by the 'operatorfunc' option. The '[ mark is positioned at the start of the text moved over by {motion}, the '] mark on the last character of the text. The function is called with one String argument: "line" {motion} was linewise "char" {motion} was charwise "block" {motion} was blockwise-visual The type can be forced, see forced-motion.
Here is an example that counts the number of spaces with <F4>:
nnoremap <expr> <F4> CountSpaces()
xnoremap <expr> <F4> CountSpaces()
" doubling <F4> works on a line
nnoremap <expr> <F4><F4> CountSpaces() .. '_'
function CountSpaces(context = {}, type = '') abort
  if a:type == ''
    let context = #{
      \ dot_command: v:false,
      \ extend_block: '',
      \ virtualedit: [&l:virtualedit, &g:virtualedit],
      \ }
    let &operatorfunc = function('CountSpaces', [context])
    set virtualedit=block
    return 'g@'
  let save = #{
    \ clipboard: &clipboard,
    \ selection: &selection,
    \ virtualedit: [&l:virtualedit, &g:virtualedit],
    \ register: getreginfo('"'),
    \ visual_marks: [getpos("'<"), getpos("'>")],
    \ }
    set clipboard= selection=inclusive virtualedit=
    let commands = #{
      \ line: "'[V']",
      \ char: "`[v`]",
      \ block: "`[\<C-V>`]",
      \ }[a:type]
    let [_, _, col, off] = getpos("']")
    if off != 0
      let vcol = getline("'[")->strpart(0, col + off)->strdisplaywidth()
      if vcol >= [line("'["), '$']->virtcol() - 1
        let a:context.extend_block = '$'
        let a:context.extend_block = vcol .. '|'
    if a:context.extend_block != ''
      let commands ..= 'oO' .. a:context.extend_block
    let commands ..= 'y'
    execute 'silent noautocmd keepjumps normal! ' .. commands
    echomsg getreg('"')->count(' ')
    call setreg('"', save.register)
    call setpos("'<", save.visual_marks[0])
    call setpos("'>", save.visual_marks[1])
    let &clipboard = save.clipboard
    let &selection = save.selection
    let [&l:virtualedit, &g:virtualedit] = get(a:context.dot_command ? save : a:context, 'virtualedit')
    let a:context.dot_command = v:true
An <expr> mapping is used to be able to fetch any prefixed count and register. This also avoids using a command line, which would trigger CmdlineEnter and CmdlineLeave autocommands.
Note that the 'selection' option is temporarily set to "inclusive" to be able to yank exactly the right text by using Visual mode from the '[ to the '] mark.
Also note that the 'clipboard' option is temporarily emptied to avoid clobbering the "* or "+ registers, if its value contains the item unnamed or unnamedplus.
The mode() function will return the state as it will be after applying the operator.
Here is an example for using a lambda function to create a normal-mode operator to add quotes around text in the current line:
nnoremap <F4> <Cmd>let &opfunc='{t ->
                        \ getline(".")
                        \ ->split("\\zs")
                        \ ->insert("\"", col("'']"))
                        \ ->insert("\"", col("''[") - 1)
                        \ ->join("")
                        \ ->setline(".")}'<CR>g@

==============================================================================2. Abbreviations abbreviation abbreviations Abbreviations

Abbreviations are used in Insert mode, Replace mode and Command-line mode. If you enter a word that is an abbreviation, it is replaced with the word it stands for. This can be used to save typing for often used long words. And you can use it to automatically correct obvious spelling errors. Examples:
:iab ms Microsoft :iab tihs this
There are three types of abbreviations:
full-id The "full-id" type consists entirely of keyword characters (letters and characters from 'iskeyword' option). This is the most common abbreviation.
Examples: "foo", "g3", "-1"
end-id The "end-id" type ends in a keyword character, but all the other characters are not keyword characters.
Examples: "#i", "..f", "$/7"
non-id The "non-id" type ends in a non-keyword character, the other characters may be of any type, excluding space and tab.
Examples: "def#", "4/7$"
Examples of strings that cannot be abbreviations: "a.b", "#def", "a b", "_$r"
An abbreviation is only recognized when you type a non-keyword character. This can also be the <Esc> that ends Insert mode or the <CR> that ends a command. The non-keyword character which ends the abbreviation is inserted after the expanded abbreviation. An exception to this is the character <C-]>, which is used to expand an abbreviation without inserting any extra characters.
:ab hh        hello
"hh<Space>" is expanded to "hello<Space>" "hh<C-]>" is expanded to "hello"
The characters before the cursor must match the abbreviation. Each type has an additional rule:
full-id In front of the match is a non-keyword character, or this is where the line or insertion starts. Exception: When the abbreviation is only one character, it is not recognized if there is a non-keyword character in front of it, other than a space or a tab. However, for the command line "'<,'>" (or any other marks) is ignored, as if the command line starts after it.
end-id In front of the match is a keyword character, or a space or a tab, or this is where the line or insertion starts.
non-id In front of the match is a space, tab or the start of the line or the insertion.
Examples: ({CURSOR} is where you type a non-keyword character)
:ab foo   four old otters
" foo{CURSOR}" is expanded to " four old otters" " foobar{CURSOR}" is not expanded "barfoo{CURSOR}" is not expanded
:ab #i #include
"#i{CURSOR}" is expanded to "#include" ">#i{CURSOR}" is not expanded
:ab ;; <endofline>
"test;;" is not expanded "test ;;" is expanded to "test <endofline>"
To avoid the abbreviation in Insert mode: Type CTRL-V before the character that would trigger the abbreviation. E.g. CTRL-V <Space>. Or type part of the abbreviation, exit insert mode with <Esc>, re-enter insert mode with "a" and type the rest.
To avoid the abbreviation in Command-line mode: Type CTRL-V twice somewhere in the abbreviation to avoid it to be replaced. A CTRL-V in front of a normal character is mostly ignored otherwise.
It is possible to move the cursor after an abbreviation:
:iab if if ()<Left>
You can even do more complicated things. For example, to consume the space typed after an abbreviation:
func Eatchar(pat)
   let c = nr2char(getchar(0))
   return (c =~ a:pat) ? '' : c
iabbr <silent> if if ()<Left><C-R>=Eatchar('\s')<CR>
There are no default abbreviations.
Abbreviations are never recursive. You can use ":ab f f-o-o" without any problem. But abbreviations can be mapped.
:abbreviate-local :abbreviate-<buffer> Just like mappings, abbreviations can be local to a buffer. This is mostly used in a filetype-plugin file. Example for a C plugin file:
:abb <buffer> FF  for (i = 0; i < ; ++i)
:ab :abbreviate :ab[breviate] list all abbreviations. The character in the first column indicates the mode where the abbreviation is used: 'i' for insert mode, 'c' for Command-line mode, '!' for both. These are the same as for mappings, see map-listing.
:abbreviate-verbose When 'verbose' is non-zero, listing an abbreviation will also display where it was last defined. Example:
:verbose abbreviate
!  teh                 the
        Last set from /home/abcd/vim/abbr.vim
See :verbose-cmd for more information.
:ab[breviate] {lhs} list the abbreviations that start with {lhs} You may need to insert a CTRL-V (type it twice) to avoid that a typed {lhs} is expanded, since command-line abbreviations apply here.
:ab[breviate] [<expr>] [<buffer>] {lhs} {rhs} add abbreviation for {lhs} to {rhs}. If {lhs} already existed it is replaced with the new {rhs}. {rhs} may contain spaces. See :map-<expr> for the optional <expr> argument. See :map-<buffer> for the optional <buffer> argument.
:una :unabbreviate :una[bbreviate] [<buffer>] {lhs} Remove abbreviation for {lhs} from the list. If none is found, remove abbreviations in which {lhs} matches with the {rhs}. This is done so that you can even remove abbreviations after expansion. To avoid expansion insert a CTRL-V (type it twice).
:norea :noreabbrev :norea[bbrev] [<expr>] [<buffer>] [lhs] [rhs] same as ":ab", but no remapping for this {rhs}
:ca :cab :cabbrev :ca[bbrev] [<expr>] [<buffer>] [lhs] [rhs] same as ":ab", but for Command-line mode only.
:cuna :cunabbrev :cuna[bbrev] [<buffer>] {lhs} Same as ":una", but for Command-line mode only.
:cnorea :cnoreabbrev :cnorea[bbrev] [<expr>] [<buffer>] [lhs] [rhs] same as ":ab", but for Command-line mode only and no remapping for this {rhs}
:ia :iabbrev :ia[bbrev] [<expr>] [<buffer>] [lhs] [rhs] same as ":ab", but for Insert mode only.
:iuna :iunabbrev :iuna[bbrev] [<buffer>] {lhs} Same as ":una", but for insert mode only.
:inorea :inoreabbrev :inorea[bbrev] [<expr>] [<buffer>] [lhs] [rhs] same as ":ab", but for Insert mode only and no remapping for this {rhs}
:abc :abclear :abc[lear] [<buffer>] Remove all abbreviations.
:iabc :iabclear :iabc[lear] [<buffer>] Remove all abbreviations for Insert mode.
:cabc :cabclear :cabc[lear] [<buffer>] Remove all abbreviations for Command-line mode.
using_CTRL-V It is possible to use special characters in the rhs of an abbreviation. CTRL-V has to be used to avoid the special meaning of most non printable characters. How many CTRL-Vs need to be typed depends on how you enter the abbreviation. This also applies to mappings. Let's use an example here.
Suppose you want to abbreviate "esc" to enter an <Esc> character. When you type the ":ab" command in Vim, you have to enter this: (here ^V is a CTRL-V and ^[ is <Esc>)
You type: ab esc ^V^V^V^V^V^[
All keyboard input is subjected to ^V quote interpretation, so the first, third, and fifth ^V characters simply allow the second, and fourth ^Vs, and the ^[, to be entered into the command-line.
You see: ab esc ^V^V^[
The command-line contains two actual ^Vs before the ^[. This is how it should appear in your vimrc file, if you choose to go that route. The first ^V is there to quote the second ^V; the :ab command uses ^V as its own quote character, so you can include quoted whitespace or the | character in the abbreviation. The :ab command doesn't do anything special with the ^[ character, so it doesn't need to be quoted. (Although quoting isn't harmful; that's why typing 7 [but not 8!] ^Vs works.)
Stored as: esc ^V^[
After parsing, the abbreviation's short form ("esc") and long form (the two characters "^V^[") are stored in the abbreviation table. If you give the :ab command with no arguments, this is how the abbreviation will be displayed.
Later, when the abbreviation is expanded because the user typed in the word "esc", the long form is subjected to the same type of ^V interpretation as keyboard input. So the ^V protects the ^[ character from being interpreted as the "exit Insert mode" character. Instead, the ^[ is inserted into the text.
Expands to: ^[
[example given by Steve Kirkendall]
When using several Vim script files, there is the danger that mappings and functions used in one script use the same name as in other scripts. To avoid this, they can be made local to the script.
<SID> <SNR> E81 The string "<SID>" can be used in a mapping or menu. This is useful if you have a script-local function that you want to call from a mapping in the same script. When executing the map command, Vim will replace "<SID>" with the special key code <SNR>, followed by a number that's unique for the script, and an underscore. Example:
:map <SID>Add
would define a mapping "<SNR>23_Add".
When defining a function in a script, "s:" can be prepended to the name to make it local to the script. But when a mapping is executed from outside of the script, it doesn't know in which script the function was defined. To avoid this problem, use "<SID>" instead of "s:". The same translation is done as for mappings. This makes it possible to define a call to the function in a mapping.
When a local function is executed, it runs in the context of the script it was defined in. This means that new functions and mappings it defines can also use "s:" or "<SID>" and it will use the same unique number as when the function itself was defined. Also, the "s:var" local script variables can be used.
When executing an autocommand or a user command, it will run in the context of the script it was defined in. This makes it possible that the command calls a local function or uses a local mapping.
In case the value is used in a context where <SID> cannot be correctly expanded, use the expand() function:
let &includexpr = expand('<SID>') .. 'My_includeexpr()'
Otherwise, using "<SID>" outside of a script context is an error.
If you need to get the script number to use in a complicated script, you can use this function:
func s:ScriptNumber()
  return matchstr(expand('<SID>'), '<SNR>\zs\d\+\ze_')
The "<SNR>" will be shown when listing functions and mappings. This is useful to find out what they are defined to.
The :scriptnames command can be used to see which scripts have been sourced and what their <SNR> number is.
It is possible to define your own Ex commands. A user-defined command can act just like a built-in command (it can have a range or arguments, arguments can be completed as filenames or buffer names, etc), except that when the command is executed, it is transformed into a normal Ex command and then executed.
For starters: See section 40.2 in the user manual.
E183 E841 user-cmd-ambiguous All user defined commands must start with an uppercase letter, to avoid confusion with builtin commands. Exceptions are these builtin commands: :Next They cannot be used for a user defined command.
The other characters of the user command can be uppercase letters, lowercase letters or digits. When using digits, note that other commands that take a numeric argument may become ambiguous. For example, the command ":Cc2" could be the user command ":Cc2" without an argument, or the command ":Cc" with argument "2". It is advised to put a space between the command name and the argument to avoid these problems.
When using a user-defined command, the command can be abbreviated. However, if an abbreviation is not unique, an error will be issued. Furthermore, a built-in command will always take precedence.
:command Rename ...
:command Renumber ...
:Rena                                " Means "Rename"
:Renu                                " Means "Renumber"
:Ren                                " Error - ambiguous
:command Paste ...
It is recommended that full names for user-defined commands are used in scripts.
:com[mand] :com :command List all user-defined commands. When listing commands, the characters in the first columns are: ! Command has the -bang attribute " Command has the -register attribute | Command has the -bar attribute b Command is local to current buffer (see below for details on attributes) The list can be filtered on command name with :filter, e.g., to list all commands with "Pyth" in the name:
filter Pyth command
:com[mand] {cmd} List the user-defined commands that start with {cmd}
:command-verbose When 'verbose' is non-zero, listing a command will also display where it was last defined and any completion argument. Example:
:verbose command TOhtml
See :verbose-cmd for more information.
E174 E182 :com[mand][!] [{attr}...] {cmd} {repl} Define a user command. The name of the command is {cmd} and its replacement text is {repl}. The command's attributes (see below) are {attr}. If the command already exists, an error is reported, unless a ! is specified, in which case the command is redefined. There is one exception: When sourcing a script again, a command that was previously defined in that script will be silently replaced.
:delc[ommand] {cmd} :delc :delcommand E184 Delete the user-defined command {cmd}.
:delc[ommand] -buffer {cmd} E1237 Delete the user-defined command {cmd} that was defined for the current buffer.
:comc[lear] :comc :comclear Delete all user-defined commands.
command-attributes User-defined commands are treated by Nvim just like any other Ex commands. They can have arguments, or have a range specified. Arguments are subject to completion as filenames, buffers, etc. Exactly how this works depends upon the command's attributes, which are specified when the command is defined.
When defining a user command in a script, it will be able to call functions local to the script and use mappings local to the script. When the user invokes the user command, it will run in the context of the script it was defined in. This matters if <SID> is used in a command.
There are a number of attributes, split into four categories: argument handling, completion behavior, range handling, and special cases. The attributes are described below, by category.
E175 E176 :command-nargs By default, a user defined command will take no arguments (and an error is reported if any are supplied). However, it is possible to specify that the command can take arguments, using the -nargs attribute. Valid cases are:
-nargs=0 No arguments are allowed (the default) -nargs=1 Exactly one argument is required, it includes spaces -nargs=* Any number of arguments are allowed (0, 1, or many), separated by white space -nargs=? 0 or 1 arguments are allowed -nargs=+ Arguments must be supplied, but any number are allowed
Arguments are considered to be separated by (unescaped) spaces or tabs in this context, except when there is one argument, then the white space is part of the argument.
Note that arguments are used as text, not as expressions. Specifically, "s:var" will use the script-local variable in the script where the command was defined, not where it is invoked! Example: script1.vim:
:let s:error = "None"
:command -nargs=1 Error echoerr <args>
:source script1.vim
:let s:error = "Wrong!"
:Error s:error
Executing script2.vim will result in "None" being echoed. Not what you intended! Calling a function may be an alternative.
:command-completion E179 E180 E181 :command-complete By default, the arguments of user defined commands do not undergo completion. However, by specifying one or the other of the following attributes, argument completion can be enabled:
-complete=arglist file names in argument list -complete=augroup autocmd groups -complete=buffer buffer names -complete=color color schemes -complete=command Ex command (and arguments) -complete=compiler compilers -complete=dir directory names -complete=environment environment variable names -complete=event autocommand events -complete=expression Vim expression -complete=file file and directory names -complete=file_in_path file and directory names in 'path' -complete=filetype filetype names 'filetype' -complete=function function name -complete=help help subjects -complete=highlight highlight groups -complete=history :history suboptions -complete=keymap keyboard mappings -complete=locale locale names (as output of locale -a) -complete=lua Lua expression :lua -complete=mapclear buffer argument -complete=mapping mapping name -complete=menu menus -complete=messages :messages suboptions -complete=option options -complete=packadd optional package pack-add names -complete=shellcmd Shell command -complete=sign :sign suboptions -complete=syntax syntax file names 'syntax' -complete=syntime :syntime suboptions -complete=tag tags -complete=tag_listfiles tags, file names are shown when CTRL-D is hit -complete=user user names -complete=var user variables -complete=custom,{func} custom completion, defined via {func} -complete=customlist,{func} custom completion, defined via {func}
If you specify completion while there is nothing to complete (-nargs=0, the default) then you get error E1208 . Note: That some completion methods might expand environment variables.
:command-completion-custom :command-completion-customlist E467 E468 It is possible to define customized completion schemes via the "custom,{func}" or the "customlist,{func}" completion argument. The {func} part should be a function with the following signature:
:function {func}(ArgLead, CmdLine, CursorPos)
The function need not use all these arguments. The function should provide the completion candidates as the return value.
For the "custom" argument, the function should return the completion candidates one per line in a newline separated string.
For the "customlist" argument, the function should return the completion candidates as a Vim List. Non-string items in the list are ignored.
The function arguments are: ArgLead the leading portion of the argument currently being completed on CmdLine the entire command line CursorPos the cursor position in it (byte index) The function may use these for determining context. For the "custom" argument, it is not necessary to filter candidates against the (implicit pattern in) ArgLead. Vim will filter the candidates with its regexp engine after function return, and this is probably more efficient in most cases. If 'wildoptions' contains "fuzzy", then the candidates will be filtered using fuzzy-matching. For the "customlist" argument, Vim will not filter the returned completion candidates and the user supplied function should filter the candidates.
The following example lists user names to a Finger command
:com -complete=custom,ListUsers -nargs=1 Finger !finger <args>
:fun ListUsers(A,L,P)
:    return system("cut -d: -f1 /etc/passwd")
The following example completes filenames from the directories specified in the 'path' option:
:com -nargs=1 -bang -complete=customlist,EditFileComplete
                    \ EditFile edit<bang> <args>
:fun EditFileComplete(A,L,P)
:    return split(globpath(&path, a:A), "\n")
This example does not work for file names with spaces!
E177 E178 :command-range :command-count By default, user-defined commands do not accept a line number range. However, it is possible to specify that the command does take a range (the -range attribute), or that it takes an arbitrary count value, either in the line number position (-range=N, like the :split command) or as a "count" argument (-count=N, like the :Next command). The count will then be available in the argument with <count>.
Possible attributes are:
-range Range allowed, default is current line -range=% Range allowed, default is whole file (1,$) -range=N A count (default N) which is specified in the line number position (like :split); allows for zero line number. -count=N A count (default N) which is specified either in the line number position, or as an initial argument (like :Next). -count Acts like -count=0
Note that -range=N and -count=N are mutually exclusive - only one should be specified.
:command-addr It is possible that the special characters in the range like ., $ or % which by default correspond to the current line, last line and the whole buffer, relate to arguments, (loaded) buffers, windows or tab pages.
Possible values are (second column is the short name used in listing): -addr=lines Range of lines (this is the default for -range) -addr=arguments arg Range for arguments -addr=buffers buf Range for buffers (also not loaded buffers) -addr=loaded_buffers load Range for loaded buffers -addr=windows win Range for windows -addr=tabs tab Range for tab pages -addr=quickfix qf Range for quickfix entries -addr=other ? Other kind of range; can use ".", "$" and "%" as with "lines" (this is the default for -count)
:command-preview {nvim-api} Commands can show an 'inccommand' (as-you-type) preview by defining a preview handler (only from Lua, see nvim_create_user_command()).
Before the preview callback is executed, Nvim will temporarily disable 'cursorline' and 'cursorcolumn' to avoid highlighting issues.
The preview callback must be a Lua function with this signature:
function cmdpreview(opts, ns, buf)
where "opts" has the same form as that given to nvim_create_user_command() callbacks, "ns" is the preview namespace id for highlights, and "buf" is the buffer that your preview routine will directly modify to show the previewed results (for "inccommand=split", or nil for "inccommand=nosplit").
Your command preview routine must implement this protocol:
1. Modify the target buffers as required for the preview (see nvim_buf_set_text() and nvim_buf_set_lines()). 2. If preview buffer is provided, add necessary text to the preview buffer. 3. Add required highlights to the target buffers. If preview buffer is provided, add required highlights to the preview buffer as well. All highlights must be added to the preview namespace which is provided as an argument to the preview callback (see nvim_buf_add_highlight() and nvim_buf_set_extmark() for help on how to add highlights to a namespace). 4. Return an integer (0, 1, 2) which controls how Nvim behaves as follows: 0: No preview is shown. 1: Preview is shown without preview window (even with "inccommand=split"). 2: Preview is shown and preview window is opened (if "inccommand=split"). For "inccommand=nosplit" this is the same as 1.
After preview ends, Nvim discards all changes to all buffers made during the preview and clears all highlights in the preview namespace.
Here's an example of a command to trim trailing whitespace from lines that supports incremental command preview:
-- If invoked as a preview callback, performs 'inccommand' preview by
-- highlighting trailing whitespace in the current buffer.
local function trim_space_preview(opts, preview_ns, preview_buf)
  vim.cmd('hi clear Whitespace')
  local line1 = opts.line1
  local line2 = opts.line2
  local buf = vim.api.nvim_get_current_buf()
  local lines = vim.api.nvim_buf_get_lines(buf, line1 - 1, line2, false)
  local preview_buf_line = 0
  for i, line in ipairs(lines) do
    local start_idx, end_idx = string.find(line, '%s+$')
    if start_idx then
      -- Highlight the match
        line1 + i - 2,
        start_idx - 1,
      -- Add lines and set highlights in the preview buffer
      -- if inccommand=split
      if preview_buf then
        local prefix = string.format('|%d| ', line1 + i - 1)
          { prefix .. line }
          #prefix + start_idx - 1,
          #prefix + end_idx
        preview_buf_line = preview_buf_line + 1
  -- Return the value of the preview type
  return 2
-- Trims all trailing whitespace in the current buffer.
local function trim_space(opts)
  local line1 = opts.line1
  local line2 = opts.line2
  local buf = vim.api.nvim_get_current_buf()
  local lines = vim.api.nvim_buf_get_lines(buf, line1 - 1, line2, false)
  local new_lines = {}
  for i, line in ipairs(lines) do
    new_lines[i] = string.gsub(line, '%s+$', '')
  vim.api.nvim_buf_set_lines(buf, line1 - 1, line2, false, new_lines)
-- Create the user command
  { nargs = '?', range = '%', addr = 'lines', preview = trim_space_preview }
-bang The command can take a ! modifier (like :q or :w) -bar The command can be followed by a "|" and another command. A "|" inside the command argument is not allowed then. Also checks for a " to start a comment. -register The first argument to the command can be an optional register name (like :del, :put, :yank). -buffer The command will only be available in the current buffer. -keepscript Do not use the location of where the user command was defined for verbose messages, use the location of where the user command was invoked.
In the cases of the -count and -register attributes, if the optional argument is supplied, it is removed from the argument list and is available to the replacement text separately. Note that these arguments can be abbreviated, but that is a deprecated feature. Use the full name for new scripts.
The replacement text {repl} for a user defined command is scanned for special escape sequences, using <...> notation. Escape sequences are replaced with values from the entered command line, and all other text is copied unchanged. The resulting string is executed as an Ex command. To avoid the replacement use <lt> in place of the initial <. Thus to include "<bang>" literally use "<lt>bang>".
The valid escape sequences are
<line1> <line1> The starting line of the command range. <line2> <line2> The final line of the command range. <range> <range> The number of items in the command range: 0, 1 or 2 <count> <count> Any count supplied (as described for the '-range' and '-count' attributes). <bang> <bang> (See the '-bang' attribute) Expands to a ! if the command was executed with a ! modifier, otherwise expands to nothing. <mods> <q-mods> :command-modifiers <mods> The command modifiers, if specified. Otherwise, expands to nothing. Supported modifiers are :aboveleft, :belowright, :botright, :browse, :confirm, :hide, :horizontal, :keepalt, :keepjumps, :keepmarks, :keeppatterns, :leftabove, :lockmarks, :noautocmd, :noswapfile :rightbelow, :sandbox, :silent, :tab, :topleft, :unsilent, :verbose, and :vertical. Note that :filter is not supported. Examples:
command! -nargs=+ -complete=file MyEdit
            \ for f in expand(<q-args>, 0, 1) |
            \ exe '<mods> split ' .. f |
            \ endfor
function! SpecialEdit(files, mods)
    for f in expand(a:files, 0, 1)
        exe a:mods .. ' split ' .. f
command! -nargs=+ -complete=file Sedit
            \ call SpecialEdit(<q-args>, <q-mods>)
<reg> <register> <reg> (See the '-register' attribute) The optional register, if specified. Otherwise, expands to nothing. <register> is a synonym for this. <args> <args> The command arguments, exactly as supplied (but as noted above, any count or register can consume some of the arguments, which are then not part of <args>). <lt> A single '<' (Less-Than) character. This is needed if you want to get a literal copy of one of these escape sequences into the expansion - for example, to get <bang>, use <lt>bang>.
<q-args> If the first two characters of an escape sequence are "q-" (for example, <q-args>) then the value is quoted in such a way as to make it a valid value for use in an expression. This uses the argument as one single value. When there is no argument <q-args> is an empty string. See the q-args-example below. <f-args> To allow commands to pass their arguments on to a user-defined function, there is a special form <f-args> ("function args"). This splits the command arguments at spaces and tabs, quotes each argument individually, and the <f-args> sequence is replaced by the comma-separated list of quoted arguments. See the Mycmd example below. If no arguments are given <f-args> is removed. To embed whitespace into an argument of <f-args>, prepend a backslash. <f-args> replaces every pair of backslashes (\\) with one backslash. A backslash followed by a character other than white space or a backslash remains unmodified. Also see f-args-example below. Overview:
XX ab "ab" XX a\b 'a\b' XX a\ b 'a b' XX a\ b 'a ', 'b' XX a\\b 'a\b' XX a\\ b 'a\', 'b' XX a\\\b 'a\\b' XX a\\\ b 'a\ b' XX a\\\\b 'a\\b' XX a\\\\ b 'a\\', 'b' XX [nothing]
Note that if the "no arguments" situation is to be handled, you have to make sure that the function can be called without arguments.
Examples for user commands:
" Delete everything after here to the end
:com Ddel +,$d
" Rename the current buffer
:com -nargs=1 -bang -complete=file Ren f <args>|w<bang>
" Replace a range with the contents of a file
" (Enter this all as one line)
:com -range -nargs=1 -complete=file
      Replace <line1>-pu_|<line1>,<line2>d|r <args>|<line1>d
" Count the number of lines in the range
:com! -range -nargs=0 Lines  echo <line2> - <line1> + 1 "lines"
f-args-example Call a user function (example of <f-args>)
:com -nargs=* Mycmd call Myfunc(<f-args>)
When executed as:
:Mycmd arg1 arg2
This will invoke:
:call Myfunc("arg1","arg2")
q-args-example A more substantial example:
:function Allargs(command)
:   let i = 0
:   while i < argc()
:          if filereadable(argv(i))
:             execute "e " .. argv(i)
:             execute a:command
:      endif
:      let i = i + 1
:   endwhile
:command -nargs=+ -complete=command Allargs call Allargs(<q-args>)
The command Allargs takes any Vim command(s) as argument and executes it on all files in the argument list. Usage example (note use of the "e" flag to ignore errors and the "update" command to write modified buffers):
:Allargs %s/foo/bar/ge|update
This will invoke:
:call Allargs("%s/foo/bar/ge|update")
Commands index
Quick reference