path: root/runtime/doc/map.txt

*map.txt*       For Vim version 7.2.  Last change: 2008 Aug 09


		  VIM REFERENCE MANUAL    by Bram Moolenaar


Key mapping, abbreviations and user-defined commands.

This subject is introduced in sections |05.3|, |24.7| and |40.1| of the user
manual.

1. Key mapping			|key-mapping|
   1.1 MAP COMMANDS			|:map-commands|
   1.2 Special arguments		|:map-arguments|
   1.3 Mapping and modes		|:map-modes|
   1.4 Listing mappings			|map-listing|
   1.5 Mapping special keys		|:map-special-keys|
   1.6 Special characters		|:map-special-chars|
   1.7 What keys to map			|map-which-keys|
   1.8 Examples				|map-examples|
   1.9 Using mappings			|map-typing|
   1.10 Mapping alt-keys		|:map-alt-keys|
   1.11 Mapping an operator		|:map-operator|
2. Abbreviations		|abbreviations|
3. Local mappings and functions	|script-local|
4. User-defined commands	|user-commands|

==============================================================================
1. Key mapping				*key-mapping* *mapping* *macro*

Key mapping is used to change the meaning of typed keys.  The most common use
is to define a sequence 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 |<>|).


1.1 MAP COMMANDS					*:map-commands*

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* *:lmap*
:cm[ap]	{lhs} {rhs}		|mapmode-c|		*:cm* *:cmap*
			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.


:no[remap]  {lhs} {rhs}		|mapmode-nvo|		*:no*  *:noremap*
: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* *:snoremap*
:ono[remap] {lhs} {rhs}		|mapmode-o|		*:ono* *:onoremap*
:no[remap]! {lhs} {rhs}		|mapmode-ic|		*:no!* *:noremap!*
:ino[remap] {lhs} {rhs}		|mapmode-i|		*:ino* *:inoremap*
:ln[oremap] {lhs} {rhs}		|mapmode-l|		*:ln*  *:lnoremap*
:cno[remap] {lhs} {rhs}		|mapmode-c|		*:cno* *:cnoremap*
			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.  {not in Vi}


: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*   *:cunmap*
			Remove the mapping of {lhs} for the modes where the
			map command applies.  The mapping may remain defined
			for other modes where it applies.
			Note: Trailing spaces are included in the {lhs}.  This
			unmap does NOT work: >
				:map @@ foo
				:unmap @@ | print

: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*
			Remove ALL mappings for the modes where the map
			command applies.  {not in Vi}
			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|
			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*
			List the key mappings for the key sequences starting
			with {lhs} in the modes where the map command applies.
			{not in Vi}

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.
{Vi does not allow ambiguous mappings}


1.2 SPECIAL ARGUMENTS					*:map-arguments*

"<buffer>", "<silent>", "<special>", "<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>* *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.
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.

						*: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>
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-<special>* *:map-special*
Define a mapping with <> notation for special keys, even though the "<" flag
may appear in 'cpoptions'.  This is useful if the side effect of setting
'cpoptions' is not desired.  Example: >
	:map <special> <F12> /Header<CR>
<
						*: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>* *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> . InsertDot()
The result of the InsertDot() function will be inserted.  It could check the
text before the cursor and start omni completion when some condition is met.

Be very careful about side effects!  The expression is evaluated while
obtaining characters, you may very well make the command dysfunctional.
For this reason the following is blocked:
- Changing the buffer text |textlock|.
- Editing another buffer.
- The |:normal| command.
- Moving the cursor is allowed, but it is restored afterwards.
- You can use getchar(), but the existing typeahead isn't seen and new
  typeahead is discarded.
If you want the mapping to do any of these let the returned characters do
that.

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 . '. '
	endfunc

	func ListReset()
	  let g:counter = 0
	  return ''
	endfunc

CTRL-L inserts the next number, CTRL-R resets the count.  CTRL-R returns an
empty string, so that nothing is inserted.

Note that there are some tricks to make special keys work and escape CSI bytes
in the text.  The |:map| command also does this, thus you must avoid that it
is done twice.  This does not work: >
	:imap <expr> <F3> "<Char-0x611B>"
Because the <Char- sequence is escaped for being a |:imap| argument and then
again for using <expr>.  This does work: >
	:imap <expr> <F3> "\u611B"
Using 0x80 as a single byte before other text does not work, it will be seen
as a special key.


1.3 MAPPING AND MODES					*:map-modes*
			*mapmode-nvo* *mapmode-n* *mapmode-v* *mapmode-o*

There are five sets of mappings
- For Normal mode: When typing commands.
- For Visual mode: When typing commands while the Visual area is highlighted.
- 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.

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:

    commands:				      modes: ~
				       Normal  Visual+Select  Operator-pending ~
: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|

    commands:				      modes: ~
					  Visual    Select ~
: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:

    commands:				      modes: ~
					  Insert  Command-line	Lang-Arg ~
:map!  :noremap!  :unmap!  :mapclear!	    yes	       yes	   -
:imap  :inoremap  :iunmap  :imapclear	    yes		-	   -
:cmap  :cnoremap  :cunmap  :cmapclear	     -	       yes	   -
:lmap  :lnoremap  :lunmap  :lmapclear	    yes*       yes*	  yes*

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-^|.  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.


1.4 LISTING MAPPINGS					*map-listing*

When listing mappings the characters in the first two columns are:

      CHAR	MODE	~
     <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

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 |'<|.

							*: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 /home/abcd/.vimrc

See |:verbose-cmd| for more information.


1.5 MAPPING SPECIAL KEYS				*:map-special-keys*

There are three ways to map a special key:
1. The Vi-compatible method: Map the key code.  Often this is a sequence that
   starts with <Esc>.  To enter a mapping like this you type ":map " and then
   you have to type CTRL-V before hitting the function key.  Note that when
   the key code for the key is in the termcap (the t_ options), it will
   automatically be translated into the internal code and become the second
   way of mapping (unless the 'k' flag is included in 'cpoptions').
2. The second method is to use the internal code for the function key.  To
   enter such a mapping type CTRL-K and then hit the function key, or use
   the form "#1", "#2", .. "#9", "#0", "<Up>", "<S-Down>", "<S-F7>", etc.
   (see table of keys |key-notation|, all keys from <Up> can be used).  The
   first ten function keys can be defined in two ways: Just the number, like
   "#2", and with "<F>", like "<F2>".  Both stand for function key 2.  "#0"
   refers to function key 10, defined with option 't_f10', which may be
   function key zero on some keyboards.  The <> form cannot be used when
   'cpoptions' includes the '<' flag.
3. Use the termcap entry, with the form <t_xx>, where "xx" is the name of the
   termcap entry.  Any string entry can be used.  For example: >
     :map <t_F3> G
<  Maps function key 13 to "G".  This does not work if 'cpoptions' includes
   the '<' flag.

The advantage of the second and third method is that the mapping will work on
different terminals without modification (the function key will be
translated into the same internal code or the actual key code, no matter what
terminal you are using.  The termcap must be correct for this to work, and you
must use the same mappings).

DETAIL: Vim first checks if a sequence from the keyboard is mapped.  If it
isn't the terminal key codes are tried (see |terminal-options|).  If a
terminal code is found it is replaced with the internal code.  Then the check
for a mapping is done again (so you can map an internal code to something
else).  What is written into the script file depends on what is recognized.
If the terminal key code was recognized as a mapping the key code itself is
written to the script file.  If it was recognized as a terminal code the
internal code is written to the script file.


1.6 SPECIAL CHARACTERS					*:map-special-chars*
							*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*
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*
To include a space in {lhs} precede it with a CTRL-V (type two CTRL-Vs for
each space).
							*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*
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>*
A easier way to get a mapping that doesn't produce anything, is to use "<Nop>"
for the {rhs}.  This only works when the |<>| notation is enabled.  For
example, to make sure that function key 8 does nothing at all: >
	: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,
otherwise it would be impossible to type the á character.

					*<Leader>* *mapleader*
To define a mapping which uses the "mapleader" variable, the special string
"<Leader>" can be used.  It is replaced with the string value of "mapleader".
If "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 "mapleader" is used at the moment the mapping is
defined.  Changing "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 <LocalLeader>q  \DoItNow
<
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>|.

							*<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
This is useful to specify a (multi-byte) 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}.

							*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:
   use	     works when			   example	~
   <Bar>     '<' is not in 'cpoptions'	   :map _l :!ls <Bar> more^M
   \|	     'b' is not in 'cpoptions'	   :map _l :!ls \| more^M
   ^V|	     always, in Vim and Vi	   :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*
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 %<CR>:echo "the end"<CR>

To avoid mapping of the characters you type in insert or Command-line mode,
type a CTRL-V first.  The mapping in Insert mode is disabled if the 'paste'
option is on.

Note that when an error is encountered (that causes an error message or 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.


1.7 WHAT KEYS TO MAP					*map-which-keys*

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). |: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.

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).


1.8 EXAMPLES						*map-examples*

A few examples (given as you type them, for "<CR>" you type four characters;
the '<' flag must not be present in 'cpoptions' for this to work). >

   :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. |@=|


1.9 USING MAPPINGS					*map-typing*

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-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.

Another problem may be that when keeping ALT or Meta pressed the terminal
prepends ESC instead of setting the 8th bit.  See |:map-alt-keys|.

						*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, when the 'remap' option is on, 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).


1.10 MAPPING ALT-KEYS					*:map-alt-keys*

In the GUI Vim handles the Alt key itself, thus mapping keys with ALT should
always work.  But in a terminal Vim gets a sequence of bytes and has to figure
out whether ALT was pressed or not.

By default Vim assumes that pressing the ALT key sets the 8th bit of a typed
character.  Most decent terminals can work that way, such as xterm, aterm and
rxvt.  If your <A-k> mappings don't work it might be that the terminal is
prefixing the character with an ESC character.  But you can just as well type
ESC before a character, thus Vim doesn't know what happened (except for
checking the delay between characters, which is not reliable).

As of this writing, some mainstream terminals like gnome-terminal and konsole
use the ESC prefix.  There doesn't appear a way to have them use the 8th bit
instead.  Xterm should work well by default.  Aterm and rxvt should work well
when started with the "--meta8" argument.  You can also tweak resources like
"metaSendsEscape", "eightBitInput" and "eightBitOutput".

On the Linux console, this behavior can be toggled with the "setmetamode"
command.  Bear in mind that not using an ESC prefix could get you in trouble
with other programs.  You should make sure that bash has the "convert-meta"
option set to "on" in order for your Meta keybindings to still work on it
(it's the default readline behavior, unless changed by specific system
configuration).  For that, you can add the line: >

	set convert-meta on

to your ~/.inputrc file. If you're creating the file, you might want to use: >

	$include /etc/inputrc

as the first line, if that file exists on your system, to keep global options.
This may cause a problem for entering special characters, such as the umlaut.
Then you should use CTRL-V before that character.

Bear in mind that convert-meta has been reported to have troubles when used in
UTF-8 locales.  On terminals like xterm, the "metaSendsEscape" resource can be
toggled on the fly through the "Main Options" menu, by pressing Ctrl-LeftClick
on the terminal; that's a good last resource in case you want to send ESC when
using other applications but not when inside VIM.


1.11 MAPPING AN OPERATOR				*:map-operator*

An operator is used before a {motion} command.  To define your own operator
you must create 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 |characterwise|
			    "block"	{motion} was |blockwise-visual||
			Although "block" would rarely appear, since it can
			only result from Visual mode where "g@" is not useful.
			{not available when compiled without the +eval
			feature}

Here is an example that counts the number of spaces with <F4>: >

	nmap <silent> <F4> :set opfunc=CountSpaces<CR>g@
	vmap <silent> <F4> :<C-U>call CountSpaces(visualmode(), 1)<CR>

	function! CountSpaces(type, ...)
	  let sel_save = &selection
	  let &selection = "inclusive"
	  let reg_save = @@

	  if a:0  " Invoked from Visual mode, use '< and '> marks.
	    silent exe "normal! `<" . a:type . "`>y"
	  elseif a:type == 'line'
	    silent exe "normal! '[V']y"
	  elseif a:type == 'block'
	    silent exe "normal! `[\<C-V>`]y"
	  else
	    silent exe "normal! `[v`]y"
	  endif

	  echomsg strlen(substitute(@@, '[^ ]', '', 'g'))

	  let &selection = sel_save
	  let @@ = reg_save
	endfunction

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 there is a separate mapping for Visual mode.  It removes the
"'<,'>" range that ":" inserts in Visual mode and invokes the function with
visualmode() and an extra argument.

==============================================================================
2. Abbreviations			*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.  {this type
	  is not supported by Vi}

	  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.

Example: >
   :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.

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 part of the abbreviation, exit
insert mode with <Esc>, re-enter insert mode with "a" and type the rest.  Or
type CTRL-V before the character after the abbreviation.
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>
This does not work if 'cpoptions' includes the '<' flag. |<>|

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
   endfunc
   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.  {some versions of Vi support
recursive abbreviations, for no apparent reason}

Abbreviations are disabled if the 'paste' option is on.

				*: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>] {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.

						*:una* *:unabbreviate*
:una[bbreviate] {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>] [lhs] [rhs]
			same as ":ab", but no remapping for this {rhs} {not
			in Vi}

						*:ca* *:cabbrev*
:ca[bbrev] [<expr>] [lhs] [rhs]
			same as ":ab", but for Command-line mode only.  {not
			in Vi}

						*:cuna* *:cunabbrev*
:cuna[bbrev] {lhs}	same as ":una", but for Command-line mode only.  {not
			in Vi}

						*:cnorea* *:cnoreabbrev*
:cnorea[bbrev] [<expr>] [lhs] [rhs]
			same as ":ab", but for Command-line mode only and no
			remapping for this {rhs} {not in Vi}

						*:ia* *:iabbrev*
:ia[bbrev] [<expr>] [lhs] [rhs]
			same as ":ab", but for Insert mode only.  {not in Vi}

						*:iuna* *:iunabbrev*
:iuna[bbrev] {lhs}	same as ":una", but for insert mode only.  {not in
			Vi}

						*:inorea* *:inoreabbrev*
:inorea[bbrev] [<expr>] [lhs] [rhs]
			same as ":ab", but for Insert mode only and no
			remapping for this {rhs} {not in Vi}

							*:abc* *:abclear*
:abc[lear]		Remove all abbreviations.  {not in Vi}

							*:iabc* *:iabclear*
:iabc[lear]		Remove all abbreviations for Insert mode.  {not in Vi}

							*:cabc* *:cabclear*
:cabc[lear]		Remove all abbreviations for Command-line mode.  {not
			in Vi}

							*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 .exrc 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]

==============================================================================
3. Local mappings and functions				*script-local*

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 requires that the
'<' flag is not present in 'cpoptions'.
   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
could 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.

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: >
	function s:SID()
	  return matchstr(expand('<sfile>'), '<SNR>\zs\d\+\ze_SID$')
	endfun

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.

This is all {not in Vi} and {not available when compiled without the +eval
feature}.

==============================================================================
4. User-defined commands				*user-commands*

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* *user-cmd-ambiguous*
All user defined commands must start with an uppercase letter, to avoid
confusion with builtin commands.  (There are a few builtin commands, notably
:Next, :Print and :X, which do start with an uppercase letter.  The builtin
will always take precedence in these cases).  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.

Example: >
	:command Rename ...
	:command Renumber ...
	:Rena				" Means "Rename"
	:Renu				" Means "Renumber"
	:Ren				" Error - ambiguous
	:command Paste ...
	:P				" The built-in :Print

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 two columns are
			    !	Command has the -bang attribute
			    "	Command has the -register attribute
			    b	Command is local to current buffer
			(see below for details on attributes)

: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. Example: >

    :verbose command TOhtml
<	Name	    Args Range Complete  Definition ~
	TOhtml	    0	 %		 :call Convert2HTML(<line1>, <line2>) ~
	    Last set from /usr/share/vim/vim-7.0/plugin/tohtml.vim ~

See |:verbose-cmd| for more information.

							*E174* *E182*
:com[mand][!] [{attr}...] {cmd} {rep}
			Define a user command.  The name of the command is
			{cmd} and its replacement text is {rep}.  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.

:delc[ommand] {cmd}				*:delc* *:delcommand* *E184*
			Delete the user-defined command {cmd}.

:comc[lear]						*:comc* *:comclear*
			Delete all user-defined commands.

Command attributes

User-defined commands are treated by Vim 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.

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.

Argument handling				*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
	-nargs=*    Any number of arguments are allowed (0, 1, or many)
	-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.

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>
<   script2.vim: >
	: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.

Completion behavior				*: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=augroup	autocmd groups
	-complete=buffer	buffer names
	-complete=command	Ex command (and arguments)
	-complete=dir		directory names
	-complete=environment	environment variable names
	-complete=event		autocommand events
	-complete=expression	Vim expression
	-complete=file		file and directory names
	-complete=shellcmd	Shell command
	-complete=function	function name
	-complete=help		help subjects
	-complete=highlight	highlight groups
	-complete=mapping	mapping name
	-complete=menu		menus
	-complete=option	options
	-complete=tag		tags
	-complete=tag_listfiles	tags, file names are shown when CTRL-D is hit
	-complete=var		user variables
	-complete=custom,{func} custom completion, defined via {func}
	-complete=customlist,{func} custom completion, defined via {func}


Custom completion			*: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 do filter the candidates with its regexp engine
after function return, and this is probably more efficient in most cases. 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")
    :endfun

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:ArgLead), "\n")
    :endfun
<

Range handling				*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|)
	-count=N    A count (default N) which is specified either in the line
		    number position, or as an initial argument (like |:Next|).
		    Specifying -count (without a default) acts like -count=0

Note that -range=N and -count=N are mutually exclusive - only one should be
specified.

Special cases				*:command-bang* *:command-bar*
					*:command-register* *:command-buffer*
There are some special cases as well:

	-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.

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.

Replacement text

The replacement text 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.
						*<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.
						*<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.
							*<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.  Overview:

	command		   <f-args> ~
	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'

Examples >

   " 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"

   " 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")

   :" 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
   :endfunction
   :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")
<
When defining an 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.

 vim:tw=78:ts=8:ft=help:norl: