Vim inspired — Faster as in fewer keystrokes — Multiple selections — Orthogonal design
git clone http://github.com/mawww/kakoune.git cd kakoune/src make ./kak
See http://github.com/mawww/golf for kakoune solutions to vimgolf challenges, regularly beating the best vim solution.
See the design document for more information on Kakoune philosophy and design.
Kakoune is a code editor heavily inspired by Vim, as such most of its commands are similar to vi’s ones, and it shares Vi’s "keystrokes as a text editing language" model.
Kakoune can operate in two modes, normal and insertion. In insertion mode, keys are directly inserted into the current buffer. In normal mode, keys are used to manipulate the current selection and to enter insertion mode.
Kakoune has a strong focus on interactivity, most commands provide immediate and incremental results, while still being competitive (as in keystroke count) with Vim.
Kakoune works on selections, which are oriented, inclusive range of characters, selections have an anchor and a cursor character. Most commands move both of them, except when extending selection where the anchor character stays fixed and the cursor one moves around.
Join us on freenode IRC #Kakoune
-
Multiple selections as a central way of interacting
-
Powerful selection manipulation primitives
-
Select all regex matches in current selections
-
Keep selections containing/not containing a match for a given regex
-
Split current selections with a regex
-
Text objects (paragraph, sentence, nestable blocks)
-
-
Powerful text manipulation primitives
-
Align selections
-
Rotate selection contents
-
Case manipulation
-
Indentation
-
Piping each selection to external filter
-
-
Client-Server architecture
-
Multiple clients on the same editing session
-
Use tmux or your X11 window manager to manage windows
-
-
Simple interaction with external programs
-
Automatic contextual help
-
Automatic as you type completion
-
Macros
-
Hooks
-
Syntax Highlighting
-
Supports multiple languages in the same buffer
-
Highlight a buffer differently in different windows
-
Kakoune dependencies are:
-
A C++14 compliant compiler (GCC >= 5 or clang >= 3.6) along with its associated C++ standard library (libstdc++ or libc++)
-
boost (>= 1.50)
-
ncurses with wide-characters support (>= 5.3, generally referred to as libncursesw)
-
asciidoc (for the
a2k
tool), to generate man pages
To build, just type make in the src directory. To generate man pages, type make doc in the src directory.
Kakoune can be built on Linux, MacOS, and Cygwin. Due to Kakoune relying heavily on being in a Unix-like environment, no native Windows version is planned.
In order to install kak on your system, rather than running it directly from
its source directory, type make install, you can specify the PREFIX
and
DESTDIR
if needed.
Tip
|
Homebrew (OSX)
brew install ncurses Then, to install, brew install --HEAD https://raw.githubusercontent.com/mawww/kakoune/master/contrib/kakoune.rb To update kakoune, brew upgrade --fetch-HEAD kakoune |
Tip
|
Fedora 22/23/24/Rawhide
Use the copr repository. dnf copr enable jkonecny/kakoune dnf install kakoune |
Tip
|
Arch Linux
A PKGBUILD kakoune-git to install Kakoune is available in the AUR. # For example build and install Kakoune via yaourt yaourt -Sy kakoune-git |
Tip
|
Gentoo
Kakoune is found in portage as app-editors/kakoune |
Tip
|
Exherbo
cave resolve -x repository/mawww cave resolve -x kakoune |
Tip
|
openSUSE
kakoune can be found in the editors devel project. Make sure to adjust the link below to point to the repository of your openSUSE version. #Example for Tumbleweed: sudo zypper addrepo http://download.opensuse.org/repositories/editors/openSUSE_Factory/editors.repo sudo zypper refresh sudo zypper install kakoune |
Just running kak launch a new kak session with a client on local terminal. kak accepts some switches:
-
-c <session>
: connect to given session, sessions are unix sockets/tmp/kakoune/<user>/<session>
,<user>/<session>
can be used as well to connect to another user’s session, provided the socket permissions have been changed to allow it. -
-e <commands>
: execute commands on startup -
-n
: ignore kakrc file -
-s <session>
: set the session name, by default it will be the pid of the initial kak process. -
-d
: run Kakoune in daemon mode, without user interface. This requires the session name to be specified with -s. In this mode, the Kakoune server will keep running even if there is no connected client, and will quit when receiving SIGTERM. -
-p <session>
: read stdin, and then send its content to the given session acting as a remote control. -
-f <keys>
: Work as a filter, read every file given on the command line and stdin if piped in, and apply given keys on each. -
-ui <userinterface>
: use given user interface,<userinterface>
can be-
ncurses
: default terminal user interface -
dummy
: empty user interface not displaying anything -
json
: json-rpc based user interface that writes json on stdout and read keystrokes as json on stdin.
-
-
-l
: list existing sessions, and check the dead ones -
-clear
: clear dead session’s socket files -
-ro
: prevent modifications to all buffers from being saved to disk -
+line[:column]
: specify a target line and column for the first file
There are two directories containing Kakoune’s scripts:
-
runtime
: located in../share/kak/
relative to thekak
binary contains the system scripts, installed with Kakoune. -
userconf
: located in$XDG_CONFIG_HOME/kak/
, which defaults to$HOME/.config/kak/
on most systems, containing the user configuration.
Unless -n
is specified, Kakoune will load its startup script located
at ${runtime}/kakrc
relative to the kak
binary. This startup script
is responsible for loading the user configuration.
First, Kakoune will search recursively for .kak
files in the autoload
directory. It will first look for an autoload
directory at
${userconf}/autoload
and will fallback to ${runtime}/autoload
if
it does not exist.
Once all those files are loaded, Kakoune will try to source
${runtime}/kakrc.local
which is expected to contain distribution provided
configuration.
And finally, the user configuration will be loaded from ${userconf}/kakrc
.
Note
|
If you create a user autoload directory in ${userconf}/autoload ,
the system one at ${runtime}/autoload will not be loaded anymore. You can
add a symbolic link to it (or to individual scripts) inside
${userconf}/autoload to keep loading system scripts.
|
The main concept in Kakoune is the selection. A selection is an inclusive, directed range of character. A selection has two ends, the anchor and the cursor.
There is always at least one selection, and a selection is always at least one character (in which case the anchor and cursor of the selections are on the same character).
In normal mode, keys are not inserted directly inside the buffer, but are editing commands. These commands provide ways to manipulate either the selections themselves, or the selected text.
When entering insert mode, keys are now directly inserted before each selection’s cursor. Some additional keys are recognised in insert mode:
-
<esc>
: leave insert mode -
<backspace>
: delete characters before cursors -
<del>
: delete characters under cursors -
<left>, <right>, <up>, <down>
: move the cursors in given direction -
<home>
: move cursors to line begin -
<end>
: move cursors to end of line -
<c-n>
: select next completion candidate -
<c-p>
: select previous completion candidate -
<c-x>
: explicit insert completion query, followed by:-
f
: explicit file completion -
w
: explicit word completion -
l
: explicit line completion
-
-
<c-o>
: disable automatic completion for this insert session -
<c-r>
: insert contents of the register given by next key -
<c-v>
: insert next keystroke directly into the buffer, without interpreting it. -
<c-u>
: commit changes up to now as a single undo group. -
<a-;>
: escape to normal mode for a single command
-
h
: select the character on the left of selection end -
j
: select the character below the selection end -
k
: select the character above the selection end -
l
: select the character on the right of selection end -
w
: select the word and following whitespaces on the right of selection end -
b
: select preceding whitespaces and the word on the left of selection end -
e
: select preceding whitespaces and the word on the right of selection end -
<a-[wbe]>
: same as [wbe] but select WORD instead of word -
f
: select to the next occurence of given character -
t
: select until the next occurence of given character -
<a-[ft]>
: same as [ft] but in the other direction -
m
: select to matching character -
M
: extend selection to matching character -
x
: select line on which selection end lies (or next line when end lies on an end-of-line) -
X
: similar tox
, except the current selection is extended -
<a-x>
: expand selections to contain full lines (including end-of-lines) -
<a-X>
: trim selections to only contain full lines (not including last end-of-line) -
%
: select whole buffer -
<a-h>
: select to line begin -
<a-l>
: select to line end -
/
: search (select next match) -
<a-/>
: search (select previous match) -
?
: search (extend to next match) -
<a-?>
: search (extend to previous match) -
n
: select next match -
N
: add a new selection with next match -
<a-n>
: select previous match -
<a-N>
: add a new selection with previous match -
pageup
: scroll up -
pagedown
: scroll down -
'
: rotate selections (the main selection becomes the next one) -
<a-'>
: rotate selections backwards -
;
: reduce selections to their cursor -
<a-;>
: flip the selections' direction -
<a-:>
: ensure selections are in forward direction (cursor after anchor) -
<a-.>
: repeat last object orf
/t
selection command.
A word is a sequence of alphanumeric characters or underscore, a WORD is a sequence of non whitespace characters.
for most selection commands, using shift permits extending current selection
instead of replacing it. for example, wWW
selects 3 consecutive words
Most selection commands also support counts, which are entered before the command itself.
For example, 3W
selects 3 consecutive words and 3w
select the third word on
the right of selection end.
Any normal mode command can be prefixed with \
which will disable hook execution
for the duration for the command (including the duration of modes the command could
move to, so \i
will disable hooks for the whole insert session).
As autoindentation is implemented in terms of hooks, this can be used to disable it when pasting text.
-
i
: enter insert mode before current selection -
a
: enter insert mode after current selection -
d
: yank and delete current selection -
c
: yank and delete current selection and enter insert mode -
.
: repeat last insert mode change (i
,a
, orc
, including the inserted text) -
<a-d>
: delete current selection -
<a-c>
: delete current selection and enter insert mode -
I
: enter insert mode at current selection begin line start -
A
: enter insert mode at current selection end line end -
o
: enter insert mode in one (or given count) new lines below current selection end -
O
: enter insert mode in one (or given count) new lines above current selection begin -
<a-o>
: add an empty line below cursor -
<a-O>
: add an empty line above cursor -
y
: yank selections -
p
: paste after current selection end -
P
: paste before current selection begin -
<a-p>
: paste all after current selection end, and select each pasted string. -
<a-P>
: paste all before current selection begin, and select each pasted string. -
R
: replace current selection with yanked text -
<a-R>
: replace current selection with every yanked text -
r
: replace each character with the next entered one -
<a-j>
: join selected lines -
<a-J>
: join selected lines and select spaces inserted in place of line breaks -
<a-m>
: merge contiguous selections together (works across lines as well) -
<gt> (>)
: indent selected lines -
<a-gt>
: indent selected lines, including empty lines -
<lt> (<)
: deindent selected lines -
<a-lt>
: deindent selected lines, do not remove incomplete indent (3 leading spaces when indent is 4) -
|
: pipe each selection through the given external filter program and replace the selection with it’s output. -
<a-|>
: pipe each selection through the given external filter program and ignore its output -
!
: insert command output before selection -
<a-!>
: append command output after selection -
u
: undo last change -
<a-u>
: move backward in history -
U
: redo last change -
<a-U>
: move forward in history -
&
: align selection, align the cursor of selections by inserting spaces before the first character of the selection -
<a-&>
: copy indent, copy the indentation of the main selection (or the count one if a count is given) to all other ones -
`
: to lower case -
~
: to upper case -
<a-
>`: swap case -
@
: convert tabs to spaces in current selections, uses the buffer tabstop option or the count parameter for tabstop. -
<a-@>
: convert spaces to tabs in current selections, uses the buffer tabstop option or the count parameter for tabstop. -
<a-">
: rotate selections content, if specified, the count groups selections, so3<a-">
rotate (1, 2, 3) and (3, 4, 6) independently.
Commands beginning with g are used to goto certain position and or buffer:
-
gh
: select to line begin -
gl
: select to line end -
gi
: select to line begin (non blank) -
gg
,gk
: go to the first line -
gj
: go to the last line -
ge
: go to last char of last line -
gt
: go to the first displayed line -
gc
: go to the middle displayed line -
gb
: go to the last displayed line -
ga
: go to the previous (alternate) buffer -
gf
: open the file whose name is selected -
g.
: go to last buffer modification position
If a count is given prior to hitting g
, g
will jump to the given line.
Using G
will extend the selection rather than jump.
Some commands, all beginning with v permit to manipulate the current view.
-
vv
orvc
: center the main selection in the window -
vt
: scroll to put the main selection on the top line of the window -
vb
: scroll to put the main selection on the bottom line of the window -
vh
: scroll the window count columns left -
vj
: scroll the window count line downward -
vk
: scroll the window count line upward -
vl
: scroll the window count columns right
Using V
will lock view mode until <esc>
is hit
Current selections position can be saved in a register and restored later on.
By default, marks use the '^' register, but using the register can be set
using "<reg>
prefix.
Z
will save the current selections to the register.
<a-Z>
will combine the current selections to the register.
z
will restore the selections from the register.
<a-z>
will combine the selections from the register with the existing ones.
When combining selections, kakoune will prompt for a combining mode:
+
will append selections from both lists into a single list
<
will select the selection with the leftmost cursor for each pair
>
will select the selection with the rightmost cursor for each pair
Some commands, like the goto commands, buffer switch or search commands, push the previous selections to the client’s jump list. It is possible to forward or backward in the jump list using:
-
<c-i>
: Jump forward -
<c-o>
: Jump backward -
<c-s>
: save current selections
Kak was designed from the start to handle multiple selections.
One way to get a multiselection is via the s
key.
For example, to change all occurrences of word 'roger' to word 'marcel' in a paragraph, here is what can be done:
-
select the paragraph with enough
x
-
press
s
and enter roger, then enter -
now paragraph selection was replaced with multiselection of each roger in the paragraph
-
press
c
and marcel<esc> to replace rogers with marcels
A multiselection can also be obtained with S
, which splits the current
selection according to the regex entered. To split a comma separated list,
use S
then ', *'
The regex syntax supported by Kakoune is the Perl one and is described here Regex syntax.
s
and S
share the search pattern with /
, and hence entering an empty
pattern uses the last one.
As a convenience, <a-s>
allows you to split the current selections on
line boundaries.
To clear multiple selections, use space
. To keep only the nth selection
use n
followed by space
, in order to remove a selection, use <a-space>
.
<a-k>
allows you to enter a regex and keep only the selections that
contains a match for this regex. Using <a-K>
you can keep the selections
not containing a match.
C
copies the current selection to the next line (or lines if a count is given)
<a-C>
does the same to previous lines.
$
allows you to enter a shell command and pipe each selection to it.
Selections whose shell command returns 0 will be kept, other will be dropped.
Some keys allow you to select a text object:
-
<a-a>
: selects the whole object -
<a-i>
: selects the inner object, that is the object excluding its surrounder. For example, for a quoted string, this will not select the quote, and for a word this will not select trailing spaces. -
[
: selects to object start -
]
: selects to object end -
{
: extends selections to object start -
}
: extends selections to object end
After this key, you need to enter a second key in order to specify which object you want.
-
b
,(
or)
: select the enclosing parenthesis -
B
,{
or}
: select the enclosing {} block -
r
,[
or]
: select the enclosing [] block -
a
,<
or>
: select the enclosing <> block -
"
orQ
: select the enclosing double quoted string -
'
orq
: select the enclosing single quoted string -
`
org
: select the enclosing grave quoted string -
w
: select the whole word -
W
: select the whole WORD -
s
: select the sentence -
p
: select the paragraph -
␣
: select the whitespaces -
i
: select the current indentation block -
n
: select the number -
u
: select the argument -
:
: select user defined object, will prompt for open and close text.
For nestable objects, a count can be used in order to specify which surrounding level to select.
When pressing :
in normal mode, Kakoune will open a prompt to enter a command.
Commands are used for non editing tasks, such as opening a buffer, writing the current one, quitting, etc.
A few keys are recognized by prompt mode to help edit a command:
-
<ret>
: validate prompt -
<esc>
: abandon without -
<left> or <a-h>
: move cursor to previous character -
<right> or <a-l>
: move cursor to previous character -
<home>
: move cursor to first character -
<end>
: move cursor past the last character -
<backspace> or <a-x>
: erase character before cursor -
<del> or <a-d>
: erase character under cursor -
<c-w>
: advance to next word begin -
<c-a-w>
: advance to next WORD begin -
<c-b>
: go back to previous word begin -
<c-a-b>
: go back to previous WORD begin -
<c-e>
: advance to next word end -
<c-a-e>
: advance to next word end -
<up> or <c-p>
: select previous entry in history -
<down> or <c-n>
: select next entry in history -
<tab>
: select next completion candidate -
<backtab>
: select previous completion candidate -
<c-r>
: insert then content of the register given by next key. -
<c-v>
: insert next keystroke without interpreting it -
<c-o>
: disable auto completion for this prompt
Commands starting with horizontal whitespace (e.g. a space) will not be saved in the command history.
Some commands take an exclamation mark (!
), which can be used to force
the execution of the command (i.e. to quit a modified buffer, the
command q!
has to be used).
-
cd [<directory>]
: change the current directory to<directory>
, or the home directory is unspecified -
doc <topic>
: display documentation about a topic. The completion list displays the available topics. -
e[dit][!] <filename> [<line> [<column>]]
: open buffer on file, go to given line and column. If file is already opened, just switch to this file. Use edit! to force reloading. -
w[rite][!] [<filename>]
: write buffer to <filename> or use its name if filename is not given. If the file is write-protected, its permissions are temporarily changed to allow saving the buffer and restored afterwards when the write! command is used. -
w[rite]a[ll]
: write all buffers that are associated to a file. -
q[uit][!] [<exit status>]
: exit Kakoune, use quit! to force quitting even if there is some unsaved buffers remaining. If specified, the client exit status will be set to <exit status>. -
w[a]q[!] [<exit status>]
: write the current buffer (or all buffers whenwaq
is used) and quit. If specified, the client exit status will be set to <exit status>. -
kill[!]
: terminate the current session, all the clients as well as the server, use kill! to ignore unsaved buffers -
b[uffer] <name>
: switch to buffer <name> -
b[uffer]n[ext]
: switch to the next buffer -
b[uffer]p[rev]
: switch to the previous buffer -
d[el]b[uf][!] [<name>]
: delete the buffer <name> -
source <filename>
: execute commands in <filename> -
colorscheme <name>
: load named colorscheme. -
rename-client <name>
: set current client name -
rename-buffer <name>
: set current buffer name -
rename-session <name>
: set current session name -
echo [options] <text>
: show <text> in status line, with the following options:-
-markup
: expand the markup strings in <text> -
-debug
: print the given text to the*debug*
buffer
-
-
nop
: does nothing, but as with every other commands, arguments may be evaluated. So nop can be used for example to execute a shell command while being sure that it’s output will not be interpreted by kak.:%sh{ echo echo tchou }
will echo tchou in Kakoune, whereas:nop %sh{ echo echo tchou }
will not, but both will execute the shell command.
Multiple commands can be separated either by new lines or by semicolons,
as such a semicolon must be escaped with \;
to be considered as a literal
semicolon argument.
When entering a command, parameters are separated by whitespace (shell like), if you want to give parameters with spaces, you should quote them.
Kakoune support three string syntax:
-
'strings'
: uninterpreted strings, you can use\'
to escape the separator, every other char is itself. -
"strings"
: expanded strings, % strings (see Expansions) contained are expended. Use \% to escape a % inside them, and \\ to escape a slash. -
%{strings}
: these strings are very useful when entering commands-
the
{
and}
delimiters are configurable: you can use any non alphanumeric character, e.g.%[string]
,%<string>
,%(string)
,%~string~
,%!string!
. -
if the character following the % is one of {[(<, then the closing one is the matching }])> and the delimiters are not escapable but are nestable. For example
%{ roger {}; }
is a valid string,%{ marcel \}
as well.
-
A special kind of %{strings}
can be used, with a type between
%
and the opening delimiter (which cannot be alphanumeric). These
strings are expanded according to their type.
For example %opt{autoinfo}
is of type 'opt'. 'opt' expansions are replaced
by the value of the given option (here autoinfo
).
Supported types are:
-
sh
: shell expansion, similar to posix shell $(…) construct, see Shell expansion for more details. -
reg
: register expansion, will be replaced by the content of the given register. -
opt
: option expansion, will be replaced with the value of the given option -
val
: value expansion, gives access to the environment variable available to the Shell expansion. Thekak_
prefix is not used there. -
arg
: argument expansion, gives access to the arguments of the current command, the content can be a number, or@
for all arguments.
For example, you can display last search pattern with
:echo %reg{/}
The %sh{…}
expansion replaces its content with the output of the shell
commands in it. It is similar to the shell $(…) syntax and is evaluated
only when needed.
For example: %sh{ ls }
is replaced with the output of the ls command.
Some of Kakoune state is available through environment variables:
-
kak_selection
: content of the main selection -
kak_selections
: content of the selection separated by colons, colons and backslashes in the selection contents are escaped with a backslash. -
kak_selection_desc
: range of the main selection, represented asanchor,cursor
; anchor and cursor are in this format:line.column
-
kak_selections_desc
: range of the selecations separated by colons -
kak_bufname
: name of the current buffer -
kak_buffile
: full path of the file or same askak_bufname
when there’s no associated file -
kak_buflist
: the current buffer list, each buffer separated by a colon -
kak_buf_line_count
: the current buffer line count -
kak_timestamp
: timestamp of the current buffer, the timestamp is an integer value which is incremented each time the buffer is modified. -
kak_history_id
: history id of the current buffer, the history id is an integer value which is used to reference a specific buffer version in the undo tree -
kak_runtime
: directory containing the kak binary -
kak_count
: count parameter passed to the command -
kak_opt_<name>
: value of option <name> -
kak_reg_<r>
: value of register <r> -
kak_session
: name of the current session -
kak_client
: name of current client -
kak_client_pid
: pid of the current client -
kak_modified
: buffer has modifications not saved -
kak_source
: path of the file currently getting executed (through the source command) -
kak_cursor_line
: line of the end of the main selection -
kak_cursor_column
: column of the end of the main selection (in byte) -
kak_cursor_char_column
: column of the end of the main selection (in character) -
kak_cursor_byte_offset
: offset of the main selection from the beginning of the buffer (in byte). -
kak_window_width
: width of the current kakoune window -
kak_window_height
: height of the current kakoune window -
kak_hook_param
: filtering text passed to the currently executing hook -
kak_hook_param_capture_N
: text captured by the hook filter regex capture N -
kak_client_env_<name>
: value of the <name> variable in the client environment. Example: $kak_client_env_SHELL is the SHELL variable
Note that in order to make only needed information available, Kakoune needs
to find the environment variable reference in the shell script executed.
Hence, %sh{ ./script.sh }
with script.sh
referencing an environment
variable will not work.
For example, you can print informations on the current file in the status line using:
:echo -- "%sh{ ls -l $kak_bufname }"
In certain context, kakoune can take a markup string, which is a string containing
formatting informations. In these strings, syntax {facename}
will enable the
face facename until another face gets activated (or the end of the string.
Literal {
shall be written \{
, and literal \
that precede a {
shall
be written \\
If not launched with the -n
switch, Kakoune will source the
../share/kak/kakrc
file relative to the kak
binary, which
will source additional files:
If the $XDG_CONFIG_HOME/kak/autoload
directory exists, load every
*.kak
files in it, and load recursively any subdirectory.
If it does not exist, falls back to the site wide autoload directory
in ../share/kak/autoload/
.
After that, if it exists, source the $XDG_CONFIG_HOME/kak/kakrc
file
which should be used for user configuration.
In order to continue autoloading site-wide files with a local autoload
directory, just add a symbolic link to ../share/kak/autoload/
into
your local autoload directory.
For user configuration, Kakoune supports options.
Options are typed, their type can be
-
int
: an integer number -
bool
: a boolean value,yes/true
orno/false
-
str
: a string, some freeform text -
coord
: a line,column pair (separated by comma) -
regex
: as a string but theset
commands will complain if the entered text is not a valid regex. -
{int,str}-list
: a list, elements are separated by a colon (:) if an element needs to contain a colon, it can be escaped with a backslash. -
range-specs
: a:
separated list of a pair of a buffer range (<begin line>.<begin column>,<end line>.<end column>
or<begin line>.<end line>+<length>
) and a string (separated by|
), except for the first element which is just the timestamp of the buffer. -
line-flags
: a:
separated list of a line number and a corresponding flag (<line>|<flag text>
), except for the first element which is just the timestamp of the buffer. -
completions
: a:
separated list of<text>|<docstring>|<menu text>
candidates, except for the first element which follows the<line>.<column>[+<length>]@<timestamp>
format to define where the completion apply in the buffer. Markup can be used in the menu text. -
enum(value1|value2|…)
: an enum, taking on of the given values -
flags(value1|value2|…)
: a set of flags, taking a combination of the given values joined by|
.
Options value can be changed using the set
commands:
:set [global,buffer,window] <option> <value> # buffer, window, or global scope
Option values can be different by scope, an option can have a global value, a buffer value and a window value. The effective value of an option depends on the current context. If we have a window in the context (interactive edition for example), then the window value (if any) is used, if not we try the buffer value (if we have a buffer in the context), and if not we use the global value.
That means that two windows on the same buffer can use different options (like different filetype, or different tabstop). However, some options might end up ignored if their scope is not in the command context:
Writing a file never uses the window options for example, so any
options related to writing won’t be taken into account if set in the
window scope (BOM
or eolformat
for example).
New options can be declared using the :decl
command:
:decl [-hidden] <type> <name> [<value>]
The -hidden
parameter makes the option invisible in completion, but
still modifiable.
Some options are built in Kakoune, and can be used to control its behaviour:
-
tabstop
int: width of a tab character. -
indentwidth
int: width (in spaces) used for indentation. 0 means a tab character. -
scrolloff
coord: number of lines,columns to keep visible around the cursor when scrolling. -
eolformat
enum(lf|crlf): the format of end of lines when writing a buffer, this is autodetected on load; values of this option assigned to thewindow
scope are ignored -
BOM
enum(none|utf8): define if the file should be written with a unicode byte order mark. Values of this option assigned to thewindow
scope are ignored -
readonly
bool: prevent modifications from being saved to disk, all buffers if set totrue
in theglobal
scope, or current buffer if set in thebuffer
scope; values of this option assigned to thewindow
scope are ignored -
incsearch
bool: execute search as it is typed -
aligntab
bool: use tabs for alignment command -
autoinfo
flags(command|onkey|normal): display automatic information box in the enabled contexts. -
autoshowcompl
bool: automatically display possible completions when editing a prompt. -
ignored_files
regex: filenames matching this regex won’t be considered as candidates on filename completion (except if the text being completed already matches it). -
disabled_hooks
regex: hooks whose group matches this regex won’t be executed. For example indentation hooks can be disabled with '.*-indent'. -
filetype
str: arbitrary string defining the type of the file filetype dependant actions should hook on this option changing for activation/deactivation. -
path
str-list: directories to search for gf command. -
completers
str-list: completion systems to use for insert mode completion. The given completers are tried in order until one generate some completion candidates. Existing completers are:-
word=all
orword=buffer
which complete using words in all buffers (word=all
) or only the current one (word=buffer
) -
filename
which tries to detect when a filename is being entered and provides completion based on local filesystem. -
option=<opt-name>
where <opt-name> is a completions option.
-
-
static_words
str-list: list of words that are always added to completion candidates when completing words in insert mode. -
extra_word_chars
codepoint-list: list of all additional codepoints that should be considered as word character for the purpose of insert mode completion. -
autoreload
enum(yes|no|ask): auto reload the buffers when an external modification is detected. -
debug
flags(hooks|shell|profile|keys|commands): dump various debug information in thedebug
buffer. -
idle_timeout
int: timeout, in milliseconds, with no user input that will trigger thePromptIdle
,InsertIdle
andNormalIdle
hooks. -
fs_checkout_timeout
int: timeout, in milliseconds, between checks in normal mode of modifications of the file associated with the current buffer on the filesystem. -
modelinefmt
string: A format string used to generate the mode line, that string is first expanded as a command line would be (expanding%…{…}
strings), then markup tags are applied (see Markup strings). Two special atom are available as markup:{{mode_info}}
with information about the current mode (exampleinsert 3 sel
), and{{context_info}}
with information such as if the file has been modified (with[+]
), or if it is new (with[new file]
). -
ui_options
: colon separated list of key=value pairs that are forwarded to the user interface implementation. The NCurses UI support the following options:-
ncurses_set_title
: ifyes
ortrue
, the terminal emulator title will be changed. -
ncurses_status_on_top
: ifyes
, ortrue
the status line will be placed at the top of the terminal rather than at the bottom. -
ncurses_assistant
: specify the nice assistant you get in info boxes, can be 'clippy' (the default), 'cat', 'dilbert' or 'none' -
ncurses_enable_mouse
: boolean option that enables mouse support -
ncurses_change_colors
: boolean option that can disable color palette changing if the terminfo enables it but the terminal does not support it. -
ncurses_wheel_down_button
andncurses_wheel_up_button
: specify which button send for wheel down/up events.
-
A Face refers how the specified text is displayed. A face has a foreground color, a background color, and some attributes.
Faces can be defined and modified with the face command:
:face <name> <facespec>
Any place requiring a face can take either a face name defined with the face
command or a direct face description (called facespec) with the following
syntax:
fg_color[,bg_color][+attributes]
fg_color and bg_color can be:
-
A named color:
black, red, green, yellow, blue, magenta, cyan, white
. -
default
, which keeps the existing color -
An rgb color:
rgb:RRGGBB
, with RRGGBB the hexadecimal value of the color.
Not specifying bg_color uses default
attributes
is a string of letters each defining an attribute:
-
u
: Underline -
r
: Reverse -
b
: Bold -
B
: Blink -
d
: Dim -
i
: Italic -
e
: Exclusive, override previous faces instead of merging with them
Using named faces instead of facespec permits to change the effective faces afterwards.
There are some builtins faces used by internal Kakoune functionalities:
-
Default
: default colors -
PrimarySelection
: main selection face for every selected character except the cursor -
SecondarySelection
: secondary selection face for every selected character except the cursor -
PrimaryCursor
: cursor of the primary selection -
SecondaryCursor
: cursor of the secondary selection -
LineNumbers
: face used by the number_lines highlighter -
LineNumberCursor
: face used to highlight the line number of the main selection -
LineNumbersWrapped
: face used to highlight the line number of wrapped lines -
MenuForeground
: face for the selected element in menus -
MenuBackground
: face for the not selected elements in menus -
Information
: face for the information windows and information messages -
Error
: face of error messages -
StatusLine
: face used for the status line -
StatusCursor
: face used for the status line cursor -
Prompt
: face used prompt displayed on the status line -
MatchingChar
: face used by the show_matching highlighter -
Search
: face used to highlight search results -
BufferPadding
: face applied on the characters that follow the last line of a buffer -
Whitespace
: face used by the show_whitespaces highlighter
Registers are named lists of text. They are used for various purposes, like storing the last yanked text, or the captured groups associated with the selections.
Yanking and pasting uses the register "
, however most commands using a register
can have their default register overridden by using the "
key followed by the
register. For example "sy
will yank (y
command) in the s
register. "sp
will paste from the s
register.
While in insert mode or in a prompt, <c-r>
followed by a register name
(one character) inserts it.
For example, <c-r>
followed by " will insert the currently yanked text.
<c-r>
followed by 2 will insert the second capture group from the last regex
selection.
Registers are lists, instead of simply text in order to interact well with multiselection. Each selection has its own captures or yank buffer.
Non alphanumeric registers have an alternative name that can be used in contexts where only alphanumeric identifiers are possible.
Some registers are not general purposes, they cannot be written to, but they contain some special data:
-
%
(percent
): current buffer name -
.
(dot
): current selection contents -
#
(hash
): selection indices (first selection has 1, second has 2, …) -
_
(underscore
): null register, always empty -
:
(colon
): last entered command
Most commands using a register default to a specific one if not specified:
-
"
(dquote
): default yank register, used by yanking and pasting commands likey
,p
andR
-
/
(slash
): default search register, used by regex based commands likes
,*
or/
-
@
(arobase
): default macro register, used byq
andQ
-
^
(caret
): default mark register, used byz
andZ
-
|
(pipe
): default shell command register, used by command that spawn a subshell such as|
,<a-|>
,!
or<a-!>
Kakoune can record and replay a sequence of key presses.
Macros are recorded with the Q
key, and are stored by default in the @
register. Another register can be chosen by with hitting "<reg>
before
the Q
key.
To replay a macro, use the q
key.
Using the *
key, you can set the search pattern to the current selection.
This tries to be intelligent. It will for example detect if the current selection
begins and/or ends at word boundaries and set the search pattern accordingly.
With <a-*>
you can set the search pattern to the current selection without
Kakoune trying to be smart.
The regex syntax supported by Kakoune is the Perl syntax currently provided by Boost : Perl Regular Expression Syntax.
The :exec
and :eval
commands can be used for running Kakoune commands.
:exec
runs keys as if they were pressed, whereas :eval
executes its given
parameters as if they were entered in the command prompt. By default,
they do their execution in the context of the current client.
These two commands also save the following registers, who are then restored
when the commands have been executed: /
, "
, |
, ^
, @
.
Some parameters provide a way to change the context of execution:
-
-client <name>
: execute in the context of the client named <name> -
-try-client <name>
: execute in the context of the client named <name> if such client exists, or else in the current context. -
-draft
: execute in a copy of the context of the selected client modifications to the selections or input state will not affect the client. This permits to make some modification to the buffer without modifying the user’s selection. -
-itersel
(requires-draft
): execute once per selection, in a context with only the considered selection. This permits avoiding cases where the selections may get merged. -
-buffer <names>
: execute in the context of each buffers in the comma separated list <names>, '*' as a name can be used to iterate on all buffers. -
-no-hooks
: disable hook execution while executing the keys/commands -
-with-maps
: use user key mapping in:exec
instead of built in keys. -
-save-regs <regs>
: regs is a string of registers to be restored after execution (overwrites the list of registers saved by default)
The execution stops when the last key/command is reached, or an error is raised.
Key parameters get concatenated, so the following commands are equivalent:
:exec otest<space>1 :exec o test <space> 1
Kakoune can propose completions while inserting text, the completers
option
controls automatic completion, which kicks in when a certain idle timeout is
reached (100 milliseconds). Insert mode completion can be explicitly triggered
using <c-x>
, followed, by:
-
f : filename completion
-
w : buffer word completion
-
l : buffer line completion
Completion candidates can be selected using <c-n>
and <c-p>
.
From insert mode, pressing <a-;>
allows you to execute a single normal mode
command. This provides a few advantages:
-
The selections are not modified: when leaving insert mode using
<esc>
the selections can change, for example when insert mode was entered witha
the cursor will go back one char. Or if on an end of line the cursor will go back left (if possible). -
The modes are nested: that means the normal mode can enter prompt (with
:
), or any other modes (using:on-key
or:menu
for example), and these modes will get back to the insert mode afterwards.
This feature is tailored for scripting/macros, as it provides a more predictable
behaviour than leaving insert mode with <esc>
, executing normal mode command
and entering back insert mode (with which binding ?)
Manipulation of the displayed text is done through highlighters, which can be added or removed with the command
:add-highlighter <highlighter_name> <highlighter_parameters...>
and
:remove-highlighter <highlighter_id>
highlighter_id
is a name generated by the highlighter specified with highlighter_name
,
possibly dependent on the parameters. Use command completion on remove-highlighter to see the existing
highlighters' id.
General highlighters are:
-
regex <ex> <capture_id>:<face>…
: highlight a regex, takes the regex as first parameter, followed by any number of face parameters. For example::add-highlighter regex (\hTODO:)?[^\n] 0:cyan 1:yellow,red
will highlight C++ style comments in cyan, with an eventual 'TODO:' in yellow on red background. -
dynregex
: Similar to regex, but expand (like a command parameter would) the given expression before building a regex from the result. -
flag_lines <face> <option_name>
: add a column in front of the buffer, and display the flags specified in <option_name>, using <face> -
show_matching
: highlight matching char of the character under the selections' cursor usingMatchingChar
face. -
show_whitespaces <-tab <separator> <-tabpad <separator> <-lf <separator> <-spc <separator> <-nbsp <separator>
: display symbols on top of whitespaces to make them more explicit using the Whitespace face. -
number_lines <-relative> <-hlcursor> <-separator <separator text>
: show line numbers. The -relative switch will show line numbers relative to the main cursor line, the -hlcursor switch will highlight the cursor line with a separate face. With the -separator switch one can specify a string to separate the line numbers column with the rest of the buffer, default is|
. -
wrap <-word> <-width <max_width>
: Soft wrap buffer content to the smallest of window width and max_width. Wrap at word boundaries if-word
is specified. -
fill <face>
: fill using given face, mostly useful with Regions highlighters -
ranges <option_name>
: use the data in the range-specs option of the given name to highlight the buffer. The string part of the is interpretted as a face to apply to the range. -
replace-ranges <option_name>
: use the data in the range-specs option of the given name to highlight the buffer. The string part of the is interpretted as a display line to display in place of the range.
The group
highlighter is a container for other highlighters. You can add
a group to the current window using
add-highlighter group <name>
and then the -group
switch of add-highlighter
provides a mean to add highlighters
inside this group.
add-highlighter -group <name> <type> <params>...
Groups can contain other groups, the -group
switch can be used to define a path.
add-highlighter -group <name> group <subname> add-highlighter -group <name>/<subname> <type> <params>...
A special highlighter provides a way to segment the buffer into regions, which are to be highlighted differently.
A region is defined by 4 parameters:
<name> <opening> <closing> <recurse>
name
is user defined, opening
, closing
and recurse
are regexes.
-
opening
defines the region start text -
closing
defines the region end text -
recurse
defines the text that matches recursively an end token into the region.
recurse
is useful for regions that can be nested, for example the %sh{ … }
construct in kakoune accept nested { … }
so %sh{ … { … } … }
is valid.
This region can be defined with:
shell_expand %sh\{ \} \{
Regions are used in the regions
highlighter which can take any number
of regions.
add-highlighter regions <name> <region_name1> <opening1> <closing1> <recurse1> \ <region_name2> <opening2> <closing2> <recurse2>...
The above command defines multiple regions in which other highlighters can be added as follows:
add-highlighter -group <name>/<region_name> ...
Regions are matched using the left-most rule: the left-most region opening starts a new region. When a region closes, the closest next opening start another region.
That matches the rule governing most programming language parsing.
regions
also supports a -default <default_region>
switch to define the
default region, when no other region matches the current buffer range.
Most programming languages can then be properly highlighted using a regions
highlighter as root:
add-highlighter regions -default code <lang> \ string <str_opening> <str_closing> <str_recurse> \ comment <comment_opening> <comment_closing> <comment_recurse> add-highlighter -group <lang>/code ... add-highlighter -group <lang>/string ... add-highlighter -group <lang>/comment ...
Highlighters are often defined for a specific filetype, and it makes then sense to share the highlighters between all the windows on the same filetypes.
A shared highlighter can be defined with the :add-highlighter
command
add-highlighter -group /<group_name> ...
When the group switch values starts with a '/', it references a group in the shared highlighters, rather than the window highlighters.
The common case would be to create a named shared group, and then fill it with highlighters:
add-highlighter -group / group <name> add-highlighter -group /name regex ...
It can then be referenced in a window using the ref
highlighter.
add-highlighter ref <name>
The ref
can reference any named highlighter in the shared namespace.
Commands can be registered to be executed when certain events arise. To register a hook use the hook command.
:hook [-group <group>] <scope> <hook_name> <filtering_regex> <commands>
<scope>
can be either global, buffer or window (or any of their prefixes).
Scopes are hierarchical, meaning that a Window calling a hook will
execute its own, the buffer ones and the global ones.
<command>
is a string containing the commands to execute when the hook is
called.
For example to automatically use line numbering with .cc files, use the following command:
:hook global WinCreate .*\.cc %{ add-highlighter number_lines }
If <group>
is given, make this hook part of the named group. groups
are used for removing hooks with the remove-hooks
command:
remove-hooks <scope> <group>
The above remove every hooks in <scope>
that are part of the given group.
Existing hooks are:
-
NormalIdle
: A certain duration has passed since last key was pressed in normal mode. -
NormalBegin
: Entering normal mode -
NormalEnd
: Leaving normal mode -
NormalKey
: A key is received in normal mode, the key is used for filtering -
InsertIdle
: A certain duration has passed since last key was pressed in insert mode. -
InsertBegin
: Entering insert mode -
InsertEnd
: Leaving insert mode -
InsertKey
: A key is received in insert mode, the key is used for filtering -
InsertChar
: A character is inserted in insert mode, the character is used for filtering -
InsertDelete
: A character is deleted in insert mode, the character deleted by the main selection is used for filtering -
InsertMove
: The cursor moved (without inserting) in insert mode, the key that triggered the move is used for filtering -
PromptIdle
: A certain duration has passed since last key was pressed in prompt mode. -
WinCreate
: A window was created, the filtering text is the buffer name -
WinClose
: A window was destroyed, the filtering text is the buffer name -
WinDisplay
: A window was bound a client, the filtering text is the buffer name -
WinResize
: A window resized, the filtering text is '<line>.<column>' -
WinSetOption
: An option was set in a window context, the filtering text is '<option_name>=<new_value>' -
BufSetOption
: An option was set in a buffer context, the filtering text is '<option_name>=<new_value>' -
BufNewFile
: A buffer for a new file has been created, filename is used for filtering -
BufOpenFile
: A buffer for an existing file has been created, filename is used for filtering -
BufCreate
: A buffer has been created, filename is used for filtering -
BufWritePre
: Executed just before a buffer is written, filename is used for filtering. -
BufWritePost
: Executed just after a buffer is written, filename is used for filtering. -
BufClose
: Executed when a buffer is deleted, while it is still valid. -
BufOpenFifo
: Executed when a buffer opens a fifo. -
BufReadFifo
: Executed after some data has been read from a fifo and inserted in the buffer. -
BufCloseFifo
: Executed when a fifo buffer closes its fifo file descriptor either because the buffer is being deleted, or because the writing end has been closed. -
RuntimeError
: an error was encountered while executing a user command the error message is used for filtering -
KakBegin
: Kakoune started, this is called just after reading the user configuration files -
KakEnd
: Kakoune is quitting. -
FocusIn
: On supported clients, triggered when the client gets focused. The filtering text is the client name. -
FocusOut
: On supported clients, triggered when the client gets unfocused. The filtering text is the client name. -
InsertCompletionShow
: Triggered when the insert completion menu gets displayed. -
InsertCompletionHide
: Triggered when the insert completion menu gets hidden. -
RawKey
: Triggered whenever a key is pressed by the user, the key is used for filtering.
When not specified, the filtering text is an empty string.
You can redefine a key’s meaning using the map command:
:map <scope> <mode> <key> <keys>
scope
can be one of global
, buffer
or window
(or any prefix),
mode one of insert
, normal
, prompt
, menu
or user
(or any prefix),
key
a single key name and keys
a list of keys.
user
mode allows for user mapping behind the ,
key. Keys will be
executed in normal mode.
An optional -docstring switch followed by a string can be used to describe what the mapping does. This docstring will be used in autoinfo boxes.
Mappings can be removed with the unmap command
:unmap <scope> <mode> <key> [<expected>]
If <expected>
is specified, unmapping will only proceed if the current
mapping matches the expected keys.
New commands can be defined using the :def
command.
:def <command_name> <commands>
<commands>
is a string containing the commands to execute.
def
can also take some flags:
-
-params <num>
: the command accept <num> parameters, with <num> either a number, or of the form <min>..<max>, with both <min> and <max> omittable. -
-file-completion
: try file completion on any parameter passed to this command -
-client-completion
: try client name completion on any parameter passed to this command -
-buffer-completion
: try buffer name completion on any parameter passed to this command -
-shell-completion
: following string is a shell command which takes parameters as positional params and output one completion candidate per line. -
-allow-override
: allow the new command to replace an existing one with the same name. -
-hidden
: do not show the command in command name completions -
-docstring
: define the documentation string for the command
Using shell expansion permits defining complex commands or accessing Kakoune state:
:def print_selection %{ echo %sh{ ${kak_selection} } }
Some helper commands can be used to define composite commands:
-
prompt <prompt> <command>
: prompt the user for a string, when the user validates, executes <command>. The entered text is available in thetext
value accessible through$kak_text
in shells or%val{text}
in commands. -
on-key <command>
: wait for next key from user, then execute <command>, the key is available through thekey
value, accessible through$kak_key
. -
menu <label1> <commands1> <label2> <commands2>…
: display a menu using labels, the selected label’s commands are executed.menu
can take a -auto-single argument, to automatically run commands when only one choice is provided. And a -select-cmds argument, in which case menu takes three argument per item, the last one being a command to execute when the item is selected (but not validated). -
info <text>
: display text in an information box, at can take a -anchor option, which acceptsleft
,right
andcursor
as value, in order to specify where the info box should be anchored relative to the main selection. -
try <commands> catch <on_error_commands>
: prevent an error in <commands> from aborting the command execution, execute <on_error_commands> instead. If nothing is to be done on error, the catch part can be ommitted. -
reg <name> <content>
: set register <name> to <content> -
select <anchor_line>.<anchor_column>,<cursor_line>.<cursor_column>:…
: replace the current selections with the one described in the argument -
debug {info,buffers,options,memory,shared-strings}
: print some debug information in thedebug
buffer
Note that these commands are available in interactive command mode, but are not that useful in this context.
With :alias
commands can be given additional names. Aliases are scoped, so
that an alias can refer to one command for a buffer, and to another for another
buffer. The following command defines <alias>
as an alias for <command>
:
:alias <scope> <alias> <command>
<scope>
can be one of global
, buffer
or window
.
:unalias <scope> <alias> [<expected>]
Will remove the given alias in the given scope. If <expected>
is specified
the alias will only be removed if its current value is <expected>
.
The :edit
command can take a -fifo
parameter:
:edit -fifo <filename> [-scroll] <buffername>
In this case, a buffer named <buffername>
is created which reads its content
from fifo <filename>
. When the fifo is written to, the buffer is automatically
updated.
If the -scroll
switch is specified, the initial cursor position will be made
such as the window displaying the buffer will scroll as new data is read.
This is very useful for running some commands asynchronously while displaying
their result in a buffer. See rc/make.kak
and rc/grep.kak
for examples.
When the buffer is deleted, the fifo will be closed, so any program writing
to it will receive SIGPIPE
. This is useful as it permits to stop the writing
program when the buffer is deleted.