table of contents
| keyd(1) | General Commands Manual | keyd(1) |
NAME¶
keyd - A key remapping daemon.
SYNOPSIS¶
keyd [command] [options]
COMMANDS¶
monitor [-t]
listen
bind reset|<binding> [<binding>...]
reload
list-keys
input [-t <timeout>] <text> [<text>...]
do [-t <timeout>] [<exp>]
OPTIONS¶
-v, --version
-h, --help
DESCRIPTION¶
keyd is a system wide key remapping daemon which supports features like layering, oneshot modifiers, and macros. In its most basic form it can be used to define a custom key layout that persists across display server boundaries (e.g wayland/X/tty).
The program runs in the foreground, printing diagnostic information to the standard output streams, and is intended to be run as a single instance managed by the init system.
NOTE:
Because keyd modifies your primary input device, it is possible to render your machine unusable with a bad config file. If you find yourself in this situation the panic sequence <backspace>+<escape>+<enter> will force keyd to terminate.
CONFIGURATION¶
Configuration files loosely follow an INI style format consisting of headers of the form [section_name] followed by a set of bindings. Lines beginning with a hash are ignored.
Config files are stored in /etc/keyd/ and loaded upon initialization. The reload command can be used to update the working set of config files (e.g sudo keyd reload).
A valid config file has the extension .conf and must begin with an [ids] section that has one of the following forms:
[ids] <id 1 (obtained via keyd monitor)> <id 2> ...
or
[ids] * -<id 1> -<id 2> ...
The first form specifies a list of ids to be explicitly matched, while the second matches any id which has not been explicitly excluded.
For example:
[ids] * -0123:4567
Will match all keyboards which do not(2) have the id 0123:4567, while:
[ids] 0123:4567
will exclusively match any devices which do. Device ids can be obtained from the monitor command (see COMMANDS). Note that a device id may only be listed in a single config file.
Each subsequent section of the file corresponds to a layer (with the exception of [global] (see GLOBALS).
Config errors will appear in the log output and can be accessed in the usual way using your system's service manager (e.g sudo journalctl -eu keyd).
If an id matches more than one device type, the prefix k: may be used to exclusively match keyboards and the prefix m: may be used to exclusively match mice. (E.g m:046d:b01d)
Note: All keyboards defined within a given config file will share the same state. This is useful for linking separate input devices together (e.g foot pedals).
Note 2: The wildcard will only match devices which keyd identifies as keyboards. keyd is also capable of managing mice (e.g to facilitate clearing of oneshot modifiers on click), but their ids must be explicitly listed.
Note 3: Mouse support is currently experimental, and is mostly confined to traditional mice (e.g touchpads). Adding some mice to your ids section may break your pointer. It may also be necessary to explicitly blacklist mice which are misidentified as keyboards (e.g if you find your moused misbehaving).
Layers¶
A layer is a collection of bindings, each of which specifies the behaviour of a particular key. Multiple layers may be active at any given time, forming a stack of occluding keymaps consulted in activation order. The default layer is called main and is where common bindings should be defined.
For example, the following config snippet defines a layer called nav and creates a toggle for it in the main layer:
[main] capslock = layer(nav) [nav] h = left k = up j = down l = right
When capslock is held, the nav layer occludes the main layer causing hjkl to function as the corresponding arrow keys.
Unlike most other remapping tools, keyd provides first class support for modifiers. A layer name may optionally end with a ':' followed by a set of modifiers to emulate in the absence of an explicit mapping. Layer names may consist of a maximum of 64 characters (including all modifiers).
These layers play nicely with other modifiers and preserve existing stacking semantics.
For example:
[main] capslock = layer(capslock) [capslock:C] j = down
will cause capslock to behave as control, except in the case of capslock+j, which will emit down. This makes it trivial to define custom modifiers which don't interfere with one another.
Note that bindings are not affected by the modifiers of the layer in which they are defined. Thus capslock+j will produce an unmodified down keypress, while shift+capslock+j will produce shift+down as expected.
Formally, each layer heading has the following form:
"[" <layer name>[:<modifier set>] "]"
Where <modifier_set> has the form:
and each modifier is one of:
M - Meta/Super
A - Alt
S - Shift
G - AltGr
Finally, each layer heading is followed by a set of bindings which take the form:
for a description of <action> and <macro> see ACTIONS and MACROS.
By default, each key is bound to itself within the main layer. The exception to this are the modifier keys, which are instead bound to eponymously named layers with the corresponding modifiers.
For example, meta is actually bound to layer(meta), where meta is internally defined as meta:M.
The full set of modifier bindings are as follows:
control = layer(control) meta = layer(meta) shift = layer(shift) leftalt = layer(alt) rightalt = layer(altgr)
A consequence of this is that overriding modifier keys is a simple matter of adding the desired bindings to an appropriate pre-defined layer.
Thus
[ids] * [control] j = down
is a completely valid config, which does what the benighted user might expect. Internally, the full config actually looks something like this:
[ids] * [main] leftcontrol = layer(control) rightcontrol = layer(control) [control:C] j = down
If multiple bindings for the same key are present, the most recent one takes precedence.
A layer heading may also appear multiple times, in which case the layer will contain the sum of all bindings. Note that the layer type may not be reassigned.
That is:
[mylayer:A] a = b c = d [mylayer:C] a = x b = c
is equivalent to:
[mylayer:A] a = x b = c c = d
Composite Layers¶
A special kind of layer called a composite layer can be defined by creating a layer with a name consisting of existing layers delimited by +. The resultant layer will be activated and given precedence when all of its constituents are activated. Composite layers are not allowed to have modifiers attached and cannot be explicitly assigned.
E.g.
[control+alt] h = left
will cause the sequence control+alt+h to produce left (ignoring the control and alt modifiers attached to the active control and alt layers), while pressing control+alt+f1 preserves those modifiers, emitting exactly what was pressed, as there is no explicit binding for f1 on the composite layer.
[main] capslock = layer(capslock) [capslock:C] [capslock+shift] h = left
Will cause the sequence capslock+shift+h to produce left, while preserving the expected functionality of capslock and shift in isolation.
Note: composite layers must always be defined after the layers of which they are comprised.
That is:
[layer1] [layer2] [layer1+layer2]
and not
[layer1+layer2] [layer1] [layer2]
Layouts¶
A layout is a special kind of layer intended for modifying alpha keys. Unlike layers, layouts cannot have any associated modifiers, and only one layout may be active at a given time. The default layout is called 'main', and can be changed using the setlayout() action.
For convenience, keyd ships with a number of common letter layouts in /usr/share/keyd/layouts. Before including these, it is instructive to inspect them. Non-english layouts include a dedicated shift layer (making order of inclusion important) and require the use of keyd's compose definitions (see Unicode Support)
E.g.
# Include the shipped colemak layout. include layouts/colemak [global] default_layout = mylayout [mylayout:layout] a = b b = c #etc... [control] 1 = setlayout(customlayout) 2 = setlayout(colemak)
Chording¶
Chords are groups of keys which are treated as a unit when simultaneously depressed. A chord can be defined by using a group of + delimited key names as a left hand value. The corresponding action will be activated if all keys are struck within the chording interval (chord_timeout).
E.g
j+k = esc
will cause esc to be produced if both j and k are simultaneously depressed.
Note: It may be desirable to change the default chording interval (50ms) to account for the physical characteristics of your keyboard.
Unicode Support¶
If keyd encounters a valid UTF8 sequence as a right hand value, it will try and translate that sequence into a macro which emits a keyd-specific XKB sequence.
In order for this to work, the sequences defined in the compose file shipped with keyd (/usr/share/keyd/keyd.compose) must be accessible. This can be achieved globally by copying the file to the appropriate location in /usr/share/X11/locale, or on a per-user basis by symlinking it to ~/.XCompose.
E.g.
Additionally you will need to be using the default US layout on your display server. Users of non-english layouts are advised to set their layout within keyd (see Layouts) to avoid conflicts between the display server layout and keyd's unicode functionality.
Note: You may have to restart your applications for this to take effect.
Note 2: The generated compose sequences are affected by modifiers in the normal way. If you want shift to produce a different symbol, you will need to define a custom shift layer (see the included layout files for an example).
Note 3: GTK4 currently has a bug which causes it to crash in the presence of large XCompose files (like /usr/share/keyd/keyd.compose).
Aliases¶
Each key may optionally be assigned an alias. This alias may be used in place of the key as a valid left hand value. Multiple keys may be bound to the same alias, but only one alias may be assigned to a key at a given time.
For example, the keys 'leftmeta' and 'rightmeta' are bound to the alias meta by default. Thus the binding 'meta = a' is equivalent to the bindings 'leftmeta = a' and 'rightmeta = a'.
Aliases are defined in a special section called 'aliases' where each line takes the form:
where <key> must be a valid key name.
Note that <name> may itself be a valid key name, in which case all references to the key within the config file will be replaced with the new key. Additionally, if the assigned alias is a valid key name, the corresponding keycode will be assigned to the key by default. This makes it possible to redefine keys before any bindings are applied and is particularly useful in conjunction with the include mechanism to account for variations in hardware.
For example:
/etc/keyd/common: meta = oneshot(meta) alt = oneshot(alt) a = a s = o # etc.. /etc/keyd/default.conf: [ids] * [main] include common /etc/keyd/magic_keyboard.conf: [ids] 004c:0267 [aliases] leftalt = meta rightalt = meta rightmeta = alt leftmeta = alt [main] include common
Allows the user to define a set of universal bindings in /etc/keyd/common without having to explicitly account for the transposed meta and alt keys within the included config snippet.
File Inclusion¶
Config files may include other files located within the config directory using the include keyword. A line of the form include <file> may appear at any point after the [ids] section. The resultant config will behave as though the contents of the included file appear in place of the include statement.
Making strategic use of these statements makes it possible to share common functionality between configs.
Include paths are relative and must be placed in one of the following directories:
E.g.
/etc/keyd/default.conf: [ids] * # Add our shared custom bindings. include common # Appends bindings to the main layer # defined in /etc/keyd/common (order matters) [main] capslock = layer(capslock) [capslock] 1 = setlayout(colemak) 2 = setlayout(dvorak) /etc/keyd/common: [main] rightmeta = layer(nav) [nav] h = left j = down k = up l = right /usr/share/keyd/layouts/dvorak: a = a s = o ...
Limitations:
GLOBALS¶
A special section called [global] may be defined in the file and can contain any of the following options:
macro_repeat_timeout: The time separating successive executions of a macro. (default: 50)
layer_indicator: If set, this will turn the capslock light on whenever a layer is active. Note: Some wayland compositors will aggressively toggle LED state rendering this option unusable.
(default: 0)
macro_sequence_timeout: If set, this will add a timeout (in microseconds) between each emitted key in a macro sequence. This is useful to avoid overflowing the input buffer on some systems.
chord_timeout: The maximum time between successive keys interpreted as part of a chord. (default: 50)
chord_hold_timeout: The length of time a chord must be held before being activated. (default: 0)
oneshot_timeout: If non-zero, timeout a oneshot layer activation after the supplied number of milliseconds. (default: 0)
disable_modifier_guard: By default, keyd will inject additional control keypresses where necessary in order to prevent programs from seeing additional modifier taps (E.g alt in firefox). If set, this option disables that behaviour. (default: 0)
overload_tap_timeout: If non-zero, ignore the tap behaviour of an overloaded key if it is held for the given number of miliseconds. (default: 0).
Note: Unicode characters and key sequences are treated as macros, and are consequently affected by the corresponding timeout options.
MACROS¶
Various keyd actions accept macro expressions.
A macro expression has one of the following forms:
Where <char> is a valid unicode character and <exp> has the form <token1> [<token2>...] and each token is one of:
The following are all valid macro expressions:
Splitting into smaller tokens serves as an escaping mechanism: macro(space) inserts a space but macro(s pace) writes "space". Likewise, macro(3+5) depresses the 3 and 5 keys as a unit while macro(3 + 5) writes "3+5".
Some prerequisites are needed for non-ASCII characters to work, see Unicode Support.
ACTIONS¶
A key may optionally be bound to an action which accepts zero or more arguments.
layer(<layer>)
oneshot(<layer>)
swap(<layer>)
[control] x = swap(xlayer) [xlayer] s = C-s b = S-insert
You probably don't need to use this unless you are trying to do something quite involved. Think hard about whether or not what you are trying to achieve can be done by other means, as it is easy to end up in states which are impossible to exit.
setlayout(<layout>)
clear()
toggle(<layer>)
layerm(<layer>, <macro>)
oneshotm(<layer>, <macro>)
swapm(<layer>, <macro>)
togglem(<layer>, <macro>)
clearm(<macro>)
Key overloading¶
overload(<layer>, <action>)
overloadt(<layer>, <action>, <timeout>)
Note that this will add a visual delay when typing, since overlapping keys will be queued until the timeout expires or the bound key is released.
overloadt2(<layer>, <action>, <timeout>)
overloadi(<action 1>, <action 2>, <idle timeout>)
This can be used in combination with other overload timeouts and is particularly useful for overloading letter keys (often called 'homerow mods').
For example:
a = overloadi(a, overloadt2(control, a, 200), 150)
This reduces the visual latency by immediately resolving the key as a letter when typed midword, but also facilitates its use as a layer trigger if it is held for a long enough period with no intervening symbols.
Since this is a common usecase, a macro called lettermod (defined below) has been defined to facilitate such definitions.
lettermod(<layer>, <key>, <idle timeout>, <hold timeout>)
timeout(<action 1>, <timeout>, <action 2>)
E.g.
timeout(a, 500, layer(control))
Will cause the assigned key to behave as control if it is held for more than 500 ms.
macro2(<timeout>, <repeat timeout>, <macro>)
Note that <macro> must be a valid macro expression.
E.g.
macro2(400, 50, macro(Hello space World)) macro2(120, 80, left)
E.g.
command(brightness down)
NOTE: Commands are executed by the user running the keyd process (probably root), use this feature judiciously.
noop
IPC¶
To facilitate extensibility, keyd employs a client-server model accessible through the use of -e. The keymap can thus be conceived of as a living entity that can be modified at run time.
In addition to allowing the user to try new bindings on the fly, this enables the user to fully leverage keyd's expressive power from other programs without incurring a performance penalty.
For instance, the user may use this functionality to write a script which alters the keymap when they switch between different tmux sessions.
The application remapping tool (keyd-application-mapper(1)) which ships with keyd is a good example of this. It is a small python script which performs event detection for the various display servers (e.g X/sway/gnome, etc) and feeds the desired mappings to the core using -e.
NOTE: Users with access to the keyd socket should be considered privileged (i.e assumed to have access to the entire system.).
Bindings¶
The bind command accepts one or more bindings, each of which must have the following form:
Where <layer> is the name of an (existing) layer in which the key is to be bound.
As a special case, the string "reset" may be used in place of a binding, in which case the current keymap will revert to its original state (all dynamically applied bindings will be dropped).
Examples:
$ keyd bind '- = C-c' $ keyd bind reset '+ = C-c' # Reset the keyboard before applying the '+' binding (drops the previous '-' binding)
By default expressions apply to the most recently active keyboard.
EXAMPLES¶
Example 1¶
Make esc+q toggle the dvorak letter layout.
[ids] * [main] esc = layer(esc) [dvorak] a = a s = o ... [esc] q = toggle(dvorak)
Example 2¶
Invert the behaviour of the shift key without breaking modifier behaviour.
[ids] * [main] 1 = ! 2 = @ 3 = # 4 = $ 5 = % 6 = ^ 7 = & 8 = * 9 = ( 0 = ) [shift] 0 = 0 1 = 1 2 = 2 3 = 3 4 = 4 5 = 5 6 = 6 7 = 7 8 = 8 9 = 9
Example 3¶
Tapping control once causes it to apply to the next key, tapping it twice activates it until it is pressed again, and holding it produces expected behaviour.
[main] control = oneshot(control) [control] control = toggle(control)
Example 4¶
Meta behaves as normal except when ` is pressed, after which the alt_tab layer is activated for the duration of the leftmeta keypress. Subsequent actuations will thus produce A-tab instead of M-\.
[meta] ` = swap(alt_tab, A-tab) [alt_tab:A] tab = A-S-tab ` = A-tab
Example 5¶
# Uses the compose key functionality of the display server to generate # international glyphs. # For this to work 'setxkbmap -option # compose:menu' must # be run after keyd has started. # A list of sequences can be found in /usr/share/X11/locale/en_US.UTF-8/Compose # on most systems. [main] rightalt = layer(dia) [dia] # Map o to ö o = macro(compose o ") # Map e to € e = macro(compose c =)
Example 6¶
# Tapping both shift keys will activate capslock. [shift] leftshift = capslock rightshift = capslock
Example 7¶
Capslock will behave as control in all instances except when used in conjunction with 'h/j/k/l' in which case it will produce arrow keys. If tapped, it will function as escape.
[main] capslock = overload(capslock, esc) esc = capslock [capslock:C] h = left k = up j = down l = right
Example 8¶
Disables the esc and end keys.
[main] esc = noop end = noop
ENVIRONMENT VARIABLES¶
KEYD_DEBUG
AUTHOR¶
Written by Raheman Vaiya (2017-).
BUGS¶
Please file any bugs or feature requests at the following url:
| 2024-07-27 |