QMK keymaps are defined inside a C source file. The data structure is an array of arrays. The outer array is a list of layer arrays while the inner layer array is a list of keys. Most keyboards define a KEYMAP() macro to help you create this array of arrays.
In QMK, const uint16_t PROGMEM keymaps[][MATRIX_ROWS][MATRIX_COLS] holds multiple layers of keymap information in 16 bit data holding the action code. You can define 32 layers at most.
For trivial key definitions, the higher 8 bits of the action code are all 0 and the lower 8 bits holds the USB HID usage code generated by the key as keycode.
Respective layers can be validated simultaneously. Layers are indexed with 0 to 31 and higher layer has precedence.
Keymap: 32 Layers Layer: action code matrix
----------------- ---------------------
stack of layers array_of_action_code[row][column]
____________ precedence _______________________
/ / | high / ESC / F1 / F2 / F3 ....
31 /___________// | /-----/-----/-----/-----
30 /___________// | / TAB / Q / W / E ....
29 /___________/ | /-----/-----/-----/-----
: _:_:_:_:_:__ | : /LCtrl/ A / S / D ....
: / : : : : : / | : / : : : :
2 /___________// | 2 `--------------------------
1 /___________// | 1 `--------------------------
0 /___________/ V low 0 `--------------------------
Sometimes, the action code stored in keymap may be referred as keycode in some documents due to the TMK history.
The state of the Keymap layer is determined by two 32 bit parameters:
default_layer_stateindicates a base keymap layer (0-31) which is always valid and to be referred (the default layer).layer_statehas current on/off status of each layer in its bits.
Keymap layer '0' is usually default_layer, wither other layers initially off after booting up the firmware, although this can configured differently in config.h. It is useful to change default_layer when you completely switch a key layout, for example, if you want to switch to Colemak instead of Qwerty.
Initial state of Keymap Change base layout
----------------------- ------------------
31 31
30 30
29 29
: :
: : ____________
2 ____________ 2 / /
1 / / ,->1 /___________/
,->0 /___________/ | 0
| |
`--- default_layer = 0 `--- default_layer = 1
layer_state = 0x00000001 layer_state = 0x00000002
On the other hand, you can change layer_state to overlay the base layer with other layers for features such as navigation keys, function keys (F1-F12), media keys, and/or special actions.
Overlay feature layer
--------------------- bit|status
____________ ---+------
31 / / 31 | 0
30 /___________// -----> 30 | 1
29 /___________/ -----> 29 | 1
: : | :
: ____________ : | :
2 / / 2 | 0
,->1 /___________/ -----> 1 | 1
| 0 0 | 0
| +
`--- default_layer = 1 |
layer_state = 0x60000002 <-'
Note that higher layer has higher priority on stack of layers, namely firmware falls down from top layer to bottom to look up keycode. Once it spots keycode other than KC_TRNS(transparent) on a layer it stops searching and lower layers aren't referred.
You can place KC_TRANS on overlay layer changes just part of layout to fall back on lower or base layer.
Key with KC_TRANS (KC_TRNS and _______ are the alias) doesn't has its own keycode and refers to lower valid layers for keycode, instead.
For this example we will walk through the default Clueboard 66% keymap. You'll find it helpful to open that file in another browser window so you can look at everything in context.
There are 3 main sections of a keymap.c file you'll want to concern yourself with:
At the top of the file you'll find this:
#include "clueboard.h"
// Helpful defines
#define GRAVE_MODS (MOD_BIT(KC_LSHIFT)|MOD_BIT(KC_RSHIFT)|MOD_BIT(KC_LGUI)|MOD_BIT(KC_RGUI)|MOD_BIT(KC_LALT)|MOD_BIT(KC_RALT))
#define _______ KC_TRNS
// Each layer gets a name for readability.
// The underscores don't mean anything - you can
// have a layer called STUFF or any other name.
// Layer names don't all need to be of the same
// length, and you can also skip them entirely
// and just use numbers.
#define _BL 0
#define _FL 1
#define _CL 2
These are some handy definitions we can use when building our keymap and our custom function. The GRAVE_MODS definition will be used later in our custom function. The _______ define makes it easier to see what keys a layer is overriding, while the _BL, _FL, and _CL defines make it easier to refer to each of our layers.
The main part of this file is the keymaps[] definition. This is where you list your layers and the contents of those layers. This part of the file begins with this definition:
const uint16_t PROGMEM keymaps[][MATRIX_ROWS][MATRIX_COLS] = {
After this you'll find a list of KEYMAP() macros. A KEYMAP() is simply a list of keys to define a single layer. Typically you'll have one or more "base layers" (such as QWERTY, Dvorak, or Colemak) and then you'll layer on top of that one or more "function" layers. Due to the way layers are processed you can't overlay a "lower" layer on top of a "higher" layer.
keymaps[][MATRIX_ROWS][MATRIX_COLS] in QMK holds the 16 bit action code (sometimes referred as the quantum keycode) in it. For the keycode representing typical keys, its high byte is 0 and its low byte is the USB HID usage ID for keyboard.
TMK from which QMK was forked uses
const uint8_t PROGMEM keymaps[][MATRIX_ROWS][MATRIX_COLS]instead and holds the 8 bit keycode. Some keycode values are reserved to induce execution of certain action codes via thefn_actions[]array.
Here is an example of the Clueboard's base layer:
/* Keymap _BL: Base Layer (Default Layer)
*/
[_BL] = KEYMAP(
F(0), KC_1, KC_2, KC_3, KC_4, KC_5, KC_6, KC_7, KC_8, KC_9, KC_0, KC_MINS, KC_EQL, KC_GRV, KC_BSPC, KC_PGUP, \
KC_TAB, KC_Q, KC_W, KC_E, KC_R, KC_T, KC_Y, KC_U, KC_I, KC_O, KC_P, KC_LBRC, KC_RBRC, KC_BSLS, KC_PGDN, \
KC_CAPS, KC_A, KC_S, KC_D, KC_F, KC_G, KC_H, KC_J, KC_K, KC_L, KC_SCLN, KC_QUOT, KC_NUHS, KC_ENT, \
KC_LSFT, KC_NUBS, KC_Z, KC_X, KC_C, KC_V, KC_B, KC_N, KC_M, KC_COMM, KC_DOT, KC_SLSH, KC_RO, KC_RSFT, KC_UP, \
KC_LCTL, KC_LGUI, KC_LALT, KC_MHEN, KC_SPC,KC_SPC, KC_HENK, KC_RALT, KC_RCTL, MO(_FL), KC_LEFT, KC_DOWN, KC_RGHT),
Some interesting things to note about this:
- From a C source point of view it's only a single array, but we have embedded whitespace to more easily visualize where each key is on the physical device.
- Plain keyboard scancodes are prefixed with KC_, while "special" keys are not.
- The upper left key activates custom function 0 (
F(0)) - The "Fn" key is defined with
MO(_FL), which moves to the_FLlayer while that key is being held down.
Our function layer is, from a code point of view, no different from the base layer. Conceptually, however, you will build that layer as an overlay, not a replacement. For many people this distinction does not matter, but as you build more complicated layering setups it matters more and more.
[_FL] = KEYMAP(
KC_GRV, KC_F1, KC_F2, KC_F3, KC_F4, KC_F5, KC_F6, KC_F7, KC_F8, KC_F9, KC_F10, KC_F11, KC_F12, _______, KC_DEL, BL_STEP, \
_______, _______, _______,_______,_______,_______,_______,_______,KC_PSCR,KC_SLCK, KC_PAUS, _______, _______, _______, _______, \
_______, _______, MO(_CL),_______,_______,_______,_______,_______,_______,_______, _______, _______, _______, _______, \
_______, _______, _______,_______,_______,_______,_______,_______,_______,_______, _______, _______, _______, _______, KC_PGUP, \
_______, _______, _______, _______, _______,_______, _______, _______, _______, MO(_FL), KC_HOME, KC_PGDN, KC_END),
Some interesting things to note:
- We have used our
_______definition to turnKC_TRNSinto_______. This makes it easier to spot the keys that have changed on this layer. - While in this layer if you press one of the
_______keys it will activate the key in the next lowest active layer.
At the bottom of the file we've defined a single custom function. This function defines a key that sends KC_ESC when pressed without modifiers and KC_GRAVE when modifiers are held. There are a couple pieces that need to be in place for this to work, and we will go over both of them.
We define the fn_actions[] array to point to custom functions. F(N) in a keymap will call element N of that array. For the Clueboard's that looks like this:
const uint16_t PROGMEM fn_actions[] = {
[0] = ACTION_FUNCTION(0), // Calls action_function()
};
In this case we've instructed QMK to call the ACTION_FUNCTION callback, which we will define in the next section.
This
fn_actions[]interface is mostly for backward compatibility. In QMK, you don't need to usefn_actions[]. You can directly useACTION_FUNCTION(N)or any other action code value itself normally generated by the macro inkeymaps[][MATRIX_ROWS][MATRIX_COLS]. N inF(N)can only be 0 to 31. Use of the action code directly inkeymapsunlocks this limitation.
To actually handle the keypress event we define an action_function(). This function will be called when the key is pressed, and then again when the key is released. We have to handle both situations within our code, as well as determining whether to send/release KC_ESC or KC_GRAVE.
void action_function(keyrecord_t *record, uint8_t id, uint8_t opt) {
static uint8_t mods_pressed;
switch (id) {
case 0:
/* Handle the combined Grave/Esc key
*/
mods_pressed = get_mods()&GRAVE_MODS; // Check to see what mods are pressed
if (record->event.pressed) {
/* The key is being pressed.
*/
if (mods_pressed) {
add_key(KC_GRV);
send_keyboard_report();
} else {
add_key(KC_ESC);
send_keyboard_report();
}
} else {
/* The key is being released.
*/
if (mods_pressed) {
del_key(KC_GRV);
send_keyboard_report();
} else {
del_key(KC_ESC);
send_keyboard_report();
}
}
break;
}
}
This should have given you a basic overview for creating your own keymap. For more details see the following resources:
We are actively working to improve these docs. If you have suggestions for how they could be made better please file an issue!