mirror of
https://github.com/qmk/qmk_firmware.git
synced 2024-12-22 17:39:55 +00:00
641 lines
32 KiB
C
641 lines
32 KiB
C
/* Copyright 2021 Yonatan Zunger
|
||
*
|
||
* This program is free software: you can redistribute it and/or modify
|
||
* it under the terms of the GNU General Public License as published by
|
||
* the Free Software Foundation, either version 2 of the License, or
|
||
* (at your option) any later version.
|
||
*
|
||
* This program is distributed in the hope that it will be useful,
|
||
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
||
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
||
* GNU General Public License for more details.
|
||
*
|
||
* You should have received a copy of the GNU General Public License
|
||
* along with this program. If not, see <http://www.gnu.org/licenses/>.
|
||
*/
|
||
#include QMK_KEYBOARD_H
|
||
#include <assert.h>
|
||
|
||
// This keymap is designed to make it easy to type in a wide variety of languages, as well as
|
||
// generate mathematical symbols (à la Space Cadet), without relying on the host OS to do
|
||
// key mappings or handle accents. Why? Because different OS's do this in radically different
|
||
// ways, and don't support all of the features one often needs.
|
||
//
|
||
// LAYER MAGIC (aka, typing in many alphabets)
|
||
//
|
||
// This keyboard has three sets of "polyglot" layers: GREEK, CADET, and YIDDISH. Each of these
|
||
// is actually a pair of layers, FOO and SHIFTFOO, which are full of Unicode points that let you
|
||
// type in them. (The Greek and Yiddish keymaps selected here are very canted towards use on a
|
||
// QWERTY layout, rather than the "standard" layouts often used for those languages in monolingual
|
||
// environments. This is useful if your keyboard doesn't have legends for all of them, which in
|
||
// most cases it won't. Of course, you could easily add more.)
|
||
//
|
||
// These each have their own layer select key, which can act as a held modifier key (GREEK+s to
|
||
// produce sigma, etc). There's also a "layer lock" key; layer lock + modifier switches you into
|
||
// that layer until you hit "layer lock" again to bounce back to QWERTY.
|
||
//
|
||
// ACCENT MAGIC
|
||
//
|
||
// We want to support easy typing of diacriticals, again without relying on the host OS. (On
|
||
// MacOS, if you want Unicode to work you have to lose all the normal accent combining keys, and
|
||
// if you're in a multi-OS world, each OS has a totally different input method)
|
||
//
|
||
// The real nuance comes from the three different ways Unicode represents these. Many common
|
||
// accent + letter combinations like é have their own dedicated code points (the combined
|
||
// normal form). One can also place a "combining accent mark" after the letter's code point to
|
||
// form the decomposed normal form (NFKD); this often renders the same as the combined form, but
|
||
// many less-sophisticated apps won't realize it's the same thing as the combined form (thus messing
|
||
// up string matching), and if you backspace you need to backspace *twice* to remove the character,
|
||
// because it's literally two characters. Finally, if you want to render just the accent mark as a
|
||
// symbol of its own, that's a *third* code point. If you're simply typing, you don't want to think
|
||
// about any of this!
|
||
//
|
||
// We thus have a bunch of special keycodes for "accent requests," which live on the FUNCTION
|
||
// layer. Accent requests don't do anything immediate, but when the *next* non-modifier key is hit,
|
||
// we generate a combined code point (if possible), two uncombined points (in cases where combined
|
||
// points don't exist), or the isolated accent followed by the next character typed (in cases where
|
||
// what you typed next isn't a letterform at all, e.g. you hit the space bar). You can also hit
|
||
// shift-<accent request> to just generate the uncombined accent on its own.
|
||
//
|
||
// The current accent request codes are modeled on the ones in MacOS.
|
||
//
|
||
// fn+` Grave accent (`)
|
||
// fn+e Acute accent (´)
|
||
// fn+i Circumflex (^)
|
||
// fn+u Diaresis / umlaut / trema (¨)
|
||
// fn+c Cedilla (¸)
|
||
// fn+n Tilde (˜)
|
||
//
|
||
// Together, these functions make for a nice "polyglot" keyboard: one that can easily type in a wide
|
||
// variety of languages, which is very useful for people who, well, need to type in a bunch of
|
||
// languages.
|
||
//
|
||
// The major TODOs are:
|
||
// - Add accent support for Hebrew accents.
|
||
// - Factor the code below so that the data layers are more clearly separated from the code logic,
|
||
// so that other users of this keymap can easily add whichever alphabets they need without
|
||
// having to deeply understand the implementation. Probably something similar to
|
||
// users/drashna/keyrecords/unicode.c, but I want to see if I can do some preprocessor magic
|
||
// so that we can actually have the rendered *character* sitting in the code instead of just the
|
||
// hex code point!
|
||
//
|
||
// PLATFORM MAGIC (aka, working well on both Mac and Windows)
|
||
//
|
||
// Finally, this keyboard can switch between Mac and Windows modes, changing various macro
|
||
// combinations, the Unicode mode, and the position of the ALT and GUI keys.
|
||
|
||
|
||
enum custom_keycodes {
|
||
// We provide special layer management keys:
|
||
// GREEK triggers the Greek (aka "Front") layer, or the SHIFTGREEK layer when shift is held.
|
||
// (Because we use Unicode, we need to implement shift-handling at the firmware level,
|
||
// rather than the OS level like we do in the QWERTY layer)
|
||
// CADET or GREEK+ALT triggers the Cadet (aka "Top") layer, or the SHIFTCADET layer when
|
||
// shift is held.
|
||
// YIDDISH triggers a keymap designed for easy Hebrew and Yiddish, based loosely on QWERTY
|
||
// layouts.
|
||
// LAYER_LOCK locks the "base" layer (i.e., QWERTY, GREEK, or CADET) to the value which is
|
||
// pressed at the moment that it is being released. When a layer lock is set, the
|
||
// analogous layer modifier key is reversed; e.g., if you lock the GREEK layer, then the
|
||
// GREEK button bounces you back to QWERTY.
|
||
//
|
||
// We also parse the shift, alt, and caps lock keys to provide management of those which is
|
||
// compatible with these various layers.
|
||
KC_GREEK = SAFE_RANGE,
|
||
KC_CADET,
|
||
KC_YIDDISH,
|
||
KC_LAYER_LOCK,
|
||
KC_PLATFORM, // Platform select
|
||
|
||
// OS-dependent macros
|
||
KC_VC_MUTE, // Video conference mute
|
||
KC_VC_HAND, // Video conference hand-raise
|
||
KC_SCRNSHT, // Screenshot (gui-shift-S on Windows, gui-shift-4 on Mac)
|
||
|
||
// These are the keycodes generated by the various "accent request" keystrokes.
|
||
KC_ACCENT_START,
|
||
KC_CGRV = KC_ACCENT_START, // Grave accent
|
||
KC_CAGU, // Acute accent
|
||
KC_CDIA, // Diaresis / umlaut / trema
|
||
KC_CCIR, // Circumflex
|
||
KC_CCED, // Cedilla
|
||
KC_CTIL, // Tilde
|
||
KC_ACCENT_END,
|
||
};
|
||
|
||
enum layers_keymap {
|
||
_QWERTY = 0,
|
||
|
||
_FIRST_LANGUAGE_LAYER,
|
||
_YIDDISH = _FIRST_LANGUAGE_LAYER,
|
||
_SHIFTYIDDISH,
|
||
_GREEK,
|
||
_SHIFTGREEK,
|
||
_CADET,
|
||
_SHIFTCADET,
|
||
_LAST_LANGUAGE_LAYER,
|
||
|
||
// Function goes last.
|
||
_FUNCTION = _LAST_LANGUAGE_LAYER,
|
||
};
|
||
|
||
// We manage our OS mode internally, and store it in a static, rather than EEPROM, bit. That's
|
||
// because it changes as we flip machines, and there's no good reason to wear out the memory.
|
||
enum os_modes {
|
||
_WINDOWS = 0,
|
||
_MAC = 1,
|
||
_OS_MODES_MAX = 2,
|
||
};
|
||
static uint8_t os_mode = _MAC;
|
||
|
||
// Key types matter for accent handling. If there's a pending accent request and another key is
|
||
// pressed:
|
||
// - If it's a normal key, we trigger all our magic accent handling.
|
||
// - If it's a modifier key, we do nothing and let the accent request hold until the next keypress.
|
||
// - If it's a special key, we drop the accent request but don't handle it.
|
||
enum key_types {
|
||
_NORMAL_KEY,
|
||
_MODIFIER_KEY,
|
||
_SPECIAL_KEY,
|
||
};
|
||
|
||
// msec to hold the platform key to trigger a switch
|
||
#define PLATFORM_HOLD_DURATION 750
|
||
|
||
// This is so that H(xxxx) has the same width as _______, which makes the grids more legible.
|
||
#define H(x) UC(0x##x)
|
||
#define MO_FN MO(_FUNCTION)
|
||
#define KC_LLCK KC_LAYER_LOCK
|
||
|
||
// Values for our OS-dependent keys, as arrays keyed by OS mode. Use Meet shortcuts on Mac, Teams on Windows
|
||
const char *VC_MUTE_VALUES[_OS_MODES_MAX] = {SS_LCTL(SS_LSFT("m")), SS_LCMD("d")};
|
||
const char *VC_HAND_VALUES[_OS_MODES_MAX] = {SS_LCTL(SS_LSFT("k")), SS_LCTL(SS_LCMD("h"))};
|
||
const char *SCRNSHT_VALUES[_OS_MODES_MAX] = {SS_LGUI(SS_LSFT("s")), SS_LCMD(SS_LSFT("4"))};
|
||
|
||
const uint16_t PROGMEM keymaps[][MATRIX_ROWS][MATRIX_COLS] = {
|
||
[_QWERTY] = LAYOUT_all(
|
||
KC_ESC, KC_LLCK, KC_GRAVE,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_BSPC, KC_HOME,
|
||
KC_PLATFORM, KC_MPLY, 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_PGUP,
|
||
KC_SCRNSHT, KC_YIDDISH, KC_LCTL, KC_A, KC_S, KC_D, KC_F, KC_G, KC_H, KC_J, KC_K, KC_L, KC_SCLN, KC_QUOT, KC_ENT, KC_PGDN,
|
||
KC_GREEK, KC_CADET, KC_LSFT, KC_Z, KC_X, KC_C, KC_V, KC_B, KC_N, KC_M, KC_COMM, KC_DOT, KC_SLSH, KC_RSFT, KC_UP, KC_END,
|
||
KC_VC_HAND, KC_VC_MUTE, KC_LCTL, KC_LGUI, KC_LALT, KC_SPC, MO_FN, KC_LEFT, KC_DOWN, KC_RGHT),
|
||
|
||
/* The Greek and Cadet layers. Tab, backspace, the nav and modifier keys, and the control block
|
||
* are always transparent.
|
||
*
|
||
* QWERTY GREEK SGREEK CADET SCADET YID SYID
|
||
* ` ׳ 05f3 ״ 05f4
|
||
* 1 ₁ 2081 ¹ 00b9 ¡ 00a1 ¿ 00bf [transparent]
|
||
* 2 ₂ 2082 ² 00b2 « 00ab » 00bb [transparent]
|
||
* 3 ₃ 2083 ³ 00b3 £ 00a3 € 20ac [transparent]
|
||
* 4 ₄ 2084 ⁴ 2074 [transparent]
|
||
* 5 ₅ 2085 ⁵ 2075 [transparent]
|
||
* 6 ₆ 2086 ⁶ 2076 [transparent]
|
||
* 7 ₇ 2087 ⁷ 2077 [transparent]
|
||
* 8 ₈ 2088 ⁸ 2078 ° 00b0 ⊗ 2297 [transparent]
|
||
* 9 ₉ 2089 ⁹ 2079 [transparent]
|
||
* 0 ₀ 2080 ⁰ 2070 ∅ 2205 [transparent]
|
||
* - ₋ 208b ⁻ 207b ¬ 00ac ⊖ 2296 [transparent]
|
||
* = ₊ 208a ₋ 208b ≠ 2260 ⊕ 2295 [transparent]
|
||
* q θ 03b8 Θ 0398 ℚ 211a ק 05e7
|
||
* w ω 03c9 Ω 03a9 ש 05e9
|
||
* e ε 03b5 Ε 0395 ∃ 2203 ∄ 2204 ע 05e2
|
||
* r ρ 03c1 Ρ 03a1 ℝ 211d ר 05e8
|
||
* t τ 03c4 Τ 03a4 ט 05d8 תּ fb4a
|
||
* y ψ 03c8 Ψ 03a8 ∨ 2228 ∧ 2227 ײ 05f2 ײַ fb1f
|
||
* u υ 03c5 Υ 03a5 ∪ 222a ∩ 2229 ו 05d5 ױ 05f1
|
||
* i ι 03b9 Ι 0399 ∞ 221e ℵ 2135 י 05d9
|
||
* o ο 03bf Ο 039f ו 05d5 אָ fb2f
|
||
* p π 03c0 Π 03a0 ≡ 2261 ≢ 2262 פ 05e4 ף 05e3
|
||
* [ ± 00b1 ∓ 2213
|
||
* ] ≈ 2248 ≉ 2249
|
||
* \ ∼ 223c ≁ 2241
|
||
* a α 03b1 Α 0391 ∀ 2200 Å 212b א 05d0 אַ fb2e
|
||
* s σ 03c3 Σ 03a3 ∈ 2208 ∉ 2209 ס 05e1 ת 05ea
|
||
* d δ 03b4 Δ 0394 ⊂ 2282 ⊄ 2284 ד 05d3
|
||
* f φ 03c6 Φ 03a6 ⊆ 2286 ⊈ 2288 פֿ fb4e
|
||
* g γ 03b3 Γ 0393 ⊇ 2287 ⊉ 2289 ג 05d2
|
||
* h η 03b7 Η 0397 ← 2190 ⇐ 21d0 ה 05d4
|
||
* j ϑ 03d1 ↓ 2193 ⇓ 21d3 ח 05d7 כֿ fb4d
|
||
* k κ 03ba Κ 039a ↑ 2191 ⇑ 21d1 כ 05db ך 05da
|
||
* l λ 03bb Λ 039b → 2192 ⇒ 21d2 ל 05dc
|
||
* ; … 2026 ⋯ 22ef ↔ 2194 ⇔ 21d4
|
||
* ' · 00b7 • 2022 ∴ 2234 ⊙ 2299
|
||
* z ζ 03b6 Ζ 0396 ℤ 2124 ז 05d6
|
||
* x ξ 03be Ξ 039e ✘ 2718 צ 05e6 ץ 05e5
|
||
* c χ 03c7 Χ 03a7 ℂ 2102 כ 05db ך 05da
|
||
* v ς 03c2 ✔ 2714 √ 221a װ 05f0 בֿ fb4c
|
||
* b β 03b2 Β 0392 ב 05d1
|
||
* n ν 03bd Ν 039d ℕ 2115 נ 05e0 ן 05df
|
||
* m μ 03bc Μ 039c מ 05de ם 05dd
|
||
* , ≪ 226a ≫ 226b ∂ 2202 ∫ 222b
|
||
* . ≲ 2272 ≳ 2273 ≰ 2270 ≱ 2271
|
||
* / ⊘ 2298
|
||
*/
|
||
[_YIDDISH] = LAYOUT_all(
|
||
KC_TRNS, KC_TRNS, H(05f3), KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, H(05e7), H(05e9), H(05e2), H(05e8), H(05d8), H(05f2), H(05d5), H(05d9), H(05d5), H(05e4), KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, H(05d0), H(05e1), H(05d3), H(fb4e), H(05d2), H(05d4), H(05d7), H(05db), H(05dc), KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, H(05d6), H(05e6), H(05db), H(05f0), H(05d1), H(05e0), H(05de), KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS),
|
||
[_SHIFTYIDDISH] = LAYOUT_all(
|
||
KC_TRNS, KC_TRNS, H(05f4), KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, H(fb4a), H(fb1f), H(05f1), KC_TRNS, H(fb2f), H(05e3), KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, H(fb2e), H(05ea), KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, H(fb4d), H(05da), KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, H(05e5), H(05da), H(fb4c), KC_TRNS, H(05df), H(05dd), KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS),
|
||
[_GREEK] = LAYOUT_all(
|
||
KC_TRNS, KC_TRNS, XXXXXXX, H(2081), H(2082), H(2083), H(2084), H(2085), H(2086), H(2087), H(2088), H(2089), H(2080), H(208b), H(208a), KC_TRNS, KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, H(03b8), H(03c9), H(03b5), H(03c1), H(03c4), H(03c8), H(03c5), H(03b9), H(03bf), H(03c0), XXXXXXX, XXXXXXX, XXXXXXX, KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, H(03b1), H(03c3), H(03b4), H(03c6), H(03b3), H(03b7), XXXXXXX, H(03ba), H(03bb), H(2026), H(00b7), KC_TRNS, KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, H(03b6), H(03be), H(03c7), XXXXXXX, H(03b2), H(03bd), H(03bc), H(226a), H(2272), XXXXXXX, KC_TRNS, KC_TRNS, KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS),
|
||
[_SHIFTGREEK] = LAYOUT_all(
|
||
KC_TRNS, KC_TRNS, XXXXXXX, H(00b9), H(00b2), H(00b3), H(2074), H(2075), H(2076), H(2077), H(2078), H(2079), H(2070), H(207b), H(208b), KC_TRNS, KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, H(0398), H(03a9), H(0395), H(03a1), H(03a4), H(03a8), H(03a5), H(0399), H(039f), H(03a0), XXXXXXX, XXXXXXX, XXXXXXX, KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, H(0391), H(03a3), H(0394), H(03a6), H(0393), H(0397), H(03d1), H(039a), H(039b), H(22ef), H(2022), KC_TRNS, KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, H(0396), H(039e), H(03a7), H(03c2), H(0392), H(039d), H(039c), H(226b), H(2273), XXXXXXX, KC_TRNS, KC_TRNS, KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS),
|
||
[_CADET] = LAYOUT_all(
|
||
KC_TRNS, KC_TRNS, XXXXXXX, H(00a1), H(00ab), H(00a3), XXXXXXX, XXXXXXX, XXXXXXX, XXXXXXX, H(00b0), XXXXXXX, H(2205), H(00ac), H(2260), KC_TRNS, KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, H(211a), XXXXXXX, H(2203), H(211d), XXXXXXX, H(2228), H(222a), H(221e), XXXXXXX, H(2261), H(00b1), H(2248), H(223c), KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, H(2200), H(2208), H(2282), H(2286), H(2287), H(2190), H(2193), H(2191), H(2192), H(2194), H(2234), KC_TRNS, KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, H(2124), H(2718), H(2102), H(2714), XXXXXXX, H(2115), XXXXXXX, H(2202), H(2270), XXXXXXX, KC_TRNS, KC_TRNS, KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS),
|
||
[_SHIFTCADET] = LAYOUT_all(
|
||
KC_TRNS, KC_TRNS, XXXXXXX, H(00bf), H(00bb), H(20ac), XXXXXXX, XXXXXXX, XXXXXXX, XXXXXXX, H(2297), XXXXXXX, XXXXXXX, H(2296), H(2295), KC_TRNS, KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, XXXXXXX, XXXXXXX, H(2204), XXXXXXX, XXXXXXX, H(2227), H(2229), H(2135), XXXXXXX, H(2262), H(2213), H(2249), H(2241), KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, H(212b), H(2209), H(2284), H(2288), H(2289), H(21d0), H(21d3), H(21d1), H(21d2), H(21d4), H(2299), KC_TRNS, KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, XXXXXXX, XXXXXXX, XXXXXXX, H(221a), XXXXXXX, XXXXXXX, XXXXXXX, H(222b), H(2271), H(2298), KC_TRNS, KC_TRNS, KC_TRNS,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS, KC_TRNS),
|
||
// The function layer mostly contains the accent marks, but also has a few meta-control
|
||
// operations. The accent marks are placed by analogy with Mac OS.
|
||
[_FUNCTION] = LAYOUT_all(
|
||
QK_BOOT, KC_TRNS, KC_CGRV, KC_F1, KC_F2, KC_F3, KC_F4, KC_F5, KC_F6, KC_F7, KC_F8, KC_F9, KC_F10, KC_F11, KC_F12, XXXXXXX, XXXXXXX,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, XXXXXXX, XXXXXXX, KC_CAGU, XXXXXXX, XXXXXXX, XXXXXXX, KC_CDIA, KC_CCIR, XXXXXXX, XXXXXXX, XXXXXXX, XXXXXXX, XXXXXXX, XXXXXXX,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, XXXXXXX, XXXXXXX, XXXXXXX, XXXXXXX, XXXXXXX, XXXXXXX, XXXXXXX, XXXXXXX, XXXXXXX, XXXXXXX, XXXXXXX, XXXXXXX, XXXXXXX,
|
||
KC_TRNS, KC_TRNS, KC_TRNS, XXXXXXX, XXXXXXX, KC_CCED, XXXXXXX, XXXXXXX, KC_CTIL, XXXXXXX, XXXXXXX, XXXXXXX, XXXXXXX, KC_TRNS, XXXXXXX, XXXXXXX,
|
||
KC_TRNS, KC_TRNS, KC_RCTL, KC_RGUI, KC_RALT, KC_TRNS, KC_TRNS, XXXXXXX, XXXXXXX, XXXXXXX),
|
||
};
|
||
|
||
////////////////////////////////////////////////////////////////////////////////////////////////////
|
||
// Accent implementation
|
||
//
|
||
// In the body of process_record_user, we store an "accent_request", which is the accent keycode if
|
||
// one was just selected, or zero otherwise. When the *next* key is hit, we look up whether the
|
||
// accent request plus that next keycode (plus the state of the shift key) together amount to an
|
||
// interesting combined (NFKC) character, and if so, emit it; otherwise, we emit the accent as a
|
||
// separate character and then process the next key normally. The resulting UI behavior is similar
|
||
// to that of the combining accent keys in MacOS.
|
||
//
|
||
// We store two arrays, depending on whether shift is or isn't held. Each is two-dimensional, with
|
||
// its outer key by the next keycode struck, and the inner key by the accent requested. The outer
|
||
// array has KC_Z + 1 as its upper bound, so that we can save memory by only coding alphabetic keys.
|
||
// The contents are either Unicode code points, or zero to indicate that we don't have a point for
|
||
// this combination.
|
||
|
||
#define KC_NUM_ACCENTS (KC_ACCENT_END - KC_ACCENT_START)
|
||
#define KC_NUM_SLOTS (KC_Z + 1)
|
||
|
||
const uint16_t PROGMEM unshifted_accents[KC_NUM_SLOTS][KC_NUM_ACCENTS] = {
|
||
// KC_CGRV, KC_CAGU, KC_CDIA, KC_CCIR, KC_CCED, KC_CTIL
|
||
[KC_A] = { 0x00e0, 0x00e1, 0x00e4, 0x00e2, 0, 0x00e3 },
|
||
[KC_E] = { 0x00e8, 0x00e9, 0x00eb, 0x00ea, 0, 0 },
|
||
[KC_I] = { 0x00ec, 0x00ed, 0x00ef, 0x00ee, 0, 0 },
|
||
[KC_O] = { 0x00f2, 0x00f3, 0x00f6, 0x00f4, 0, 0x00f5 },
|
||
[KC_U] = { 0x00f9, 0x00fa, 0x00fc, 0x00fb, 0, 0 },
|
||
[KC_Y] = { 0, 0, 0x00ff, 0, 0, 0 },
|
||
[KC_N] = { 0, 0, 0, 0, 0, 0x00f1 },
|
||
[KC_C] = { 0, 0, 0, 0, 0x00e7, 0 },
|
||
};
|
||
|
||
const uint16_t PROGMEM shifted_accents[KC_NUM_SLOTS][KC_NUM_ACCENTS] = {
|
||
// KC_CGRV, KC_CAGU, KC_CDIA, KC_CCIR, KC_CCED, KC_CTIL
|
||
[KC_A] = { 0x00c0, 0x00c1, 0x00c4, 0x00c2, 0, 0x00c3 },
|
||
[KC_E] = { 0x00c8, 0x00c9, 0x00cb, 0x00ca, 0, 0 },
|
||
[KC_I] = { 0x00cc, 0x00cd, 0x00cf, 0x00ce, 0, 0 },
|
||
[KC_O] = { 0x00d2, 0x00d3, 0x00d6, 0x00d4, 0, 0x00d5 },
|
||
[KC_U] = { 0x00d9, 0x00da, 0x00dc, 0x00db, 0, 0 },
|
||
[KC_Y] = { 0, 0, 0x00df, 0, 0, 0 },
|
||
[KC_N] = { 0, 0, 0, 0, 0, 0x00d1 },
|
||
[KC_C] = { 0, 0, 0, 0, 0x00c7, 0 },
|
||
};
|
||
|
||
// The uncombined and combined forms of the accents, for when we want to emit them as single
|
||
// characters.
|
||
const uint16_t PROGMEM uncombined_accents[KC_NUM_ACCENTS] = {
|
||
[KC_CGRV - KC_ACCENT_START] = 0x0060,
|
||
[KC_CAGU - KC_ACCENT_START] = 0x00b4,
|
||
[KC_CDIA - KC_ACCENT_START] = 0x00a8,
|
||
[KC_CCIR - KC_ACCENT_START] = 0x005e,
|
||
[KC_CCED - KC_ACCENT_START] = 0x00b8,
|
||
[KC_CTIL - KC_ACCENT_START] = 0x02dc,
|
||
};
|
||
|
||
const uint16_t PROGMEM combined_accents[KC_NUM_ACCENTS] = {
|
||
[KC_CGRV - KC_ACCENT_START] = 0x0300,
|
||
[KC_CAGU - KC_ACCENT_START] = 0x0301,
|
||
[KC_CDIA - KC_ACCENT_START] = 0x0308,
|
||
[KC_CCIR - KC_ACCENT_START] = 0x0302,
|
||
[KC_CCED - KC_ACCENT_START] = 0x0327,
|
||
[KC_CTIL - KC_ACCENT_START] = 0x0303,
|
||
};
|
||
|
||
// This function manages keypresses that happen after an accent has been selected by an earlier
|
||
// keypress.
|
||
// Args:
|
||
// accent_key: The accent key which was earlier selected. This must be in the range
|
||
// [KC_ACCENT_START, KC_ACCENT_END).
|
||
// keycode: The keycode which was just pressed.
|
||
// is_shifted: The current shift state (as set by a combination of shift and caps lock)
|
||
//
|
||
// Returns true if the keycode has been completely handled by this function (and so should not be
|
||
// processed further by process_record_user) or false otherwise.
|
||
bool process_key_after_accent(
|
||
uint16_t accent_key,
|
||
uint16_t keycode,
|
||
bool is_shifted
|
||
) {
|
||
assert(accent_key >= KC_ACCENT_START);
|
||
assert(accent_key < KC_ACCENT_END);
|
||
const int accent_index = accent_key - KC_ACCENT_START;
|
||
|
||
// If the keycode is outside A..Z, we know we shouldn't even bother with a table lookup.
|
||
if (keycode <= KC_Z) {
|
||
// Pick the correct array. Because this is progmem, we're going to need to do the
|
||
// two-dimensional array indexing by hand, and so we just cast it to a single-dimensional array.
|
||
const uint16_t *points = (const uint16_t*)(is_shifted ? shifted_accents : unshifted_accents);
|
||
const uint16_t code_point = pgm_read_word(points + KC_NUM_ACCENTS * keycode + accent_index);
|
||
if (code_point) {
|
||
register_unicode(code_point);
|
||
return true;
|
||
}
|
||
}
|
||
|
||
// If we get here, there was no accent match. Emit the accent as its own character (i.e. a
|
||
// Unicode combining accent mark) and return false so that process_record_user also registers
|
||
// whatever is appropriate for the keycode after that. The host can figure out what to do with
|
||
// combining Unicode.
|
||
register_unicode(pgm_read_word(uncombined_accents + accent_index));
|
||
return false;
|
||
}
|
||
|
||
// This is a bitmask which selects the activation bits for layers *other* than our language
|
||
// selectors.
|
||
#define NON_LANGUAGE_LAYERS ~(((1UL << _LAST_LANGUAGE_LAYER) - 1) - ((1UL << _FIRST_LANGUAGE_LAYER) - 1))
|
||
|
||
// Update the current layer state and return the layer we're in.
|
||
uint8_t update_layer(
|
||
uint8_t layer_lock,
|
||
uint8_t layer_select_held,
|
||
bool shifted
|
||
) {
|
||
uint8_t current_layer = layer_lock;
|
||
layer_state_t language_layers = 0;
|
||
|
||
// If there's a layer select being held right now, then it updates the current layer.
|
||
// (If it's the layer select for the currently locked layer, then instead it's a toggle
|
||
// back to _QWERTY!)
|
||
if (layer_select_held != _QWERTY) {
|
||
current_layer = (layer_lock == layer_select_held ? _QWERTY : layer_select_held);
|
||
}
|
||
|
||
language_layers |= (1UL << current_layer);
|
||
|
||
// If we're shifted (with either shift or caps lock), and we're in one of our special
|
||
// layers, bump up to the SHIFTED version of that layer. We don't do this for QWERTY;
|
||
// there we just emit USB HID codes and let the host deal with shift.
|
||
if (shifted && current_layer != _QWERTY) {
|
||
++current_layer;
|
||
language_layers |= (1UL << current_layer);
|
||
}
|
||
|
||
// Update the QMK layer state by stomping just the language layer bits.
|
||
const layer_state_t new_layer_state = (layer_state & NON_LANGUAGE_LAYERS) | language_layers;
|
||
if (new_layer_state != layer_state) {
|
||
layer_state_set(new_layer_state);
|
||
}
|
||
|
||
return current_layer;
|
||
}
|
||
|
||
void set_os_mode(uint8_t new_mode) {
|
||
os_mode = new_mode;
|
||
// NB: We set unicode_config.input_mode directly, rather than calling
|
||
// set_unicode_input_mode, because we don't want to persist this and so we shouldn't put
|
||
// extra load on the EEPROMs.
|
||
unicode_config.input_mode = (os_mode == _MAC ? UNICODE_MODE_MACOS : UNICODE_MODE_WINCOMPOSE);
|
||
// Swap LALT and LGUI depending on Mac/Windows.
|
||
keymap_config.swap_lalt_lgui = (os_mode == _MAC);
|
||
// This would be a great moment for some auditory or visual feedback, but this keyboard
|
||
// doesn't support it. :(
|
||
}
|
||
|
||
void toggle_os_mode(void) {
|
||
set_os_mode((os_mode + 1) % _OS_MODES_MAX);
|
||
}
|
||
|
||
void keyboard_post_init_user() {
|
||
set_os_mode(_WINDOWS);
|
||
}
|
||
|
||
bool process_record_user(uint16_t keycode, keyrecord_t *record) {
|
||
// We track these persistent globals and manage them on our own, rather than trying to rely on
|
||
// get_mods or the like, because this function is called *before* that's updated!
|
||
static bool shift_held = false;
|
||
static bool alt_held = false;
|
||
static bool ctrl_held = false;
|
||
static bool super_held = false;
|
||
|
||
// These are where we remember the values of lock states.
|
||
static bool shift_lock = false;
|
||
static uint8_t layer_lock = _QWERTY; // The currently locked layer
|
||
static uint8_t next_layer_lock = _QWERTY; // Used when layer_lock is held
|
||
// Which layer select key is currently being held down. _QWERTY is equivalent to "none."
|
||
static uint8_t layer_select_held = _QWERTY;
|
||
|
||
// When the hold on the platform key started
|
||
static uint16_t platform_hold_start = 0;
|
||
|
||
// The accent request, or zero if there isn't one.
|
||
static uint16_t accent_request = 0;
|
||
|
||
// What kind of key we're striking right now, so that we know what to do if any accent requests
|
||
// are hanging around.
|
||
uint8_t key_type = _NORMAL_KEY;
|
||
|
||
// The layer selection and locking logic is:
|
||
// * By default, the current layer is given by saved value layer_lock.
|
||
// * If a layer select key is held down, we update the current layer to that value.
|
||
// (But special thing: If the current layer lock is <layer> and you hit the select key
|
||
// for <layer>, it instead toggles the current layer back to _QWERTY! That way you can
|
||
// insert some QWERTY keys in the midst of other-layer text.)
|
||
// * If the KC_LAYER_LOCK key is held down and a layer select key gets pressed, we update
|
||
// next_layer_lock to that selected layer. When KC_LAYER_LOCK is released, we update
|
||
// layer_lock to next_layer_lock. Note that that simply tapping KC_LAYER_LOCK resets
|
||
// layer_lock to _QWERTY.
|
||
// * After all of this is done, we check if shift is held (via either shift or caps lock);
|
||
// if it is, and our current layer isn't _QWERTY, then we bump the current layer ID by 1
|
||
// to get the shifted layer.
|
||
|
||
// Step 1: Process various interesting keycodes, especially ones that update our running
|
||
// state variables.
|
||
switch (keycode) {
|
||
// Monitoring the modifier keys, because we'll need them for our logic!
|
||
case KC_LSFT:
|
||
case KC_RSFT:
|
||
shift_held = record->event.pressed;
|
||
key_type = _MODIFIER_KEY;
|
||
break;
|
||
case KC_CAPS:
|
||
// If we're in QWERTY mode, caps lock is already going to be managed by the host OS, but by
|
||
// tracking it ourselves we can also usefully apply it to the GREEK and CADET layers.
|
||
shift_lock = !shift_lock;
|
||
key_type = _MODIFIER_KEY;
|
||
break;
|
||
case KC_LALT:
|
||
case KC_RALT:
|
||
alt_held = record->event.pressed;
|
||
key_type = _MODIFIER_KEY;
|
||
break;
|
||
case KC_LCTL:
|
||
case KC_RCTL:
|
||
ctrl_held = record->event.pressed;
|
||
key_type = _MODIFIER_KEY;
|
||
break;
|
||
case KC_LGUI:
|
||
case KC_RGUI:
|
||
super_held = record->event.pressed;
|
||
key_type = _MODIFIER_KEY;
|
||
break;
|
||
|
||
case KC_LAYER_LOCK:
|
||
if (record->event.pressed) {
|
||
// On press, get ready for a layer selection.
|
||
next_layer_lock = _QWERTY;
|
||
} else {
|
||
// On release, propagate next_layer_lock to layer_lock.
|
||
layer_lock = next_layer_lock;
|
||
}
|
||
key_type = _MODIFIER_KEY;
|
||
break;
|
||
|
||
// Layer selectors
|
||
case KC_GREEK:
|
||
if (record->event.pressed) {
|
||
layer_select_held = _GREEK;
|
||
next_layer_lock = _GREEK;
|
||
} else {
|
||
layer_select_held = _QWERTY;
|
||
}
|
||
key_type = _MODIFIER_KEY;
|
||
break;
|
||
case KC_CADET:
|
||
if (record->event.pressed) {
|
||
layer_select_held = _CADET;
|
||
next_layer_lock = _CADET;
|
||
} else {
|
||
layer_select_held = _QWERTY;
|
||
}
|
||
key_type = _MODIFIER_KEY;
|
||
break;
|
||
case KC_YIDDISH:
|
||
if (record->event.pressed) {
|
||
layer_select_held = _YIDDISH;
|
||
next_layer_lock = _YIDDISH;
|
||
} else {
|
||
layer_select_held = _QWERTY;
|
||
}
|
||
key_type = _MODIFIER_KEY;
|
||
break;
|
||
|
||
// Accent selectors
|
||
case KC_CGRV:
|
||
case KC_CAGU:
|
||
case KC_CDIA:
|
||
case KC_CCIR:
|
||
case KC_CCED:
|
||
case KC_CTIL:
|
||
// The accent request keys normally update accent_request (whose effect will trigger the next
|
||
// time we see a "normal" key pressed). However, shift+accent request will instead immediately
|
||
// generate the Unicode combining accent symbol instead.
|
||
if (shift_held) {
|
||
register_unicode(pgm_read_word(combined_accents + keycode - KC_ACCENT_START));
|
||
return false;
|
||
} else {
|
||
accent_request = keycode;
|
||
}
|
||
key_type = _MODIFIER_KEY;
|
||
break;
|
||
|
||
// Our special keycodes
|
||
case KC_PLATFORM:
|
||
if (record->event.pressed) {
|
||
platform_hold_start = record->event.time;
|
||
} else if (platform_hold_start != 0 && record->event.time - platform_hold_start > PLATFORM_HOLD_DURATION) {
|
||
toggle_os_mode();
|
||
}
|
||
key_type = _SPECIAL_KEY;
|
||
return true;
|
||
|
||
case KC_VC_MUTE:
|
||
if (record->event.pressed) {
|
||
send_string(VC_MUTE_VALUES[os_mode]);
|
||
return true;
|
||
}
|
||
key_type = _SPECIAL_KEY;
|
||
break;
|
||
|
||
case KC_VC_HAND:
|
||
if (record->event.pressed) {
|
||
send_string(VC_HAND_VALUES[os_mode]);
|
||
return true;
|
||
}
|
||
key_type = _SPECIAL_KEY;
|
||
break;
|
||
|
||
case KC_SCRNSHT:
|
||
if (record->event.pressed) {
|
||
send_string(SCRNSHT_VALUES[os_mode]);
|
||
return true;
|
||
}
|
||
key_type = _SPECIAL_KEY;
|
||
break;
|
||
|
||
case QK_BOOT:
|
||
key_type = _SPECIAL_KEY;
|
||
break;
|
||
}
|
||
|
||
// Step 2: Finalize current_layer and update the QMK layer state.
|
||
const bool shifted = (shift_held != shift_lock);
|
||
const uint8_t current_layer = update_layer(layer_lock, layer_select_held, shifted);
|
||
|
||
// Step 3: Handle accents.
|
||
bool handled = false;
|
||
if (accent_request && record->event.pressed) {
|
||
// If we're in any layer other than _QWERTY, or a modifier key is being held down,
|
||
// then we're actually generating a special key, not a normal one.
|
||
if (key_type == _NORMAL_KEY &&
|
||
(current_layer != _QWERTY || ctrl_held || super_held || alt_held)) {
|
||
key_type = _SPECIAL_KEY;
|
||
}
|
||
|
||
switch (key_type) {
|
||
case _NORMAL_KEY:
|
||
handled = process_key_after_accent(accent_request, keycode, shifted);
|
||
accent_request = 0;
|
||
break;
|
||
|
||
case _SPECIAL_KEY:
|
||
accent_request = 0;
|
||
break;
|
||
|
||
case _MODIFIER_KEY:
|
||
break;
|
||
}
|
||
}
|
||
|
||
return !handled;
|
||
}
|