feat: add dutch corpus (#136)

Inclues US/QWERTY layout (used in the Netherlands) ad AZERTY layout (used in Belgium).

packages/keybr-content-words/lib/load.ts

@@ -66,6 +66,13 @@ export async function loadWordList(language: Language): Promise<WordList> {
           "./data/words-it.json"
         )
       ).default as unknown as WordList;
+    case Language.NL:
+      return (
+        await import(
+          /* webpackChunkName: "words-nl" */
+          "./data/words-nl.json"
+        )
+      ).default as unknown as WordList;
     case Language.PL:
       return (
         await import(

packages/keybr-keyboard-generator/lib/index.ts

@@ -73,6 +73,14 @@ const files: readonly [input: string | LayoutConfig, output: string][] = [
     "cldr-keyboards-43.0/keyboards/windows/it-t-k0-windows.xml",
     "../keybr-keyboard/lib/data/layout/it_it-win.ts",
   ],
+  [
+    "cldr-keyboards-43.0/keyboards/windows/fr-t-k0-windows.xml",
+    "../keybr-keyboard/lib/data/layout/nl_be-win.ts",
+  ],
+  [
+    "cldr-keyboards-43.0/keyboards/windows/en-t-k0-windows.xml",
+    "../keybr-keyboard/lib/data/layout/nl_nl-win.ts",
+  ],
   [
     "cldr-keyboards-43.0/keyboards/windows/pl-t-k0-windows-extended.xml",
     "../keybr-keyboard/lib/data/layout/pl_pl-win.ts",

packages/keybr-keyboard/lib/data/layout.ts

@@ -24,6 +24,8 @@ export { LAYOUT_FR_OPTIMOT_ERGO } from "./layout/fr_optimot_ergo.ts";
 export { LAYOUT_HE_IL_WIN } from "./layout/he_il-win.ts";
 export { LAYOUT_IT_IT_WIN } from "./layout/it_it-win.ts";
 export { LAYOUT_JA_JP_JIS } from "./layout/ja_jp_jis.ts";
+export { LAYOUT_NL_BE_WIN } from "./layout/nl_be-win.ts";
+export { LAYOUT_NL_NL_WIN } from "./layout/nl_nl-win.ts";
 export { LAYOUT_PL_PL_WIN } from "./layout/pl_pl-win.ts";
 export { LAYOUT_PT_BR_WIN } from "./layout/pt_br-win.ts";
 export { LAYOUT_PT_PT_WIN } from "./layout/pt_pt-win.ts";

packages/keybr-keyboard/lib/data/layout/nl_be-win.ts

@@ -0,0 +1,55 @@
+// Generated file, do not edit.
+
+import { type CodePointDict } from "../../types.ts";
+
+// prettier-ignore
+export const LAYOUT_NL_BE_WIN: CodePointDict = {
+  Backquote: [/* ² */ 0x00b2],
+  Digit1: [/* & */ 0x0026, /* 1 */ 0x0031],
+  Digit2: [/* é */ 0x00e9, /* 2 */ 0x0032, /* DEAD TILDE */ 0x0303],
+  Digit3: [/* " */ 0x0022, /* 3 */ 0x0033, /* # */ 0x0023],
+  Digit4: [/* ' */ 0x0027, /* 4 */ 0x0034, /* { */ 0x007b],
+  Digit5: [/* ( */ 0x0028, /* 5 */ 0x0035, /* [ */ 0x005b],
+  Digit6: [/* - */ 0x002d, /* 6 */ 0x0036, /* | */ 0x007c],
+  Digit7: [/* è */ 0x00e8, /* 7 */ 0x0037, /* DEAD GRAVE ACCENT */ 0x0300],
+  Digit8: [/* _ */ 0x005f, /* 8 */ 0x0038, /* \ */ 0x005c],
+  Digit9: [/* ç */ 0x00e7, /* 9 */ 0x0039, /* ^ */ 0x005e],
+  Digit0: [/* à */ 0x00e0, /* 0 */ 0x0030, /* @ */ 0x0040],
+  Minus: [/* ) */ 0x0029, /* ° */ 0x00b0, /* ] */ 0x005d],
+  Equal: [/* = */ 0x003d, /* + */ 0x002b, /* } */ 0x007d],
+  KeyQ: [/* a */ 0x0061, /* A */ 0x0041],
+  KeyW: [/* z */ 0x007a, /* Z */ 0x005a],
+  KeyE: [/* e */ 0x0065, /* E */ 0x0045, /* € */ 0x20ac],
+  KeyR: [/* r */ 0x0072, /* R */ 0x0052],
+  KeyT: [/* t */ 0x0074, /* T */ 0x0054],
+  KeyY: [/* y */ 0x0079, /* Y */ 0x0059],
+  KeyU: [/* u */ 0x0075, /* U */ 0x0055],
+  KeyI: [/* i */ 0x0069, /* I */ 0x0049],
+  KeyO: [/* o */ 0x006f, /* O */ 0x004f],
+  KeyP: [/* p */ 0x0070, /* P */ 0x0050],
+  BracketLeft: [/* DEAD CIRCUMFLEX ACCENT */ 0x0302, /* DEAD DIAERESIS */ 0x0308],
+  BracketRight: [/* $ */ 0x0024, /* £ */ 0x00a3, /* ¤ */ 0x00a4],
+  Backslash: [/* * */ 0x002a, /* µ */ 0x00b5],
+  KeyA: [/* q */ 0x0071, /* Q */ 0x0051],
+  KeyS: [/* s */ 0x0073, /* S */ 0x0053],
+  KeyD: [/* d */ 0x0064, /* D */ 0x0044],
+  KeyF: [/* f */ 0x0066, /* F */ 0x0046],
+  KeyG: [/* g */ 0x0067, /* G */ 0x0047],
+  KeyH: [/* h */ 0x0068, /* H */ 0x0048],
+  KeyJ: [/* j */ 0x006a, /* J */ 0x004a],
+  KeyK: [/* k */ 0x006b, /* K */ 0x004b],
+  KeyL: [/* l */ 0x006c, /* L */ 0x004c],
+  Semicolon: [/* m */ 0x006d, /* M */ 0x004d],
+  Quote: [/* ù */ 0x00f9, /* % */ 0x0025],
+  IntlBackslash: [/* < */ 0x003c, /* > */ 0x003e],
+  KeyZ: [/* w */ 0x0077, /* W */ 0x0057],
+  KeyX: [/* x */ 0x0078, /* X */ 0x0058],
+  KeyC: [/* c */ 0x0063, /* C */ 0x0043],
+  KeyV: [/* v */ 0x0076, /* V */ 0x0056],
+  KeyB: [/* b */ 0x0062, /* B */ 0x0042],
+  KeyN: [/* n */ 0x006e, /* N */ 0x004e],
+  KeyM: [/* , */ 0x002c, /* ? */ 0x003f],
+  Comma: [/* ; */ 0x003b, /* . */ 0x002e],
+  Period: [/* : */ 0x003a, /* / */ 0x002f],
+  Slash: [/* ! */ 0x0021, /* § */ 0x00a7],
+};

packages/keybr-keyboard/lib/data/layout/nl_nl-win.ts

@@ -0,0 +1,55 @@
+// Generated file, do not edit.
+
+import { type CodePointDict } from "../../types.ts";
+
+// prettier-ignore
+export const LAYOUT_NL_NL_WIN: CodePointDict = {
+  Backquote: [/* ` */ 0x0060, /* ~ */ 0x007e],
+  Digit1: [/* 1 */ 0x0031, /* ! */ 0x0021],
+  Digit2: [/* 2 */ 0x0032, /* @ */ 0x0040],
+  Digit3: [/* 3 */ 0x0033, /* # */ 0x0023],
+  Digit4: [/* 4 */ 0x0034, /* $ */ 0x0024],
+  Digit5: [/* 5 */ 0x0035, /* % */ 0x0025],
+  Digit6: [/* 6 */ 0x0036, /* ^ */ 0x005e],
+  Digit7: [/* 7 */ 0x0037, /* & */ 0x0026],
+  Digit8: [/* 8 */ 0x0038, /* * */ 0x002a],
+  Digit9: [/* 9 */ 0x0039, /* ( */ 0x0028],
+  Digit0: [/* 0 */ 0x0030, /* ) */ 0x0029],
+  Minus: [/* - */ 0x002d, /* _ */ 0x005f],
+  Equal: [/* = */ 0x003d, /* + */ 0x002b],
+  KeyQ: [/* q */ 0x0071, /* Q */ 0x0051],
+  KeyW: [/* w */ 0x0077, /* W */ 0x0057],
+  KeyE: [/* e */ 0x0065, /* E */ 0x0045],
+  KeyR: [/* r */ 0x0072, /* R */ 0x0052],
+  KeyT: [/* t */ 0x0074, /* T */ 0x0054],
+  KeyY: [/* y */ 0x0079, /* Y */ 0x0059],
+  KeyU: [/* u */ 0x0075, /* U */ 0x0055],
+  KeyI: [/* i */ 0x0069, /* I */ 0x0049],
+  KeyO: [/* o */ 0x006f, /* O */ 0x004f],
+  KeyP: [/* p */ 0x0070, /* P */ 0x0050],
+  BracketLeft: [/* [ */ 0x005b, /* { */ 0x007b],
+  BracketRight: [/* ] */ 0x005d, /* } */ 0x007d],
+  Backslash: [/* \ */ 0x005c, /* | */ 0x007c],
+  KeyA: [/* a */ 0x0061, /* A */ 0x0041],
+  KeyS: [/* s */ 0x0073, /* S */ 0x0053],
+  KeyD: [/* d */ 0x0064, /* D */ 0x0044],
+  KeyF: [/* f */ 0x0066, /* F */ 0x0046],
+  KeyG: [/* g */ 0x0067, /* G */ 0x0047],
+  KeyH: [/* h */ 0x0068, /* H */ 0x0048],
+  KeyJ: [/* j */ 0x006a, /* J */ 0x004a],
+  KeyK: [/* k */ 0x006b, /* K */ 0x004b],
+  KeyL: [/* l */ 0x006c, /* L */ 0x004c],
+  Semicolon: [/* ; */ 0x003b, /* : */ 0x003a],
+  Quote: [/* ' */ 0x0027, /* " */ 0x0022],
+  IntlBackslash: [/* \ */ 0x005c, /* | */ 0x007c],
+  KeyZ: [/* z */ 0x007a, /* Z */ 0x005a],
+  KeyX: [/* x */ 0x0078, /* X */ 0x0058],
+  KeyC: [/* c */ 0x0063, /* C */ 0x0043],
+  KeyV: [/* v */ 0x0076, /* V */ 0x0056],
+  KeyB: [/* b */ 0x0062, /* B */ 0x0042],
+  KeyN: [/* n */ 0x006e, /* N */ 0x004e],
+  KeyM: [/* m */ 0x006d, /* M */ 0x004d],
+  Comma: [/* , */ 0x002c, /* < */ 0x003c],
+  Period: [/* . */ 0x002e, /* > */ 0x003e],
+  Slash: [/* / */ 0x002f, /* ? */ 0x003f],
+};

packages/keybr-keyboard/lib/load.ts

@@ -39,6 +39,8 @@ import {
   LAYOUT_HE_IL_WIN,
   LAYOUT_IT_IT_WIN,
   LAYOUT_JA_JP_JIS,
+  LAYOUT_NL_BE_WIN,
+  LAYOUT_NL_NL_WIN,
   LAYOUT_PL_PL_WIN,
   LAYOUT_PT_BR_WIN,
   LAYOUT_PT_PT_WIN,
@@ -77,6 +79,8 @@ const layoutDict = new Map<Layout, CodePointDict>([
   [Layout.FR_OPTIMOT_ERGO, LAYOUT_FR_OPTIMOT_ERGO],
   [Layout.HE_IL, LAYOUT_HE_IL_WIN],
   [Layout.IT_IT, LAYOUT_IT_IT_WIN],
+  [Layout.NL_BE, LAYOUT_NL_BE_WIN],
+  [Layout.NL_NL, LAYOUT_NL_NL_WIN],
   [Layout.PL_PL, LAYOUT_PL_PL_WIN],
   [Layout.PT_BR, LAYOUT_PT_BR_WIN],
   [Layout.PT_PT, LAYOUT_PT_PT_WIN],

packages/keybr-layout/lib/language.ts

@@ -10,6 +10,7 @@ export class Language implements EnumItem {
   static readonly FR = new Language("fr", "latin", "ltr");
   static readonly HE = new Language("he", "hebrew", "rtl");
   static readonly IT = new Language("it", "latin", "ltr");
+  static readonly NL = new Language("nl", "latin", "ltr");
   static readonly PL = new Language("pl", "latin", "ltr");
   static readonly PT = new Language("pt", "latin", "ltr");
   static readonly RU = new Language("ru", "cyrillic", "ltr");
@@ -27,6 +28,7 @@ export class Language implements EnumItem {
     Language.FR,
     Language.HE,
     Language.IT,
+    Language.NL,
     Language.PL,
     Language.PT,
     Language.RU,

packages/keybr-layout/lib/layout.ts

@@ -304,6 +304,36 @@ export class Layout implements XEnumItem {
       Geometry.MATRIX,
     ),
   );
+  static readonly NL_BE = new Layout(
+    /* id= */ "nl-azerty",
+    /* xid= */ 0x91,
+    /* name= */ "Belgium",
+    /* family= */ LayoutFamily.AZERTY,
+    /* language= */ Language.NL,
+    /* emulate= */ false,
+    /* geometries= */ new Enum(
+      Geometry.STANDARD_101,
+      Geometry.STANDARD_101_FULL,
+      Geometry.STANDARD_102,
+      Geometry.STANDARD_102_FULL,
+      Geometry.MATRIX,
+    ),
+  );
+  static readonly NL_NL = new Layout(
+    /* id= */ "nl",
+    /* xid= */ 0x90,
+    /* name= */ "US",
+    /* family= */ LayoutFamily.QWERTY,
+    /* language= */ Language.NL,
+    /* emulate= */ false,
+    /* geometries= */ new Enum(
+      Geometry.STANDARD_101,
+      Geometry.STANDARD_101_FULL,
+      Geometry.STANDARD_102,
+      Geometry.STANDARD_102_FULL,
+      Geometry.MATRIX,
+    ),
+  );
   static readonly ES_ES = new Layout(
     /* id= */ "es",
     /* xid= */ 0x50,
@@ -514,6 +544,8 @@ export class Layout implements XEnumItem {
     Layout.FR_OPTIMOT_ERGO,
     Layout.HE_IL,
     Layout.IT_IT,
+    Layout.NL_BE,
+    Layout.NL_NL,
     Layout.ES_ES,
     Layout.PL_PL,
     Layout.PT_BR,

packages/keybr-phonetic-model-loader/lib/assets.ts

@@ -8,6 +8,7 @@ import langEsData from "@keybr/phonetic-model/assets/lang-es.data";
 import langFrData from "@keybr/phonetic-model/assets/lang-fr.data";
 import langHeData from "@keybr/phonetic-model/assets/lang-he.data";
 import langItData from "@keybr/phonetic-model/assets/lang-it.data";
+import langNlData from "@keybr/phonetic-model/assets/lang-nl.data";
 import langPlData from "@keybr/phonetic-model/assets/lang-pl.data";
 import langPtData from "@keybr/phonetic-model/assets/lang-pt.data";
 import langRuData from "@keybr/phonetic-model/assets/lang-ru.data";
@@ -35,6 +36,8 @@ export function modelAssetPath(language: Language): string {
       return langHeData;
     case Language.IT:
       return langItData;
+    case Language.NL:
+      return langNlData;
     case Language.PL:
       return langPlData;
     case Language.PT:

packages/keybr-phonetic-model/lib/blacklist/blacklist-nl.json

@@ -0,0 +1 @@
+[]

packages/keybr-phonetic-model/lib/blacklist/blacklist.ts

@@ -5,6 +5,7 @@ import blacklistEn from "./blacklist-en.json";
 import blacklistEs from "./blacklist-es.json";
 import blacklistFr from "./blacklist-fr.json";
 import blacklistIt from "./blacklist-it.json";
+import blacklistNl from "./blacklist-nl.json";
 import blacklistPl from "./blacklist-pl.json";
 import blacklistPt from "./blacklist-pt.json";
 import blacklistRu from "./blacklist-ru.json";
@@ -25,6 +26,7 @@ const blacklistByLanguage = ((items: [Language, string[]][]) =>
   [Language.ES, blacklistEs],
   [Language.FR, blacklistFr],
   [Language.IT, blacklistIt],
+  [Language.NL, blacklistNl],
   [Language.PL, blacklistPl],
   [Language.PT, blacklistPt],
   [Language.RU, blacklistRu],

packages/keybr-phonetic-model/lib/generate/languages.ts

@@ -15,6 +15,7 @@ export const languages: readonly Language[] = [
   { id: "fr", alphabet: "aàbcçdeéèfghijlmnopqrstuvxyz" },
   { id: "he", alphabet: "אבגדהוזחטיכךלמםנןסעפףצץקרשת" },
   { id: "it", alphabet: "abcdefghijklmnopqrstuvwxyz" },
+  { id: "nl", alphabet: "abcdefghijklmnopqrstuvwxyz" },
   { id: "pl", alphabet: "aąbcćdeęfghijklłmnńoóprsśtuwyzźż" },
   { id: "pt", alphabet: "aáâãàbcçdeéêfghiíjklmnoóôõpqrstuúvwxyz" },
   { id: "ru", alphabet: "абвгдежзийклмнопрстуфхцчшщъыьэюя" },