@lokascript/semantic 1.0.0

package/src/generators/profiles/korean.ts

@@ -0,0 +1,127 @@
+/**
+ * Korean Language Profile
+ *
+ * SOV word order, particles (을/를, 에, 에서, etc.), space-separated between words.
+ * Agglutinative language with particles attaching to words.
+ */
+
+import type { LanguageProfile } from './types';
+
+export const koreanProfile: LanguageProfile = {
+  code: 'ko',
+  name: 'Korean',
+  nativeName: '한국어',
+  direction: 'ltr',
+  wordOrder: 'SOV',
+  markingStrategy: 'particle',
+  usesSpaces: true, // Korean uses spaces between words, but particles attach
+  verb: {
+    position: 'end',
+    suffixes: ['다', '요', '니다', '세요'],
+    subjectDrop: true,
+  },
+  references: {
+    me: '나', // "I/me" (informal)
+    it: '그것', // "it"
+    you: '너', // "you" (informal)
+    result: '결과',
+    event: '이벤트',
+    target: '대상',
+    body: '본문',
+  },
+  possessive: {
+    marker: '의', // Possessive particle
+    markerPosition: 'between',
+    specialForms: {
+      me: '내', // Contracted form of 나의 (my)
+      it: '그것의', // "its"
+      you: '네', // Contracted form of 너의 (your)
+    },
+    keywords: {
+      내: 'me', // nae (my)
+      네: 'you', // ne (your)
+      그의: 'it', // geu-ui (its/his)
+    },
+  },
+  roleMarkers: {
+    patient: { primary: '을', alternatives: ['를'], position: 'after' },
+    destination: { primary: '에', alternatives: ['으로', '로', '에서'], position: 'after' },
+    source: { primary: '에서', alternatives: ['부터'], position: 'after' },
+    style: { primary: '로', alternatives: ['으로'], position: 'after' },
+    event: { primary: '을', alternatives: ['를'], position: 'after' }, // Event as object marker
+  },
+  keywords: {
+    // Class/Attribute operations
+    toggle: { primary: '토글', alternatives: ['전환'], normalized: 'toggle' },
+    add: { primary: '추가', normalized: 'add' },
+    remove: { primary: '제거', alternatives: ['삭제'], normalized: 'remove' },
+    // Content operations
+    put: { primary: '넣다', alternatives: ['넣기', '놓기'], normalized: 'put' },
+    append: { primary: '추가', normalized: 'append' },
+    take: { primary: '가져오다', normalized: 'take' },
+    make: { primary: '만들다', normalized: 'make' },
+    clone: { primary: '복사', normalized: 'clone' },
+    swap: { primary: '교환', alternatives: ['바꾸다'], normalized: 'swap' },
+    morph: { primary: '변형', alternatives: ['변환'], normalized: 'morph' },
+    // Variable operations
+    set: { primary: '설정', normalized: 'set' },
+    get: { primary: '얻다', alternatives: ['가져오기'], normalized: 'get' },
+    increment: { primary: '증가', normalized: 'increment' },
+    decrement: { primary: '감소', normalized: 'decrement' },
+    log: { primary: '로그', normalized: 'log' },
+    // Visibility
+    show: { primary: '보이다', alternatives: ['표시', '보이기'], normalized: 'show' },
+    hide: { primary: '숨기다', alternatives: ['숨기기'], normalized: 'hide' },
+    transition: { primary: '전환', normalized: 'transition' },
+    // Events
+    on: { primary: '에', alternatives: ['시', '때', '할 때'], normalized: 'on' },
+    trigger: { primary: '트리거', normalized: 'trigger' },
+    send: { primary: '보내다', normalized: 'send' },
+    // DOM focus
+    focus: { primary: '포커스', normalized: 'focus' },
+    blur: { primary: '블러', normalized: 'blur' },
+    // Navigation
+    go: { primary: '이동', normalized: 'go' },
+    // Async
+    wait: { primary: '대기', normalized: 'wait' },
+    fetch: { primary: '가져오기', normalized: 'fetch' },
+    settle: { primary: '안정', normalized: 'settle' },
+    // Control flow
+    if: { primary: '만약', normalized: 'if' },
+    when: { primary: '때', normalized: 'when' },
+    where: { primary: '어디', normalized: 'where' },
+    else: { primary: '아니면', normalized: 'else' },
+    repeat: { primary: '반복', normalized: 'repeat' },
+    for: { primary: '동안', normalized: 'for' },
+    while: { primary: '동안', normalized: 'while' },
+    continue: { primary: '계속', normalized: 'continue' },
+    halt: { primary: '정지', normalized: 'halt' },
+    throw: { primary: '던지다', normalized: 'throw' },
+    call: { primary: '호출', normalized: 'call' },
+    return: { primary: '반환', normalized: 'return' },
+    then: { primary: '그다음', alternatives: ['그리고', '그런후'], normalized: 'then' },
+    and: { primary: '그리고', alternatives: ['또한', '및'], normalized: 'and' },
+    end: { primary: '끝', alternatives: ['종료', '마침'], normalized: 'end' },
+    // Advanced
+    js: { primary: 'JS실행', alternatives: ['js'], normalized: 'js' },
+    async: { primary: '비동기', normalized: 'async' },
+    tell: { primary: '말하다', normalized: 'tell' },
+    default: { primary: '기본값', normalized: 'default' },
+    init: { primary: '초기화', normalized: 'init' },
+    behavior: { primary: '동작', normalized: 'behavior' },
+    install: { primary: '설치', normalized: 'install' },
+    measure: { primary: '측정', normalized: 'measure' },
+    // Modifiers
+    into: { primary: '으로', normalized: 'into' },
+    before: { primary: '전에', normalized: 'before' },
+    after: { primary: '후에', normalized: 'after' },
+    // Event modifiers (for repeat until event)
+    until: { primary: '까지', normalized: 'until' },
+    event: { primary: '이벤트', normalized: 'event' },
+    from: { primary: '에서', normalized: 'from' },
+  },
+  tokenization: {
+    particles: ['을', '를', '이', '가', '은', '는', '에', '에서', '으로', '로', '와', '과', '도'],
+    boundaryStrategy: 'space',
+  },
+};

package/src/generators/profiles/marker-templates.ts

@@ -0,0 +1,288 @@
+/**
+ * Marker Templates
+ *
+ * Shared role marker configurations for language families.
+ * These templates reduce duplication across language profiles.
+ *
+ * Usage:
+ *   import { SVO_PREPOSITION_MARKERS } from './marker-templates';
+ *   roleMarkers: { ...SVO_PREPOSITION_MARKERS, destination: { ...custom } }
+ */
+
+import type { RoleMarker, VerbConfig, PossessiveConfig, MarkingStrategy } from './types';
+import type { SemanticRole } from '../../types';
+
+// =============================================================================
+// Role Marker Templates by Word Order
+// =============================================================================
+
+/**
+ * Common role markers for SVO languages with prepositions.
+ * Used by: English, Spanish, French, German, Portuguese, Italian, Russian, Indonesian
+ */
+export const SVO_PREPOSITION_MARKERS: Partial<Record<SemanticRole, RoleMarker>> = {
+  patient: { primary: '', position: 'before' },
+  destination: { primary: 'on', position: 'before' },
+  source: { primary: 'from', position: 'before' },
+  style: { primary: 'with', position: 'before' },
+};
+
+/**
+ * Common role markers for SOV languages with particles/postpositions.
+ * Used by: Japanese, Korean, Turkish, Quechua
+ */
+export const SOV_PARTICLE_MARKERS: Partial<Record<SemanticRole, RoleMarker>> = {
+  patient: { primary: 'を', position: 'after' },
+  destination: { primary: 'に', position: 'after' },
+  source: { primary: 'から', position: 'after' },
+  style: { primary: 'で', position: 'after' },
+};
+
+/**
+ * Korean-specific role markers (vowel harmony variants).
+ */
+export const KOREAN_PARTICLE_MARKERS: Partial<Record<SemanticRole, RoleMarker>> = {
+  patient: { primary: '을', alternatives: ['를'], position: 'after' },
+  destination: { primary: '에', alternatives: ['에게', '한테'], position: 'after' },
+  source: { primary: '에서', alternatives: ['로부터'], position: 'after' },
+  style: { primary: '로', alternatives: ['으로'], position: 'after' },
+};
+
+/**
+ * Turkish-specific role markers (vowel harmony).
+ */
+export const TURKISH_SUFFIX_MARKERS: Partial<Record<SemanticRole, RoleMarker>> = {
+  patient: { primary: 'i', alternatives: ['ı', 'u', 'ü'], position: 'after' },
+  destination: { primary: 'e', alternatives: ['a', 'de', 'da'], position: 'after' },
+  source: { primary: 'den', alternatives: ['dan'], position: 'after' },
+  style: { primary: 'le', alternatives: ['la', 'ile'], position: 'after' },
+};
+
+/**
+ * Arabic role markers (VSO, RTL).
+ */
+export const ARABIC_PREPOSITION_MARKERS: Partial<Record<SemanticRole, RoleMarker>> = {
+  patient: { primary: '', position: 'before' },
+  destination: { primary: 'على', alternatives: ['إلى', 'في'], position: 'before' },
+  source: { primary: 'من', position: 'before' },
+  style: { primary: 'بـ', alternatives: ['مع'], position: 'before' },
+};
+
+// =============================================================================
+// Verb Configuration Templates
+// =============================================================================
+
+/**
+ * SVO verb configuration (verb at start, optional subject drop).
+ */
+export const SVO_VERB_CONFIG: VerbConfig = {
+  position: 'start',
+  subjectDrop: false,
+};
+
+/**
+ * SVO verb configuration for pro-drop languages (Spanish, Italian, Portuguese).
+ */
+export const SVO_PRODROP_VERB_CONFIG: VerbConfig = {
+  position: 'start',
+  subjectDrop: true,
+};
+
+/**
+ * SOV verb configuration (verb at end, usually subject drop).
+ */
+export const SOV_VERB_CONFIG: VerbConfig = {
+  position: 'end',
+  subjectDrop: true,
+};
+
+/**
+ * VSO verb configuration (Arabic).
+ */
+export const VSO_VERB_CONFIG: VerbConfig = {
+  position: 'start',
+  subjectDrop: true,
+};
+
+/**
+ * German V2 verb configuration.
+ */
+export const V2_VERB_CONFIG: VerbConfig = {
+  position: 'second',
+  subjectDrop: false,
+};
+
+// =============================================================================
+// Possessive Configuration Templates
+// =============================================================================
+
+/**
+ * English possessive configuration ('s marker).
+ */
+export const ENGLISH_POSSESSIVE: PossessiveConfig = {
+  marker: "'s",
+  markerPosition: 'after-object',
+  specialForms: { me: 'my', it: 'its', you: 'your' },
+  keywords: { my: 'me', your: 'you', its: 'it' },
+};
+
+/**
+ * Romance language possessive configuration (de + adjectives).
+ * Used by: Spanish, French, Portuguese, Italian
+ */
+export const ROMANCE_POSSESSIVE_BASE: Omit<PossessiveConfig, 'specialForms' | 'keywords'> = {
+  marker: 'de',
+  markerPosition: 'before-property',
+  usePossessiveAdjectives: true,
+};
+
+/**
+ * Japanese possessive configuration (の particle).
+ */
+export const JAPANESE_POSSESSIVE: PossessiveConfig = {
+  marker: 'の',
+  markerPosition: 'between',
+  keywords: {
+    私の: 'me',
+    あなたの: 'you',
+    その: 'it',
+  },
+};
+
+/**
+ * Korean possessive configuration (의 particle).
+ */
+export const KOREAN_POSSESSIVE: PossessiveConfig = {
+  marker: '의',
+  markerPosition: 'between',
+  keywords: {
+    내: 'me',
+    나의: 'me',
+    너의: 'you',
+    그것의: 'it',
+  },
+};
+
+/**
+ * German possessive configuration.
+ */
+export const GERMAN_POSSESSIVE: PossessiveConfig = {
+  marker: 'von',
+  markerPosition: 'before-property',
+  usePossessiveAdjectives: true,
+  specialForms: { me: 'mein', it: 'sein', you: 'dein' },
+  keywords: {
+    mein: 'me',
+    meine: 'me',
+    meinen: 'me',
+    dein: 'you',
+    deine: 'you',
+    sein: 'it',
+    seine: 'it',
+  },
+};
+
+// =============================================================================
+// Language Family Base Configurations
+// =============================================================================
+
+/**
+ * Base configuration for SVO languages with prepositions.
+ */
+export interface SVOBaseConfig {
+  wordOrder: 'SVO';
+  markingStrategy: 'preposition';
+  usesSpaces: true;
+  direction: 'ltr';
+}
+
+export const SVO_BASE_CONFIG: SVOBaseConfig = {
+  wordOrder: 'SVO',
+  markingStrategy: 'preposition',
+  usesSpaces: true,
+  direction: 'ltr',
+};
+
+/**
+ * Base configuration for SOV languages with particles.
+ */
+export interface SOVBaseConfig {
+  wordOrder: 'SOV';
+  markingStrategy: MarkingStrategy;
+  usesSpaces: boolean;
+  direction: 'ltr';
+}
+
+export const SOV_PARTICLE_CONFIG: SOVBaseConfig = {
+  wordOrder: 'SOV',
+  markingStrategy: 'particle',
+  usesSpaces: false,
+  direction: 'ltr',
+};
+
+export const SOV_SPACE_CONFIG: SOVBaseConfig = {
+  wordOrder: 'SOV',
+  markingStrategy: 'postposition',
+  usesSpaces: true,
+  direction: 'ltr',
+};
+
+/**
+ * Base configuration for VSO languages.
+ */
+export interface VSOBaseConfig {
+  wordOrder: 'VSO';
+  markingStrategy: 'preposition';
+  usesSpaces: boolean;
+  direction: 'ltr' | 'rtl';
+}
+
+export const VSO_BASE_CONFIG: VSOBaseConfig = {
+  wordOrder: 'VSO',
+  markingStrategy: 'preposition',
+  usesSpaces: true,
+  direction: 'rtl',
+};
+
+// =============================================================================
+// Helper Functions
+// =============================================================================
+
+/**
+ * Create role markers by translating from a base template.
+ * Useful for creating language-specific markers from SVO_PREPOSITION_MARKERS.
+ */
+export function createRoleMarkers(
+  translations: Partial<Record<SemanticRole, string | { primary: string; alternatives?: string[] }>>
+): Partial<Record<SemanticRole, RoleMarker>> {
+  const markers: Partial<Record<SemanticRole, RoleMarker>> = {};
+
+  for (const [role, value] of Object.entries(translations)) {
+    if (typeof value === 'string') {
+      markers[role as SemanticRole] = { primary: value, position: 'before' };
+    } else {
+      markers[role as SemanticRole] = { ...value, position: 'before' };
+    }
+  }
+
+  // Patient is always empty with position before
+  if (!markers.patient) {
+    markers.patient = { primary: '', position: 'before' };
+  }
+
+  return markers;
+}
+
+/**
+ * Create Romance possessive config with language-specific overrides.
+ */
+export function createRomancePossessive(
+  specialForms: Record<string, string>,
+  keywords: Record<string, string>
+): PossessiveConfig {
+  return {
+    ...ROMANCE_POSSESSIVE_BASE,
+    specialForms,
+    keywords,
+  };
+}

package/src/generators/profiles/ms.ts

@@ -0,0 +1,130 @@
+/**
+ * Malay Language Profile
+ *
+ * SVO word order, prepositions, space-separated.
+ */
+
+import type { LanguageProfile } from './types';
+
+export const malayProfile: LanguageProfile = {
+  code: 'ms',
+  name: 'Malay',
+  nativeName: 'Melayu',
+  direction: 'ltr',
+  wordOrder: 'SVO',
+  markingStrategy: 'preposition',
+  usesSpaces: true,
+  defaultVerbForm: 'base',
+  verb: {
+    position: 'start',
+    subjectDrop: true,
+  },
+  references: {
+    me: 'saya',
+    it: 'ia',
+    you: 'kamu',
+    result: 'hasil',
+    event: 'peristiwa',
+    target: 'sasaran',
+    body: 'badan',
+  },
+  possessive: {
+    marker: '', // Malay often uses suffix or juxtaposition
+    markerPosition: 'after-object',
+    keywords: {
+      // Full pronoun forms (also serve as possessives)
+      saya: 'me', // my (formal)
+      aku: 'me', // my (informal)
+      awak: 'you', // your (informal, Malaysian)
+      kamu: 'you', // your (informal)
+      anda: 'you', // your (formal)
+      dia: 'it', // his/her/its
+      ia: 'it', // its
+      // Suffix forms (attached to nouns)
+      '-ku': 'me', // my (suffix)
+      '-mu': 'you', // your (suffix)
+      '-nya': 'it', // his/her/its (suffix)
+    },
+  },
+  roleMarkers: {
+    destination: { primary: 'ke', alternatives: ['pada'], position: 'before' },
+    source: { primary: 'dari', position: 'before' },
+    patient: { primary: '', position: 'before' },
+    style: { primary: 'dengan', position: 'before' },
+  },
+  keywords: {
+    // Class/Attribute operations
+    toggle: { primary: 'togol', alternatives: ['tukar'], normalized: 'toggle' },
+    add: { primary: 'tambah', normalized: 'add' },
+    remove: { primary: 'buang', alternatives: ['padam'], normalized: 'remove' },
+    // Content operations
+    put: { primary: 'letak', alternatives: ['letakkan'], normalized: 'put' },
+    append: { primary: 'tambah_hujung', normalized: 'append' },
+    prepend: { primary: 'tambah_mula', normalized: 'prepend' },
+    take: { primary: 'ambil', normalized: 'take' },
+    make: { primary: 'buat', alternatives: ['cipta'], normalized: 'make' },
+    clone: { primary: 'klon', alternatives: ['salin'], normalized: 'clone' },
+    swap: { primary: 'tukar_tempat', normalized: 'swap' },
+    morph: { primary: 'ubah_bentuk', normalized: 'morph' },
+    // Variable operations
+    set: { primary: 'tetapkan', alternatives: ['setkan'], normalized: 'set' },
+    get: { primary: 'dapatkan', alternatives: ['ambil'], normalized: 'get' },
+    increment: { primary: 'tambah_satu', normalized: 'increment' },
+    decrement: { primary: 'kurang_satu', normalized: 'decrement' },
+    log: { primary: 'catat', alternatives: ['log'], normalized: 'log' },
+    // Visibility
+    show: { primary: 'tunjuk', alternatives: ['papar'], normalized: 'show' },
+    hide: { primary: 'sembunyi', alternatives: ['sorok'], normalized: 'hide' },
+    transition: { primary: 'peralihan', normalized: 'transition' },
+    // Events
+    on: { primary: 'apabila', alternatives: ['bila', 'ketika'], normalized: 'on' },
+    trigger: { primary: 'cetuskan', normalized: 'trigger' },
+    send: { primary: 'hantar', normalized: 'send' },
+    // DOM focus
+    focus: { primary: 'fokus', normalized: 'focus' },
+    blur: { primary: 'kabur', normalized: 'blur' },
+    // Navigation
+    go: { primary: 'pergi', alternatives: ['pindah'], normalized: 'go' },
+    // Async
+    wait: { primary: 'tunggu', normalized: 'wait' },
+    fetch: { primary: 'ambil_dari', alternatives: ['muat'], normalized: 'fetch' },
+    settle: { primary: 'selesai', normalized: 'settle' },
+    // Control flow
+    if: { primary: 'jika', alternatives: ['kalau'], normalized: 'if' },
+    when: { primary: 'apabila', normalized: 'when' },
+    where: { primary: 'di_mana', normalized: 'where' },
+    else: { primary: 'kalau_tidak', alternatives: ['jika_tidak'], normalized: 'else' },
+    repeat: { primary: 'ulang', normalized: 'repeat' },
+    for: { primary: 'untuk', normalized: 'for' },
+    while: { primary: 'selagi', alternatives: ['semasa'], normalized: 'while' },
+    continue: { primary: 'teruskan', normalized: 'continue' },
+    halt: { primary: 'henti', alternatives: ['berhenti'], normalized: 'halt' },
+    throw: { primary: 'lempar', normalized: 'throw' },
+    call: { primary: 'panggil', normalized: 'call' },
+    return: { primary: 'pulang', alternatives: ['kembali'], normalized: 'return' },
+    then: { primary: 'kemudian', alternatives: ['lepas_itu'], normalized: 'then' },
+    and: { primary: 'dan', normalized: 'and' },
+    end: { primary: 'tamat', alternatives: ['habis'], normalized: 'end' },
+    // Advanced
+    js: { primary: 'js', normalized: 'js' },
+    async: { primary: 'tak_segerak', normalized: 'async' },
+    tell: { primary: 'beritahu', normalized: 'tell' },
+    default: { primary: 'lalai', normalized: 'default' },
+    init: { primary: 'mula', alternatives: ['mulakan'], normalized: 'init' },
+    behavior: { primary: 'kelakuan', normalized: 'behavior' },
+    install: { primary: 'pasang', normalized: 'install' },
+    measure: { primary: 'ukur', normalized: 'measure' },
+    // Modifiers
+    into: { primary: 'ke_dalam', normalized: 'into' },
+    before: { primary: 'sebelum', normalized: 'before' },
+    after: { primary: 'selepas', alternatives: ['lepas'], normalized: 'after' },
+    // Event modifiers
+    until: { primary: 'sehingga', alternatives: ['sampai'], normalized: 'until' },
+    event: { primary: 'peristiwa', normalized: 'event' },
+    from: { primary: 'dari', normalized: 'from' },
+  },
+  eventHandler: {
+    keyword: { primary: 'apabila', alternatives: ['bila', 'ketika'], normalized: 'on' },
+    sourceMarker: { primary: 'dari', position: 'before' },
+  },
+};