@lokascript/semantic 1.0.0

package/src/generators/event-handler-generator.ts

@@ -0,0 +1,109 @@
+/**
+ * Event Handler Pattern Generator
+ *
+ * Generates event handler patterns from LanguageProfile.eventHandler config.
+ * Used for simple SVO languages that don't need complex hand-crafted patterns.
+ *
+ * Languages using this generator: German, French, Indonesian, Quechua, Swahili
+ * Languages with hand-crafted patterns: English, Japanese, Korean, Arabic, Turkish, Spanish, Portuguese, Chinese
+ */
+
+import type { LanguagePattern } from '../types';
+import type { LanguageProfile } from './profiles/types';
+
+/**
+ * Helper to create a literal token, only including alternatives if defined.
+ */
+function literalToken(
+  value: string,
+  alternatives?: string[]
+): { type: 'literal'; value: string; alternatives?: string[] } {
+  if (alternatives && alternatives.length > 0) {
+    return { type: 'literal', value, alternatives };
+  }
+  return { type: 'literal', value };
+}
+
+/**
+ * Generate event handler patterns from a language profile's eventHandler config.
+ * Returns empty array if the profile doesn't have eventHandler config.
+ */
+export function generateEventHandlerPatterns(profile: LanguageProfile): LanguagePattern[] {
+  const eh = profile.eventHandler;
+  if (!eh) return [];
+
+  const patterns: LanguagePattern[] = [];
+  const lang = profile.code;
+
+  // Pattern 1: Standard - "{keyword} {event}"
+  // e.g., "bei click", "sur click"
+  patterns.push({
+    id: `event-${lang}-standard`,
+    language: lang,
+    command: 'on',
+    priority: 100,
+    template: {
+      format: `${eh.keyword.primary} {event}`,
+      tokens: [
+        literalToken(eh.keyword.primary, eh.keyword.alternatives),
+        { type: 'role', role: 'event' },
+      ],
+    },
+    extraction: {
+      event: { position: 1 },
+    },
+  });
+
+  // Pattern 2: With source - "{keyword} {event} {sourceMarker} {source}"
+  // e.g., "bei click von #button", "sur click de #button"
+  const sourceMarkerValue = eh.sourceMarker.primary;
+  const sourceExtraction: { marker: string; markerAlternatives?: string[] } = {
+    marker: sourceMarkerValue,
+  };
+  if (eh.sourceMarker.alternatives && eh.sourceMarker.alternatives.length > 0) {
+    sourceExtraction.markerAlternatives = eh.sourceMarker.alternatives;
+  }
+
+  patterns.push({
+    id: `event-${lang}-source`,
+    language: lang,
+    command: 'on',
+    priority: 110, // Higher priority for more specific pattern
+    template: {
+      format: `${eh.keyword.primary} {event} ${sourceMarkerValue} {source}`,
+      tokens: [
+        literalToken(eh.keyword.primary, eh.keyword.alternatives),
+        { type: 'role', role: 'event' },
+        literalToken(sourceMarkerValue, eh.sourceMarker.alternatives),
+        { type: 'role', role: 'source' },
+      ],
+    },
+    extraction: {
+      event: { position: 1 },
+      source: sourceExtraction,
+    },
+  });
+
+  // Pattern 3: Conditional (if defined) - "{conditionalKeyword} {event}"
+  // e.g., "wenn click", "quand click"
+  if (eh.conditionalKeyword) {
+    patterns.push({
+      id: `event-${lang}-when`,
+      language: lang,
+      command: 'on',
+      priority: 105, // Higher than standard, lower than with-source
+      template: {
+        format: `${eh.conditionalKeyword.primary} {event}`,
+        tokens: [
+          literalToken(eh.conditionalKeyword.primary, eh.conditionalKeyword.alternatives),
+          { type: 'role', role: 'event' },
+        ],
+      },
+      extraction: {
+        event: { position: 1 },
+      },
+    });
+  }
+
+  return patterns;
+}

package/src/generators/index.ts

@@ -0,0 +1,117 @@
+/**
+ * Pattern Generators
+ *
+ * Provides tools for generating language-specific patterns from
+ * high-level schemas and language profiles.
+ */
+
+// Language profiles
+export {
+  type LanguageProfile,
+  type WordOrder,
+  type MarkingStrategy,
+  type RoleMarker,
+  type VerbConfig,
+  type VerbForm,
+  type KeywordTranslation,
+  type TokenizationConfig,
+  englishProfile,
+  japaneseProfile,
+  arabicProfile,
+  spanishProfile,
+  koreanProfile,
+  chineseProfile,
+  turkishProfile,
+  portugueseProfile,
+  frenchProfile,
+  germanProfile,
+  indonesianProfile,
+  quechuaProfile,
+  swahiliProfile,
+  languageProfiles,
+  getProfile,
+  getSupportedLanguages,
+  isLanguageSupported,
+} from './language-profiles';
+
+// Command schemas
+export {
+  type CommandSchema,
+  type CommandPrecondition,
+  type RoleSpec,
+  type CommandCategory,
+  toggleSchema,
+  addSchema,
+  removeSchema,
+  putSchema,
+  setSchema,
+  showSchema,
+  hideSchema,
+  onSchema,
+  triggerSchema,
+  waitSchema,
+  fetchSchema,
+  goSchema,
+  incrementSchema,
+  decrementSchema,
+  appendSchema,
+  prependSchema,
+  logSchema,
+  sendSchema,
+  // Tier 2: Content & variable operations
+  takeSchema,
+  makeSchema,
+  cloneSchema,
+  getCommandSchema,
+  // Tier 3: Control flow & DOM
+  callSchema,
+  returnSchema,
+  focusSchema,
+  blurSchema,
+  // Tier 4: DOM Content Manipulation
+  swapSchema,
+  morphSchema,
+  // Tier 5: Control flow & Behavior system
+  haltSchema,
+  behaviorSchema,
+  installSchema,
+  measureSchema,
+  // Registry functions
+  commandSchemas,
+  getSchema,
+  getSchemasByCategory,
+  getDefinedSchemas,
+} from './command-schemas';
+
+// Pattern generator
+export {
+  type GeneratorConfig,
+  generatePattern,
+  generateSimplePattern,
+  generatePatternVariants,
+  generatePatternsForLanguage,
+  generatePatternsForCommand,
+  generateAllPatterns,
+  getGeneratorSummary,
+  validateLanguageKeywords,
+} from './pattern-generator';
+
+// Schema validation (for build-time validation and tooling)
+export {
+  type SchemaValidationItem,
+  type SchemaValidationSeverity,
+  type SchemaValidation,
+  type SchemaValidationResult,
+  SchemaErrorCodes,
+  validateCommandSchema,
+  validateAllSchemas,
+  formatValidationResults,
+  getValidationStats,
+} from './schema-validator';
+
+export {
+  type SchemaErrorCode,
+  SchemaErrorMessages,
+  SchemaErrorSuggestions,
+  createSchemaValidationItem,
+} from './schema-error-codes';

package/src/generators/language-profiles.ts

@@ -0,0 +1,139 @@
+/**
+ * Language Profiles
+ *
+ * Re-exports from individual profile files for backwards compatibility.
+ * For minimal bundles, import specific profiles directly:
+ *
+ * @example
+ * ```typescript
+ * // Tree-shakeable import
+ * import { englishProfile } from './profiles/english';
+ *
+ * // Full import (all profiles bundled)
+ * import { englishProfile, languageProfiles } from './language-profiles';
+ * ```
+ *
+ * @generated This file is auto-generated. Do not edit manually.
+ */
+
+// Re-export types
+export type {
+  LanguageProfile,
+  WordOrder,
+  MarkingStrategy,
+  RoleMarker,
+  VerbConfig,
+  VerbForm,
+  PossessiveConfig,
+  KeywordTranslation,
+  TokenizationConfig,
+} from './profiles/types';
+
+// Re-export individual profiles
+export { arabicProfile } from './profiles/arabic';
+export { bengaliProfile } from './profiles/bengali';
+export { chineseProfile } from './profiles/chinese';
+export { englishProfile } from './profiles/english';
+export { frenchProfile } from './profiles/french';
+export { germanProfile } from './profiles/german';
+export { hindiProfile } from './profiles/hindi';
+export { indonesianProfile } from './profiles/indonesian';
+export { italianProfile } from './profiles/italian';
+export { japaneseProfile } from './profiles/japanese';
+export { koreanProfile } from './profiles/korean';
+export { malayProfile } from './profiles/ms';
+export { polishProfile } from './profiles/polish';
+export { portugueseProfile } from './profiles/portuguese';
+export { quechuaProfile } from './profiles/quechua';
+export { russianProfile } from './profiles/russian';
+export { spanishProfile } from './profiles/spanish';
+export { swahiliProfile } from './profiles/swahili';
+export { thaiProfile } from './profiles/thai';
+export { tagalogProfile } from './profiles/tl';
+export { turkishProfile } from './profiles/turkish';
+export { ukrainianProfile } from './profiles/ukrainian';
+export { vietnameseProfile } from './profiles/vietnamese';
+
+// Import for creating the combined object
+import { arabicProfile } from './profiles/arabic';
+import { bengaliProfile } from './profiles/bengali';
+import { chineseProfile } from './profiles/chinese';
+import { englishProfile } from './profiles/english';
+import { frenchProfile } from './profiles/french';
+import { germanProfile } from './profiles/german';
+import { hindiProfile } from './profiles/hindi';
+import { indonesianProfile } from './profiles/indonesian';
+import { italianProfile } from './profiles/italian';
+import { japaneseProfile } from './profiles/japanese';
+import { koreanProfile } from './profiles/korean';
+import { malayProfile } from './profiles/ms';
+import { polishProfile } from './profiles/polish';
+import { portugueseProfile } from './profiles/portuguese';
+import { quechuaProfile } from './profiles/quechua';
+import { russianProfile } from './profiles/russian';
+import { spanishProfile } from './profiles/spanish';
+import { swahiliProfile } from './profiles/swahili';
+import { thaiProfile } from './profiles/thai';
+import { tagalogProfile } from './profiles/tl';
+import { turkishProfile } from './profiles/turkish';
+import { ukrainianProfile } from './profiles/ukrainian';
+import { vietnameseProfile } from './profiles/vietnamese';
+import type { LanguageProfile } from './profiles/types';
+
+// =============================================================================
+// Profile Registry (backwards compatibility)
+// =============================================================================
+
+/**
+ * All available language profiles.
+ * @deprecated Import individual profiles for tree-shaking.
+ */
+export const languageProfiles: Record<string, LanguageProfile> = {
+  ar: arabicProfile,
+  bn: bengaliProfile,
+  zh: chineseProfile,
+  en: englishProfile,
+  fr: frenchProfile,
+  de: germanProfile,
+  hi: hindiProfile,
+  id: indonesianProfile,
+  it: italianProfile,
+  ja: japaneseProfile,
+  ko: koreanProfile,
+  ms: malayProfile,
+  pl: polishProfile,
+  pt: portugueseProfile,
+  qu: quechuaProfile,
+  ru: russianProfile,
+  es: spanishProfile,
+  sw: swahiliProfile,
+  th: thaiProfile,
+  tl: tagalogProfile,
+  tr: turkishProfile,
+  uk: ukrainianProfile,
+  vi: vietnameseProfile,
+};
+
+/**
+ * Get a language profile by code.
+ * @deprecated Use the registry's getProfile instead.
+ */
+export function getProfile(code: string): LanguageProfile | undefined {
+  return languageProfiles[code];
+}
+
+/**
+ * Get all supported language codes.
+ * @deprecated Use the registry's getRegisteredLanguages instead.
+ */
+export function getSupportedLanguages(): string[] {
+  return Object.keys(languageProfiles);
+}
+
+/**
+ * Check if a language is supported.
+ * @deprecated Use the registry's isLanguageRegistered instead.
+ */
+export function isLanguageSupported(code: string): boolean {
+  return code in languageProfiles;
+}