@lokascript/semantic 1.0.0 → 1.1.0

package/dist/{types-C4dcj53L.d.ts → types-BY3Id07j.d.ts}

@@ -502,7 +502,7 @@ interface PossessiveConfig {
  * Complete language profile for pattern generation.
  */
 interface LanguageProfile {
-    /** ISO 639-1 language code */
+    /** ISO 639-1 or BCP 47 language code (e.g., 'es' or 'es-MX') */
     readonly code: string;
     /** Human-readable language name */
     readonly name: string;
@@ -541,18 +541,33 @@ interface LanguageProfile {
      * Individual keywords can override this via KeywordTranslation.form
      */
     readonly defaultVerbForm?: VerbForm;
+    /**
+     * Base language code to extend (for regional variants).
+     * When set, this profile inherits from the base and overrides specific fields.
+     * Example: 'es-MX' profile with extends: 'es' inherits from Spanish base.
+     */
+    readonly extends?: string;
 }
 /**
  * Configuration for event handler pattern generation.
- * Used by simple SVO languages that don't need hand-crafted patterns.
+ * Supports both SVO and SOV/VSO languages.
  */
 interface EventHandlerConfig {
-    /** Primary event keyword (e.g., 'on', 'bei', 'sur') */
-    readonly keyword: KeywordTranslation;
+    /** Primary event keyword (e.g., 'on', 'bei', 'sur') for SVO */
+    readonly keyword?: KeywordTranslation;
     /** Source filter marker (e.g., 'from', 'von', 'de') */
-    readonly sourceMarker: RoleMarker;
+    readonly sourceMarker?: RoleMarker;
     /** Conditional keyword (e.g., 'when', 'wenn', 'quand') */
     readonly conditionalKeyword?: KeywordTranslation;
+    /** Event marker for SOV/VSO languages (e.g., で (Japanese), 할 때 (Korean), da (Turkish), عند (Arabic)) */
+    readonly eventMarker?: RoleMarker;
+    /** Temporal/conditional markers that can optionally appear with events */
+    readonly temporalMarkers?: string[];
+    /**
+     * Negation marker for expressing negated events (e.g., Arabic عدم = "not/lack of").
+     * Used in patterns like: عند عدم التركيز = "when not focusing" = "on blur"
+     */
+    readonly negationMarker?: RoleMarker;
 }
 /**
  * Verb form used for command keywords.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@lokascript/semantic",
-  "version": "1.0.0",
+  "version": "1.1.0",
   "description": "Semantic multilingual parser for hyperscript (23 languages)",
   "type": "module",
   "main": "dist/index.cjs",
@@ -14,7 +14,7 @@
       "require": "./dist/index.cjs"
     },
     "./browser": {
-      "types": "./dist/browser.d.ts",
+      "types": "./dist/index.d.ts",
       "default": "./dist/browser.global.js"
     },
     "./browser/en": {
@@ -41,66 +41,57 @@
     "./browser/lazy": {
       "default": "./dist/browser-lazy.lazy.global.js"
     },
-    "./languages": {
-      "types": "./dist/languages/index.d.ts",
-      "import": "./dist/languages/index.mjs"
-    },
     "./languages/en": {
       "types": "./dist/languages/en.d.ts",
-      "import": "./dist/languages/en.mjs"
+      "import": "./dist/languages/en.js"
     },
     "./languages/es": {
       "types": "./dist/languages/es.d.ts",
-      "import": "./dist/languages/es.mjs"
+      "import": "./dist/languages/es.js"
     },
     "./languages/ja": {
       "types": "./dist/languages/ja.d.ts",
-      "import": "./dist/languages/ja.mjs"
+      "import": "./dist/languages/ja.js"
     },
     "./languages/ar": {
       "types": "./dist/languages/ar.d.ts",
-      "import": "./dist/languages/ar.mjs"
+      "import": "./dist/languages/ar.js"
     },
     "./languages/ko": {
       "types": "./dist/languages/ko.d.ts",
-      "import": "./dist/languages/ko.mjs"
+      "import": "./dist/languages/ko.js"
     },
     "./languages/zh": {
       "types": "./dist/languages/zh.d.ts",
-      "import": "./dist/languages/zh.mjs"
+      "import": "./dist/languages/zh.js"
     },
     "./languages/tr": {
       "types": "./dist/languages/tr.d.ts",
-      "import": "./dist/languages/tr.mjs"
+      "import": "./dist/languages/tr.js"
     },
     "./languages/pt": {
       "types": "./dist/languages/pt.d.ts",
-      "import": "./dist/languages/pt.mjs"
+      "import": "./dist/languages/pt.js"
     },
     "./languages/fr": {
       "types": "./dist/languages/fr.d.ts",
-      "import": "./dist/languages/fr.mjs"
+      "import": "./dist/languages/fr.js"
     },
     "./languages/de": {
       "types": "./dist/languages/de.d.ts",
-      "import": "./dist/languages/de.mjs"
+      "import": "./dist/languages/de.js"
     },
     "./languages/id": {
       "types": "./dist/languages/id.d.ts",
-      "import": "./dist/languages/id.mjs"
+      "import": "./dist/languages/id.js"
     },
     "./languages/qu": {
       "types": "./dist/languages/qu.d.ts",
-      "import": "./dist/languages/qu.mjs"
+      "import": "./dist/languages/qu.js"
     },
     "./languages/sw": {
       "types": "./dist/languages/sw.d.ts",
-      "import": "./dist/languages/sw.mjs"
-    },
-    "./explicit": {
-      "types": "./dist/explicit/index.d.ts",
-      "import": "./dist/explicit/index.mjs",
-      "require": "./dist/explicit/index.js"
+      "import": "./dist/languages/sw.js"
     },
     "./browser/ja": {
       "default": "./dist/browser-ja.ja.global.js"
@@ -141,7 +132,7 @@
   },
   "scripts": {
     "build": "tsup",
-    "build:browser": "tsup src/browser.ts --format iife --global-name HyperFixiSemantic --minify",
+    "build:browser": "tsup src/browser.ts --format iife --global-name LokaScriptSemantic --minify",
     "dev": "tsup --watch",
     "test": "vitest",
     "test:run": "vitest run",
@@ -159,7 +150,7 @@
   },
   "sideEffects": [
     "./src/languages/*.ts",
-    "./dist/languages/*.mjs",
+    "./dist/languages/*.js",
     "./src/patterns/index.ts"
   ],
   "keywords": [
@@ -186,13 +177,13 @@
   ],
   "repository": {
     "type": "git",
-    "url": "git+https://github.com/hyperfixi/hyperfixi.git",
+    "url": "git+https://github.com/codetalcott/lokascript.git",
     "directory": "packages/semantic"
   },
   "bugs": {
-    "url": "https://github.com/hyperfixi/hyperfixi/issues"
+    "url": "https://github.com/codetalcott/lokascript/issues"
   },
-  "homepage": "https://github.com/hyperfixi/hyperfixi/tree/main/packages/semantic#readme",
+  "homepage": "https://github.com/codetalcott/lokascript/tree/main/packages/semantic#readme",
   "engines": {
     "node": ">=18.0.0"
   },

package/src/generators/command-schemas.ts

@@ -254,7 +254,7 @@ export const putSchema: CommandSchema = {
       required: true,
       expectedTypes: ['literal', 'selector', 'reference', 'expression'],
       svoPosition: 1,
-      sovPosition: 2,
+      sovPosition: 1, // SOV: patient comes first (を marker)
     },
     {
       role: 'destination',
@@ -262,7 +262,7 @@ export const putSchema: CommandSchema = {
       required: true,
       expectedTypes: ['selector', 'reference'],
       svoPosition: 2,
-      sovPosition: 1,
+      sovPosition: 2, // SOV: destination comes second (に/에/a marker)
     },
   ],
   // Runtime error documentation
@@ -324,13 +324,18 @@ export const setSchema: CommandSchema = {
       role: 'destination',
       description: 'The property or variable to set',
       required: true,
-      expectedTypes: ['selector', 'reference'],
+      expectedTypes: ['selector', 'reference', 'expression'],
       svoPosition: 1,
       sovPosition: 1,
       // Override destination marker for English (remove 'on', use no marker)
-      // Other languages keep their default destination markers
+      // SOV languages swap markers: variable gets patient marker (を/를/i)
+      // Arabic (VSO): no marker before variable
       markerOverride: {
         en: '', // No marker before destination in English: "set :x to 5"
+        ja: 'を', // "x を 10 に 設定" - variable gets object marker
+        ko: '를', // "x 를 10 으로 설정" - variable gets object marker
+        tr: 'i', // "x i 10 e ayarla" - variable gets accusative marker
+        ar: '', // "عيّن x إلى 10" - no marker before variable
       },
     },
     {
@@ -341,7 +346,8 @@ export const setSchema: CommandSchema = {
       svoPosition: 2,
       sovPosition: 2,
       // Override patient marker for SVO languages with their native prepositions
-      // SOV languages (Korean, Japanese, Turkish) use their default object markers
+      // SOV languages swap markers: value gets destination marker (に/으로/e)
+      // Arabic (VSO): إلى preposition before value
       markerOverride: {
         en: 'to', // "set :x to 5"
         es: 'a', // "establecer x a 10"
@@ -349,6 +355,10 @@ export const setSchema: CommandSchema = {
         fr: 'à', // "définir x à 10"
         de: 'auf', // "setze x auf 10"
         id: 'ke', // "atur x ke 10"
+        ja: 'に', // "x を 10 に 設定" - value gets destination marker
+        ko: '으로', // "x 를 10 으로 설정" - value gets manner/instrument marker
+        tr: 'e', // "x i 10 e ayarla" - value gets dative marker
+        ar: 'إلى', // "عيّن x إلى 10" - value gets preposition "to"
       },
     },
   ],
@@ -716,18 +726,19 @@ export const getCommandSchema: CommandSchema = {
       expectedTypes: ['selector', 'reference', 'expression'],
       svoPosition: 1,
       sovPosition: 2,
-      // No marker before source for simple GET pattern: "get #element" not "get from #element"
+      // Marker overrides for GET pattern
+      // SOV languages use object markers, SVO languages have no marker
       markerOverride: {
         en: '',
         es: '',
         pt: '',
         fr: '',
         de: '',
-        ja: '',
+        ja: 'を', // Japanese object marker: #element を 取得
         zh: '',
-        ko: '',
-        ar: '',
-        tr: '',
+        ko: '를', // Korean object marker: #element 를 가져오기
+        ar: 'على', // Arabic preposition: احصل على #element
+        tr: 'i', // Turkish accusative: #element i al
         id: '',
       },
     },

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

@@ -37,52 +37,58 @@ export function generateEventHandlerPatterns(profile: LanguageProfile): Language
 
   // 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 },
-    },
-  });
+  // Only generate if keyword is defined
+  if (eh.keyword) {
+    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;
-  }
+    // Pattern 2: With source - "{keyword} {event} {sourceMarker} {source}"
+    // e.g., "bei click von #button", "sur click de #button"
+    // Only generate if both keyword and sourceMarker are defined
+    if (eh.sourceMarker) {
+      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} ${sourceMarkerValue} {source}`,
-      tokens: [
-        literalToken(eh.keyword.primary, eh.keyword.alternatives),
-        { type: 'role', role: 'event' },
-        literalToken(sourceMarkerValue, eh.sourceMarker.alternatives),
-        { type: 'role', role: 'source' },
-      ],
-    },
-    extraction: {
-      event: { position: 1 },
-      source: sourceExtraction,
-    },
-  });
+      patterns.push({
+        id: `event-${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"

package/src/generators/language-profiles.ts

@@ -36,6 +36,7 @@ export { chineseProfile } from './profiles/chinese';
 export { englishProfile } from './profiles/english';
 export { frenchProfile } from './profiles/french';
 export { germanProfile } from './profiles/german';
+export { hebrewProfile } from './profiles/hebrew';
 export { hindiProfile } from './profiles/hindi';
 export { indonesianProfile } from './profiles/indonesian';
 export { italianProfile } from './profiles/italian';
@@ -47,6 +48,7 @@ export { portugueseProfile } from './profiles/portuguese';
 export { quechuaProfile } from './profiles/quechua';
 export { russianProfile } from './profiles/russian';
 export { spanishProfile } from './profiles/spanish';
+export { spanishMexicoProfile } from './profiles/spanishMexico';
 export { swahiliProfile } from './profiles/swahili';
 export { thaiProfile } from './profiles/thai';
 export { tagalogProfile } from './profiles/tl';
@@ -61,6 +63,7 @@ import { chineseProfile } from './profiles/chinese';
 import { englishProfile } from './profiles/english';
 import { frenchProfile } from './profiles/french';
 import { germanProfile } from './profiles/german';
+import { hebrewProfile } from './profiles/hebrew';
 import { hindiProfile } from './profiles/hindi';
 import { indonesianProfile } from './profiles/indonesian';
 import { italianProfile } from './profiles/italian';
@@ -72,6 +75,7 @@ import { portugueseProfile } from './profiles/portuguese';
 import { quechuaProfile } from './profiles/quechua';
 import { russianProfile } from './profiles/russian';
 import { spanishProfile } from './profiles/spanish';
+import { spanishMexicoProfile } from './profiles/spanishMexico';
 import { swahiliProfile } from './profiles/swahili';
 import { thaiProfile } from './profiles/thai';
 import { tagalogProfile } from './profiles/tl';
@@ -95,6 +99,7 @@ export const languageProfiles: Record<string, LanguageProfile> = {
   en: englishProfile,
   fr: frenchProfile,
   de: germanProfile,
+  he: hebrewProfile,
   hi: hindiProfile,
   id: indonesianProfile,
   it: italianProfile,
@@ -106,6 +111,7 @@ export const languageProfiles: Record<string, LanguageProfile> = {
   qu: quechuaProfile,
   ru: russianProfile,
   es: spanishProfile,
+  'es-MX': spanishMexicoProfile,
   sw: swahiliProfile,
   th: thaiProfile,
   tl: tagalogProfile,