@lokascript/i18n 2.5.1 → 2.7.0

package/src/grammar/index.ts

@@ -35,6 +35,7 @@ export {
   frenchProfile,
   portugueseProfile,
   indonesianProfile,
+  malayProfile,
   quechuaProfile,
   swahiliProfile,
   bengaliProfile,

package/src/grammar/profiles/index.ts

@@ -6,7 +6,8 @@
  * language-specific markers and rules.
  */
 
-import type { LanguageProfile } from '../types';
+import type { LanguageProfile, SemanticRole } from '../types';
+import { reorderRoles, insertMarkers } from '../types';
 
 // =============================================================================
 // English (Reference Language)
@@ -104,6 +105,41 @@ export const japaneseProfile: LanguageProfile = {
         insertMarkers: true,
       },
     },
+    {
+      name: 'set-to',
+      description: 'Transform set X to Y to Japanese verb-final order',
+      priority: 90,
+      match: {
+        commands: ['set', '設定'],
+        requiredRoles: ['action', 'patient', 'destination'],
+        // Skip a set whose value swept up a trailing block terminator (`set X to Y
+        // end` — the body of an inline `if … then set … end`): verb-final reorder
+        // would push the verb past `end` and break the block. These parse fine
+        // verb-medial (the default).
+        predicate: parsed => !/(^|\s)end$/i.test((parsed.original ?? '').trim()),
+      },
+      transform: {
+        // X を Y に 設定 — verb-final, matching the generated SOV set pattern
+        // (markerOverride: を on the destination / に on the value). Default reorder
+        // emitted verb-MEDIAL (`X を 設定 Y に`), unmatched by any set pattern.
+        roleOrder: ['patient', 'destination', 'action'],
+        insertMarkers: true,
+      },
+    },
+    {
+      name: 'bind-to',
+      description: 'Transform bind $var to #el to Japanese verb-final order',
+      priority: 90,
+      match: {
+        commands: ['bind', 'バインド'],
+        requiredRoles: ['action', 'patient', 'destination'],
+      },
+      transform: {
+        // $var を #el に バインド
+        roleOrder: ['patient', 'destination', 'action'],
+        insertMarkers: true,
+      },
+    },
   ],
 };
 
@@ -127,6 +163,10 @@ export const koreanProfile: LanguageProfile = {
     { form: '를', role: 'patient', position: 'postposition', required: true, alternatives: ['을'] },
     { form: '에', role: 'destination', position: 'postposition', required: true },
     { form: '에서', role: 'source', position: 'postposition', required: true },
+    // Event marker — the semantic *-event-ko-sov-* patterns anchor on 할 때
+    // (the generator profile's eventMarker); without it every ko handler
+    // emitted a bare event name no fused pattern could anchor.
+    { form: '할 때', role: 'event', position: 'postposition', required: true },
     {
       form: '로',
       role: 'style',
@@ -158,6 +198,56 @@ export const koreanProfile: LanguageProfile = {
         insertMarkers: true,
       },
     },
+    {
+      name: 'put-into',
+      description: 'Transform put X into Y to Korean verb-final order',
+      priority: 90,
+      match: {
+        commands: ['put', '넣다'],
+        requiredRoles: ['action', 'patient', 'destination'],
+        // Standalone put only (e.g. a then-chain clause): an event handler whose
+        // action is `put` must keep the event mid-stream, so don't verb-final it.
+        predicate: parsed => !parsed.roles.has('event'),
+      },
+      transform: {
+        // X 를 Y 에 넣다 (patient, destination, verb-last)
+        roleOrder: ['patient', 'destination', 'action'],
+        insertMarkers: true,
+      },
+    },
+    {
+      name: 'set-to',
+      description: 'Transform set X to Y to Korean verb-final order',
+      priority: 90,
+      match: {
+        commands: ['set', '설정'],
+        requiredRoles: ['action', 'patient', 'destination'],
+        // Skip a set whose value swept up a trailing block terminator (`set X to Y
+        // end` — the body of an inline `if … then set … end`): verb-final reorder
+        // would push the verb past `end` and break the block. These parse fine
+        // verb-medial (the default).
+        predicate: parsed => !/(^|\s)end$/i.test((parsed.original ?? '').trim()),
+      },
+      transform: {
+        // X 를 Y 에 설정 — verb-final, matching the generated SOV set pattern.
+        roleOrder: ['patient', 'destination', 'action'],
+        insertMarkers: true,
+      },
+    },
+    {
+      name: 'bind-to',
+      description: 'Transform bind $var to #el to Korean verb-final order',
+      priority: 90,
+      match: {
+        commands: ['bind', '바인드'],
+        requiredRoles: ['action', 'patient', 'destination'],
+      },
+      transform: {
+        // $var 를 #el 에 바인드
+        roleOrder: ['patient', 'destination', 'action'],
+        insertMarkers: true,
+      },
+    },
   ],
 };
 
@@ -198,20 +288,33 @@ export const chineseProfile: LanguageProfile = {
         requiredRoles: ['event', 'action'],
       },
       transform: {
-        // 当 点击 时 增加 #count
+        // 当 EVENT 时 ACTION [patient] [destination] [source] [style] [method] ...
         roleOrder: ['event', 'action', 'patient'],
         insertMarkers: true,
-        custom: (parsed, _profile) => {
-          // Handle 当...时 circumfix
+        custom: (parsed, profile) => {
+          // Handle 当...时 circumfix wrapping the event, then emit the action,
+          // then any remaining roles in canonical order with their markers.
+          // Earlier this function emitted only event/action/patient and silently
+          // dropped destination, source, style, etc. — that produced truncated
+          // output for `send X to Y`, `fetch /api with method:...`, etc.
           const event = parsed.roles.get('event');
           const action = parsed.roles.get('action');
-          const patient = parsed.roles.get('patient');
 
-          const parts = ['当', event?.translated || event?.value, '时'];
-          parts.push(action?.translated || action?.value || '');
-          if (patient) {
-            parts.push(patient.translated || patient.value);
-          }
+          // Build the post-action role list using the canonical order minus
+          // event/action, plus a safety net for any roles not in the order.
+          const remaining = new Map(parsed.roles);
+          remaining.delete('event');
+          remaining.delete('action');
+
+          const remainingOrder: SemanticRole[] = (profile.canonicalOrder || []).filter(
+            r => r !== 'event' && r !== 'action'
+          );
+          const reordered = reorderRoles(remaining, remainingOrder);
+          const markedTokens = insertMarkers(reordered, profile.markers, profile.adpositionType);
+
+          const parts: string[] = ['当', event?.translated || event?.value || '', '时'];
+          if (action) parts.push(action.translated || action.value);
+          parts.push(...markedTokens);
 
           return parts.filter(Boolean).join(' ');
         },
@@ -231,6 +334,30 @@ export const chineseProfile: LanguageProfile = {
         insertMarkers: true,
       },
     },
+    {
+      name: 'bind-to',
+      description: 'Bind binds a variable to an element: 绑定 X 到 Y (no 把)',
+      priority: 90,
+      match: {
+        commands: ['bind', '绑定'],
+        requiredRoles: ['action', 'patient', 'destination'],
+      },
+      transform: {
+        // Plain SVO without the 把 object-fronting: the bound variable is the
+        // semantic destination (not a fronted patient), and `到` marks the
+        // target element. Custom so the patient does not pick up 把.
+        roleOrder: ['action', 'patient', 'destination'],
+        custom: parsed => {
+          const action = parsed.roles.get('action');
+          const patient = parsed.roles.get('patient');
+          const destination = parsed.roles.get('destination');
+          const verb = action?.translated || action?.value || '';
+          const v = patient?.translated || patient?.value || '';
+          const d = destination?.translated || destination?.value || '';
+          return [verb, v, '到', d].filter(Boolean).join(' ');
+        },
+      },
+    },
   ],
 };
 
@@ -339,6 +466,56 @@ export const turkishProfile: LanguageProfile = {
         insertMarkers: true,
       },
     },
+    {
+      name: 'put-into',
+      description: 'Transform put X into Y to Turkish verb-final order',
+      priority: 90,
+      match: {
+        commands: ['put', 'koy'],
+        requiredRoles: ['action', 'patient', 'destination'],
+        // Standalone put only (e.g. a then-chain clause): an event handler whose
+        // action is `put` must keep the event mid-stream, so don't verb-final it.
+        predicate: parsed => !parsed.roles.has('event'),
+      },
+      transform: {
+        // X i Y e koy (patient, destination, verb-last)
+        roleOrder: ['patient', 'destination', 'action'],
+        insertMarkers: true,
+      },
+    },
+    {
+      name: 'set-to',
+      description: 'Transform set X to Y to Turkish verb-final order',
+      priority: 90,
+      match: {
+        commands: ['set', 'ayarla'],
+        requiredRoles: ['action', 'patient', 'destination'],
+        // Skip a set whose value swept up a trailing block terminator (`set X to Y
+        // end` — the body of an inline `if … then set … end`): verb-final reorder
+        // would push the verb past `end` and break the block. These parse fine
+        // verb-medial (the default).
+        predicate: parsed => !/(^|\s)end$/i.test((parsed.original ?? '').trim()),
+      },
+      transform: {
+        // X i Y e ayarla — verb-final, matching the generated SOV set pattern.
+        roleOrder: ['patient', 'destination', 'action'],
+        insertMarkers: true,
+      },
+    },
+    {
+      name: 'bind-to',
+      description: 'Transform bind $var to #el to Turkish verb-final order',
+      priority: 90,
+      match: {
+        commands: ['bind'],
+        requiredRoles: ['action', 'patient', 'destination'],
+      },
+      transform: {
+        // $var i #el e bağla
+        roleOrder: ['patient', 'destination', 'action'],
+        insertMarkers: true,
+      },
+    },
   ],
 };
 
@@ -473,6 +650,35 @@ export const indonesianProfile: LanguageProfile = {
   ],
 };
 
+// =============================================================================
+// Malay (SVO, Austronesian) — grammar mirrors Indonesian (shared function words:
+// pada / ke / dari / dengan / sebagai). Without this profile the transformer threw
+// "Unknown target locale: ms", so no ms translation could be generated at all.
+// =============================================================================
+
+export const malayProfile: LanguageProfile = {
+  code: 'ms',
+  name: 'Bahasa Melayu',
+
+  wordOrder: 'SVO',
+  adpositionType: 'preposition',
+  morphology: 'agglutinative',
+  direction: 'ltr',
+
+  canonicalOrder: ['event', 'action', 'patient', 'destination'],
+
+  markers: [
+    // ms marks the event clause with `apabila` ("when/on") — matching the semantic
+    // ms event-handler pattern (`apabila {event} {body}`) and the ms dict's
+    // `on: apabila`. (Indonesian uses `pada` here; the two diverge on the event head.)
+    { form: 'apabila', role: 'event', position: 'preposition', required: true },
+    { form: 'ke', role: 'destination', position: 'preposition', required: false },
+    { form: 'dari', role: 'source', position: 'preposition', required: false },
+    { form: 'dengan', role: 'style', position: 'preposition', required: false },
+    { form: 'sebagai', role: 'method', position: 'preposition', required: false },
+  ],
+};
+
 // =============================================================================
 // Quechua (SOV, Quechuan)
 // =============================================================================
@@ -521,7 +727,10 @@ export const swahiliProfile: LanguageProfile = {
     { form: 'kwa', role: 'destination', position: 'preposition', required: false },
     { form: 'kutoka', role: 'source', position: 'preposition', required: false },
     { form: 'na', role: 'style', position: 'preposition', required: false },
-    { form: 'kama', role: 'method', position: 'preposition', required: false }, // "as/like"
+    // `kuwa`, not `kama`: sw `kama` is the IF keyword — rendering the method
+    // role ("as X") with it grew a phantom `if` at semantic parse time. See the
+    // matching comment in dictionaries/sw.ts (modifiers.as).
+    { form: 'kuwa', role: 'method', position: 'preposition', required: false }, // "as/become"
   ],
 };
 
@@ -568,6 +777,56 @@ export const bengaliProfile: LanguageProfile = {
         insertMarkers: true,
       },
     },
+    {
+      name: 'put-into',
+      description: 'Transform put X into Y to Bengali verb-final order',
+      priority: 90,
+      match: {
+        commands: ['put', 'রাখুন'],
+        requiredRoles: ['action', 'patient', 'destination'],
+        // Standalone put only (e.g. a then-chain clause): an event handler whose
+        // action is `put` must keep the event mid-stream, so don't verb-final it.
+        predicate: parsed => !parsed.roles.has('event'),
+      },
+      transform: {
+        // X কে Y তে রাখুন (patient, destination, verb-last)
+        roleOrder: ['patient', 'destination', 'action'],
+        insertMarkers: true,
+      },
+    },
+    {
+      name: 'set-to',
+      description: 'Transform set X to Y to Bengali verb-final order',
+      priority: 90,
+      match: {
+        commands: ['set', 'সেট'],
+        requiredRoles: ['action', 'patient', 'destination'],
+        // Skip a set whose value swept up a trailing block terminator (`set X to Y
+        // end` — the body of an inline `if … then set … end`): verb-final reorder
+        // would push the verb past `end` and break the block. These parse fine
+        // verb-medial (the default).
+        predicate: parsed => !/(^|\s)end$/i.test((parsed.original ?? '').trim()),
+      },
+      transform: {
+        // X কে Y তে সেট — verb-final, matching the generated SOV set pattern.
+        roleOrder: ['patient', 'destination', 'action'],
+        insertMarkers: true,
+      },
+    },
+    {
+      name: 'bind-to',
+      description: 'Transform bind $var to #el to Bengali verb-final order',
+      priority: 90,
+      match: {
+        commands: ['bind', 'বাইন্ড'],
+        requiredRoles: ['action', 'patient', 'destination'],
+      },
+      transform: {
+        // $var কে #el তে বাইন্ড
+        roleOrder: ['patient', 'destination', 'action'],
+        insertMarkers: true,
+      },
+    },
   ],
 };
 
@@ -726,6 +985,61 @@ export const hindiProfile: LanguageProfile = {
         insertMarkers: true,
       },
     },
+    {
+      name: 'put-into',
+      description: 'Transform put X into Y to Hindi verb-final order',
+      priority: 90,
+      match: {
+        commands: ['put', 'रखें'],
+        requiredRoles: ['action', 'patient', 'destination'],
+      },
+      transform: {
+        // "hi" को #out में रखें
+        roleOrder: ['patient', 'destination', 'action'],
+        insertMarkers: true,
+      },
+    },
+    {
+      name: 'set-to',
+      description: 'Transform set X to Y to Hindi verb-final order',
+      priority: 90,
+      match: {
+        commands: ['set', 'सेट'],
+        requiredRoles: ['action', 'patient', 'destination'],
+        // Skip a set whose value swept up a trailing block terminator (`set X to Y
+        // end` — the body of an inline `if … then set … end`): verb-final reorder
+        // would push the verb past `end` and break the block. These parse fine
+        // verb-medial (the default).
+        predicate: parsed => !/(^|\s)end$/i.test((parsed.original ?? '').trim()),
+      },
+      transform: {
+        // #x.innerText को इसका.name में सेट — verb-final, matching the generated
+        // SOV set pattern (set schema markerOverride puts को on the destination /
+        // में on the value). The default reorder emitted verb-MEDIAL
+        // (`#x को सेट … में`), which no set pattern matched → bare-event fallback.
+        roleOrder: ['patient', 'destination', 'action'],
+        insertMarkers: true,
+      },
+    },
+    {
+      name: 'bind-to',
+      description: 'Transform bind $var to #el to Hindi verb-final order',
+      priority: 90,
+      match: {
+        commands: ['bind', 'बाइंड', 'बांधें'],
+        requiredRoles: ['action', 'patient', 'destination'],
+      },
+      transform: {
+        // $greeting को #name-input में bind — verb-final, matching the generated SOV
+        // bind pattern (`{destination} को {source} में bind`). The default reorder
+        // emitted verb-MEDIAL (`$greeting को bind #name-input में`), which the
+        // verb-final pattern never matched → the bare-event fallback mis-anchored
+        // the fronted `$greeting` as a phantom `on` event (the rf=0.00 bind residue).
+        // ja/ko/zh/tr/bn already carry this rule; hi was the only SOV gap.
+        roleOrder: ['patient', 'destination', 'action'],
+        insertMarkers: true,
+      },
+    },
   ],
 };
 
@@ -748,7 +1062,11 @@ export const tagalogProfile: LanguageProfile = {
   markers: [
     { form: 'kapag', role: 'event', position: 'preposition', required: true },
     { form: 'sa', role: 'destination', position: 'preposition', required: false },
-    { form: 'mula sa', role: 'source', position: 'preposition', required: false },
+    // mula_sa (underscore) matches the tl dict and the semantic profile's
+    // source marker — the spaced form emitted two tokens the generated
+    // patterns' single mula_sa literal could never match, so `remove X
+    // mula sa Y` lost its source and the schema default fabricated me.
+    { form: 'mula_sa', role: 'source', position: 'preposition', required: false },
     { form: 'nang', role: 'style', position: 'preposition', required: false },
     { form: 'bilang', role: 'method', position: 'preposition', required: false },
   ],
@@ -854,6 +1172,33 @@ export const hebrewProfile: LanguageProfile = {
     { form: 'עם', role: 'style', position: 'preposition', required: false },
     { form: 'כ', role: 'method', position: 'preposition', required: false },
   ],
+
+  rules: [
+    {
+      name: 'bind-to',
+      description: 'Bind binds a variable to an element: קשור X ל Y',
+      priority: 90,
+      match: {
+        commands: ['bind', 'קשור'],
+        requiredRoles: ['action', 'patient', 'destination'],
+      },
+      transform: {
+        // The bound variable is the semantic destination (kept bare, no את
+        // object marker), and ל marks the target element. Custom so the
+        // variable does not pick up את and the element uses ל (not על).
+        roleOrder: ['action', 'patient', 'destination'],
+        custom: parsed => {
+          const action = parsed.roles.get('action');
+          const patient = parsed.roles.get('patient');
+          const destination = parsed.roles.get('destination');
+          const verb = action?.translated || action?.value || '';
+          const v = patient?.translated || patient?.value || '';
+          const d = destination?.translated || destination?.value || '';
+          return [verb, v, 'ל', d].filter(Boolean).join(' ');
+        },
+      },
+    },
+  ],
 };
 
 // =============================================================================
@@ -872,6 +1217,7 @@ export const profiles: Record<string, LanguageProfile> = {
   fr: frenchProfile,
   pt: portugueseProfile,
   id: indonesianProfile,
+  ms: malayProfile,
   qu: quechuaProfile,
   sw: swahiliProfile,
   bn: bengaliProfile,