@lokascript/framework 2.6.0 → 2.7.0

package/dist/multilingual/builders.d.ts

@@ -0,0 +1,88 @@
+/**
+ * Framework ↔ Semantic bridge builders.
+ *
+ * Pure functions that combine an injected {@link GrammarProfileSlice} (the
+ * per-language grammar facts, typically a `@lokascript/semantic`
+ * `LanguageProfile`) with a {@link DomainVocabulary} (the only thing a domain
+ * authors per language) into the shapes `createMultilingualDSL` consumes:
+ * a `PatternGenLanguageProfile`, a `LanguageTokenizer`, or a complete
+ * `LanguageConfig`.
+ */
+import type { LanguageTokenizer } from '../core/types';
+import type { PatternGenLanguageProfile } from '../generation/pattern-generator';
+import type { LanguageConfig } from '../api/create-dsl';
+import type { LanguageProfile as GrammarProfile } from '../grammar';
+import { type ValueExtractor } from '../interfaces/value-extractor';
+import type { DomainVocabulary, GrammarProfileSlice } from './types';
+/**
+ * Build a pattern-generation profile: domain keywords + slice grammar
+ * (word order, role markers). The result feeds `generatePattern` /
+ * `generatePatternVariants` and `LanguageConfig.patternProfile`.
+ */
+export declare function buildPatternProfile(slice: GrammarProfileSlice, vocab: DomainVocabulary): PatternGenLanguageProfile;
+/** Options for {@link buildDomainTokenizer}, overriding the slice-derived defaults. */
+export interface DomainTokenizerOptions {
+    /**
+     * Recognize operator tokens (default: false, matching `createSimpleTokenizer`).
+     * Domains whose grammar contains operators (e.g. SQL's `WHERE age > 18`) pass
+     * `true`; most natural-language DSLs leave it off.
+     */
+    readonly includeOperators?: boolean;
+    /**
+     * Case-insensitive keyword matching. Defaults to true for bicameral scripts
+     * (`latin`, `cyrillic`, or when the slice omits `script`), false otherwise.
+     */
+    readonly caseInsensitive?: boolean;
+    /** Extra extractors, registered before the slice-derived ones */
+    readonly customExtractors?: readonly ValueExtractor[];
+}
+/**
+ * Build a domain tokenizer from a grammar slice + domain vocabulary, wrapping
+ * `createSimpleTokenizer`. Derived from the slice:
+ *
+ * - `direction` (rtl for e.g. Arabic/Hebrew slices)
+ * - keyword set: vocabulary verbs (primary + alternatives), role markers
+ *   (slice defaults merged with `vocab.roleMarkerOverrides`), tokenization
+ *   particles, and `vocab.tokenizerKeywords`
+ * - keyword-normalization profile (native verb → action name, marker → role)
+ * - a `LatinExtendedIdentifierExtractor` when `slice.script === 'latin'`, so
+ *   diacritics (ñ, é, ş, …) never split identifiers
+ * - case-insensitivity for bicameral scripts
+ */
+export declare function buildDomainTokenizer(slice: GrammarProfileSlice, vocab: DomainVocabulary, options?: DomainTokenizerOptions): LanguageTokenizer;
+/** Metadata / overrides for {@link buildLanguageConfig}. */
+export interface LanguageConfigMeta {
+    /** English name (default: `slice.name`, falling back to `slice.code`) */
+    readonly name?: string;
+    /** Native name (default: `slice.nativeName`, falling back to the English name) */
+    readonly nativeName?: string;
+    /** Grammar-transformation profile to attach (optional, as on `LanguageConfig`) */
+    readonly grammarProfile?: GrammarProfile;
+    /** Use this tokenizer instead of the slice-derived one */
+    readonly tokenizer?: LanguageTokenizer;
+    /** Options for the slice-derived tokenizer (ignored when `tokenizer` is given) */
+    readonly tokenizerOptions?: DomainTokenizerOptions;
+}
+/**
+ * The one-call path into `createMultilingualDSL`: build a complete
+ * `LanguageConfig` (tokenizer + pattern profile + names) from a grammar slice
+ * and a domain vocabulary.
+ */
+export declare function buildLanguageConfig(slice: GrammarProfileSlice, vocab: DomainVocabulary, meta?: LanguageConfigMeta): LanguageConfig;
+/**
+ * Derive schema `markerOverride` defaults for ONE language from the slice's
+ * role markers, given a domain-role → semantic-role mapping. Returns
+ * `{ domainRole: primaryMarker }` with entries only where the slice has a
+ * marker for the mapped semantic role.
+ *
+ * ```ts
+ * deriveRoleMarkers(jaSlice, { table: 'source', target: 'destination' });
+ * // → { table: 'から', target: 'に' }
+ * ```
+ *
+ * Explicitly authored `markerOverride` entries stay authoritative — when
+ * assembling a per-language marker map for a schema role, spread the explicit
+ * entries after the derived defaults.
+ */
+export declare function deriveRoleMarkers(slice: GrammarProfileSlice, roleMapping: Readonly<Record<string, string>>): Record<string, string>;
+//# sourceMappingURL=builders.d.ts.map

package/dist/multilingual/builders.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"builders.d.ts","sourceRoot":"","sources":["../../src/multilingual/builders.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAEH,OAAO,KAAK,EAAE,iBAAiB,EAAE,MAAM,eAAe,CAAC;AACvD,OAAO,KAAK,EAAE,yBAAyB,EAAE,MAAM,iCAAiC,CAAC;AACjF,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,mBAAmB,CAAC;AACxD,OAAO,KAAK,EAAE,eAAe,IAAI,cAAc,EAAE,MAAM,YAAY,CAAC;AAEpE,OAAO,EAEL,KAAK,cAAc,EACpB,MAAM,+BAA+B,CAAC;AACvC,OAAO,KAAK,EAAE,gBAAgB,EAAE,mBAAmB,EAAmB,MAAM,SAAS,CAAC;AA+BtF;;;;GAIG;AACH,wBAAgB,mBAAmB,CACjC,KAAK,EAAE,mBAAmB,EAC1B,KAAK,EAAE,gBAAgB,GACtB,yBAAyB,CAiB3B;AAED,uFAAuF;AACvF,MAAM,WAAW,sBAAsB;IACrC;;;;OAIG;IACH,QAAQ,CAAC,gBAAgB,CAAC,EAAE,OAAO,CAAC;IACpC;;;OAGG;IACH,QAAQ,CAAC,eAAe,CAAC,EAAE,OAAO,CAAC;IACnC,iEAAiE;IACjE,QAAQ,CAAC,gBAAgB,CAAC,EAAE,SAAS,cAAc,EAAE,CAAC;CACvD;AAMD;;;;;;;;;;;;GAYG;AACH,wBAAgB,oBAAoB,CAClC,KAAK,EAAE,mBAAmB,EAC1B,KAAK,EAAE,gBAAgB,EACvB,OAAO,GAAE,sBAA2B,GACnC,iBAAiB,CA6CnB;AAED,4DAA4D;AAC5D,MAAM,WAAW,kBAAkB;IACjC,yEAAyE;IACzE,QAAQ,CAAC,IAAI,CAAC,EAAE,MAAM,CAAC;IACvB,kFAAkF;IAClF,QAAQ,CAAC,UAAU,CAAC,EAAE,MAAM,CAAC;IAC7B,kFAAkF;IAClF,QAAQ,CAAC,cAAc,CAAC,EAAE,cAAc,CAAC;IACzC,0DAA0D;IAC1D,QAAQ,CAAC,SAAS,CAAC,EAAE,iBAAiB,CAAC;IACvC,kFAAkF;IAClF,QAAQ,CAAC,gBAAgB,CAAC,EAAE,sBAAsB,CAAC;CACpD;AAED;;;;GAIG;AACH,wBAAgB,mBAAmB,CACjC,KAAK,EAAE,mBAAmB,EAC1B,KAAK,EAAE,gBAAgB,EACvB,IAAI,GAAE,kBAAuB,GAC5B,cAAc,CAUhB;AAED;;;;;;;;;;;;;;GAcG;AACH,wBAAgB,iBAAiB,CAC/B,KAAK,EAAE,mBAAmB,EAC1B,WAAW,EAAE,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,GAC5C,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAOxB"}

package/dist/multilingual/index.d.ts

@@ -1,8 +1,12 @@
 /**
- * Multilingual support - language profiles and templates
+ * Multilingual support — the framework ↔ semantic bridge.
  *
- * TODO: Extract language profile templates from semantic package
- * For now, this module is a placeholder for future language profile utilities.
+ * Injection-based: domains import per-language grammar data (e.g. semantic's
+ * `KNOWN_PROFILES`) and inject it into these builders; framework itself never
+ * imports the semantic package. See `docs-internal/FRAMEWORK_SEMANTIC_BRIDGE_PLAN.md`.
  */
+export type { GrammarProfileSlice, DomainVocabulary, DomainKeywordTranslation, DomainKeywordEntry, RoleMarkerSlice, TokenizationSlice, VerbSlice, } from './types';
+export { buildPatternProfile, buildDomainTokenizer, buildLanguageConfig, deriveRoleMarkers, } from './builders';
+export type { DomainTokenizerOptions, LanguageConfigMeta } from './builders';
 export type { LanguageProfile, WordOrder, AdpositionType, MorphologyType } from '../grammar';
 //# sourceMappingURL=index.d.ts.map

package/dist/multilingual/index.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/multilingual/index.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAGH,YAAY,EAAE,eAAe,EAAE,SAAS,EAAE,cAAc,EAAE,cAAc,EAAE,MAAM,YAAY,CAAC"}
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/multilingual/index.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAEH,YAAY,EACV,mBAAmB,EACnB,gBAAgB,EAChB,wBAAwB,EACxB,kBAAkB,EAClB,eAAe,EACf,iBAAiB,EACjB,SAAS,GACV,MAAM,SAAS,CAAC;AAEjB,OAAO,EACL,mBAAmB,EACnB,oBAAoB,EACpB,mBAAmB,EACnB,iBAAiB,GAClB,MAAM,YAAY,CAAC;AACpB,YAAY,EAAE,sBAAsB,EAAE,kBAAkB,EAAE,MAAM,YAAY,CAAC;AAG7E,YAAY,EAAE,eAAe,EAAE,SAAS,EAAE,cAAc,EAAE,cAAc,EAAE,MAAM,YAAY,CAAC"}