@lokascript/framework 2.6.0 → 2.7.0

package/dist/index.cjs

@@ -47,7 +47,10 @@ __export(index_exports, {
   WhitespaceExtractor: () => WhitespaceExtractor,
   accumulateBlocks: () => accumulateBlocks,
   buildDisambiguation: () => buildDisambiguation,
+  buildDomainTokenizer: () => buildDomainTokenizer,
   buildFeedback: () => buildFeedback,
+  buildLanguageConfig: () => buildLanguageConfig,
+  buildPatternProfile: () => buildPatternProfile,
   buildPhrase: () => buildPhrase,
   buildTablesFromProfiles: () => buildTablesFromProfiles,
   combineClassifiers: () => combineClassifiers,
@@ -78,6 +81,7 @@ __export(index_exports, {
   createUnicodeRangeClassifier: () => createUnicodeRangeClassifier,
   defineCommand: () => import_intent6.defineCommand,
   defineRole: () => import_intent6.defineRole,
+  deriveRoleMarkers: () => deriveRoleMarkers,
   detectWordOrders: () => detectWordOrders,
   extractCssSelector: () => extractCssSelector,
   extractNumber: () => extractNumber,
@@ -5008,22 +5012,6 @@ function withDefaultExtractors(tokenizer) {
   return tokenizer;
 }
 
-// src/core/tokenization/char-classifiers.ts
-function createUnicodeRangeClassifier(ranges) {
-  return (char) => {
-    const code = char.charCodeAt(0);
-    return ranges.some(([start, end]) => code >= start && code <= end);
-  };
-}
-function combineClassifiers(...classifiers) {
-  return (char) => classifiers.some((fn) => fn(char));
-}
-function createLatinCharClassifiers(letterPattern) {
-  const isLetter = (char) => letterPattern.test(char);
-  const isIdentifierChar = (char) => isLetter(char) || /[0-9_-]/.test(char);
-  return { isLetter, isIdentifierChar };
-}
-
 // src/core/tokenization/base-tokenizer.ts
 var SIMPLE_TOKENIZER_OPERATOR_SET = new Set(DEFAULT_OPERATORS);
 var MARKER_CONCEPT_NORMALIZEDS = /* @__PURE__ */ new Set([
@@ -5785,6 +5773,119 @@ function createSimpleTokenizer(config) {
   return new SimpleTokenizer();
 }
 
+// src/multilingual/builders.ts
+function mergeRoleMarkers(slice, vocab) {
+  const merged = {};
+  const add = (role, marker) => {
+    if (!marker?.primary) {
+      delete merged[role];
+      return;
+    }
+    merged[role] = {
+      primary: marker.primary,
+      ...marker.alternatives?.length && { alternatives: [...marker.alternatives] },
+      ...marker.position && { position: marker.position }
+    };
+  };
+  for (const [role, marker] of Object.entries(slice.roleMarkers ?? {})) add(role, marker);
+  for (const [role, marker] of Object.entries(vocab.roleMarkerOverrides ?? {})) add(role, marker);
+  return merged;
+}
+function buildPatternProfile(slice, vocab) {
+  const keywords = {};
+  for (const [action, translation] of Object.entries(vocab.keywords)) {
+    keywords[action] = {
+      primary: translation.primary,
+      ...translation.alternatives?.length && { alternatives: [...translation.alternatives] }
+    };
+  }
+  const roleMarkers = mergeRoleMarkers(slice, vocab);
+  return {
+    code: slice.code,
+    wordOrder: slice.wordOrder,
+    keywords,
+    ...Object.keys(roleMarkers).length > 0 && { roleMarkers }
+  };
+}
+function defaultCaseInsensitive(script) {
+  return script === void 0 || script === "latin" || script === "cyrillic";
+}
+function buildDomainTokenizer(slice, vocab, options = {}) {
+  const roleMarkers = mergeRoleMarkers(slice, vocab);
+  const keywords = /* @__PURE__ */ new Set();
+  for (const translation of Object.values(vocab.keywords)) {
+    keywords.add(translation.primary);
+    for (const alt of translation.alternatives ?? []) keywords.add(alt);
+  }
+  for (const marker of Object.values(roleMarkers)) {
+    keywords.add(marker.primary);
+    for (const alt of marker.alternatives ?? []) keywords.add(alt);
+  }
+  for (const particle of slice.tokenization?.particles ?? []) keywords.add(particle);
+  for (const extra of vocab.tokenizerKeywords ?? []) keywords.add(extra);
+  const profileKeywords = {};
+  for (const [action, translation] of Object.entries(vocab.keywords)) {
+    profileKeywords[action] = {
+      primary: translation.primary,
+      ...translation.alternatives?.length && { alternatives: [...translation.alternatives] },
+      normalized: translation.normalized ?? action
+    };
+  }
+  const keywordProfile = {
+    keywords: profileKeywords,
+    ...Object.keys(roleMarkers).length > 0 && { roleMarkers }
+  };
+  const customExtractors = [
+    ...options.customExtractors ?? [],
+    ...slice.script === "latin" ? [new LatinExtendedIdentifierExtractor()] : []
+  ];
+  return createSimpleTokenizer({
+    language: slice.code,
+    direction: slice.direction ?? "ltr",
+    keywords: [...keywords],
+    ...vocab.keywordExtras?.length && { keywordExtras: vocab.keywordExtras.map((e) => ({ ...e })) },
+    keywordProfile,
+    includeOperators: options.includeOperators ?? false,
+    caseInsensitive: options.caseInsensitive ?? defaultCaseInsensitive(slice.script),
+    ...customExtractors.length > 0 && { customExtractors }
+  });
+}
+function buildLanguageConfig(slice, vocab, meta = {}) {
+  const name = meta.name ?? slice.name ?? slice.code;
+  return {
+    code: slice.code,
+    name,
+    nativeName: meta.nativeName ?? slice.nativeName ?? name,
+    tokenizer: meta.tokenizer ?? buildDomainTokenizer(slice, vocab, meta.tokenizerOptions),
+    patternProfile: buildPatternProfile(slice, vocab),
+    ...meta.grammarProfile && { grammarProfile: meta.grammarProfile }
+  };
+}
+function deriveRoleMarkers(slice, roleMapping) {
+  const derived = {};
+  for (const [domainRole, semanticRole] of Object.entries(roleMapping)) {
+    const marker = slice.roleMarkers?.[semanticRole];
+    if (marker?.primary) derived[domainRole] = marker.primary;
+  }
+  return derived;
+}
+
+// src/core/tokenization/char-classifiers.ts
+function createUnicodeRangeClassifier(ranges) {
+  return (char) => {
+    const code = char.charCodeAt(0);
+    return ranges.some(([start, end]) => code >= start && code <= end);
+  };
+}
+function combineClassifiers(...classifiers) {
+  return (char) => classifiers.some((fn) => fn(char));
+}
+function createLatinCharClassifiers(letterPattern) {
+  const isLetter = (char) => letterPattern.test(char);
+  const isIdentifierChar = (char) => isLetter(char) || /[0-9_-]/.test(char);
+  return { isLetter, isIdentifierChar };
+}
+
 // src/core/tokenization/morphology/types.ts
 function noChange(word) {
   return { stem: word, confidence: 1 };
@@ -6351,7 +6452,10 @@ function accumulateIndented(statements, config) {
   WhitespaceExtractor,
   accumulateBlocks,
   buildDisambiguation,
+  buildDomainTokenizer,
   buildFeedback,
+  buildLanguageConfig,
+  buildPatternProfile,
   buildPhrase,
   buildTablesFromProfiles,
   combineClassifiers,
@@ -6382,6 +6486,7 @@ function accumulateIndented(statements, config) {
   createUnicodeRangeClassifier,
   defineCommand,
   defineRole,
+  deriveRoleMarkers,
   detectWordOrders,
   extractCssSelector,
   extractNumber,