@xema/omni-protocol 0.1.127 → 0.1.128

package/dist/index.js

@@ -160,7 +160,8 @@ export function localizedText(text, languages) {
         return undefined;
     const byTag = new Map(Object.entries(text).map(([tag, value]) => [tag.toLowerCase(), value]));
     for (const language of languages) {
-        for (let tag = language.toLowerCase(); tag !== ""; tag = tag.includes("-") ? tag.slice(0, tag.lastIndexOf("-")) : "") {
+        // RFC 4647 Lookup: drop subtags from the end, and a single-letter extension marker with the one before it.
+        for (let tag = language.toLowerCase(); tag !== ""; tag = shorterTag(tag)) {
             const found = byTag.get(tag);
             if (found !== undefined)
                 return found;
@@ -168,6 +169,12 @@ export function localizedText(text, languages) {
     }
     return undefined;
 }
+function shorterTag(tag) {
+    if (!tag.includes("-"))
+        return "";
+    const shorter = tag.slice(0, tag.lastIndexOf("-"));
+    return /-[a-z0-9]$/.test(shorter) ? shorter.slice(0, -2) : shorter;
+}
 // ---------------------------------------------------------------------------
 // Utilities.
 // ---------------------------------------------------------------------------

package/dist/validation.js

@@ -561,8 +561,17 @@ export function validateManifest(manifest, path = "manifest") {
     }
     return into.violations;
 }
-/** Language tags as BCP 47 writes them: a language, then subtags of up to eight letters or digits. */
-const LANGUAGE_TAG = /^[A-Za-z]{2,3}(?:-[A-Za-z0-9]{1,8})*$/;
+/** A well-formed BCP 47 tag as ECMA-402 reads one, a Unicode locale identifier: judged by its spelling, never by what the machine has installed. */
+function isLanguageTag(value) {
+    if (typeof value !== "string" || value === "")
+        return false;
+    try {
+        return Intl.getCanonicalLocales(value).length === 1;
+    }
+    catch {
+        return false;
+    }
+}
 /** The agent's languages, as a host states them on both contexts: BCP 47 tags, most preferred first, at least one. */
 export function validateLanguages(value, path = "context.languages") {
     const into = new Collector();
@@ -572,7 +581,7 @@ export function validateLanguages(value, path = "context.languages") {
     }
     const seen = new Set();
     value.forEach((tag, index) => {
-        if (typeof tag !== "string" || !LANGUAGE_TAG.test(tag)) {
+        if (!isLanguageTag(tag)) {
             into.add("context.languages.tag", `${path}[${index}]`, `${describeValue(tag)} is not a BCP 47 language tag`);
             return;
         }
@@ -633,7 +642,7 @@ function validateLocalizedText(value, rule, path, into) {
     }
     const seen = new Set();
     for (const [tag, text] of Object.entries(value)) {
-        if (!into.require(LANGUAGE_TAG.test(tag), `${rule}.language`, `${path}.${tag}`, `${describeValue(tag)} is not a language tag such as en, hi or en-IN`))
+        if (!into.require(isLanguageTag(tag), `${rule}.language`, `${path}.${tag}`, `${describeValue(tag)} is not a language tag such as en, hi or en-IN`))
             continue;
         into.require(!seen.has(tag.toLowerCase()), `${rule}.language.unique`, `${path}.${tag}`, "language tags are case-insensitive, and each language is stated once");
         seen.add(tag.toLowerCase());

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@xema/omni-protocol",
-  "version": "0.1.127",
+  "version": "0.1.128",
   "description": "The Omni protocol: the contract every provider adapter implements",
   "type": "module",
   "license": "MIT",