wikitongues-db 0.2.2 → 0.2.4

package/dist/index.mjs

@@ -217,7 +217,13 @@ var MULTILINGUAL_ALIASES = {
   sw: "swh",
   so: "som",
   zu: "zul",
-  xh: "xho"
+  xh: "xho",
+  et: "est",
+  az: "aze",
+  lv: "lav",
+  ne: "nep",
+  ff: "ful",
+  mg: "mlg"
 };
 var MACROLANGUAGE_EXPANSIONS = {
   fas: ["fas", "pes", "prs"],
@@ -266,7 +272,10 @@ var MACROLANGUAGE_EXPANSIONS = {
   est: ["est", "ekk", "vro"],
   aka: ["aka", "twi", "fat"],
   nep: ["nep", "npi"],
-  ful: ["ful", "fuf", "fuh", "fub", "fuq"]
+  ful: ["ful", "fuf", "fuh", "fub", "fuq"],
+  lav: ["lav", "lvs", "ltg"],
+  mon: ["mon", "khk", "mvf"],
+  mlg: ["mlg", "plt", "skg", "tkg"]
 };
 var LanguageResolver = class {
   isoToName = /* @__PURE__ */ new Map();
@@ -308,6 +317,14 @@ var LanguageResolver = class {
         this.aliases.set(prefix, iso);
       }
     }
+    const part1 = lang.standards.iso639_3.part1?.toLowerCase();
+    if (part1 && !this.aliases.has(part1)) {
+      this.aliases.set(part1, iso);
+    }
+    const macro = lang.standards.bcp47.macrolanguage?.toLowerCase();
+    if (macro && !this.aliases.has(macro)) {
+      this.aliases.set(macro, iso);
+    }
     const gc = lang.glottocode;
     this.glottoToIso.set(gc, iso);
     this.isoToGlotto.set(iso, gc);
@@ -373,7 +390,7 @@ var LanguageResolver = class {
         }
       }
     }
-    if (matchedIsos.size === 0) {
+    if (matchedIsos.size === 0 && norm.length >= 4) {
       for (const [autoKey, isoSet] of this.autonymToIso.entries()) {
         if (autoKey.includes(norm)) {
           for (const iso of isoSet) {
@@ -9900,6 +9917,7 @@ var ReferenceHydrator = class {
     }
     const resolved = { tag: raw, primarySubtag: primary, variantSubtags: [] };
     if (lang.description) resolved.description = lang.description;
+    if (lang.macrolanguage) resolved.macrolanguage = lang.macrolanguage;
     let state = 0;
     for (const sub of parts.slice(1)) {
       const lower = sub.toLowerCase();
@@ -9997,6 +10015,21 @@ var Language = class _Language {
     if (this.speakerClaim) out.push(this.speakerClaim);
     return out;
   }
+  /**
+   * Check if this language matches the given code by ISO 639-3, ISO 639-1 part1,
+   * BCP-47 (exact or prefix), macrolanguage, or Glottocode.
+   */
+  matchesCode(code) {
+    const c = code.trim().toLowerCase();
+    if (!c) return false;
+    if (this.iso639_3.toLowerCase() === c) return true;
+    if (this.standards.iso639_3.part1?.toLowerCase() === c) return true;
+    if (this.standards.bcp47.macrolanguage?.toLowerCase() === c) return true;
+    const bcp = this.bcp47.toLowerCase();
+    if (bcp === c || bcp.startsWith(`${c}-`)) return true;
+    if (this.glottocode.toLowerCase() === c) return true;
+    return false;
+  }
   toDict() {
     return {
       standards: {
@@ -10181,6 +10214,8 @@ var Video = class _Video {
     const set = /* @__PURE__ */ new Set();
     for (const lang of this.allLanguages) {
       if (lang.iso639_3) set.add(lang.iso639_3);
+      if (lang.standards.iso639_3.part1) set.add(lang.standards.iso639_3.part1);
+      if (lang.standards.bcp47.macrolanguage) set.add(lang.standards.bcp47.macrolanguage);
     }
     return set;
   }
@@ -10191,6 +10226,9 @@ var Video = class _Video {
     const set = /* @__PURE__ */ new Set();
     for (const lang of this.allLanguages) {
       if (lang.bcp47) set.add(lang.bcp47);
+      if (lang.iso639_3) set.add(lang.iso639_3);
+      if (lang.standards.iso639_3.part1) set.add(lang.standards.iso639_3.part1);
+      if (lang.standards.bcp47.macrolanguage) set.add(lang.standards.bcp47.macrolanguage);
     }
     return set;
   }
@@ -10228,6 +10266,8 @@ var Video = class _Video {
     const qNorm = normalizeText(q);
     for (const lang of this.allLanguages) {
       if (lang.iso639_3 === qLower) return true;
+      if (lang.standards.iso639_3.part1?.toLowerCase() === qLower) return true;
+      if (lang.standards.bcp47.macrolanguage?.toLowerCase() === qLower) return true;
       const bcp = lang.bcp47.toLowerCase();
       if (bcp === qLower || bcp.startsWith(`${qLower}-`)) return true;
       if (lang.glottocode === qLower) return true;
@@ -10236,7 +10276,8 @@ var Video = class _Video {
         const labelNorm = normalizeText(label);
         if (labelNorm === qNorm || ` ${labelNorm} `.includes(` ${qNorm} `)) return true;
       }
-      if (lang.autonym.toLowerCase().includes(qLower)) return true;
+      const autoLower = lang.autonym.toLowerCase();
+      if (autoLower === qLower || qLower.length >= 4 && autoLower.includes(qLower)) return true;
     }
     return false;
   }
@@ -10344,6 +10385,12 @@ var VideoCollection = class _VideoCollection {
     }
     return this._videos.some((v) => v.id === item.id);
   }
+  /**
+   * Check if any video in the collection matches the given language query.
+   */
+  hasLanguage(query) {
+    return this._videos.some((v) => v.hasLanguage(query));
+  }
   // -------------------------------------------------------------------------
   // Aggregation & Summary Properties
   // -------------------------------------------------------------------------
@@ -10649,7 +10696,7 @@ var DatasetIndex = class {
     const list = map.get(key);
     if (!list) {
       map.set(key, [video]);
-    } else {
+    } else if (!list.some((v) => v.id === video.id)) {
       list.push(video);
     }
   }
@@ -10662,12 +10709,30 @@ var DatasetIndex = class {
       if (plIso) {
         this.appendToMap(this.byIsoPrimary, plIso, video);
         this.appendToMap(this.byIso, plIso, video);
+        const part1 = video.primaryLanguage.standards.iso639_3.part1?.toLowerCase();
+        if (part1 && part1 !== plIso) {
+          this.appendToMap(this.byIsoPrimary, part1, video);
+          this.appendToMap(this.byIso, part1, video);
+        }
+        const macro = video.primaryLanguage.standards.bcp47.macrolanguage?.toLowerCase();
+        if (macro && macro !== plIso && macro !== part1) {
+          this.appendToMap(this.byIsoPrimary, macro, video);
+          this.appendToMap(this.byIso, macro, video);
+        }
       }
       for (const addLang of video.additionalLanguages) {
         const aIso = addLang.iso639_3;
-        if (aIso && aIso !== plIso) {
+        if (aIso) {
           this.appendToMap(this.byIso, aIso, video);
         }
+        const part1 = addLang.standards.iso639_3.part1?.toLowerCase();
+        if (part1 && part1 !== aIso) {
+          this.appendToMap(this.byIso, part1, video);
+        }
+        const macro = addLang.standards.bcp47.macrolanguage?.toLowerCase();
+        if (macro && macro !== aIso && macro !== part1) {
+          this.appendToMap(this.byIso, macro, video);
+        }
       }
       for (const lang of video.allLanguages) {
         const bcp = lang.bcp47.toLowerCase().trim();
@@ -10678,6 +10743,18 @@ var DatasetIndex = class {
             this.appendToMap(this.byBcp47, prefix, video);
           }
         }
+        const iso = lang.iso639_3?.toLowerCase().trim();
+        if (iso && iso !== bcp) {
+          this.appendToMap(this.byBcp47, iso, video);
+        }
+        const part1 = lang.standards.iso639_3.part1?.toLowerCase().trim();
+        if (part1 && part1 !== bcp && part1 !== iso) {
+          this.appendToMap(this.byBcp47, part1, video);
+        }
+        const macro = lang.standards.bcp47.macrolanguage?.toLowerCase().trim();
+        if (macro && macro !== bcp && macro !== iso && macro !== part1) {
+          this.appendToMap(this.byBcp47, macro, video);
+        }
       }
       for (const lang of video.allLanguages) {
         this.appendToMap(this.byGlottocode, lang.glottocode, video);
@@ -10920,6 +10997,14 @@ var QueryBuilder = class _QueryBuilder {
       if (matchedIsos.has(lIso) || lIso === rawLower) {
         return true;
       }
+      const part1 = lang.standards.iso639_3.part1?.toLowerCase();
+      if (part1 && (matchedIsos.has(part1) || part1 === rawLower)) {
+        return true;
+      }
+      const macro = lang.standards.bcp47.macrolanguage?.toLowerCase();
+      if (macro && (matchedIsos.has(macro) || macro === rawLower)) {
+        return true;
+      }
       const lBcp = lang.bcp47.toLowerCase();
       if (lBcp === rawLower || lBcp.startsWith(`${rawLower}-`)) {
         return true;
@@ -10932,7 +11017,9 @@ var QueryBuilder = class _QueryBuilder {
           const lNorm = normalizeText(label);
           if (lNorm === norm || pattern && pattern.test(lNorm)) return true;
         }
-        if (lang.autonym.toLowerCase().includes(rawLower)) return true;
+        const autoLower = lang.autonym.toLowerCase();
+        if (autoLower === rawLower) return true;
+        if (rawLower.length >= 4 && autoLower.includes(rawLower)) return true;
       }
       return false;
     };
@@ -10955,13 +11042,17 @@ var QueryBuilder = class _QueryBuilder {
   iso(code, includeAdditional = true) {
     const codeClean = code.trim().toLowerCase();
     const predicate = (v) => {
-      if (v.primaryLanguage.iso639_3.toLowerCase() === codeClean) {
+      const matchIso = (l) => {
+        if (l.iso639_3.toLowerCase() === codeClean) return true;
+        if (l.standards.iso639_3.part1?.toLowerCase() === codeClean) return true;
+        if (l.standards.bcp47.macrolanguage?.toLowerCase() === codeClean) return true;
+        return false;
+      };
+      if (matchIso(v.primaryLanguage)) {
         return true;
       }
       if (includeAdditional) {
-        return v.additionalLanguages.some(
-          (al) => al.iso639_3.toLowerCase() === codeClean
-        );
+        return v.additionalLanguages.some(matchIso);
       }
       return false;
     };
@@ -10973,10 +11064,13 @@ var QueryBuilder = class _QueryBuilder {
     const predicate = (v) => {
       for (const lang of v.allLanguages) {
         const lTag = lang.bcp47.toLowerCase();
+        const lIso = lang.iso639_3.toLowerCase();
+        const part1 = lang.standards.iso639_3.part1?.toLowerCase();
+        const macro = lang.standards.bcp47.macrolanguage?.toLowerCase();
         if (exact) {
-          if (lTag === tagClean) return true;
+          if (lTag === tagClean || lIso === tagClean || part1 === tagClean || macro === tagClean) return true;
         } else {
-          if (lTag === tagClean || lTag.startsWith(`${tagClean}-`)) return true;
+          if (lTag === tagClean || lTag.startsWith(`${tagClean}-`) || lIso === tagClean || part1 === tagClean || macro === tagClean) return true;
         }
       }
       return false;
@@ -66507,6 +66601,16 @@ var WikitonguesDB = class _WikitonguesDB {
   findByLanguage(languageQuery, includeAdditional = true) {
     return this.query().language(languageQuery, includeAdditional).all();
   }
+  /**
+   * Check if the database contains any video matching the language query
+   * (by ISO 639-3, ISO 639-1, BCP 47, macrolanguage, or name).
+   */
+  hasLanguage(query) {
+    const qClean = query.trim().toLowerCase();
+    if (this.index.byIso.has(qClean)) return true;
+    if (this.index.byBcp47.has(qClean)) return true;
+    return this.findByLanguage(query).length > 0;
+  }
   /**
    * Shortcut method to filter videos by common criteria.
    */