wikitongues-db 0.2.1 → 0.2.3

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) {
@@ -486,6 +503,12 @@ var reference_default = {
       type: "L",
       invertedName: "Albanian, Tosk"
     },
+    alt: {
+      name: "Southern Altai",
+      scope: "I",
+      type: "L",
+      invertedName: "Altai, Southern"
+    },
     amh: {
       name: "Amharic",
       scope: "I",
@@ -1653,6 +1676,12 @@ var reference_default = {
       type: "L",
       invertedName: "Kabiy\xE8"
     },
+    kcg: {
+      name: "Tyap",
+      scope: "I",
+      type: "L",
+      invertedName: "Tyap"
+    },
     kdd: {
       name: "Yankunytjatjara",
       scope: "I",
@@ -1741,6 +1770,12 @@ var reference_default = {
       type: "L",
       invertedName: "Khmu"
     },
+    kls: {
+      name: "Kalasha",
+      scope: "I",
+      type: "L",
+      invertedName: "Kalasha"
+    },
     klu: {
       name: "Klao",
       scope: "I",
@@ -5221,6 +5256,14 @@ var reference_default = {
       longitude: 47.776611,
       familyId: "nakh1245"
     },
+    kala1372: {
+      name: "Chitral Kalasha",
+      level: "language",
+      macroarea: "Eurasia",
+      latitude: 35.6415,
+      longitude: 71.6657,
+      familyId: "indo1319"
+    },
     kala1399: {
       name: "Kalaallisut",
       level: "language",
@@ -6539,6 +6582,15 @@ var reference_default = {
       latitude: 16.74515934758395,
       longitude: -94.92819268907601
     },
+    ower1238: {
+      name: "Owerri",
+      level: "dialect",
+      macroarea: "Africa",
+      latitude: 4.62705,
+      longitude: 7.23441,
+      familyId: "atla1278",
+      parentLanguageId: "nucl1417"
+    },
     paip1241: {
       name: "Paipai",
       level: "language",
@@ -7004,6 +7056,15 @@ var reference_default = {
       familyId: "sino1245",
       parentLanguageId: "wuch1236"
     },
+    shet1241: {
+      name: "Shaetlan",
+      level: "dialect",
+      macroarea: "Eurasia",
+      latitude: 55.886945,
+      longitude: -3.710133,
+      familyId: "indo1319",
+      parentLanguageId: "scot1243"
+    },
     shin1264: {
       name: "Shina",
       level: "language",
@@ -7184,6 +7245,14 @@ var reference_default = {
       longitude: 27,
       familyId: "ural1272"
     },
+    sout2694: {
+      name: "Southern Altai",
+      level: "language",
+      macroarea: "Eurasia",
+      latitude: 50.1347,
+      longitude: 86.8705,
+      familyId: "turk1311"
+    },
     sout2772: {
       name: "Southern Daly",
       level: "family",
@@ -7704,6 +7773,14 @@ var reference_default = {
       familyId: "atla1278",
       parentLanguageId: "akan1250"
     },
+    tyap1238: {
+      name: "Tyap",
+      level: "language",
+      macroarea: "Africa",
+      latitude: 9.75568,
+      longitude: 8.3082,
+      familyId: "atla1278"
+    },
     uigh1240: {
       name: "Uighur",
       level: "language",
@@ -8069,6 +8146,9 @@ var reference_default = {
         description: "Tosk Albanian",
         macrolanguage: "sq"
       },
+      alt: {
+        description: "Southern Altai"
+      },
       am: {
         description: "Amharic",
         suppressScript: "Ethi"
@@ -8687,6 +8767,9 @@ var reference_default = {
       kbp: {
         description: "Kabiy\xE8"
       },
+      kcg: {
+        description: "Tyap"
+      },
       kdd: {
         description: "Yankunytjatjara"
       },
@@ -8732,6 +8815,9 @@ var reference_default = {
         description: "Kalaallisut; Greenlandic",
         suppressScript: "Latn"
       },
+      kls: {
+        description: "Kalasha"
+      },
       klu: {
         description: "Klao"
       },
@@ -9718,6 +9804,9 @@ var reference_default = {
       sy: {
         description: "Syrian Arab Republic"
       },
+      tw: {
+        description: "Taiwan, Province of China"
+      },
       us: {
         description: "United States"
       }
@@ -9828,6 +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();
@@ -10109,6 +10199,8 @@ 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;
   }
@@ -10119,6 +10211,9 @@ 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;
   }
@@ -10156,6 +10251,8 @@ 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;
@@ -10164,7 +10261,8 @@ 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;
   }
@@ -10272,6 +10370,12 @@ 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
   // -------------------------------------------------------------------------
@@ -10577,7 +10681,7 @@ 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);
     }
   }
@@ -10590,12 +10694,30 @@ 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();
@@ -10606,6 +10728,18 @@ 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);
@@ -10848,6 +10982,14 @@ 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;
@@ -10860,7 +11002,9 @@ 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;
     };
@@ -10883,13 +11027,17 @@ 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;
     };
@@ -10901,10 +11049,13 @@ 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;
@@ -63950,140 +64101,2500 @@ Best wishes to everybody at home, and I'm looking forward to be surrounded by re
         "2019 International Year of Indigenous Languages"
       ]
     }
-  }
-];
-
-// src/dataset.ts
-var dataset = wikitongues_normalized_default;
-
-// src/client.ts
-var WikitonguesDB = class _WikitonguesDB {
-  _videos = [];
-  hydrator;
-  resolver;
-  index;
-  searchEngine;
-  /**
-   * Loads and hydrates every record. Throws HydrationError on the first record whose
-   * standards do not resolve against the reference tables (see CLASSIFICATION_RULES.md §3.1.1).
-   */
-  constructor(options) {
-    this.hydrator = new ReferenceHydrator(options?.reference);
-    this.resolver = new LanguageResolver();
-    const rawRecords = options?.data || options?.records || dataset;
-    for (const record of rawRecords) {
-      const video = record instanceof Video ? record : Video.fromDict(record, this.hydrator);
-      this._videos.push(video);
-      for (const lang of video.allLanguages) {
-        this.resolver.registerDatasetLanguage(lang);
+  },
+  {
+    id: "MxvdCaFjJZo",
+    url: "https://www.youtube.com/watch?v=MxvdCaFjJZo",
+    duration_seconds: 25,
+    upload_date: "2026-09-18",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "zul",
+        glottocode: "zulu1248",
+        bcp47: "zu"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Zulu",
+      wikitongues_lineage: null,
+      autonym: "isiZulu"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Halala Shezi (Lolly)",
+        role: "native",
+        origin: "South Africa"
       }
+    ],
+    provenance: {
+      country_code: "ZA",
+      country_name: "South Africa",
+      region: null,
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Lolly speaking Zulu",
+      tags: [
+        "Wikitongues"
+      ]
     }
-    this.index = new DatasetIndex(this._videos);
-    this.searchEngine = new SearchEngine(this._videos, this.resolver);
-  }
-  get length() {
-    return this._videos.length;
-  }
-  // -------------------------------------------------------------------------
-  // Static Factory Methods
-  // -------------------------------------------------------------------------
-  /**
-   * Create a WikitonguesDB instance directly from an in-memory list of dicts or Videos.
-   */
-  static fromRecords(records) {
-    return new _WikitonguesDB({ records });
-  }
-  /**
-   * Create a WikitonguesDB instance from a JSON string or JSON array.
-   */
-  static fromJSON(json) {
-    const data = typeof json === "string" ? JSON.parse(json) : json;
-    return new _WikitonguesDB({ data });
-  }
-  /**
-   * Create a WikitonguesDB instance from a JSON or JSONL file path (Node.js environments).
-   */
-  static fromFile(filepath) {
-    try {
-      const fs = __require("fs");
-      const content = fs.readFileSync(filepath, "utf-8");
-      const trimmed = content.trim();
-      let records = [];
-      if (trimmed.startsWith("[")) {
-        records = JSON.parse(trimmed);
-      } else {
-        records = trimmed.split("\n").filter((line) => line.trim()).map((line) => JSON.parse(line));
+  },
+  {
+    id: "WgbfWSs0YaY",
+    url: "https://www.youtube.com/watch?v=WgbfWSs0YaY",
+    duration_seconds: 26,
+    upload_date: "2026-09-16",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "zul",
+        glottocode: "zulu1248",
+        bcp47: "zu"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Zulu",
+      wikitongues_lineage: null,
+      autonym: "isiZulu"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Halala Shezi (Lolly)",
+        role: "native",
+        origin: "South Africa"
       }
-      return new _WikitonguesDB({ data: records });
-    } catch (err) {
-      throw new Error(`Failed to load dataset from file '${filepath}': ${err.message}`);
+    ],
+    provenance: {
+      country_code: "ZA",
+      country_name: "South Africa",
+      region: null,
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Lolly speaking Zulu",
+      tags: [
+        "Wikitongues"
+      ]
     }
-  }
-  // -------------------------------------------------------------------------
-  // Direct Lookups (O(1))
-  // -------------------------------------------------------------------------
-  /**
-   * Lookup a video by its YouTube ID in O(1) time.
-   */
-  get(videoId) {
-    return this.index.byId.get(videoId.trim()) || null;
-  }
-  /**
-   * Lookup videos matching a 3-letter SIL ISO 639-3 code in O(1) time.
-   */
-  getByIso(code) {
-    const list = this.index.byIso.get(code.trim().toLowerCase());
-    return new VideoCollection(list || []);
-  }
-  /**
-   * Lookup videos matching a BCP 47 language tag in O(1) time.
-   */
-  getByBcp47(tag) {
-    const list = this.index.byBcp47.get(tag.trim().toLowerCase());
-    return new VideoCollection(list || []);
-  }
-  /**
-   * Lookup videos matching a Glottolog code in O(1) time. A language node also matches videos
-   * classified under one of its dialect nodes (e.g. `port1283` returns `braz1246` recordings).
-   */
-  getByGlottocode(glottocode) {
-    const list = this.index.byGlottocode.get(glottocode.trim().toLowerCase());
-    return new VideoCollection(list || []);
-  }
-  /**
-   * Lookup videos matching an ISO 3166-1 alpha-2 code or country name in O(1) time.
-   */
-  getByCountry(countryCodeOrName) {
-    const c = countryCodeOrName.trim();
-    if (c.length === 2 && this.index.byCountryCode.has(c.toUpperCase())) {
-      return new VideoCollection(this.index.byCountryCode.get(c.toUpperCase()) || []);
+  },
+  {
+    id: "kF-HQw0BQNk",
+    url: "https://www.youtube.com/watch?v=kF-HQw0BQNk",
+    duration_seconds: 26,
+    upload_date: "2026-09-14",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "quc",
+        glottocode: "kich1262",
+        bcp47: "quc"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "K'iche'",
+      wikitongues_lineage: null,
+      autonym: "Qatzijob'al"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Lorenzo",
+        role: "native",
+        origin: "Guatemala"
+      }
+    ],
+    provenance: {
+      country_code: "GT",
+      country_name: "Guatemala",
+      region: null,
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Lorenzo speaking K'iche",
+      tags: [
+        "Wikitongues"
+      ]
     }
-    const normName = normalizeText(c);
-    if (this.index.byCountryName.has(normName)) {
-      return new VideoCollection(this.index.byCountryName.get(normName) || []);
+  },
+  {
+    id: "mp6Tj1hE55o",
+    url: "https://www.youtube.com/watch?v=mp6Tj1hE55o",
+    duration_seconds: 23,
+    upload_date: "2026-09-11",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "quc",
+        glottocode: "kich1262",
+        bcp47: "quc"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "K'iche'",
+      wikitongues_lineage: null,
+      autonym: "Qatzijob'al"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Lorenzo",
+        role: "native",
+        origin: "Guatemala"
+      }
+    ],
+    provenance: {
+      country_code: "GT",
+      country_name: "Guatemala",
+      region: null,
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Lorenzo speaking K'iche",
+      tags: [
+        "Wikitongues"
+      ]
     }
-    return this.query().country(c).all();
-  }
-  /**
-   * Lookup videos for a given speaker name.
-   */
-  getBySpeaker(name) {
-    const normName = normalizeText(name);
-    if (this.index.bySpeakerName.has(normName)) {
-      return new VideoCollection(this.index.bySpeakerName.get(normName) || []);
+  },
+  {
+    id: "LIcFLAdGTuw",
+    url: "https://www.youtube.com/watch?v=LIcFLAdGTuw",
+    duration_seconds: 38,
+    upload_date: "2026-09-09",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "quc",
+        glottocode: "kich1262",
+        bcp47: "quc"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "K'iche'",
+      wikitongues_lineage: null,
+      autonym: "Qatzijob'al"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Lorenzo",
+        role: "native",
+        origin: "Guatemala"
+      }
+    ],
+    provenance: {
+      country_code: "GT",
+      country_name: "Guatemala",
+      region: null,
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Lorenzo speaking K'iche",
+      tags: [
+        "Wikitongues"
+      ]
     }
-    return this.query().speaker({ name }).all();
-  }
-  // -------------------------------------------------------------------------
-  // Smart Lookups & Shortcut Queries
-  // -------------------------------------------------------------------------
-  /**
-   * Intelligent language search: resolves natural names, multilingual aliases ('russe'),
-   * ISO 639-3 ('rus'), BCP 47 ('ru'), Glottocodes, autonyms ('Русский'), Wikitongues
-   * classifications, Glottolog / ISO names and speaker claims.
-   */
-  findByLanguage(languageQuery, includeAdditional = true) {
-    return this.query().language(languageQuery, includeAdditional).all();
+  },
+  {
+    id: "KK1boSHTmIk",
+    url: "https://www.youtube.com/watch?v=KK1boSHTmIk",
+    duration_seconds: 46,
+    upload_date: "2026-04-23",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "cmn",
+        glottocode: "mand1415",
+        bcp47: "cmn"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Mandarin",
+      wikitongues_lineage: null,
+      autonym: "\u666E\u901A\u8BDD"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Wanyu",
+        role: "native",
+        origin: "China"
+      }
+    ],
+    provenance: {
+      country_code: "CN",
+      country_name: "China",
+      region: null,
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Wanyu speaking Mandarin",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages"
+      ]
+    }
+  },
+  {
+    id: "oGz_R-r-HLY",
+    url: "https://www.youtube.com/watch?v=oGz_R-r-HLY",
+    duration_seconds: 58,
+    upload_date: "2026-04-21",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "cmn",
+        glottocode: "mand1415",
+        bcp47: "cmn"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Mandarin",
+      wikitongues_lineage: null,
+      autonym: "\u666E\u901A\u8BDD"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Wanyu",
+        role: "native",
+        origin: "China"
+      }
+    ],
+    provenance: {
+      country_code: "CN",
+      country_name: "China",
+      region: null,
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Wanyu speaking Mandarin",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages"
+      ]
+    }
+  },
+  {
+    id: "Tg2w5WKPIOw",
+    url: "https://www.youtube.com/watch?v=Tg2w5WKPIOw",
+    duration_seconds: 39,
+    upload_date: "2026-04-19",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "eng",
+        glottocode: "stan1293",
+        bcp47: "en-BM"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Bermudian English",
+      wikitongues_lineage: null,
+      autonym: "English"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Trey",
+        role: "native",
+        origin: "Bermuda"
+      }
+    ],
+    provenance: {
+      country_code: "BM",
+      country_name: "Bermuda",
+      region: null,
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Trey speaking Bermudian English",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages"
+      ]
+    }
+  },
+  {
+    id: "-cbdIMW-cpU",
+    url: "https://www.youtube.com/watch?v=-cbdIMW-cpU",
+    duration_seconds: 33,
+    upload_date: "2026-04-17",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "eng",
+        glottocode: "stan1293",
+        bcp47: "en-BM"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Bermudian English",
+      wikitongues_lineage: null,
+      autonym: "English"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Trey",
+        role: "native",
+        origin: "Bermuda"
+      }
+    ],
+    provenance: {
+      country_code: "BM",
+      country_name: "Bermuda",
+      region: null,
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Trey speaking Bermudian English",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages"
+      ]
+    }
+  },
+  {
+    id: "inmozOx5Gos",
+    url: "https://www.youtube.com/watch?v=inmozOx5Gos",
+    duration_seconds: 78,
+    upload_date: "2026-04-15",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "eng",
+        glottocode: "stan1293",
+        bcp47: "en-BM"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Bermudian English",
+      wikitongues_lineage: null,
+      autonym: "English"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Trey",
+        role: "native",
+        origin: "Bermuda"
+      }
+    ],
+    provenance: {
+      country_code: "BM",
+      country_name: "Bermuda",
+      region: null,
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Trey speaks Bermudian English",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages"
+      ]
+    }
+  },
+  {
+    id: "2S-P_O-_1N8",
+    url: "https://www.youtube.com/watch?v=2S-P_O-_1N8",
+    duration_seconds: 45,
+    upload_date: "2026-03-22",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "cym",
+        glottocode: "wels1247",
+        bcp47: "cy"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Welsh",
+      wikitongues_lineage: null,
+      autonym: "Cymraeg"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Hywel",
+        role: "native",
+        origin: "Wales"
+      }
+    ],
+    provenance: {
+      country_code: "GB",
+      country_name: "United Kingdom",
+      region: "Wales",
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Hywel speaking Welsh",
+      tags: []
+    }
+  },
+  {
+    id: "N4q_mynBip0",
+    url: "https://www.youtube.com/watch?v=N4q_mynBip0",
+    duration_seconds: 30,
+    upload_date: "2026-03-20",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "cym",
+        glottocode: "wels1247",
+        bcp47: "cy"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Welsh",
+      wikitongues_lineage: null,
+      autonym: "Cymraeg"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Hywel",
+        role: "native",
+        origin: "Wales"
+      }
+    ],
+    provenance: {
+      country_code: "GB",
+      country_name: "United Kingdom",
+      region: "Wales",
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Hywel speaking Welsh",
+      tags: []
+    }
+  },
+  {
+    id: "FyWJnhCa-ng",
+    url: "https://www.youtube.com/watch?v=FyWJnhCa-ng",
+    duration_seconds: 25,
+    upload_date: "2026-03-16",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "cym",
+        glottocode: "wels1247",
+        bcp47: "cy"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Welsh",
+      wikitongues_lineage: null,
+      autonym: "Cymraeg"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Hywel",
+        role: "native",
+        origin: "Wales"
+      }
+    ],
+    provenance: {
+      country_code: "GB",
+      country_name: "United Kingdom",
+      region: "Wales",
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Hywel speaking Welsh",
+      tags: []
+    }
+  },
+  {
+    id: "gB6x-lINQXM",
+    url: "https://www.youtube.com/watch?v=gB6x-lINQXM",
+    duration_seconds: 40,
+    upload_date: "2026-03-14",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "deu",
+        glottocode: "stan1295",
+        bcp47: "de-US"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Texas German",
+      wikitongues_lineage: null,
+      autonym: "Texasdeutsch"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Vernell",
+        role: "native",
+        origin: "Texas"
+      }
+    ],
+    provenance: {
+      country_code: "US",
+      country_name: "United States",
+      region: "Texas",
+      city: "Fredericksburg",
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Vernell speaks Texas German",
+      tags: []
+    }
+  },
+  {
+    id: "3L_akwiAvPI",
+    url: "https://www.youtube.com/watch?v=3L_akwiAvPI",
+    duration_seconds: 29,
+    upload_date: "2026-03-12",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "deu",
+        glottocode: "stan1295",
+        bcp47: "de-US"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Texas German",
+      wikitongues_lineage: null,
+      autonym: "Texasdeutsch"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Vernell",
+        role: "native",
+        origin: "Texas"
+      }
+    ],
+    provenance: {
+      country_code: "US",
+      country_name: "United States",
+      region: "Texas",
+      city: "Fredericksburg",
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Vernell speaks Texas German",
+      tags: []
+    }
+  },
+  {
+    id: "ACkloeSPfFA",
+    url: "https://www.youtube.com/watch?v=ACkloeSPfFA",
+    duration_seconds: 73,
+    upload_date: "2026-03-11",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "frc",
+        glottocode: "caju1236",
+        bcp47: "frc"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Louisiana French",
+      wikitongues_lineage: null,
+      autonym: "Fran\xE7ais louisianais"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Sam Craft",
+        role: "heritage",
+        origin: "Louisiana"
+      }
+    ],
+    provenance: {
+      country_code: "US",
+      country_name: "United States",
+      region: "Louisiana",
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Sam speaking Louisiana French (Part 3)",
+      tags: []
+    }
+  },
+  {
+    id: "TwwLkiLNkBg",
+    url: "https://www.youtube.com/watch?v=TwwLkiLNkBg",
+    duration_seconds: 58,
+    upload_date: "2026-03-10",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "frc",
+        glottocode: "caju1236",
+        bcp47: "frc"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Louisiana French",
+      wikitongues_lineage: null,
+      autonym: "Fran\xE7ais louisianais"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Sam Craft",
+        role: "heritage",
+        origin: "Louisiana"
+      }
+    ],
+    provenance: {
+      country_code: "US",
+      country_name: "United States",
+      region: "Louisiana",
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Sam speaking Louisiana French (Part 2)",
+      tags: [
+        "Wikitongues"
+      ]
+    }
+  },
+  {
+    id: "su4ynRj560Q",
+    url: "https://www.youtube.com/watch?v=su4ynRj560Q",
+    duration_seconds: 87,
+    upload_date: "2026-03-10",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "frc",
+        glottocode: "caju1236",
+        bcp47: "frc"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Louisiana French",
+      wikitongues_lineage: null,
+      autonym: "Fran\xE7ais louisianais"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Sam Craft",
+        role: "heritage",
+        origin: "Louisiana"
+      }
+    ],
+    provenance: {
+      country_code: "US",
+      country_name: "United States",
+      region: "Louisiana",
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Sam speaking Louisiana French (Part 1)",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages"
+      ]
+    }
+  },
+  {
+    id: "JFC9iGDrDTA",
+    url: "https://www.youtube.com/watch?v=JFC9iGDrDTA",
+    duration_seconds: 103,
+    upload_date: "2026-02-13",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "reading_or_song",
+    primary_language: {
+      standards: {
+        iso639_3: "sco",
+        glottocode: "shet1241",
+        bcp47: "sco"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Shetlandic",
+      wikitongues_lineage: null,
+      autonym: "Shaetlan"
+    },
+    additional_languages: [
+      {
+        standards: {
+          iso639_3: "eng",
+          glottocode: "stan1293",
+          bcp47: "en"
+        },
+        speaker_claim: null,
+        wikitongues_classification: "English",
+        wikitongues_lineage: null,
+        autonym: "English"
+      }
+    ],
+    speakers: [
+      {
+        name: "Christine De Luca",
+        role: "native",
+        origin: "Shetland"
+      }
+    ],
+    provenance: {
+      country_code: "GB",
+      country_name: "United Kingdom",
+      region: "Scotland",
+      city: "Edinburgh",
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Christine reads a poem in English and Shetlandic",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages"
+      ]
+    }
+  },
+  {
+    id: "p6QqAeuzuGY",
+    url: "https://www.youtube.com/watch?v=p6QqAeuzuGY",
+    duration_seconds: 78,
+    upload_date: "2026-02-11",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "sco",
+        glottocode: "shet1241",
+        bcp47: "sco"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Shetlandic",
+      wikitongues_lineage: null,
+      autonym: "Shaetlan"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Christine De Luca",
+        role: "native",
+        origin: "Shetland"
+      }
+    ],
+    provenance: {
+      country_code: "GB",
+      country_name: "United Kingdom",
+      region: "Scotland",
+      city: "Edinburgh",
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Christine was taught English. Writing led her back to her mother tongue.",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages"
+      ]
+    }
+  },
+  {
+    id: "FDYuB3KfViY",
+    url: "https://www.youtube.com/watch?v=FDYuB3KfViY",
+    duration_seconds: 65,
+    upload_date: "2026-02-09",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "sco",
+        glottocode: "shet1241",
+        bcp47: "sco"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Shetlandic",
+      wikitongues_lineage: null,
+      autonym: "Shaetlan"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Christine De Luca",
+        role: "native",
+        origin: "Shetland"
+      }
+    ],
+    provenance: {
+      country_code: "GB",
+      country_name: "United Kingdom",
+      region: "Scotland",
+      city: "Edinburgh",
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: `Edinburgh's Poet Laureate (2014-2017) challenges the idea that Shetlandic is "less than"`,
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages"
+      ]
+    }
+  },
+  {
+    id: "bicQILtQ8is",
+    url: "https://www.youtube.com/watch?v=bicQILtQ8is",
+    duration_seconds: 82,
+    upload_date: "2026-02-06",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "reading_or_song",
+    primary_language: {
+      standards: {
+        iso639_3: "gul",
+        glottocode: "gull1241",
+        bcp47: "gul-US"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Gullah",
+      wikitongues_lineage: null,
+      autonym: "Gullah"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Caroline",
+        role: "native",
+        origin: "South Carolina"
+      }
+    ],
+    provenance: {
+      country_code: "US",
+      country_name: "United States",
+      region: "South Carolina",
+      city: "Charleston",
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "The New Testament, read in Gullah",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages"
+      ]
+    }
+  },
+  {
+    id: "ig9niswAGCk",
+    url: "https://www.youtube.com/watch?v=ig9niswAGCk",
+    duration_seconds: 75,
+    upload_date: "2026-02-04",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "gul",
+        glottocode: "gull1241",
+        bcp47: "gul-US"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Gullah",
+      wikitongues_lineage: null,
+      autonym: "Gullah"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Caroline",
+        role: "native",
+        origin: "South Carolina"
+      }
+    ],
+    provenance: {
+      country_code: "US",
+      country_name: "United States",
+      region: "South Carolina",
+      city: "Charleston",
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Caroline used to be corrected for speaking Gullah, now she teaches it.",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages"
+      ]
+    }
+  },
+  {
+    id: "2Y5aRpz4k30",
+    url: "https://www.youtube.com/watch?v=2Y5aRpz4k30",
+    duration_seconds: 54,
+    upload_date: "2026-02-02",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "language_lesson",
+    primary_language: {
+      standards: {
+        iso639_3: "gul",
+        glottocode: "gull1241",
+        bcp47: "gul-US"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Gullah",
+      wikitongues_lineage: null,
+      autonym: "Gullah"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Caroline",
+        role: "native",
+        origin: "South Carolina"
+      }
+    ],
+    provenance: {
+      country_code: "US",
+      country_name: "United States",
+      region: "South Carolina",
+      city: "Charleston",
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Gullah: a living language from the U.S. South",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages"
+      ]
+    }
+  },
+  {
+    id: "nGeLETLlFXA",
+    url: "https://www.youtube.com/watch?v=nGeLETLlFXA",
+    duration_seconds: 100,
+    upload_date: "2025-12-10",
+    license: "CC-BY-NC-4.0",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "quh",
+        glottocode: "sout2991",
+        bcp47: "quh"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "South Bolivian Quechua",
+      wikitongues_lineage: "Quechuan",
+      autonym: "Qhichwa"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Nicaela",
+        role: "native",
+        origin: "Bolivia"
+      }
+    ],
+    provenance: {
+      country_code: "BO",
+      country_name: "Bolivia",
+      region: null,
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Nicaela speaking South Bolivian Quechua | Quechuan | Language of the Americas | Wikitongues",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages"
+      ]
+    }
+  },
+  {
+    id: "0jzN1GREMv0",
+    url: "https://www.youtube.com/watch?v=0jzN1GREMv0",
+    duration_seconds: 48,
+    upload_date: "2025-02-21",
+    license: "CC-BY-NC-4.0",
+    content_type: "language_lesson",
+    primary_language: {
+      standards: {
+        iso639_3: "gle",
+        glottocode: "iris1253",
+        bcp47: "ga"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Irish",
+      wikitongues_lineage: null,
+      autonym: "Gaeilge"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Ahmad",
+        role: "second_language",
+        origin: "Syria"
+      }
+    ],
+    provenance: {
+      country_code: null,
+      country_name: null,
+      region: null,
+      city: null,
+      recorded_by: "Wikitongues",
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Learn Irish with Ahmad | International Mother Language Day | Wikitongues",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages"
+      ]
+    }
+  },
+  {
+    id: "7Se_-1IX2j4",
+    url: "https://www.youtube.com/watch?v=7Se_-1IX2j4",
+    duration_seconds: 71,
+    upload_date: "2025-02-21",
+    license: "CC-BY-NC-4.0",
+    content_type: "language_lesson",
+    primary_language: {
+      standards: {
+        iso639_3: "yor",
+        glottocode: "yoru1245",
+        bcp47: "yo"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Yoruba",
+      wikitongues_lineage: null,
+      autonym: "\xC8d\xE8 Yor\xF9b\xE1"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Hanifa",
+        role: "native",
+        origin: "Nigeria"
+      }
+    ],
+    provenance: {
+      country_code: "NG",
+      country_name: "Nigeria",
+      region: null,
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Yoruba with Hanifa | International Mother Language Day | Wikitongues",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages"
+      ]
+    }
+  },
+  {
+    id: "rHVxTCtfqrw",
+    url: "https://www.youtube.com/watch?v=rHVxTCtfqrw",
+    duration_seconds: 96,
+    upload_date: "2025-02-21",
+    license: "CC-BY-NC-4.0",
+    content_type: "language_lesson",
+    primary_language: {
+      standards: {
+        iso639_3: "ibo",
+        glottocode: "nucl1417",
+        bcp47: "ig"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Central Igbo",
+      wikitongues_lineage: null,
+      autonym: "As\u1EE5s\u1EE5 Igbo"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Oluchukwu",
+        role: "native",
+        origin: "Nigeria"
+      }
+    ],
+    provenance: {
+      country_code: "NG",
+      country_name: "Nigeria",
+      region: null,
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Learn Central Igbo with Oluchukwu | International Mother Language Day | Wikitongues",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages"
+      ]
+    }
+  },
+  {
+    id: "6i0neXrFZfg",
+    url: "https://www.youtube.com/watch?v=6i0neXrFZfg",
+    duration_seconds: 57,
+    upload_date: "2025-02-21",
+    license: "CC-BY-SA-4.0",
+    content_type: "language_lesson",
+    primary_language: {
+      standards: {
+        iso639_3: "kcg",
+        glottocode: "tyap1238",
+        bcp47: "kcg"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Tyap",
+      wikitongues_lineage: null,
+      autonym: "Tyap"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Kambai",
+        role: "native",
+        origin: "Nigeria"
+      }
+    ],
+    provenance: {
+      country_code: "NG",
+      country_name: "Nigeria",
+      region: null,
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Learn Tyap with Kambai | International Mother Language Day | Wikitongues",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages"
+      ]
+    }
+  },
+  {
+    id: "B1Yu1EuUyUQ",
+    url: "https://www.youtube.com/watch?v=B1Yu1EuUyUQ",
+    duration_seconds: 42,
+    upload_date: "2025-02-21",
+    license: "CC-BY-SA-4.0",
+    content_type: "language_lesson",
+    primary_language: {
+      standards: {
+        iso639_3: "ind",
+        glottocode: "stin1234",
+        bcp47: "id"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Bahasa Indonesia",
+      wikitongues_lineage: null,
+      autonym: "Bahasa Indonesia"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Neysa",
+        role: "native",
+        origin: "Indonesia"
+      }
+    ],
+    provenance: {
+      country_code: "ID",
+      country_name: "Indonesia",
+      region: null,
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Learn Bahasa Indonesia with Neysa | International Mother Language Day | Wikitongues",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages"
+      ]
+    }
+  },
+  {
+    id: "5WKzRb2_Tt0",
+    url: "https://www.youtube.com/watch?v=5WKzRb2_Tt0",
+    duration_seconds: 60,
+    upload_date: "2025-02-21",
+    license: "CC-BY-SA-4.0",
+    content_type: "language_lesson",
+    primary_language: {
+      standards: {
+        iso639_3: "ibo",
+        glottocode: "ower1238",
+        bcp47: "ig"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Owerri Igbo",
+      wikitongues_lineage: null,
+      autonym: "As\u1EE5s\u1EE5 Igbo"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Lucy",
+        role: "native",
+        origin: "Nigeria"
+      }
+    ],
+    provenance: {
+      country_code: "NG",
+      country_name: "Nigeria",
+      region: null,
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Learn Owerri Igbo with Lucy | International Mother Language Day | Wikitongues",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages"
+      ]
+    }
+  },
+  {
+    id: "W4CM82mvMv8",
+    url: "https://www.youtube.com/watch?v=W4CM82mvMv8",
+    duration_seconds: 52,
+    upload_date: "2025-02-21",
+    license: "CC-BY-NC-4.0",
+    content_type: "language_lesson",
+    primary_language: {
+      standards: {
+        iso639_3: "fin",
+        glottocode: "finn1318",
+        bcp47: "fi"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Finnish",
+      wikitongues_lineage: null,
+      autonym: "Suomi"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Kristen",
+        role: "second_language",
+        origin: null
+      }
+    ],
+    provenance: {
+      country_code: null,
+      country_name: null,
+      region: null,
+      city: null,
+      recorded_by: "Wikitongues",
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Learn Finnish with Kristen | International Mother Language Day | Wikitongues",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages"
+      ]
+    }
+  },
+  {
+    id: "l3F1h1XyXEY",
+    url: "https://www.youtube.com/watch?v=l3F1h1XyXEY",
+    duration_seconds: 60,
+    upload_date: "2025-02-21",
+    license: "CC-BY-SA-4.0",
+    content_type: "language_lesson",
+    primary_language: {
+      standards: {
+        iso639_3: "hau",
+        glottocode: "haus1257",
+        bcp47: "ha"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Hausa",
+      wikitongues_lineage: null,
+      autonym: "Harshen Hausa"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Sadam",
+        role: "native",
+        origin: "Nigeria"
+      }
+    ],
+    provenance: {
+      country_code: "NG",
+      country_name: "Nigeria",
+      region: null,
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Learn Hausa with Sadam | International Mother Language Day | Wikitongues",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages"
+      ]
+    }
+  },
+  {
+    id: "yasjvWgLkqU",
+    url: "https://www.youtube.com/watch?v=yasjvWgLkqU",
+    duration_seconds: 60,
+    upload_date: "2024-11-19",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "meta",
+    primary_language: {
+      standards: {
+        iso639_3: "eng",
+        glottocode: "stan1293",
+        bcp47: "en"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "English",
+      wikitongues_lineage: null,
+      autonym: "English"
+    },
+    additional_languages: [],
+    speakers: [],
+    provenance: {
+      country_code: null,
+      country_name: null,
+      region: null,
+      city: null,
+      recorded_by: "Wikitongues",
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Discover N\xFCshu: The secret script created by women, for women!",
+      tags: []
+    }
+  },
+  {
+    id: "5gZ7muELrXw",
+    url: "https://www.youtube.com/watch?v=5gZ7muELrXw",
+    duration_seconds: 1697,
+    upload_date: "2022-11-05",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "meta",
+    primary_language: {
+      standards: {
+        iso639_3: "eng",
+        glottocode: "stan1293",
+        bcp47: "en"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "English",
+      wikitongues_lineage: null,
+      autonym: "English"
+    },
+    additional_languages: [],
+    speakers: [],
+    provenance: {
+      country_code: null,
+      country_name: null,
+      region: null,
+      city: null,
+      recorded_by: "Wikitongues",
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Language Revitalization Accelerator Q&A",
+      tags: []
+    }
+  },
+  {
+    id: "faidns4sK3Q",
+    url: "https://www.youtube.com/watch?v=faidns4sK3Q",
+    duration_seconds: 1611,
+    upload_date: "2022-04-14",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "meta",
+    primary_language: {
+      standards: {
+        iso639_3: "eng",
+        glottocode: "stan1293",
+        bcp47: "en"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "English",
+      wikitongues_lineage: null,
+      autonym: "English"
+    },
+    additional_languages: [],
+    speakers: [],
+    provenance: {
+      country_code: null,
+      country_name: null,
+      region: null,
+      city: null,
+      recorded_by: "Wikitongues",
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Announcing the 2022 grant recipients",
+      tags: []
+    }
+  },
+  {
+    id: "6UctiHFsUrE",
+    url: "https://www.youtube.com/watch?v=6UctiHFsUrE",
+    duration_seconds: 3921,
+    upload_date: "2020-11-11",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "meta",
+    primary_language: {
+      standards: {
+        iso639_3: "eng",
+        glottocode: "stan1293",
+        bcp47: "en"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "English",
+      wikitongues_lineage: null,
+      autonym: "English"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Cordel Green",
+        role: "fellow_activist",
+        origin: "Jamaica"
+      },
+      {
+        name: "Daniel Sikazwe",
+        role: "fellow_activist",
+        origin: "Zambia"
+      }
+    ],
+    provenance: {
+      country_code: null,
+      country_name: null,
+      region: null,
+      city: null,
+      recorded_by: "Wikitongues",
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Radio in Language Diversity | Conversations from Jamaica and Zambia | Wikitongues",
+      tags: [
+        "language diversity in media",
+        "radio",
+        "language diversity",
+        "linguistic diversity",
+        "language toolkit",
+        "radio for language",
+        "multilingual radio",
+        "multilingual radio stations",
+        "jamaica",
+        "jamaican radio",
+        "jamaican languages",
+        "zambia",
+        "zambian radio",
+        "zambian languages",
+        "language education",
+        "jamaicain patois lessons",
+        "radio jam upchurch",
+        "speaking on the radio",
+        "all walks of life",
+        "sociolinguistics",
+        "jamaican music",
+        "jamaican patois lessons",
+        "indigenous langauges",
+        "how to promote language",
+        "gullah",
+        "wikitongues"
+      ]
+    }
+  },
+  {
+    id: "iDPp2yORw94",
+    url: "https://www.youtube.com/watch?v=iDPp2yORw94",
+    duration_seconds: 4151,
+    upload_date: "2020-10-11",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "meta",
+    primary_language: {
+      standards: {
+        iso639_3: "eng",
+        glottocode: "stan1293",
+        bcp47: "en"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "English",
+      wikitongues_lineage: null,
+      autonym: "English"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Ryan Andrew Milov-Cordoba (Call Me C\xF3rdoba)",
+        role: "fellow_activist",
+        origin: "United States"
+      }
+    ],
+    provenance: {
+      country_code: null,
+      country_name: null,
+      region: null,
+      city: null,
+      recorded_by: "Wikitongues",
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Music in Language Diversity | Conversation and music with Call Me C\xF3rdoba | Wikitongues",
+      tags: [
+        "music in language diversity",
+        "language diversity in education",
+        "linguistic diversity",
+        "indigenous peoples day",
+        "call me cordoba",
+        "arabic contemporary music",
+        "Save my language",
+        "global issues",
+        "language learning",
+        "language revitalization",
+        "language preservation",
+        "endangered languages",
+        "endangered languages project",
+        "Living Tongues Institute",
+        "Wikitongues",
+        "David Harrison",
+        "Mother tongue",
+        "Linguistics",
+        "Indigenous languages",
+        "arabic language learning",
+        "\u0627\u0644\u062A\u0646\u0648\u0639 \u0627\u0644\u0644\u063A\u0648\u064A",
+        "bahau"
+      ]
+    }
+  },
+  {
+    id: "HaYhISDoVyM",
+    url: "https://www.youtube.com/watch?v=HaYhISDoVyM",
+    duration_seconds: 1576,
+    upload_date: "2020-09-28",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "meta",
+    primary_language: {
+      standards: {
+        iso639_3: "eng",
+        glottocode: "stan1293",
+        bcp47: "en"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "English",
+      wikitongues_lineage: null,
+      autonym: "English"
+    },
+    additional_languages: [],
+    speakers: [],
+    provenance: {
+      country_code: null,
+      country_name: null,
+      region: null,
+      city: null,
+      recorded_by: "Wikitongues",
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Wikitongues Updates, September 2020",
+      tags: []
+    }
+  },
+  {
+    id: "2UmHjJvPGec",
+    url: "https://www.youtube.com/watch?v=2UmHjJvPGec",
+    duration_seconds: 10161,
+    upload_date: "2020-08-02",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "meta",
+    primary_language: {
+      standards: {
+        iso639_3: "eng",
+        glottocode: "stan1293",
+        bcp47: "en"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "English",
+      wikitongues_lineage: null,
+      autonym: "English"
+    },
+    additional_languages: [],
+    speakers: [],
+    provenance: {
+      country_code: null,
+      country_name: null,
+      region: null,
+      city: null,
+      recorded_by: "Wikitongues",
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Wikitongues Virtual Birthday Party",
+      tags: [
+        "birthday party",
+        "birthday party coronavirus",
+        "zoom university",
+        "virtual birthday party",
+        "Wikitongues",
+        "wikitongues birthday party",
+        "virtual birthday party ideas",
+        "wikitongues virtual birthday party",
+        "international coronavirus response",
+        "how to have fun in pandemic",
+        "everything is online now",
+        "annual language celebration",
+        "good things that came from coronavirus",
+        "how to have fun when the world is ending",
+        "New York",
+        "eastern europe",
+        "jakarta",
+        "sociolinguistics",
+        "language variation and change",
+        "gullah"
+      ]
+    }
+  },
+  {
+    id: "1UutaDfzaqc",
+    url: "https://www.youtube.com/watch?v=1UutaDfzaqc",
+    duration_seconds: 3906,
+    upload_date: "2020-06-26",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "meta",
+    primary_language: {
+      standards: {
+        iso639_3: "eng",
+        glottocode: "stan1293",
+        bcp47: "en"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "English",
+      wikitongues_lineage: null,
+      autonym: "English"
+    },
+    additional_languages: [],
+    speakers: [],
+    provenance: {
+      country_code: null,
+      country_name: null,
+      region: null,
+      city: null,
+      recorded_by: "Wikitongues",
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "How do I save my language? Introducing the Language Sustainability Toolkit",
+      tags: [
+        "Save my language",
+        "My language",
+        "global issues",
+        "Language",
+        "language learning",
+        "language learning apps",
+        "language revitalization canada",
+        "hawaiian language revitalization",
+        "native american language revitalization",
+        "ojibwe language revitalization",
+        "language revitalization",
+        "language preservation",
+        "endangered languages",
+        "endangered languages project",
+        "Living Tongues Institute",
+        "Wikitongues",
+        "David Harrison",
+        "Anna Luisa Daigneault",
+        "Mother tongue",
+        "Linguistics",
+        "Indigenous languages",
+        "Sustainability"
+      ]
+    }
+  },
+  {
+    id: "pEU_EC7OcdE",
+    url: "https://www.youtube.com/watch?v=pEU_EC7OcdE",
+    duration_seconds: 3878,
+    upload_date: "2020-06-24",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "meta",
+    primary_language: {
+      standards: {
+        iso639_3: "eng",
+        glottocode: "stan1293",
+        bcp47: "en"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "English",
+      wikitongues_lineage: null,
+      autonym: "English"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Tim Brooks",
+        role: "fellow_activist",
+        origin: "United States"
+      }
+    ],
+    provenance: {
+      country_code: null,
+      country_name: null,
+      region: null,
+      city: null,
+      recorded_by: "Wikitongues",
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "Tim Brooks from Endangered Alphabets | FIRST-EVER Wikitongues live-stream!",
+      tags: [
+        "Tim Brooks",
+        "Endangered Alphabets",
+        "Alphabets",
+        "alphabet in spanish",
+        "writing systems",
+        "Writing systems of the world",
+        "writing systems in europe",
+        "Wikitongues",
+        "sociolinguistics",
+        "balinese language",
+        "languages of the philippines",
+        "Easter Island",
+        "Rapa Nui",
+        "rongorongo script",
+        "rongorongo language",
+        "semitic abjads alphabet",
+        "Hebrew alphabet",
+        "samaritans",
+        "israel and palestine",
+        "israel and philippines relationship",
+        "Louisiana French",
+        "english conversation",
+        "british accent vs american accent",
+        "Mongolian"
+      ]
+    }
+  },
+  {
+    id: "pd0YZ9o22QI",
+    url: "https://www.youtube.com/watch?v=pd0YZ9o22QI",
+    duration_seconds: 33,
+    upload_date: "2020-02-27",
+    license: "CC-BY-NC-4.0",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "kls",
+        glottocode: "kala1372",
+        bcp47: "kls"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Kalasha",
+      wikitongues_lineage: null,
+      autonym: "Kalashamondr"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Najib",
+        role: "native",
+        origin: "Pakistan"
+      }
+    ],
+    provenance: {
+      country_code: "PK",
+      country_name: "Pakistan",
+      region: "Khyber Pakhtunkhwa",
+      city: "Chitral",
+      recorded_by: "Rizwan Hussain",
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "WIKITONGUES: Najib speaking Kalasha",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages",
+        "Kalasha language",
+        "Kalasha-mun",
+        "Pakistan",
+        "Pakistani people",
+        "South Asian people",
+        "Dardic languages"
+      ]
+    }
+  },
+  {
+    id: "gUWkNeLUZsg",
+    url: "https://www.youtube.com/watch?v=gUWkNeLUZsg",
+    duration_seconds: 30,
+    upload_date: "2020-02-16",
+    license: "CC-BY-SA-4.0",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "mlt",
+        glottocode: "malt1254",
+        bcp47: "mt"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Maltese",
+      wikitongues_lineage: null,
+      autonym: "Malti"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Keith",
+        role: "native",
+        origin: "Malta"
+      }
+    ],
+    provenance: {
+      country_code: "MT",
+      country_name: "Malta",
+      region: null,
+      city: "Valletta",
+      recorded_by: "Irena Dahl",
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "WIKITONGUES: Keith speaking Maltese",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages",
+        "Maltese",
+        "Maltese language",
+        "Maltese people",
+        "Malta",
+        "Semitic languages",
+        "Arabic dialects",
+        "European languages"
+      ]
+    }
+  },
+  {
+    id: "gE11A0fmkR0",
+    url: "https://www.youtube.com/watch?v=gE11A0fmkR0",
+    duration_seconds: 21,
+    upload_date: "2019-08-30",
+    license: "CC-BY-SA-4.0",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "alt",
+        glottocode: "sout2694",
+        bcp47: "alt"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Altai",
+      wikitongues_lineage: null,
+      autonym: "\u0410\u043B\u0442\u0430\u0439 \u0442\u0438\u043B"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Dmitry",
+        role: "native",
+        origin: "Russia"
+      }
+    ],
+    provenance: {
+      country_code: "RU",
+      country_name: "Russia",
+      region: "Altai Republic",
+      city: null,
+      recorded_by: "Archyn Termishev",
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "WIKITONGUES: Dmitry speaking Altai",
+      tags: [
+        "Wikitongues",
+        "IYIL2019",
+        "2019 International Year of Indigenous Languages"
+      ]
+    }
+  },
+  {
+    id: "TotFWZLrQAY",
+    url: "https://www.youtube.com/watch?v=TotFWZLrQAY",
+    duration_seconds: 55,
+    upload_date: "2014-08-07",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "epo",
+        glottocode: "espe1235",
+        bcp47: "eo"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Esperanto",
+      wikitongues_lineage: null,
+      autonym: "Esperanto"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Hilliard (Hiljardo)",
+        role: "second_language",
+        origin: "United States"
+      }
+    ],
+    provenance: {
+      country_code: "US",
+      country_name: "United States",
+      region: "Connecticut",
+      city: null,
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: "Saluton. Mia nomo estas Hiljardo, en Esperanto, kaj mi parolis Esperanton dum du monatoj. Mi havas dek kvar de jaroj. Mi lernis Esperanton por lerni pri aliaj kulturoj kaj aliaj popoloj. Esperanto helpas min multe. Mi ofte konversacias kun homoj de la mondo. Mi lo\u011Das en Konektikuto en Usono. Mi amas lerni pri lingvoj kaj historio. Mi anka\u016D parolas anglan lingvon, francan lingvon, rusan lingvon, nederlandan lingvon, kaj afrikanson. Mi amas lerni novajn lingvojn , \u0109ar \u011Di estas sana, kaj \u011Di estas bona ekzerco por la cerbo. Dankon por rigardi.",
+      english_translation: 'Hi. Mi name is "Hiljardo" [Hilliard], in Esperanto, and I have spoken Esperanto for two months. I am fourteen years old. I learned Esperanto to learn about foreign cultures and foreign peoples. Esperanto helps me a lot. I often converse with people from the world. I live in Connecticut in the United States of America. I like to learn about languages and history. I also speak English, French, Russian, Dutch, and Afrikaans. I like to learn new languages, because it is healthy, and it is a good exercise for the brain. Thanks for watching.'
+    },
+    raw_metadata: {
+      title: "WIKITONGUES: Hiljiardo speaking Esperanto",
+      tags: [
+        "Wikitongues",
+        "Esperanto Language (Human Language)"
+      ]
+    }
+  },
+  {
+    id: "--DnU743JLU",
+    url: "https://www.youtube.com/watch?v=--DnU743JLU",
+    duration_seconds: 40,
+    upload_date: "2013-02-18",
+    license: "ALL_RIGHTS_RESERVED",
+    content_type: "oral_history",
+    primary_language: {
+      standards: {
+        iso639_3: "nan",
+        glottocode: "minn1241",
+        bcp47: "nan-TW"
+      },
+      speaker_claim: null,
+      wikitongues_classification: "Taiwanese Min Nan",
+      wikitongues_lineage: null,
+      autonym: "\u81FA\u7063\u8A71"
+    },
+    additional_languages: [],
+    speakers: [
+      {
+        name: "Liang Cheng Chia (Sunny)",
+        role: "native",
+        origin: "Taiwan"
+      }
+    ],
+    provenance: {
+      country_code: "TW",
+      country_name: "Taiwan",
+      region: null,
+      city: "Kaohsiung",
+      recorded_by: null,
+      recording_date: null
+    },
+    transcription: {
+      has_subtitles: false,
+      available_subtitles: [],
+      native_text: null,
+      english_translation: null
+    },
+    raw_metadata: {
+      title: "WIKITONGUES: Sunny speaking Taiwanese Min Nan",
+      tags: [
+        "Taiwanese",
+        "Hokkien",
+        "Taiwan",
+        "nation",
+        "country",
+        "Vietnam",
+        "USA",
+        "languages of Taiwan",
+        "Min Nan",
+        "Chinese"
+      ]
+    }
+  }
+];
+
+// src/dataset.ts
+var dataset = wikitongues_normalized_default;
+
+// src/client.ts
+var WikitonguesDB = class _WikitonguesDB {
+  _videos = [];
+  hydrator;
+  resolver;
+  index;
+  searchEngine;
+  /**
+   * Loads and hydrates every record. Throws HydrationError on the first record whose
+   * standards do not resolve against the reference tables (see CLASSIFICATION_RULES.md §3.1.1).
+   */
+  constructor(options) {
+    this.hydrator = new ReferenceHydrator(options?.reference);
+    this.resolver = new LanguageResolver();
+    const rawRecords = options?.data || options?.records || dataset;
+    for (const record of rawRecords) {
+      const video = record instanceof Video ? record : Video.fromDict(record, this.hydrator);
+      this._videos.push(video);
+      for (const lang of video.allLanguages) {
+        this.resolver.registerDatasetLanguage(lang);
+      }
+    }
+    this.index = new DatasetIndex(this._videos);
+    this.searchEngine = new SearchEngine(this._videos, this.resolver);
+  }
+  get length() {
+    return this._videos.length;
+  }
+  // -------------------------------------------------------------------------
+  // Static Factory Methods
+  // -------------------------------------------------------------------------
+  /**
+   * Create a WikitonguesDB instance directly from an in-memory list of dicts or Videos.
+   */
+  static fromRecords(records) {
+    return new _WikitonguesDB({ records });
+  }
+  /**
+   * Create a WikitonguesDB instance from a JSON string or JSON array.
+   */
+  static fromJSON(json) {
+    const data = typeof json === "string" ? JSON.parse(json) : json;
+    return new _WikitonguesDB({ data });
+  }
+  /**
+   * Create a WikitonguesDB instance from a JSON or JSONL file path (Node.js environments).
+   */
+  static fromFile(filepath) {
+    try {
+      const fs = __require("fs");
+      const content = fs.readFileSync(filepath, "utf-8");
+      const trimmed = content.trim();
+      let records = [];
+      if (trimmed.startsWith("[")) {
+        records = JSON.parse(trimmed);
+      } else {
+        records = trimmed.split("\n").filter((line) => line.trim()).map((line) => JSON.parse(line));
+      }
+      return new _WikitonguesDB({ data: records });
+    } catch (err) {
+      throw new Error(`Failed to load dataset from file '${filepath}': ${err.message}`);
+    }
+  }
+  // -------------------------------------------------------------------------
+  // Direct Lookups (O(1))
+  // -------------------------------------------------------------------------
+  /**
+   * Lookup a video by its YouTube ID in O(1) time.
+   */
+  get(videoId) {
+    return this.index.byId.get(videoId.trim()) || null;
+  }
+  /**
+   * Lookup videos matching a 3-letter SIL ISO 639-3 code in O(1) time.
+   */
+  getByIso(code) {
+    const list = this.index.byIso.get(code.trim().toLowerCase());
+    return new VideoCollection(list || []);
+  }
+  /**
+   * Lookup videos matching a BCP 47 language tag in O(1) time.
+   */
+  getByBcp47(tag) {
+    const list = this.index.byBcp47.get(tag.trim().toLowerCase());
+    return new VideoCollection(list || []);
+  }
+  /**
+   * Lookup videos matching a Glottolog code in O(1) time. A language node also matches videos
+   * classified under one of its dialect nodes (e.g. `port1283` returns `braz1246` recordings).
+   */
+  getByGlottocode(glottocode) {
+    const list = this.index.byGlottocode.get(glottocode.trim().toLowerCase());
+    return new VideoCollection(list || []);
+  }
+  /**
+   * Lookup videos matching an ISO 3166-1 alpha-2 code or country name in O(1) time.
+   */
+  getByCountry(countryCodeOrName) {
+    const c = countryCodeOrName.trim();
+    if (c.length === 2 && this.index.byCountryCode.has(c.toUpperCase())) {
+      return new VideoCollection(this.index.byCountryCode.get(c.toUpperCase()) || []);
+    }
+    const normName = normalizeText(c);
+    if (this.index.byCountryName.has(normName)) {
+      return new VideoCollection(this.index.byCountryName.get(normName) || []);
+    }
+    return this.query().country(c).all();
+  }
+  /**
+   * Lookup videos for a given speaker name.
+   */
+  getBySpeaker(name) {
+    const normName = normalizeText(name);
+    if (this.index.bySpeakerName.has(normName)) {
+      return new VideoCollection(this.index.bySpeakerName.get(normName) || []);
+    }
+    return this.query().speaker({ name }).all();
+  }
+  // -------------------------------------------------------------------------
+  // Smart Lookups & Shortcut Queries
+  // -------------------------------------------------------------------------
+  /**
+   * Intelligent language search: resolves natural names, multilingual aliases ('russe'),
+   * ISO 639-3 ('rus'), BCP 47 ('ru'), Glottocodes, autonyms ('Русский'), Wikitongues
+   * classifications, Glottolog / ISO names and speaker claims.
+   */
+  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.