ochre-sdk 1.3.5 → 1.3.7

package/dist/languages.d.mts

@@ -0,0 +1,684 @@
+//#region src/languages.d.ts
+/**
+ * What an OCHRE language code tags
+ *
+ * - `language`: text in a natural language
+ * - `transliteration`: a language written in another script (`ota-xlt`)
+ * - `vocabulary`: codes or notation that are not a language (`rel`)
+ * - `non-linguistic`: `zxx`, which OCHRE publishes for aliases and for
+ *   labels that are not in any language
+ */
+export type OchreLanguageKind = "language" | "transliteration" | "vocabulary" | "non-linguistic";
+/**
+ * The standard an OCHRE language code is taken from
+ *
+ * - `iso639-2b`: the bibliographic ISO 639-2 code, which is OCHRE's rule
+ * - `iso639-2t`: the terminology ISO 639-2 code, which OCHRE uses for Dutch
+ * - `iso639-3`: the ISO 639-3 identifier, for languages ISO 639-2 has no
+ *   code for
+ * - `bcp47`: a BCP 47 tag (`zh-Latn-pinyin`)
+ * - `ochre`: a code of OCHRE's own, which may collide with an unrelated
+ *   ISO 639-3 language (`lmc` is OCHRE's Late Middle Chinese, ISO's Limilngan)
+ */
+export type OchreLanguageStandard = "iso639-2b" | "iso639-2t" | "iso639-3" | "bcp47" | "ochre";
+/**
+ * A language code OCHRE publishes, with the standard codes it stands for
+ */
+export type OchreLanguage = {
+  /**
+   * The English name of the language
+   */
+  name: string;
+  /**
+   * What the code tags
+   */
+  kind: OchreLanguageKind;
+  /**
+   * The standard the OCHRE code itself is taken from
+   */
+  standard: OchreLanguageStandard;
+  /**
+   * The ISO 639-3 identifier of the language, or null when the code tags
+   * no language, or a language ISO 639-3 does not code
+   */
+  iso639_3: string | null;
+  /**
+   * The BCP 47 tag to hand to `Intl` and HTML `lang` attributes, or null
+   * when no standard tag fits
+   */
+  bcp47: string | null;
+};
+/**
+ * Every language code OCHRE publishes, mapped to the ISO 639-3 and BCP 47
+ * codes it stands for
+ *
+ * OCHRE publishes ISO 639-2 codes in their bibliographic (B) form, the
+ * Library of Congress MARC convention (`per`, not ISO 639-3 `fas`), with
+ * three exceptions: the ISO 639-3 identifier where ISO 639-2 has no code
+ * (`yua`), the terminology (T) form for Dutch (`nld`, not `dut`), and codes
+ * of its own for transliterations and vocabularies (`ota-xlt`, `rel`), some
+ * of which collide with unrelated ISO 639-3 languages. Key everything by
+ * these codes, because they are what OCHRE serves, and convert through this
+ * table when a standard code is needed.
+ */
+export declare const OCHRE_LANGUAGES: {
+  readonly akk: {
+    readonly name: "Akkadian";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "akk";
+    readonly bcp47: "akk";
+  };
+  readonly xno: {
+    readonly name: "Anglo-Norman";
+    readonly kind: "language";
+    readonly standard: "iso639-3";
+    readonly iso639_3: "xno";
+    readonly bcp47: "xno";
+  };
+  readonly ara: {
+    readonly name: "Arabic";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "ara";
+    readonly bcp47: "ar";
+  };
+  readonly arc: {
+    readonly name: "Aramaic";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "arc";
+    readonly bcp47: "arc";
+  };
+  readonly arm: {
+    readonly name: "Armenian";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "hye";
+    readonly bcp47: "hy";
+  };
+  readonly bel: {
+    readonly name: "Belarusian";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "bel";
+    readonly bcp47: "be";
+  };
+  readonly ben: {
+    readonly name: "Bengali";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "ben";
+    readonly bcp47: "bn";
+  };
+  readonly btc: {
+    readonly name: "Betacode";
+    readonly kind: "vocabulary";
+    readonly standard: "ochre";
+    readonly iso639_3: null;
+    readonly bcp47: "grc-Latn-x-betacode";
+  };
+  readonly bos: {
+    readonly name: "Bosnian";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "bos";
+    readonly bcp47: "bs";
+  };
+  readonly bul: {
+    readonly name: "Bulgarian";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "bul";
+    readonly bcp47: "bg";
+  };
+  readonly cnn: {
+    readonly name: "Canaanite";
+    readonly kind: "language";
+    readonly standard: "ochre";
+    readonly iso639_3: null;
+    readonly bcp47: null;
+  };
+  readonly yue: {
+    readonly name: "Cantonese";
+    readonly kind: "language";
+    readonly standard: "iso639-3";
+    readonly iso639_3: "yue";
+    readonly bcp47: "yue";
+  };
+  readonly cat: {
+    readonly name: "Catalan";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "cat";
+    readonly bcp47: "ca";
+  };
+  readonly chi: {
+    readonly name: "Chinese";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "zho";
+    readonly bcp47: "zh";
+  };
+  readonly chu: {
+    readonly name: "Church Slavonic";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "chu";
+    readonly bcp47: "cu";
+  };
+  readonly lzh: {
+    readonly name: "Classical Chinese";
+    readonly kind: "language";
+    readonly standard: "iso639-3";
+    readonly iso639_3: "lzh";
+    readonly bcp47: "lzh";
+  };
+  readonly cop: {
+    readonly name: "Coptic";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "cop";
+    readonly bcp47: "cop";
+  };
+  readonly hrv: {
+    readonly name: "Croatian";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "hrv";
+    readonly bcp47: "hr";
+  };
+  readonly cze: {
+    readonly name: "Czech";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "ces";
+    readonly bcp47: "cs";
+  };
+  readonly egyd: {
+    readonly name: "Demotic";
+    readonly kind: "language";
+    readonly standard: "ochre";
+    readonly iso639_3: "egy";
+    readonly bcp47: "egy-Egyd";
+  };
+  readonly nld: {
+    readonly name: "Dutch";
+    readonly kind: "language";
+    readonly standard: "iso639-2t";
+    readonly iso639_3: "nld";
+    readonly bcp47: "nl";
+  };
+  readonly dum: {
+    readonly name: "Dutch (Middle)";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "dum";
+    readonly bcp47: "dum";
+  };
+  readonly egy: {
+    readonly name: "Egyptian";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "egy";
+    readonly bcp47: "egy";
+  };
+  readonly elx: {
+    readonly name: "Elamite";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "elx";
+    readonly bcp47: "elx";
+  };
+  readonly eng: {
+    readonly name: "English";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "eng";
+    readonly bcp47: "en";
+  };
+  readonly enm: {
+    readonly name: "English (Middle)";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "enm";
+    readonly bcp47: "enm";
+  };
+  readonly ecy: {
+    readonly name: "Eteocypriot";
+    readonly kind: "language";
+    readonly standard: "iso639-3";
+    readonly iso639_3: "ecy";
+    readonly bcp47: "ecy";
+  };
+  readonly fre: {
+    readonly name: "French";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "fra";
+    readonly bcp47: "fr";
+  };
+  readonly frm: {
+    readonly name: "French (Middle)";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "frm";
+    readonly bcp47: "frm";
+  };
+  readonly fro: {
+    readonly name: "French (Old)";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "fro";
+    readonly bcp47: "fro";
+  };
+  readonly geo: {
+    readonly name: "Georgian";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "kat";
+    readonly bcp47: "ka";
+  };
+  readonly ger: {
+    readonly name: "German";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "deu";
+    readonly bcp47: "de";
+  };
+  readonly nds: {
+    readonly name: "German (Low)";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "nds";
+    readonly bcp47: "nds";
+  };
+  readonly gmh: {
+    readonly name: "German (Middle High)";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "gmh";
+    readonly bcp47: "gmh";
+  };
+  readonly goh: {
+    readonly name: "German (Old High)";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "goh";
+    readonly bcp47: "goh";
+  };
+  readonly grc: {
+    readonly name: "Greek";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "grc";
+    readonly bcp47: "grc";
+  };
+  readonly guj: {
+    readonly name: "Gujarati";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "guj";
+    readonly bcp47: "gu";
+  };
+  readonly "zh-Latn-pinyin": {
+    readonly name: "Hanyu Pinyin";
+    readonly kind: "transliteration";
+    readonly standard: "bcp47";
+    readonly iso639_3: "cmn";
+    readonly bcp47: "zh-Latn-pinyin";
+  };
+  readonly htt: {
+    readonly name: "Hattian";
+    readonly kind: "language";
+    readonly standard: "ochre";
+    readonly iso639_3: "xht";
+    readonly bcp47: "xht";
+  };
+  readonly heb: {
+    readonly name: "Hebrew";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "heb";
+    readonly bcp47: "he";
+  };
+  readonly hin: {
+    readonly name: "Hindi";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "hin";
+    readonly bcp47: "hi";
+  };
+  readonly hit: {
+    readonly name: "Hittite";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "hit";
+    readonly bcp47: "hit";
+  };
+  readonly xhu: {
+    readonly name: "Hurrian";
+    readonly kind: "language";
+    readonly standard: "iso639-3";
+    readonly iso639_3: "xhu";
+    readonly bcp47: "xhu";
+  };
+  readonly ita: {
+    readonly name: "Italian";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "ita";
+    readonly bcp47: "it";
+  };
+  readonly jpn: {
+    readonly name: "Japanese";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "jpn";
+    readonly bcp47: "ja";
+  };
+  readonly "jpn-xlt": {
+    readonly name: "Japanese XLit";
+    readonly kind: "transliteration";
+    readonly standard: "ochre";
+    readonly iso639_3: "jpn";
+    readonly bcp47: "ja-Latn-t-ja";
+  };
+  readonly kaz: {
+    readonly name: "Kazakh";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "kaz";
+    readonly bcp47: "kk";
+  };
+  readonly kor: {
+    readonly name: "Korean";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "kor";
+    readonly bcp47: "ko";
+  };
+  readonly lmc: {
+    readonly name: "Late Middle Chinese";
+    readonly kind: "language";
+    readonly standard: "ochre";
+    readonly iso639_3: "ltc";
+    readonly bcp47: "ltc";
+  };
+  readonly lat: {
+    readonly name: "Latin";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "lat";
+    readonly bcp47: "la";
+  };
+  readonly xlt: {
+    readonly name: "Latin XLit";
+    readonly kind: "transliteration";
+    readonly standard: "ochre";
+    readonly iso639_3: null;
+    readonly bcp47: "und-Latn";
+  };
+  readonly luw: {
+    readonly name: "Luwian";
+    readonly kind: "language";
+    readonly standard: "ochre";
+    readonly iso639_3: null;
+    readonly bcp47: null;
+  };
+  readonly cmn: {
+    readonly name: "Mandarin";
+    readonly kind: "language";
+    readonly standard: "iso639-3";
+    readonly iso639_3: "cmn";
+    readonly bcp47: "cmn";
+  };
+  readonly mar: {
+    readonly name: "Marathi";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "mar";
+    readonly bcp47: "mr";
+  };
+  readonly met: {
+    readonly name: "Metrical";
+    readonly kind: "vocabulary";
+    readonly standard: "ochre";
+    readonly iso639_3: null;
+    readonly bcp47: null;
+  };
+  readonly zxx: {
+    readonly name: "No linguistic content";
+    readonly kind: "non-linguistic";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "zxx";
+    readonly bcp47: "zxx";
+  };
+  readonly och: {
+    readonly name: "Old Chinese";
+    readonly kind: "language";
+    readonly standard: "iso639-3";
+    readonly iso639_3: "och";
+    readonly bcp47: "och";
+  };
+  readonly peo: {
+    readonly name: "Old Persian";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "peo";
+    readonly bcp47: "peo";
+  };
+  readonly ota: {
+    readonly name: "Ottoman Turkish";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "ota";
+    readonly bcp47: "ota";
+  };
+  readonly "ota-xlt": {
+    readonly name: "Ottoman Turkish XLit";
+    readonly kind: "transliteration";
+    readonly standard: "ochre";
+    readonly iso639_3: "ota";
+    readonly bcp47: "ota-Latn-t-ota-arab";
+  };
+  readonly pla: {
+    readonly name: "Palaic";
+    readonly kind: "language";
+    readonly standard: "ochre";
+    readonly iso639_3: "plq";
+    readonly bcp47: "plq";
+  };
+  readonly per: {
+    readonly name: "Persian";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "fas";
+    readonly bcp47: "fa";
+  };
+  readonly phn: {
+    readonly name: "Phoenician";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "phn";
+    readonly bcp47: "phn";
+  };
+  readonly xpg: {
+    readonly name: "Phrygian";
+    readonly kind: "language";
+    readonly standard: "iso639-3";
+    readonly iso639_3: "xpg";
+    readonly bcp47: "xpg";
+  };
+  readonly pol: {
+    readonly name: "Polish";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "pol";
+    readonly bcp47: "pl";
+  };
+  readonly por: {
+    readonly name: "Portuguese";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "por";
+    readonly bcp47: "pt";
+  };
+  readonly pro: {
+    readonly name: "Pronunciation";
+    readonly kind: "vocabulary";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: null;
+    readonly bcp47: null;
+  };
+  readonly rel: {
+    readonly name: "Relator";
+    readonly kind: "vocabulary";
+    readonly standard: "ochre";
+    readonly iso639_3: null;
+    readonly bcp47: "zxx";
+  };
+  readonly rus: {
+    readonly name: "Russian";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "rus";
+    readonly bcp47: "ru";
+  };
+  readonly san: {
+    readonly name: "Sanskrit";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "san";
+    readonly bcp47: "sa";
+  };
+  readonly srp: {
+    readonly name: "Serbian";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "srp";
+    readonly bcp47: "sr";
+  };
+  readonly spa: {
+    readonly name: "Spanish";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "spa";
+    readonly bcp47: "es";
+  };
+  readonly sux: {
+    readonly name: "Sumerian";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "sux";
+    readonly bcp47: "sux";
+  };
+  readonly sup: {
+    readonly name: "Supplemental";
+    readonly kind: "vocabulary";
+    readonly standard: "ochre";
+    readonly iso639_3: null;
+    readonly bcp47: null;
+  };
+  readonly syc: {
+    readonly name: "Syriac";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "syc";
+    readonly bcp47: "syc";
+  };
+  readonly tei: {
+    readonly name: "TEI";
+    readonly kind: "vocabulary";
+    readonly standard: "ochre";
+    readonly iso639_3: null;
+    readonly bcp47: null;
+  };
+  readonly tur: {
+    readonly name: "Turkish";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "tur";
+    readonly bcp47: "tr";
+  };
+  readonly uga: {
+    readonly name: "Ugaritic";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "uga";
+    readonly bcp47: "uga";
+  };
+  readonly ukr: {
+    readonly name: "Ukrainian";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "ukr";
+    readonly bcp47: "uk";
+  };
+  readonly urd: {
+    readonly name: "Urdu";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "urd";
+    readonly bcp47: "ur";
+  };
+  readonly uzb: {
+    readonly name: "Uzbek";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "uzb";
+    readonly bcp47: "uz";
+  };
+  readonly var: {
+    readonly name: "Variation";
+    readonly kind: "vocabulary";
+    readonly standard: "ochre";
+    readonly iso639_3: null;
+    readonly bcp47: null;
+  };
+  readonly was: {
+    readonly name: "Washo";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "was";
+    readonly bcp47: "was";
+  };
+  readonly yid: {
+    readonly name: "Yiddish";
+    readonly kind: "language";
+    readonly standard: "iso639-2b";
+    readonly iso639_3: "yid";
+    readonly bcp47: "yi";
+  };
+  readonly yua: {
+    readonly name: "Yucatec Maya";
+    readonly kind: "language";
+    readonly standard: "iso639-3";
+    readonly iso639_3: "yua";
+    readonly bcp47: "yua";
+  };
+};
+/**
+ * A language code OCHRE publishes
+ */
+export type OchreLanguageCode = keyof typeof OCHRE_LANGUAGES;
+/**
+ * Check whether a string is a language code OCHRE publishes
+ * @param code - The code to check
+ * @returns Whether OCHRE publishes the code
+ */
+export declare function isOchreLanguageCode(code: string): code is OchreLanguageCode;
+/**
+ * Look up a language OCHRE publishes by its OCHRE code
+ * @param code - The OCHRE language code
+ * @returns The language, or null when OCHRE does not publish the code
+ */
+export declare function getOchreLanguage(code: string): OchreLanguage | null;
+/**
+ * Find the OCHRE code of a language from its ISO 639-3 identifier
+ * @param iso639_3 - The ISO 639-3 identifier
+ * @returns The OCHRE code of the language itself, not of a transliteration,
+ * or null when OCHRE publishes no code for it
+ */
+export declare function getOchreLanguageCode(iso639_3: string): OchreLanguageCode | null;
+//#endregion