@delali/narsil 0.1.15 → 0.2.0

package/dist/languages/kirundi.mjs

@@ -0,0 +1,92 @@
+// src/languages/kirundi.ts
+var stopWords = /* @__PURE__ */ new Set([
+  "aba",
+  "abandi",
+  "abo",
+  "aho",
+  "aka",
+  "ako",
+  "ari",
+  "ariko",
+  "aya",
+  "ayo",
+  "bari",
+  "bo",
+  "bose",
+  "bw",
+  "c",
+  "cane",
+  "canke",
+  "gusa",
+  "hagati",
+  "hanyuma",
+  "hari",
+  "hasi",
+  "hejuru",
+  "hose",
+  "ibi",
+  "ibindi",
+  "ico",
+  "iki",
+  "ikindi",
+  "imbere",
+  "inyuma",
+  "iri",
+  "iryo",
+  "ivyo",
+  "iyi",
+  "iyo",
+  "izindi",
+  "izo",
+  "jewe",
+  "kandi",
+  "ku",
+  "kubera",
+  "kuko",
+  "kuri",
+  "mu",
+  "muri",
+  "mwebwe",
+  "mwese",
+  "n",
+  "na",
+  "ndi",
+  "ni",
+  "nka",
+  "no",
+  "nta",
+  "rw",
+  "si",
+  "turi",
+  "twebwe",
+  "twese",
+  "u",
+  "ubu",
+  "uku",
+  "ukwo",
+  "uri",
+  "uru",
+  "urwo",
+  "utu",
+  "utwo",
+  "uwo",
+  "uwundi",
+  "uyu",
+  "vy",
+  "vyose",
+  "we",
+  "wewe",
+  "y",
+  "yari",
+  "yose",
+  "zose"
+]);
+var kirundi = {
+  name: "kirundi",
+  revision: "1",
+  stemmer: null,
+  stopWords,
+  tokenizer: { splitPattern: /[^a-z0-9]+/gi }
+};
+
+export { kirundi };

package/dist/languages/korean.d.ts

@@ -0,0 +1,5 @@
+import { L as LanguageModule } from '../language-CmJBHf1j.js';
+
+declare const korean: LanguageModule;
+
+export { korean };

package/dist/languages/korean.mjs

@@ -0,0 +1,252 @@
+// src/languages/korean.ts
+var stopWords = /* @__PURE__ */ new Set([
+  "\uAC00",
+  "\uAC00\uB839",
+  "\uAC01",
+  "\uAC01\uAC01",
+  "\uAC01\uC790",
+  "\uAC01\uC885",
+  "\uAC19\uB2E4",
+  "\uAC19\uC774",
+  "\uAC70\uBC14",
+  "\uAC70\uC758",
+  "\uAC83",
+  "\uAC83\uB4E4",
+  "\uACA8\uC6B0",
+  "\uACB0\uAD6D",
+  "\uACE0\uB85C",
+  "\uACE7",
+  "\uACFC",
+  "\uACFC\uC5F0",
+  "\uAD00\uD55C",
+  "\uAD6C",
+  "\uADF8",
+  "\uADF8\uB4E4",
+  "\uADF8\uB54C",
+  "\uADF8\uB798",
+  "\uADF8\uB7FC",
+  "\uADF8\uC800",
+  "\uAE30\uD0C0",
+  "\uAE4C\uC545",
+  "\uAE4C\uC9C0",
+  "\uAF48\uB2F9",
+  "\uB059\uB059",
+  "\uB07C\uC775",
+  "\uB098",
+  "\uB0A8\uB4E4",
+  "\uB0A8\uC9D3",
+  "\uB108",
+  "\uB108\uD76C",
+  "\uB124",
+  "\uB137",
+  "\uB144",
+  "\uB204\uAD6C",
+  "\uB2E4\uB978",
+  "\uB2E4\uB9CC",
+  "\uB2E4\uC12F",
+  "\uB2E4\uC18C",
+  "\uB2E4\uC218",
+  "\uB2E4\uC74C",
+  "\uB2E8\uC9C0",
+  "\uB2F5\uB2E4",
+  "\uB2F9\uC2E0",
+  "\uB2F9\uC7A5",
+  "\uB315\uADF8",
+  "\uB3D9\uC548",
+  "\uB458",
+  "\uB465\uB465",
+  "\uB4E4",
+  "\uB4F1",
+  "\uB4F1\uB4F1",
+  "\uB529\uB3D9",
+  "\uB530\uB77C",
+  "\uB530\uC704",
+  "\uB531",
+  "\uB54C",
+  "\uB610",
+  "\uB610\uD55C",
+  "\uB69D\uB69D",
+  "\uB839",
+  "\uB85C",
+  "\uB85C\uC368",
+  "\uB959",
+  "\uB97C",
+  "\uB9C8\uC800",
+  "\uB9C8\uCE58",
+  "\uB9CC\uC57D",
+  "\uB9CC\uC77C",
+  "\uB9CC\uD07C",
+  "\uB9E4",
+  "\uB9E4\uBC88",
+  "\uBA87",
+  "\uBAA8",
+  "\uBAA8\uB450",
+  "\uBB34\uB835",
+  "\uBB34\uC2A8",
+  "\uBB34\uC5C7",
+  "\uBB3C\uB860",
+  "\uBC0F",
+  "\uBC14\uB85C",
+  "\uBC84\uAE08",
+  "\uBD10",
+  "\uBD10\uB77C",
+  "\uBD80\uD130",
+  "\uBD95\uBD95",
+  "\uBE44\uB85D",
+  "\uC090\uAC71",
+  "\uC0AC",
+  "\uC0BC",
+  "\uC124\uB839",
+  "\uC124\uB9C8",
+  "\uC124\uC0AC",
+  "\uC14B",
+  "\uC18C\uC0DD",
+  "\uC18C\uC778",
+  "\uC1A8",
+  "\uC27F",
+  "\uC2DC\uAC01",
+  "\uC2DC\uAC04",
+  "\uC2E4\uB85C",
+  "\uC544",
+  "\uC544\uB2C8",
+  "\uC544\uC57C",
+  "\uC544\uC774",
+  "\uC544\uD558",
+  "\uC544\uD649",
+  "\uC557",
+  "\uC57C",
+  "\uC57D\uAC04",
+  "\uC591\uC790",
+  "\uC5B4",
+  "\uC5B4\uB290",
+  "\uC5B4\uB514",
+  "\uC5B4\uB54C",
+  "\uC5B4\uB5A4",
+  "\uC5B4\uC774",
+  "\uC5B4\uCC0C",
+  "\uC5B8\uC81C",
+  "\uC5BC\uB9C8",
+  "\uC5C9\uC5C9",
+  "\uC5D0",
+  "\uC5D0\uAC8C",
+  "\uC5D0\uC11C",
+  "\uC5EC",
+  "\uC5EC\uAE30",
+  "\uC5EC\uB35F",
+  "\uC5EC\uBD80",
+  "\uC5EC\uC12F",
+  "\uC5EC\uCC28",
+  "\uC601",
+  "\uC601\uCC28",
+  "\uC608",
+  "\uC624",
+  "\uC624\uC9C1",
+  "\uC624\uD638",
+  "\uC640",
+  "\uC640\uC544",
+  "\uC65C",
+  "\uC6B0\uB9AC",
+  "\uC6B0\uC120",
+  "\uC6B4\uC6B4",
+  "\uC6D4",
+  "\uC719\uC719",
+  "\uC721",
+  "\uC73C\uB85C",
+  "\uC744",
+  "\uC751",
+  "\uC751\uB2F9",
+  "\uC758",
+  "\uC758\uD574",
+  "\uC774",
+  "\uC774\uAC83",
+  "\uC774\uACF3",
+  "\uC774\uB54C",
+  "\uC774\uB798",
+  "\uC774\uB7F0",
+  "\uC774\uBC88",
+  "\uC774\uBD10",
+  "\uC774\uC0C1",
+  "\uC774\uC820",
+  "\uC774\uCABD",
+  "\uC778\uC820",
+  "\uC77C",
+  "\uC77C\uACF1",
+  "\uC77C\uB2E8",
+  "\uC77C\uB54C",
+  "\uC788\uB2E4",
+  "\uC790",
+  "\uC790\uAE30",
+  "\uC790\uC2E0",
+  "\uC7A0\uAE50",
+  "\uC7A0\uC2DC",
+  "\uC800",
+  "\uC800\uAC83",
+  "\uC800\uAE30",
+  "\uC800\uCABD",
+  "\uC800\uD76C",
+  "\uC804\uBD80",
+  "\uC804\uC790",
+  "\uC804\uD6C4",
+  "\uC81C",
+  "\uC870\uAE08",
+  "\uC870\uCC28",
+  "\uC878\uC878",
+  "\uC880",
+  "\uC88B\uC544",
+  "\uC88D\uC88D",
+  "\uC989",
+  "\uC989\uC2DC",
+  "\uC9C0\uB9CC",
+  "\uCC38",
+  "\uCC38\uB098",
+  "\uCCC7",
+  "\uCE60",
+  "\uCF78\uCF78",
+  "\uCF85\uCF85",
+  "\uCFF5",
+  "\uD0C0\uB2E4",
+  "\uD0C0\uC778",
+  "\uD0D5\uD0D5",
+  "\uD22D",
+  "\uD264",
+  "\uD2C8\uD0C0",
+  "\uD30D",
+  "\uD314",
+  "\uD37D",
+  "\uD384\uB801",
+  "\uD558",
+  "\uD558\uB098",
+  "\uD558\uD558",
+  "\uD55C\uB370",
+  "\uD560\uB54C",
+  "\uD560\uBFD0",
+  "\uD568\uAED8",
+  "\uD574\uC694",
+  "\uD5C8",
+  "\uD5C8\uAC71",
+  "\uD5C8\uD5C8",
+  "\uD5C9",
+  "\uD5C9\uD5C9",
+  "\uD639\uC2DC",
+  "\uD639\uC740",
+  "\uD63C\uC790",
+  "\uD6E8\uC52C",
+  "\uD718\uC775",
+  "\uD734",
+  "\uD750\uD750",
+  "\uD765"
+]);
+var korean = {
+  name: "korean",
+  revision: "1",
+  stemmer: null,
+  stopWords,
+  tokenizer: {
+    splitPattern: /[^\p{Script=Hangul}\p{Script=Han}0-9a-zA-Z]+/giu,
+    normalizeDiacritics: false,
+    minTokenLength: 1,
+    ngramSize: 2
+  }
+};
+
+export { korean };

package/dist/languages/kurmanji.d.ts

@@ -0,0 +1,5 @@
+import { L as LanguageModule } from '../language-CmJBHf1j.js';
+
+declare const kurmanji: LanguageModule;
+
+export { kurmanji };

package/dist/languages/kurmanji.mjs

@@ -0,0 +1,50 @@
+// src/languages/kurmanji.ts
+var stopWords = /* @__PURE__ */ new Set([
+  "bi",
+  "\xE7awa",
+  "\xE7end",
+  "\xE7i",
+  "\xE7ima",
+  "\xE7iqas",
+  "da",
+  "de",
+  "di",
+  "em",
+  "ev",
+  "ew",
+  "ez",
+  "gelek",
+  "hem\xFB",
+  "her",
+  "hin",
+  "h\xFBn",
+  "ji",
+  "k\xEA",
+  "keng\xEE",
+  "kes",
+  "k\xEE",
+  "ku",
+  "li",
+  "me",
+  "min",
+  "te",
+  "ti\u015Ft",
+  "tu",
+  "\xFB",
+  "va",
+  "v\xEA",
+  "v\xEE",
+  "wan",
+  "we",
+  "w\xEA",
+  "w\xEE"
+]);
+var kurmanji = {
+  name: "kurmanji",
+  revision: "1",
+  stemmer: null,
+  stopWords,
+  tokenizer: { splitPattern: /[^a-z0-9çêîşû]+/gi }
+};
+
+export { kurmanji };

package/dist/languages/kyrgyz.d.ts

@@ -0,0 +1,5 @@
+import { L as LanguageModule } from '../language-CmJBHf1j.js';
+
+declare const kyrgyz: LanguageModule;
+
+export { kyrgyz };

package/dist/languages/kyrgyz.mjs

@@ -0,0 +1,108 @@
+// src/languages/kyrgyz.ts
+var stopWords = /* @__PURE__ */ new Set([
+  "\u0430\u0433\u0430",
+  "\u0430\u0434\u0430\u043C",
+  "\u0430\u0439\u0442\u0442\u044B",
+  "\u0430\u0439\u0442\u044B\u043C\u044B\u043D\u0434\u0430",
+  "\u0430\u0439\u0442\u044B\u043F",
+  "\u0430\u043B",
+  "\u0430\u043B\u0430\u0440",
+  "\u0430\u043B\u0430\u0440\u0434\u044B\u043D",
+  "\u0430\u043B\u0433\u0430\u043D",
+  "\u0430\u043B\u0443\u0443",
+  "\u0430\u043B\u044B\u043F",
+  "\u0430\u043D\u0434\u0430",
+  "\u0430\u043D\u0434\u0430\u043D",
+  "\u0430\u043D\u044B",
+  "\u0430\u043D\u044B\u043D",
+  "\u0430\u0440",
+  "\u0431\u0430\u0440",
+  "\u0431\u0430\u0441\u043C\u0430",
+  "\u0431\u0430\u0448",
+  "\u0431\u0430\u0448\u043A\u0430",
+  "\u0431\u0430\u0448\u043A\u044B",
+  "\u0431\u0430\u0448\u0447\u044B\u0441\u044B",
+  "\u0431\u0435\u0440\u0433\u0435\u043D",
+  "\u0431\u0438\u0437",
+  "\u0431\u0438\u043B\u0434\u0438\u0440\u0433\u0435\u043D",
+  "\u0431\u0438\u043B\u0434\u0438\u0440\u0434\u0438",
+  "\u0431\u0438\u0440",
+  "\u0431\u0438\u0440\u0438\u043D\u0447\u0438",
+  "\u0431\u0438\u0440\u043E\u043A",
+  "\u0431\u0438\u0448\u043A\u0435\u043A",
+  "\u0431\u043E\u043B\u0433\u043E\u043D",
+  "\u0431\u043E\u043B\u043E\u0442",
+  "\u0431\u043E\u043B\u0441\u043E",
+  "\u0431\u043E\u043B\u0443\u043F",
+  "\u0431\u043E\u044E\u043D\u0447\u0430",
+  "\u0431\u0443\u0433\u0430",
+  "\u0431\u0443\u043B",
+  "\u0433\u0430\u043D\u0430",
+  "\u0434\u0430",
+  "\u0434\u0430\u0433\u044B",
+  "\u0434\u0435\u0433\u0435\u043D",
+  "\u0434\u0435\u0434\u0438",
+  "\u0434\u0435\u043F",
+  "\u0436\u0430\u043D\u0430",
+  "\u0436\u0430\u04A3\u044B",
+  "\u0436\u0430\u0442\u0430\u0442",
+  "\u0436\u0430\u0442\u043A\u0430\u043D",
+  "\u0436\u0435",
+  "\u0436\u043E\u0433\u043E\u0440\u043A\u0443",
+  "\u0436\u043E\u043A",
+  "\u0436\u043E\u043B",
+  "\u0436\u043E\u043B\u0443",
+  "\u043A\u0430\u0431\u044B\u043B",
+  "\u043A\u0430\u043B\u0433\u0430\u043D",
+  "\u043A\u0430\u043D\u0434\u0430\u0439",
+  "\u043A\u0430\u0440\u0430\u0442\u0430",
+  "\u043A\u0430\u0440\u0448\u044B",
+  "\u043A\u0430\u0442\u0430\u0440\u044B",
+  "\u043A\u0435\u043B\u0433\u0435\u043D",
+  "\u043A\u0435\u0440\u0435\u043A",
+  "\u043A\u0438\u0439\u0438\u043D",
+  "\u043A\u043E\u043B",
+  "\u043A\u04E9\u043F",
+  "\u043A\u04AF\u043D\u04AF",
+  "\u043A\u044B\u043B\u043C\u044B\u0448",
+  "\u043A\u044B\u0440\u0433\u044B\u0437",
+  "\u043C\u0430\u0430\u043B\u044B\u043C\u0430\u0442",
+  "\u043C\u0430\u043C\u043B\u0435\u043A\u0435\u0442\u0442\u0438\u043A",
+  "\u043C\u0435\u043D",
+  "\u043C\u0435\u043D\u0435\u043D",
+  "\u043C\u0438\u04A3",
+  "\u043C\u0443\u0440\u0434\u0430\u0433\u044B",
+  "\u043C\u04AF\u043C\u043A\u04AF\u043D",
+  "\u043C\u044B\u0439\u0437\u0430\u043C",
+  "\u043C\u044B\u043D\u0434\u0430\u0439",
+  "\u043E\u0448\u043E\u043B",
+  "\u043E\u0448\u043E\u043D\u0434\u043E\u0439",
+  "\u04E9\u0437",
+  "\u0441\u04E9\u0437",
+  "\u0441\u04AF\u0440\u04E9\u0442",
+  "\u0442\u0430\u0440\u0430\u0431\u044B\u043D\u0430\u043D",
+  "\u0442\u0443\u0440\u0433\u0430\u043D",
+  "\u0442\u0443\u0443\u0440\u0430\u043B\u0443\u0443",
+  "\u0443\u043A\u0443\u043A",
+  "\u0443\u0447\u0443\u0440\u0434\u0430",
+  "\u04AF\u0447",
+  "\u04AF\u0447\u04AF\u043D",
+  "\u0447\u0435\u0439\u0438\u043D",
+  "\u0447\u0435\u043A",
+  "\u044D\u043A\u0435\u043D\u0438\u043D",
+  "\u044D\u043A\u0438",
+  "\u044D\u043B",
+  "\u044D\u043B\u0435",
+  "\u044D\u043C\u0435\u0441",
+  "\u044D\u043C\u0438",
+  "\u044D\u0447"
+]);
+var kyrgyz = {
+  name: "kyrgyz",
+  revision: "1",
+  stemmer: null,
+  stopWords,
+  tokenizer: { splitPattern: /[^a-z0-9а-яңөү]+/gi }
+};
+
+export { kyrgyz };

package/dist/languages/lao.d.ts

@@ -0,0 +1,5 @@
+import { L as LanguageModule } from '../language-CmJBHf1j.js';
+
+declare const lao: LanguageModule;
+
+export { lao };

package/dist/languages/lao.mjs

@@ -0,0 +1,63 @@
+// src/languages/lao.ts
+var stopWords = /* @__PURE__ */ new Set([
+  "\u0E81\u0ECD",
+  "\u0E81\u0EB1\u0E99",
+  "\u0E81\u0EB1\u0E9A",
+  "\u0E82\u0ECD",
+  "\u0E82\u0EB6\u0EC9\u0E99",
+  "\u0E84\u0EB1\u0EC9\u0E87",
+  "\u0E84\u0EB7",
+  "\u0E84\u0EBB\u0E87",
+  "\u0E88\u0EB0",
+  "\u0E88\u0EB1\u0E94",
+  "\u0E88\u0EB6\u0EC8\u0E87",
+  "\u0EAA\u0EB4\u0EC8\u0E87",
+  "\u0EAA\u0EB8\u0E94",
+  "\u0EAA\u0EBB\u0EC8\u0E87",
+  "\u0E8D\u0EB1\u0E87",
+  "\u0E94\u0EB1\u0E87",
+  "\u0EC4\u0E94\u0EC9",
+  "\u0E95\u0ECD\u0EC8",
+  "\u0E95\u0EB1\u0EC9\u0E87",
+  "\u0EC1\u0E95\u0EC8",
+  "\u0E96\u0EC9\u0EB2",
+  "\u0E96\u0EB7\u0E81",
+  "\u0E97\u0EB1\u0E87",
+  "\u0E97\u0EB5\u0EC8",
+  "\u0E97\u0EB8\u0E81",
+  "\u0E99\u0EB1\u0E81",
+  "\u0E99\u0EB1\u0EC9\u0E99",
+  "\u0E99\u0ECD\u0EB2",
+  "\u0E99\u0EB5\u0EC9",
+  "\u0EC3\u0E99",
+  "\u0E9A\u0ECD\u0EC8",
+  "\u0EC4\u0E9B",
+  "\u0E9C\u0EBB\u0E99",
+  "\u0E9E\u0EBB\u0E9A",
+  "\u0EA1\u0EB2",
+  "\u0EA1\u0EB5",
+  "\u0EA1\u0EB7\u0EC9",
+  "\u0EA2\u0EB9\u0EC8",
+  "\u0EA5\u0EBB\u0E87",
+  "\u0EA7\u0EC8\u0EB2",
+  "\u0EC4\u0EA7\u0EC9",
+  "\u0EAB\u0EA5\u0EB7",
+  "\u0EAB\u0EBC\u0EB7",
+  "\u0EC3\u0EAB\u0EC9",
+  "\u0EAD\u0EB5\u0E81",
+  "\u0EAE\u0EB1\u0E9A"
+]);
+var lao = {
+  name: "lao",
+  revision: "1",
+  stemmer: null,
+  stopWords,
+  tokenizer: {
+    splitPattern: /[^ກ-ໟa-z0-9]+/giu,
+    normalizeDiacritics: false,
+    minTokenLength: 1,
+    ngramSize: 2
+  }
+};
+
+export { lao };

package/dist/languages/latin.d.ts

@@ -0,0 +1,5 @@
+import { L as LanguageModule } from '../language-CmJBHf1j.js';
+
+declare const latin: LanguageModule;
+
+export { latin };

package/dist/languages/latin.mjs

@@ -0,0 +1,69 @@
+import { removeMarks } from '../chunk-KH3YGMVT.mjs';
+import { withNormalisedSpellings } from '../chunk-GHJWC2E5.mjs';
+
+// src/languages/latin.ts
+var stopWords = /* @__PURE__ */ new Set([
+  "a",
+  "ab",
+  "ac",
+  "ad",
+  "at",
+  "atque",
+  "aut",
+  "autem",
+  "cum",
+  "de",
+  "dum",
+  "e",
+  "erant",
+  "erat",
+  "est",
+  "et",
+  "etiam",
+  "ex",
+  "haec",
+  "hic",
+  "hoc",
+  "in",
+  "ita",
+  "me",
+  "nec",
+  "neque",
+  "non",
+  "per",
+  "qua",
+  "quae",
+  "quam",
+  "qui",
+  "quibus",
+  "quidem",
+  "quo",
+  "quod",
+  "re",
+  "rebus",
+  "rem",
+  "res",
+  "sed",
+  "si",
+  "sic",
+  "sunt",
+  "tamen",
+  "tandem",
+  "te",
+  "ut",
+  "vel"
+]);
+var MACRON = /\u0304/g;
+function normalize(token) {
+  return removeMarks(token, MACRON);
+}
+var latin = {
+  name: "latin",
+  revision: "1",
+  stemmer: null,
+  stopWords: withNormalisedSpellings(stopWords, normalize),
+  normalizer: normalize,
+  tokenizer: { splitPattern: /[^a-z0-9āēīōū]+/gi }
+};
+
+export { latin };

package/dist/languages/latvian.d.ts

@@ -0,0 +1,5 @@
+import { L as LanguageModule } from '../language-CmJBHf1j.js';
+
+declare const latvian: LanguageModule;
+
+export { latvian };