@delali/narsil 0.1.0

package/dist/languages/japanese.mjs

@@ -0,0 +1,210 @@
+// src/languages/japanese.ts
+var stopWords = /* @__PURE__ */ new Set([
+  "\u306E",
+  "\u306B",
+  "\u306F",
+  "\u3092",
+  "\u305F",
+  "\u304C",
+  "\u3067",
+  "\u3066",
+  "\u3068",
+  "\u3057",
+  "\u308C",
+  "\u3055",
+  "\u3042\u308B",
+  "\u3044\u308B",
+  "\u3059\u308B",
+  "\u306A\u308B",
+  "\u3067\u304D\u308B",
+  "\u306A\u3044",
+  "\u3082",
+  "\u306A",
+  "\u3084",
+  "\u304B",
+  "\u3078",
+  "\u3070",
+  "\u3088",
+  "\u306D",
+  "\u307E\u3067",
+  "\u304B\u3089",
+  "\u3088\u308A",
+  "\u307B\u3069",
+  "\u3060\u3051",
+  "\u3057\u304B",
+  "\u3070\u304B\u308A",
+  "\u304F\u3089\u3044",
+  "\u3050\u3089\u3044",
+  "\u307E\u307E",
+  "\u306A\u304C\u3089",
+  "\u3064\u3064",
+  "\u306E\u3067",
+  "\u306E\u306B",
+  "\u3051\u308C\u3069",
+  "\u3051\u308C\u3069\u3082",
+  "\u3051\u3069",
+  "\u3060\u304C",
+  "\u3057\u304B\u3057",
+  "\u3068\u3053\u308D\u304C",
+  "\u3067\u3082",
+  "\u305D\u308C\u3067\u3082",
+  "\u3060\u304B\u3089",
+  "\u3057\u305F\u304C\u3063\u3066",
+  "\u305D\u306E\u305F\u3081",
+  "\u3086\u3048\u306B",
+  "\u305D\u3057\u3066",
+  "\u307E\u305F",
+  "\u304A\u3088\u3073",
+  "\u306A\u3089\u3073\u306B",
+  "\u3042\u308B\u3044\u306F",
+  "\u307E\u305F\u306F",
+  "\u3082\u3057\u304F\u306F",
+  "\u305D\u308C\u3068\u3082",
+  "\u3064\u307E\u308A",
+  "\u3059\u306A\u308F\u3061",
+  "\u305F\u3068\u3048\u3070",
+  "\u306A\u304A",
+  "\u305F\u3060\u3057",
+  "\u3082\u3063\u3068\u3082",
+  "\u3055\u3066",
+  "\u3068\u3053\u308D\u3067",
+  "\u305D\u308C\u3067\u306F",
+  "\u3067\u306F",
+  "\u3055\u3089\u306B",
+  "\u3057\u304B\u3082",
+  "\u305D\u306E\u4E0A",
+  "\u305D\u308C\u306B",
+  "\u3053\u308C",
+  "\u305D\u308C",
+  "\u3042\u308C",
+  "\u3069\u308C",
+  "\u3053\u306E",
+  "\u305D\u306E",
+  "\u3042\u306E",
+  "\u3069\u306E",
+  "\u3053\u3053",
+  "\u305D\u3053",
+  "\u3042\u305D\u3053",
+  "\u3069\u3053",
+  "\u3053\u3046",
+  "\u305D\u3046",
+  "\u3042\u3042",
+  "\u3069\u3046",
+  "\u3053\u3093\u306A",
+  "\u305D\u3093\u306A",
+  "\u3042\u3093\u306A",
+  "\u3069\u3093\u306A",
+  "\u3053\u3061\u3089",
+  "\u305D\u3061\u3089",
+  "\u3042\u3061\u3089",
+  "\u3069\u3061\u3089",
+  "\u79C1",
+  "\u50D5",
+  "\u4FFA",
+  "\u81EA\u5206",
+  "\u3042\u306A\u305F",
+  "\u5F7C",
+  "\u5F7C\u5973",
+  "\u6211\u3005",
+  "\u79C1\u9054",
+  "\u5F7C\u3089",
+  "\u8AB0",
+  "\u4F55",
+  "\u3069\u306A\u305F",
+  "\u3044\u3064",
+  "\u306A\u305C",
+  "\u3069\u3046\u3057\u3066",
+  "\u3044\u304B\u304C",
+  "\u3067\u3059",
+  "\u307E\u3059",
+  "\u3067\u3057\u305F",
+  "\u307E\u3057\u305F",
+  "\u3067\u3057\u3087\u3046",
+  "\u3060\u308D\u3046",
+  "\u3060\u3063\u305F",
+  "\u3067\u3042\u308B",
+  "\u3067\u3042\u3063\u305F",
+  "\u3067\u3042\u308A",
+  "\u3067\u3042\u308D\u3046",
+  "\u3066\u3044\u308B",
+  "\u3066\u3044\u305F",
+  "\u3066\u304A\u304F",
+  "\u3066\u3042\u308B",
+  "\u3066\u3057\u307E\u3046",
+  "\u3066\u3044\u304F",
+  "\u3066\u304F\u308B",
+  "\u3066\u307F\u308B",
+  "\u3088\u3046\u3060",
+  "\u3089\u3057\u3044",
+  "\u305D\u3046\u3060",
+  "\u307F\u305F\u3044\u3060",
+  "\u3079\u304D",
+  "\u306F\u305A",
+  "\u308F\u3051",
+  "\u3053\u3068",
+  "\u3082\u306E",
+  "\u3068\u3053\u308D",
+  "\u3068\u304D",
+  "\u3068\u3082",
+  "\u305F\u3081",
+  "\u3046\u3061",
+  "\u307B\u3046",
+  "\u305F\u3061",
+  "\u3089",
+  "\u306A\u3069",
+  "\u306A\u3093\u304B",
+  "\u306A\u3093\u3066",
+  "\u3063\u3066",
+  "\u3068\u3044\u3046",
+  "\u3068\u3044\u3063\u305F",
+  "\u3068\u3057\u3066",
+  "\u306B\u3068\u3063\u3066",
+  "\u306B\u3064\u3044\u3066",
+  "\u306B\u95A2\u3057\u3066",
+  "\u306B\u5BFE\u3057\u3066",
+  "\u306B\u304A\u3044\u3066",
+  "\u306B\u304A\u3051\u308B",
+  "\u306B\u3088\u3063\u3066",
+  "\u306B\u3088\u308B",
+  "\u306B\u3088\u308A",
+  "\u3092\u3082\u3063\u3066",
+  "\u306B\u95A2\u3059\u308B",
+  "\u306B\u5BFE\u3059\u308B",
+  "\u305D\u308C\u305E\u308C",
+  "\u3059\u3079\u3066",
+  "\u307F\u3093\u306A",
+  "\u307B\u3068\u3093\u3069",
+  "\u305F\u304F\u3055\u3093",
+  "\u3044\u304F\u3064",
+  "\u3044\u304F\u3089",
+  "\u3082\u3046",
+  "\u307E\u3060",
+  "\u3082\u3063\u3068",
+  "\u305A\u3063\u3068",
+  "\u3068\u3066\u3082",
+  "\u304B\u306A\u308A",
+  "\u3084\u306F\u308A",
+  "\u3084\u3063\u3071\u308A",
+  "\u305C\u3072",
+  "\u304D\u3063\u3068",
+  "\u305F\u3076\u3093",
+  "\u304A\u305D\u3089\u304F",
+  "\u7279\u306B",
+  "\u3061\u3087\u3063\u3068",
+  "\u5C11\u3057",
+  "\u3060\u3044\u305F\u3044",
+  "\u307B\u307C",
+  "\u53CA\u3073"
+]);
+var japanese = {
+  name: "japanese",
+  stemmer: null,
+  stopWords,
+  tokenizer: {
+    splitPattern: /[^\u3005\u3040-\u309f\u30a0-\u30ff\u4e00-\u9fff\u3400-\u4dbf\uf900-\ufaff0-9a-zA-Z]+/gi,
+    normalizeDiacritics: false,
+    minTokenLength: 1
+  }
+};
+
+export { japanese };

package/dist/languages/nepali.d.ts

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

package/dist/languages/nepali.mjs

@@ -0,0 +1,262 @@
+// src/languages/nepali.ts
+var POSTPOSITIONS = [
+  "\u092E\u093E\u0930\u094D\u092B\u0924",
+  "\u0926\u094D\u0935\u093E\u0930\u093E",
+  "\u092A\u094D\u0930\u0924\u093F",
+  "\u0926\u0947\u0916\u093F",
+  "\u0938\u092E\u094D\u092E",
+  "\u092A\u091B\u093F",
+  "\u0932\u093E\u0908",
+  "\u0938\u0902\u0917",
+  "\u0938\u0901\u0917",
+  "\u0915\u093E",
+  "\u0915\u0940",
+  "\u0915\u094B",
+  "\u0915\u0948",
+  "\u0932\u0947",
+  "\u092E\u093E"
+];
+var VERB_SUFFIXES = [
+  "\u0925\u093F\u090F\u0928\u0928\u094D",
+  "\u0939\u0941\u0928\u094D\u091B",
+  "\u0939\u0941\u0928\u0947\u091B",
+  "\u0928\u0947\u091B\u0938\u094D",
+  "\u0925\u093F\u090F\u0928",
+  "\u0925\u093F\u092F\u094B",
+  "\u0907\u091B\u0938\u094D",
+  "\u090F\u091B\u0938\u094D",
+  "\u0907\u091B\u0928\u094D",
+  "\u090F\u091B\u0928\u094D",
+  "\u0939\u0941\u0928\u0947",
+  "\u0939\u0941\u0928\u094D",
+  "\u091B\u093F\u0928\u094D",
+  "\u0932\u093E\u0928\u094D",
+  "\u0932\u093F\u0928\u094D",
+  "\u0925\u093F\u090F",
+  "\u091B\u0928\u094D",
+  "\u091B\u094C\u0902",
+  "\u091B\u0948\u0928",
+  "\u091B\u0938\u094D",
+  "\u0928\u0947\u091B",
+  "\u091B\u094C\u0901",
+  "\u091B\u094C",
+  "\u0907\u091B",
+  "\u090F\u091B",
+  "\u091B\u0941",
+  "\u091B\u0947",
+  "\u0925\u093E",
+  "\u0925\u0940",
+  "\u0925\u0947",
+  "\u0928\u0947",
+  "\u0928\u0941",
+  "\u0926\u0948",
+  "\u0926\u093E",
+  "\u0924\u093E",
+  "\u0924\u0940",
+  "\u0924\u0947",
+  "\u0932\u093E",
+  "\u0932\u094B",
+  "\u0908\u0902",
+  "\u092F\u093E",
+  "\u092F\u094B",
+  "\u092F\u094C",
+  "\u0906\u0909",
+  "\u0906\u090F",
+  "\u091B",
+  "\u090F",
+  "\u0908"
+];
+var NOUN_SUFFIXES = [
+  "\u0939\u0930\u0942\u0932\u093E\u0908",
+  "\u0939\u0930\u0942\u0915\u094B",
+  "\u0939\u0930\u0942\u0932\u0947",
+  "\u0939\u0930\u0942\u092E\u093E",
+  "\u0924\u093E\u0932\u093E\u0908",
+  "\u0926\u093E\u0930\u0915\u094B",
+  "\u0924\u093E\u0915\u094B",
+  "\u0924\u093E\u0932\u0947",
+  "\u0924\u093E\u092E\u093E",
+  "\u0939\u0930\u0942",
+  "\u0926\u093E\u0930",
+  "\u0924\u093E"
+];
+var CASE_ENDINGS = ["\u092A\u094D\u0930\u0924\u093F", "\u0926\u0947\u0916\u093F", "\u0938\u092E\u094D\u092E", "\u0932\u093E\u0908", "\u092C\u093E\u091F", "\u0938\u0902\u0917", "\u0938\u0901\u0917", "\u0932\u0947", "\u0915\u094B", "\u092E\u093E", "\u092E\u0948"];
+function removeLongestSuffix(word, suffixes, minLength) {
+  for (const suffix of suffixes) {
+    if (word.endsWith(suffix) && word.length - suffix.length >= minLength) {
+      return word.slice(0, word.length - suffix.length);
+    }
+  }
+  return word;
+}
+function stem(word) {
+  if (word.length < 3) return word;
+  word = removeLongestSuffix(word, POSTPOSITIONS, 2);
+  let previous = "";
+  while (previous !== word) {
+    previous = word;
+    word = removeLongestSuffix(word, VERB_SUFFIXES, 2);
+    if (word !== previous) continue;
+    word = removeLongestSuffix(word, NOUN_SUFFIXES, 2);
+    if (word !== previous) continue;
+    word = removeLongestSuffix(word, CASE_ENDINGS, 2);
+  }
+  return word;
+}
+var stopWords = /* @__PURE__ */ new Set([
+  "\u0905\u0915\u094D\u0938\u0930",
+  "\u0905\u0917\u093E\u0921\u093F",
+  "\u0905\u091D\u0948",
+  "\u0905\u0928\u0941\u0938\u093E\u0930",
+  "\u0905\u0928\u094D\u0924\u0930\u094D\u0917\u0924",
+  "\u0905\u0928\u094D\u092F",
+  "\u0905\u092C",
+  "\u0905\u0930\u0942",
+  "\u0905\u0930\u094D\u0915\u094B",
+  "\u0905\u0930\u094D\u0925\u093E\u0924",
+  "\u0905\u0932\u0917",
+  "\u0906\u090F",
+  "\u0906\u0920",
+  "\u0906\u0926\u093F",
+  "\u0906\u092B\u0942",
+  "\u0906\u092B\u094D\u0928\u094B",
+  "\u0906\u092F\u094B",
+  "\u0909\u0926\u093E\u0939\u0930\u0923",
+  "\u0909\u0928",
+  "\u0909\u0928\u0915\u094B",
+  "\u0909\u0928\u0932\u0947",
+  "\u0909\u092A",
+  "\u090F\u0915",
+  "\u090F\u0915\u0926\u092E",
+  "\u0915\u0924\u0948",
+  "\u0915\u092E",
+  "\u0915\u0938\u0930\u0940",
+  "\u0915\u0938\u0948",
+  "\u0915\u0938\u0948\u0932\u0947",
+  "\u0915\u0939\u093E\u0901\u092C\u093E\u091F",
+  "\u0915\u093E",
+  "\u0915\u093F",
+  "\u0915\u093F\u0928",
+  "\u0915\u093F\u0928\u092D\u0928\u0947",
+  "\u0915\u0941\u0928\u0948",
+  "\u0915\u0941\u0930\u093E",
+  "\u0915\u0943\u092A\u092F\u093E",
+  "\u0915\u0947",
+  "\u0915\u0947\u0939\u093F",
+  "\u0915\u094B",
+  "\u0915\u094B\u0939\u0940",
+  "\u0917\u090F",
+  "\u0917\u0930\u0940",
+  "\u0917\u0930\u0947\u0915\u094B",
+  "\u0917\u0930\u0947\u0930",
+  "\u0917\u0930\u094D\u091B",
+  "\u0917\u0930\u094D\u0928",
+  "\u0917\u0930\u094D\u0928\u0941",
+  "\u0917\u0930\u094D\u0928\u0947",
+  "\u091A\u093E\u0930",
+  "\u091B",
+  "\u091B\u0928\u094D",
+  "\u091B\u0941",
+  "\u091B\u0948\u0928",
+  "\u091B\u094C\u0902",
+  "\u091C\u092C",
+  "\u091C\u0938\u0915\u094B",
+  "\u091C\u0938\u0932\u093E\u0908",
+  "\u091C\u0938\u0932\u0947",
+  "\u091C\u0938\u094D\u0924\u0948",
+  "\u091C\u0938\u094D\u0924\u094B",
+  "\u091C\u0939\u093E\u0901",
+  "\u091C\u093E\u0928",
+  "\u091C\u0941\u0928",
+  "\u091C\u0947",
+  "\u091C\u094B",
+  "\u0920\u0940\u0915",
+  "\u0924",
+  "\u0924\u0925\u093E",
+  "\u0924\u092A\u093E\u0908\u0902",
+  "\u0924\u0930",
+  "\u0924\u0932",
+  "\u0924\u093F\u0928\u0940",
+  "\u0924\u093F\u0928\u0940\u0939\u0930\u0942",
+  "\u0924\u0940\u0928",
+  "\u0924\u094D\u092F\u0938\u092A\u091B\u093F",
+  "\u0924\u094D\u092F\u0939\u093E\u0901",
+  "\u0924\u094D\u092F\u094B",
+  "\u0925\u093F\u090F",
+  "\u0925\u093F\u092F\u094B",
+  "\u0926\u093F\u090F",
+  "\u0926\u0941\u0908",
+  "\u0926\u0947\u0916\u093F",
+  "\u0927\u0947\u0930\u0948",
+  "\u0928",
+  "\u0928\u093F",
+  "\u0928\u093F\u092E\u094D\u0924\u093F",
+  "\u0928\u0948",
+  "\u092A\u091B\u093F",
+  "\u092A\u0928\u093F",
+  "\u092A\u0930\u094D\u091B",
+  "\u092A\u0939\u093F\u0932\u0947",
+  "\u092A\u0939\u093F\u0932\u094B",
+  "\u092A\u093E\u0901\u091A",
+  "\u092A\u094D\u0930\u0924\u093F",
+  "\u092A\u094D\u0930\u0924\u094D\u092F\u0947\u0915",
+  "\u092B\u0947\u0930\u093F",
+  "\u092C\u0928\u0947",
+  "\u092C\u093E\u0930\u0947",
+  "\u092C\u093E\u0939\u0947\u0915",
+  "\u092C\u0940\u091A",
+  "\u092D\u090F",
+  "\u092D\u090F\u0915\u094B",
+  "\u092D\u0928\u094D\u0926\u093E",
+  "\u092D\u0928\u094D\u0928\u0947",
+  "\u092D\u0930",
+  "\u092D\u093F\u0924\u094D\u0930",
+  "\u092E",
+  "\u092E\u0932\u093E\u0908",
+  "\u092E\u093E",
+  "\u092E\u093E\u0924\u094D\u0930",
+  "\u092E\u093E\u0925\u093F",
+  "\u092E\u0947\u0930\u094B",
+  "\u092F\u0924\u093F",
+  "\u092F\u0926\u093F",
+  "\u092F\u0938",
+  "\u092F\u0938\u0915\u094B",
+  "\u092F\u0938\u094D\u0924\u094B",
+  "\u092F\u0939\u093E\u0901",
+  "\u092F\u093E",
+  "\u092F\u0940",
+  "\u092F\u094B",
+  "\u0930",
+  "\u0930\u0939\u0947\u0915\u094B",
+  "\u0930\u093E\u092E\u094D\u0930\u094B",
+  "\u0932\u093E\u0908",
+  "\u0932\u093E\u0917\u093F",
+  "\u0932\u0947",
+  "\u0938\u0901\u0917",
+  "\u0938\u0915\u094D\u091B",
+  "\u0938\u092C\u0948",
+  "\u0938\u092E\u092F",
+  "\u0938\u092E\u094D\u092E",
+  "\u0938\u0939\u0940",
+  "\u0938\u093E\u0924",
+  "\u0938\u093E\u0925",
+  "\u0938\u094B",
+  "\u0939\u0930\u0947",
+  "\u0939\u0930\u0947\u0915",
+  "\u0939\u093E\u092E\u0940",
+  "\u0939\u093E\u092E\u094D\u0930\u094B",
+  "\u0939\u0941\u0901",
+  "\u0939\u0941\u0928",
+  "\u0939\u0941\u0928\u0947",
+  "\u0939\u0941\u0928\u094D\u091B",
+  "\u0939\u094B",
+  "\u0939\u094B\u0907\u0928"
+]);
+var nepali = {
+  name: "nepali",
+  stemmer: stem,
+  stopWords,
+  tokenizer: { splitPattern: /[^\u0900-\u097fa-z0-9]+/gi }
+};
+
+export { nepali };

package/dist/languages/norwegian.d.ts

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