taraskevizer 5.1.4 → 5.1.6

package/dist/index.cjs

@@ -25,16 +25,17 @@ __export(src_exports, {
   Taraskevizer: () => Taraskevizer,
   VARIATION: () => VARIATION,
   __tarask__: () => __tarask__,
+  convertAlphabet: () => convertAlphabet,
   gobj: () => gobj
 });
 module.exports = __toCommonJS(src_exports);
 
 // src/dict/latin.ts
 var latinLetters = [
-  [/(?<=[аеёіоуўыэюяьʼ| ]\(?)е/, "je"],
-  [/(?<=[аеёіоуўыэюяьʼ| ]\(?)ё/, "jo"],
-  [/(?<=[аеёіоуўыэюяьʼ| ]\(?)ю/, "ju"],
-  [/(?<=[аеёіоуўыэюяьʼ| ]\(?)я/, "ja"],
+  [/(?<=[аеёіоуўыэюяьʼ| >АЕЁІОУЎЫЭЮЯЬ]|^)е/, "je"],
+  [/(?<=[eаеёіоуўыэюяьʼ| >АЕЁІОУЎЫЭЮЯЬ]|^)ё/, "jo"],
+  [/(?<=[eoаеёіоуўыэюяьʼ| >АЕЁІОУЎЫЭЮЯЬ]|^)ю/, "ju"],
+  [/(?<=[eouаеёіоуўыэюяьʼ| >АЕЁІОУЎЫЭЮЯЬ]|^)я/, "ja"],
   [/ʼі/, "ji"],
   [/ʼ/, ""],
   [/ць/, "ć"],
@@ -202,8 +203,8 @@ var chemicalElements2 = [
 chemicalElements2[1] = chemicalElements2[0] + "|айнштайн|мендзялев|сыборг|гас|флеров";
 var chemicalElements3 = " гал|бэрыл|тул|бэркл|набэл";
 var toOneLine = (str) => str.replace(/\n/g, "|");
-var ia = (word, words) => ` ${word} (?=\\(?(?:[бвгджзйклмнпстфцчшў]*[оё]|${words.replace(/\(/g, "(?:")}|${iwords}))`;
-var iwords = toOneLine(`і( 
+var ia = (word, words) => ` ${word} (?=\\(?(?:[бвгджзйклмнпстфцчшў]*[оё]|${words.replace(/\(/g, "(?:")}|i(?:${iwords})))`;
+var iwords = toOneLine(` 
 біс
 бсэн
 в[аеоы] 
@@ -240,7 +241,7 @@ var iwords = toOneLine(`і(
 нфікс
 нфімум
 ншась?ц
-нш(а[йя]?|ую|ы(мі|х|я)?) 
+нш(а[ейя]?|ась?ц|ую|ы(мі|х|я)?) 
 псілан
 рад
 рбіс
@@ -258,7 +259,7 @@ var iwords = toOneLine(`і(
 тар
 тры
 ць?він
-шыяс)`);
+шыяс`);
 var rawWordlist = [
   /* А */
   [/аахен/, "аахэн"],
@@ -1846,7 +1847,7 @@ var rawsofteners = [
   [/з(?=([бвзлмнц]|дз)[еёіюяь])/, "зь"],
   [/с(?=[бвлмнпсфц][еёіюяь])/, "сь"],
   [/ц(?=[вм][еёіюяь])/, "ць"],
-  [`( (?:б[ея]|пра|цера)?з) (\\(?)(?=[еёюяь]|([бвзйлмнпстфц]|дз)[еёіюяь]|імі? |іх(?:ні)?|${iwords})`, "$1ь $2"],
+  [`( (?:б[ея]|пра|цера)?з) (?=\\(?)(?=[еёюяь]|([бвзйлмнпстфц]|дз)[еёіюяь]|імі? |іх(?:ні)?|і(?:${iwords}))`, "$1ь "],
   [/([сз])ʼ(?=[яюеё])/, "$1ь"]
 ];
 var gwords = [
@@ -1968,12 +1969,13 @@ var getLastLetter = (word, i) => {
 };
 var NOFIX_CHAR = "  ";
 var NOFIX_REGEX = new RegExp(NOFIX_CHAR, "g");
-var OPTIONAL_WORDS_REGEX = /\(.*?\)/g;
+var OPTIONAL_WORDS_REGEX = /\([^\)]*?\)/g;
 var G_REGEX = /[Ґґ]/g;
 var ALPHABET = {
   CYRILLIC: 0,
   LATIN: 1,
-  ARABIC: 2
+  ARABIC: 2,
+  LATIN_JI: 1
 };
 var REPLACE_J = {
   NEVER: 0,
@@ -2040,7 +2042,7 @@ var restoreCase = (text, orig) => {
       text[i] = word.toUpperCase();
     } else {
       text[i] = word[0] === "(" ? word.replace(
-        /.*?(?=\))/,
+        /[^)]*?(?=\))/,
         ($0) => $0.replace(/[(|]./g, ($02) => $02.toUpperCase())
       ) : word[0].toUpperCase() + word.slice(1);
     }
@@ -2119,13 +2121,14 @@ var __tarask__ = {
   replaceWithDict,
   afterTarask
 };
+var convertAlphabet = (text, abc) => replaceWithDict(replaceWithDict(text, letters[abc]), lettersUpperCase[abc]);
 var Taraskevizer = class {
   abc = ALPHABET.CYRILLIC;
   j = REPLACE_J.NEVER;
+  doEscapeCapitalized = true;
   html = {
     g: false
   };
-  doEscapeCapitalized = true;
   nonHtml = {
     h: false,
     ansiColors: false,
@@ -2136,12 +2139,13 @@ var Taraskevizer = class {
       return;
     const general = options.general;
     if (general) {
-      if (general.abc)
-        this.abc = general.abc;
-      if (general.j)
-        this.j = general.j;
-      if ("doEscapeCapitalized" in general)
-        this.doEscapeCapitalized = general.doEscapeCapitalized;
+      for (const prop of [
+        "abc",
+        "j",
+        "doEscapeCapitalized"
+      ])
+        if (prop in general)
+          this[prop] = general[prop];
     }
     if (options.OVERRIDE_taraskevize)
       this.taraskevize = options.OVERRIDE_taraskevize;
@@ -2196,7 +2200,7 @@ var Taraskevizer = class {
     text = ` ${text.trim()} `.replace(//g, "");
     if (this.doEscapeCapitalized)
       text = text.replace(new RegExp("(?!<=\\p{Lu} )(\\p{Lu}{2,})(?!= \\p{Lu})", "gu"), "<*.$1>");
-    text = text.replace(/<(\*?)([,.]?)(.*?)>/gs, ($0, $1, $2, $3) => {
+    text = text.replace(/<(\*?)([,.]?)([^>]*?)>/gs, ($0, $1, $2, $3) => {
       if ($2 === ",")
         return LEFT_ANGLE_BRACKET + $3 + ">";
       if ($1)
@@ -2208,10 +2212,7 @@ var Taraskevizer = class {
       return NOFIX_CHAR;
     }).replace(/г'(?![еёіюя])/g, "ґ").replace(/([\n\t])/g, " $1 ").replace(/ - /g, " — ").replace(new RegExp("(\\p{P}|\\p{S}|\\d)", "gu"), " $1 ").replace(/ ['`’] (?=\S)/g, "ʼ").replace(/\(/g, "&#40");
     let splittedOrig, splitted;
-    splittedOrig = replaceWithDict(
-      replaceWithDict(text, letters[abc]),
-      lettersUpperCase[abc]
-    ).split(" ");
+    splittedOrig = convertAlphabet(text, abc).split(" ");
     text = this.taraskevize(text.toLowerCase());
     if (j)
       text = replaceIbyJ(text, j === REPLACE_J.ALWAYS);
@@ -2241,5 +2242,6 @@ var Taraskevizer = class {
   Taraskevizer,
   VARIATION,
   __tarask__,
+  convertAlphabet,
   gobj
 });

package/dist/index.d.ts

@@ -28,6 +28,7 @@ declare const ALPHABET: {
     readonly CYRILLIC: 0;
     readonly LATIN: 1;
     readonly ARABIC: 2;
+    readonly LATIN_JI: 1;
 };
 declare const REPLACE_J: {
     readonly NEVER: 0;
@@ -45,13 +46,14 @@ declare const __tarask__: {
     readonly replaceWithDict: (text: string, dict?: ExtendedDict) => string;
     readonly afterTarask: ExtendedDict;
 };
+declare const convertAlphabet: (text: string, abc: Alphabet) => string;
 declare class Taraskevizer {
     abc: Alphabet;
     j: OptionJ;
+    doEscapeCapitalized: boolean;
     html: {
         g: boolean;
     };
-    doEscapeCapitalized: boolean;
     nonHtml: {
         h: boolean;
         ansiColors: boolean;
@@ -76,4 +78,4 @@ declare const gobj: {
     readonly Ґ: "Г";
 };
 
-export { ALPHABET, type HtmlOptions, type NonHtmlOptions, REPLACE_J, type TaraskOptions, Taraskevizer, VARIATION, __tarask__, gobj };
+export { ALPHABET, type HtmlOptions, type NonHtmlOptions, REPLACE_J, type TaraskOptions, Taraskevizer, VARIATION, __tarask__, convertAlphabet, gobj };

package/dist/index.js

@@ -1,9 +1,9 @@
 // src/dict/latin.ts
 var latinLetters = [
-  [/(?<=[аеёіоуўыэюяьʼ| ]\(?)е/, "je"],
-  [/(?<=[аеёіоуўыэюяьʼ| ]\(?)ё/, "jo"],
-  [/(?<=[аеёіоуўыэюяьʼ| ]\(?)ю/, "ju"],
-  [/(?<=[аеёіоуўыэюяьʼ| ]\(?)я/, "ja"],
+  [/(?<=[аеёіоуўыэюяьʼ| >АЕЁІОУЎЫЭЮЯЬ]|^)е/, "je"],
+  [/(?<=[eаеёіоуўыэюяьʼ| >АЕЁІОУЎЫЭЮЯЬ]|^)ё/, "jo"],
+  [/(?<=[eoаеёіоуўыэюяьʼ| >АЕЁІОУЎЫЭЮЯЬ]|^)ю/, "ju"],
+  [/(?<=[eouаеёіоуўыэюяьʼ| >АЕЁІОУЎЫЭЮЯЬ]|^)я/, "ja"],
   [/ʼі/, "ji"],
   [/ʼ/, ""],
   [/ць/, "ć"],
@@ -171,8 +171,8 @@ var chemicalElements2 = [
 chemicalElements2[1] = chemicalElements2[0] + "|айнштайн|мендзялев|сыборг|гас|флеров";
 var chemicalElements3 = " гал|бэрыл|тул|бэркл|набэл";
 var toOneLine = (str) => str.replace(/\n/g, "|");
-var ia = (word, words) => ` ${word} (?=\\(?(?:[бвгджзйклмнпстфцчшў]*[оё]|${words.replace(/\(/g, "(?:")}|${iwords}))`;
-var iwords = toOneLine(`і( 
+var ia = (word, words) => ` ${word} (?=\\(?(?:[бвгджзйклмнпстфцчшў]*[оё]|${words.replace(/\(/g, "(?:")}|i(?:${iwords})))`;
+var iwords = toOneLine(` 
 біс
 бсэн
 в[аеоы] 
@@ -209,7 +209,7 @@ var iwords = toOneLine(`і(
 нфікс
 нфімум
 ншась?ц
-нш(а[йя]?|ую|ы(мі|х|я)?) 
+нш(а[ейя]?|ась?ц|ую|ы(мі|х|я)?) 
 псілан
 рад
 рбіс
@@ -227,7 +227,7 @@ var iwords = toOneLine(`і(
 тар
 тры
 ць?він
-шыяс)`);
+шыяс`);
 var rawWordlist = [
   /* А */
   [/аахен/, "аахэн"],
@@ -1815,7 +1815,7 @@ var rawsofteners = [
   [/з(?=([бвзлмнц]|дз)[еёіюяь])/, "зь"],
   [/с(?=[бвлмнпсфц][еёіюяь])/, "сь"],
   [/ц(?=[вм][еёіюяь])/, "ць"],
-  [`( (?:б[ея]|пра|цера)?з) (\\(?)(?=[еёюяь]|([бвзйлмнпстфц]|дз)[еёіюяь]|імі? |іх(?:ні)?|${iwords})`, "$1ь $2"],
+  [`( (?:б[ея]|пра|цера)?з) (?=\\(?)(?=[еёюяь]|([бвзйлмнпстфц]|дз)[еёіюяь]|імі? |іх(?:ні)?|і(?:${iwords}))`, "$1ь "],
   [/([сз])ʼ(?=[яюеё])/, "$1ь"]
 ];
 var gwords = [
@@ -1937,12 +1937,13 @@ var getLastLetter = (word, i) => {
 };
 var NOFIX_CHAR = "  ";
 var NOFIX_REGEX = new RegExp(NOFIX_CHAR, "g");
-var OPTIONAL_WORDS_REGEX = /\(.*?\)/g;
+var OPTIONAL_WORDS_REGEX = /\([^\)]*?\)/g;
 var G_REGEX = /[Ґґ]/g;
 var ALPHABET = {
   CYRILLIC: 0,
   LATIN: 1,
-  ARABIC: 2
+  ARABIC: 2,
+  LATIN_JI: 1
 };
 var REPLACE_J = {
   NEVER: 0,
@@ -2009,7 +2010,7 @@ var restoreCase = (text, orig) => {
       text[i] = word.toUpperCase();
     } else {
       text[i] = word[0] === "(" ? word.replace(
-        /.*?(?=\))/,
+        /[^)]*?(?=\))/,
         ($0) => $0.replace(/[(|]./g, ($02) => $02.toUpperCase())
       ) : word[0].toUpperCase() + word.slice(1);
     }
@@ -2088,13 +2089,14 @@ var __tarask__ = {
   replaceWithDict,
   afterTarask
 };
+var convertAlphabet = (text, abc) => replaceWithDict(replaceWithDict(text, letters[abc]), lettersUpperCase[abc]);
 var Taraskevizer = class {
   abc = ALPHABET.CYRILLIC;
   j = REPLACE_J.NEVER;
+  doEscapeCapitalized = true;
   html = {
     g: false
   };
-  doEscapeCapitalized = true;
   nonHtml = {
     h: false,
     ansiColors: false,
@@ -2105,12 +2107,13 @@ var Taraskevizer = class {
       return;
     const general = options.general;
     if (general) {
-      if (general.abc)
-        this.abc = general.abc;
-      if (general.j)
-        this.j = general.j;
-      if ("doEscapeCapitalized" in general)
-        this.doEscapeCapitalized = general.doEscapeCapitalized;
+      for (const prop of [
+        "abc",
+        "j",
+        "doEscapeCapitalized"
+      ])
+        if (prop in general)
+          this[prop] = general[prop];
     }
     if (options.OVERRIDE_taraskevize)
       this.taraskevize = options.OVERRIDE_taraskevize;
@@ -2165,7 +2168,7 @@ var Taraskevizer = class {
     text = ` ${text.trim()} `.replace(//g, "");
     if (this.doEscapeCapitalized)
       text = text.replace(new RegExp("(?!<=\\p{Lu} )(\\p{Lu}{2,})(?!= \\p{Lu})", "gu"), "<*.$1>");
-    text = text.replace(/<(\*?)([,.]?)(.*?)>/gs, ($0, $1, $2, $3) => {
+    text = text.replace(/<(\*?)([,.]?)([^>]*?)>/gs, ($0, $1, $2, $3) => {
       if ($2 === ",")
         return LEFT_ANGLE_BRACKET + $3 + ">";
       if ($1)
@@ -2177,10 +2180,7 @@ var Taraskevizer = class {
       return NOFIX_CHAR;
     }).replace(/г'(?![еёіюя])/g, "ґ").replace(/([\n\t])/g, " $1 ").replace(/ - /g, " — ").replace(new RegExp("(\\p{P}|\\p{S}|\\d)", "gu"), " $1 ").replace(/ ['`’] (?=\S)/g, "ʼ").replace(/\(/g, "&#40");
     let splittedOrig, splitted;
-    splittedOrig = replaceWithDict(
-      replaceWithDict(text, letters[abc]),
-      lettersUpperCase[abc]
-    ).split(" ");
+    splittedOrig = convertAlphabet(text, abc).split(" ");
     text = this.taraskevize(text.toLowerCase());
     if (j)
       text = replaceIbyJ(text, j === REPLACE_J.ALWAYS);
@@ -2209,5 +2209,6 @@ export {
   Taraskevizer,
   VARIATION,
   __tarask__,
+  convertAlphabet,
   gobj
 };

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "taraskevizer",
-	"version": "5.1.4",
+	"version": "5.1.6",
 	"author": "GooseOb",
 	"repository": {
 		"type": "git",