taraskevizer 8.0.11 → 8.0.13

package/dist/bin.js

@@ -102,6 +102,12 @@ const optionDict = toHashTable([
             mode = 'abcOnly';
         },
     ],
+    [
+        ['--phonetic', '-ph'],
+        () => {
+            mode = 'phonetic';
+        },
+    ],
 ]);
 let currOption;
 process.argv.reverse();

package/dist/dict/iawords.d.ts

@@ -0,0 +1,2 @@
+import type { RawDict } from './types';
+export declare const iaWords: RawDict;

package/dist/dict/iawords.js

@@ -0,0 +1,97 @@
+import { iwords } from './iwords.js';
+import { toOneLine } from './lib.js';
+const ia = (word, words) => ` ${word}(?= \\(?(?:[бвгджзйклмнпстфцчшў]*[оё]|${words.replace(/\(/g, '(?:')}|i(?:${iwords})))`;
+export const iaWords = [
+    
+    [
+        ia('не', toOneLine(`(вы)?(бач[ауы]
+бег[ла]
+блыта[юц]
+брал[аі]
+буд(зе|у(ць|чы)?) 
+ваш
+веда(л[аі]|ць|ю(ць|чы)?|ў)
+дума(л[аі]|ць|ю(ць|чы)?|ў)
+зна(л[аі]|ць|ю(ць|чы)?)
+кла(д |л|ў)
+ма(е|ючы|ю(ць|чы)?)
+ме(л|ў|ць)
+мы(ц|л[аі])
+пі(са|шу)(цц|чы|ць )?
+ў?пэўн
+ска(ж(а|уць)|за )
+спал[аі]
+трэба 
+чу[елўцю])
+веды
+веліч
+выраш
+вы[дп]ал[еіяю]
+выпад
+гор(ш|ай)
+гу(к|чн|ст)
+дрэнн
+іхн
+карт\\S{0,4}[ \\)]
+каш(ай?|амі|у|ы) 
+лёгк
+літар
+лішн
+(мен|бол|леп)(ш|ей)
+мап\\S{0,4}[ \\)]
+медз
+мякк
+наш([аы]я|ую|ых)
+прыйдзеш
+руш[аы]
+ста(ў|не|нуць|л[аі])
+таго
+тое
+тыя
+ўвод
+ўс[еёя]
+фарб(а(ў|мі)?|у|ы)? 
+ян `)),
+        ' ня',
+    ],
+    
+    [
+        ia('без', toOneLine(`(вы)?(клад[ау]
+мела 
+мытых
+піса[нц]
+ска(жа|жуць|за) 
+спал[аі]
+чу[елўцю])
+ваш
+ведаў
+велічы
+выраш
+вы[дп]але
+выпад
+гор(ш|ай)
+гу(к|чн|ст)
+дрэнн
+іхн
+карт\\S{0,4}[ \\)]
+каш(ы|аў) 
+лёгк
+літар
+лішн
+медз
+(мен|бол|леп|ш|ей)
+мап\\S{0,4}[ \\)]
+мякк
+наш(ую|ых)
+пішучы
+руша
+стаў
+таго
+ўвод
+ўс[еёя]
+фарб(аў|ы)? 
+цукр[ау] 
+ян`)),
+        ' бяз',
+    ],
+];

package/dist/dict/wordlist.js

@@ -1,5 +1,5 @@
-import { iwords } from './iwords.js';
-import { dictFrom, toOneLine } from './lib.js';
+import { iaWords } from './iawords.js';
+import { dictFrom } from './lib.js';
 const chemicalElements1 = [
     'сканд|ванад|рубід|род|ірыд|рэзэрфорд',
     'стронц|бар|цэр|лютэц|самар| тор|амэрыц|кальц|кюр|дармштат|лівэрмор|натр'
@@ -11,7 +11,6 @@ const chemicalElements2 = [
 ];
 chemicalElements2[1] = chemicalElements2[0] + '|айнштайн|мендзялев|сыборг|гас|флеров';
 const chemicalElements3 = ' гал|бэрыл|тул|бэркл|набэл';
-const ia = (word, words) => ` ${word}(?= \\(?(?:[бвгджзйклмнпстфцчшў]*[оё]|${words.replace(/\(/g, '(?:')}|i(?:${iwords})))`;
 const gwords = [
     ' ґб ',
     'ґг',
@@ -1415,92 +1414,7 @@ export const wordlist = dictFrom.raw([
     [/ фасці(?=[вў])/, ' фа(сьці|ста)'],
     
     [/ зеленалужск/, ' (зеленалуск|зялёналуск)'],
-    
-    [ia('не', toOneLine(`(вы)?(бач[ауы]
-бег[ла]
-блыта[юц]
-брал[аі]
-буд(зе|у(ць|чы)?) 
-ваш
-веда(л[аі]|ць|ю(ць|чы)?|ў)
-дума(л[аі]|ць|ю(ць|чы)?|ў)
-зна(л[аі]|ць|ю(ць|чы)?)
-кла(д |л|ў)
-ма(е|ючы|ю(ць|чы)?)
-ме(л|ў|ць)
-мы(ц|л[аі])
-пі(са|шу)(цц|чы|ць )?
-ў?пэўн
-ска(ж(а|уць)|за )
-спал[аі]
-трэба 
-чу[елўцю])
-веды
-веліч
-выраш
-вы[дп]ал[еіяю]
-выпад
-гор(ш|ай)
-гу(к|чн|ст)
-дрэнн
-іхн
-карт\\S{0,4}[ \\)]
-каш(ай?|амі|у|ы) 
-лёгк
-літар
-лішн
-(мен|бол|леп)(ш|ей)
-мап\\S{0,4}[ \\)]
-медз
-мякк
-наш([аы]я|ую|ых)
-прыйдзеш
-руш[аы]
-ста(ў|не|нуць|л[аі])
-таго
-тое
-тыя
-ўвод
-ўс[еёя]
-фарб(а(ў|мі)?|у|ы)? 
-ян `)), ' ня'],
-    
-    [ia('без', toOneLine(`(вы)?(клад[ау]
-мела 
-мытых
-піса[нц]
-ска(жа|жуць|за) 
-спал[аі]
-чу[елўцю])
-ваш
-ведаў
-велічы
-выраш
-вы[дп]але
-выпад
-гор(ш|ай)
-гу(к|чн|ст)
-дрэнн
-іхн
-карт\\S{0,4}[ \\)]
-каш(ы|аў) 
-лёгк
-літар
-лішн
-медз
-(мен|бол|леп|ш|ей)
-мап\\S{0,4}[ \\)]
-мякк
-наш(ую|ых)
-пішучы
-руша
-стаў
-таго
-ўвод
-ўс[еёя]
-фарб(аў|ы)? 
-цукр[ау] 
-ян`)), ' бяз'],
+    ...iaWords,
     
     [/(\S)([іы]) (аа|ба|генэ|кры|тэ)зіс /, '$1($2|ая) $3(зіс|за) '],
     [/ага (аа|ба|генэ|кры|тэ)зіс[ау] /, '(ага|ай) $1(зісу|зы) '],

package/dist/pipelines.d.ts

@@ -43,6 +43,7 @@ export declare const plainText: (TaraskStep<import("./steps/types").SplittedText
 export declare const html: (TaraskStep<import("./steps/types").SplittedTextStorage> | TaraskStep<import("./steps/resolve-syntax").SpecialSyntaxStorage> | TaraskStep<import("./steps/whitespaces").WhiteSpaceStorage>)[];
 /**
  * Pipeline for phonetizing.
+ * @alpha
  */
-export declare const phonetic: (TaraskStep<import("./steps/resolve-syntax").SpecialSyntaxStorage> | TaraskStep<import("./steps/whitespaces").WhiteSpaceStorage>)[];
+export declare const phonetic: (TaraskStep<import("./steps/types").SplittedTextStorage> | TaraskStep<import("./steps/resolve-syntax").SpecialSyntaxStorage> | TaraskStep<import("./steps/whitespaces").WhiteSpaceStorage>)[];
 export {};

package/dist/pipelines.js

@@ -1,5 +1,5 @@
 
-import { applyVariationsHtml, applyVariationsNonHtml, applyGHtml, applyGNonHtml, highlightDiffStep, highlightDiffStepNonHtml, applyNoFix, convertAlphabet, convertAlphabetLowerCase, joinSplittedText, prepare, replaceIbyJ, resolveSpecialSyntax, restoreCaseStep, restoreWhitespaces, storeSplittedAbcConvertedOrig, storeSplittedText, taraskevize, whitespacesToSpaces, trim, finalize, toLowerCase, } from './steps/index.js';
+import { applyVariationsHtml, applyVariationsNonHtml, applyGHtml, applyGNonHtml, highlightDiffStep, highlightDiffStepNonHtml, applyNoFix, convertAlphabet, convertAlphabetLowerCase, joinSplittedText, prepare, replaceIbyJ, resolveSpecialSyntax, restoreCaseStep, restoreWhitespaces, storeSplittedAbcConvertedOrig, storeSplittedText, taraskevize, phonetize, whitespacesToSpaces, trim, finalize, toLowerCase, iotacizeJi, } from './steps/index.js';
 import { htmlWrappers } from './lib/wrappers.js';
 const resolveSpecialSyntaxWithLAB = resolveSpecialSyntax('<');
 const finalizeWithNewLine = finalize('\n');
@@ -46,13 +46,4 @@ export const plainText = createPipeline(resolveSpecialSyntaxWithLAB, applyGNonHt
 
 export const html = createPipeline(resolveSpecialSyntax('&lt;'), applyGHtml, applyVariationsHtml, finalize('<br>'), highlightDiffStep(htmlWrappers.fix));
 
-export const phonetic = [
-    trim,
-    resolveSpecialSyntaxWithLAB,
-    prepare,
-    whitespacesToSpaces,
-    convertAlphabet,
-    restoreWhitespaces,
-    applyNoFix,
-    finalizeWithNewLine,
-];
+export const phonetic = plainText.flatMap((item) => item === taraskevize ? [phonetize, iotacizeJi] : item);

package/dist/steps/index.d.ts

@@ -22,6 +22,7 @@ export * from './restore-case';
 export * from './store-splitted-abc-converted-orig';
 export * from './store-splitted-text';
 export * from './taraskevize';
+export * from './phonetize';
 export * from './whitespaces';
 export * from './trim';
 export * from './finalize';

package/dist/steps/index.js

@@ -11,6 +11,7 @@ export * from './restore-case.js';
 export * from './store-splitted-abc-converted-orig.js';
 export * from './store-splitted-text.js';
 export * from './taraskevize.js';
+export * from './phonetize.js';
 export * from './whitespaces.js';
 export * from './trim.js';
 export * from './finalize.js';

package/dist/steps/phonetize.d.ts

@@ -0,0 +1,2 @@
+export declare const phonetize: import("./types").TaraskStep<{}>;
+export declare const iotacizeJi: import("./types").TaraskStep<{}>;

package/dist/steps/phonetize.js

@@ -0,0 +1,21 @@
+import { iaWords } from '../dict/iawords.js';
+import { noSoften, softeners } from '../dict/index.js';
+import { iwords } from '../dict/iwords.js';
+import { afterTarask, dictFrom, mutatingStep, replaceWithDict, } from '../lib/index.js';
+export const phonetize = mutatingStep(({ text }) => {
+    text = replaceWithDict(text, noSoften.concat(iaWords));
+    softening: do {
+        text = replaceWithDict(text, softeners);
+        for (const [pattern, result] of softeners)
+            if (result !== '$1дзьдз' && pattern.test(text))
+                continue softening;
+        break;
+    } while (true);
+    return replaceWithDict(text.replace(/\ue0ff/g, '').replace(/не пра/g, 'не пра'), afterTarask).replace(/не пра/g, 'не пра');
+});
+const idict = dictFrom.raw([
+    [/([аеёіоуыэюя] )і ў/, '$1й у'],
+    [/([аеёіоуыэюя] )і /, '$1й '],
+    [` і(?=${iwords})`, ' йі'],
+]);
+export const iotacizeJi = mutatingStep(({ text }) => replaceWithDict(text, idict));

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "taraskevizer",
-	"version": "8.0.11",
+	"version": "8.0.13",
 	"author": "GooseOb",
 	"repository": {
 		"type": "git",