feat(locales): add slovenian language

src/browser/utils.browser.ts

@@ -15,6 +15,8 @@ import O from '../locales/nl-NL/text.json';
 import P from '../locales/pl-PL/text.json';
 import PGW from '../locales/wes-PGW/text.json';
 import S from '../locales/es-ES/text.json';
+import SV from '../locales/sl-SI/text.json';
+import SW from '../locales/sw-KE/text.json';
 import T from '../locales/pt-BR/text.json';
 import TG from '../locales/tl-PH/text.json';
 import TK from '../locales/tr-TR/text.json';
@@ -35,6 +37,36 @@ declare global {
 }
 
 window.jw_epub_parser = {
-  languages: { CH, CHS, E, F, FI, I, J, K, KO, M, MG, O, P, PGW, S, T, TG, TK, TND, TNK, TPO, TTM, TW, U, VZ, X, Z },
+  languages: {
+    CH,
+    CHS,
+    E,
+    F,
+    FI,
+    I,
+    J,
+    K,
+    KO,
+    M,
+    MG,
+    O,
+    P,
+    PGW,
+    S,
+    SV,
+    SW,
+    T,
+    TG,
+    TK,
+    TND,
+    TNK,
+    TPO,
+    TTM,
+    TW,
+    U,
+    VZ,
+    X,
+    Z,
+  },
   path: path,
 };

src/common/date_parser.ts

@@ -58,9 +58,9 @@ option2 = `(\\d{1,2}) de (${wordWithDiacritics}) a (?:\\d{1,2}) de (?:${wordWith
 option3 = `(\\d{1,2}) de (${wordWithDiacritics}) de (\\d{4})`;
 const mwbDatePatternTPO = `${option1}|${option2}|${option3}`;
 
-// date like 1-) 4-10 de noviembre; or 2) 25 de noviembre a 1 de diciembre; or 3) 30 de diciembre de 2024 a 5 de enero de 2025
+// date like 1-) 4.-10. de noviembre; or 2) 25. noviembre - 1. de diciembre; or 3) 30. diciembre 2024-5. enero 2025
 option1 = `(\\d{1,2}).[-–](?:\\d{1,2}). (${wordWithDiacritics})`;
-option2 = `(\\d{1,2}). (${wordWithDiacritics}) [-–] (?:\\d{1,2}). (?:${wordWithDiacritics})`;
+option2 = `(\\d{1,2}). (${wordWithDiacritics}) ?[-–] ?(?:\\d{1,2}). (?:${wordWithDiacritics})`;
 option3 = `(\\d{1,2}). (${wordWithDiacritics}) (\\d{4})`;
 const mwbDatePatternX = `${option1}|${option2}|${option3}`;
 
@@ -75,6 +75,8 @@ const mwbDatePatterns: LangRegExp = {
   P: new RegExp(mwbDatePatternP, 'giu'),
   PGW: new RegExp(mwbDatePatternE, 'giu'),
   S: new RegExp(mwbDatePatternS, 'giu'),
+  SV: new RegExp(mwbDatePatternX, 'giu'),
+  SW: new RegExp(mwbDatePatternE, 'giu'),
   T: new RegExp(mwbDatePatternT, 'giu'),
   TG: new RegExp(mwbDatePatternE, 'giu'),
   TPO: new RegExp(mwbDatePatternTPO, 'giu'),
@@ -127,6 +129,7 @@ const mwbDateParsing: MWBDateParsing = {
   J: mwbParsingE,
   KO: mwbParsingE,
   PGW: mwbParsingE,
+  SW: mwbParsingE,
   TG: mwbParsingE,
   TW: mwbParsingE,
 };
@@ -181,8 +184,9 @@ const wDatePatternCommon = `${option1}|${option2}|${option3}`;
 
 // date like 1-) December 16-22, 2024; or 2) December 30, 2024-January 5, 2025
 option1 = `(${wordWithDiacritics}) (\\d{1,2})[-–](?:\\d{1,2})?, (\\d{4})`;
-option2 = `(${wordWithDiacritics}) (\\d{1,2}), (\\d{4})`;
-const wDatePatternE = `${option1}|${option2}`;
+option2 = `(${wordWithDiacritics}) (\\d{1,2})[-–](?:${wordWithDiacritics}) (?:\\d{1,2}), (\\d{4})`;
+option3 = `(${wordWithDiacritics}) (\\d{1,2}), (\\d{4})`;
+const wDatePatternE = `${option1}|${option2}|${option3}`;
 
 // date like 1-) 2024年12月16-22日; or 2) 2024年12月30日-2025年1月5日
 option1 = `(\\d{4})年(?:nián)?(\\d{1,2})月(?:yuè)?(\\d{1,2})[-–～](\\d{1,2})日`;
@@ -229,9 +233,10 @@ option3 = `(${wordWithDiacritics}) (\\d{1,2}), (\\d{4})`;
 const wDatePatternTW = `${option1}|${option2}|${option3}`;
 
 // date like 1-) 40. Studienartikel: 9. bis 15. Dezember 2024; or 2) 43. Studienartikel: 30. Dezember 2024 bis 5. Januar 2025
-option1 = `(\\d{1,2}). bis (?:\\d{1,2}).? (${wordWithDiacritics}) (\\d{4})`;
-option2 = `(\\d{1,2}). (${wordWithDiacritics}) (\\d{4})`;
-const wDatePatternX = `${option1}|${option2}`;
+option1 = `(\\d{1,2}).(?:${dateRangeSeparator})(?:\\d{1,2}).? (${wordWithDiacritics}) (\\d{4})`;
+option2 = `(\\d{1,2}). (${wordWithDiacritics})(?:${dateRangeSeparator})(?:\\d{1,2}).? (?:${wordWithDiacritics}) (\\d{4})`;
+option3 = `(\\d{1,2}). (${wordWithDiacritics}) (\\d{4})`;
+const wDatePatternX = `${option1}|${option2}|${option3}`;
 
 // date like 1-) Tutkitaan 3.–9.2.2025 ; or 2) Tutkitaan 24.2.–2.3.2025. ; or 3) Tutkitaan 30.12.2024–5.1.2025.
 option1 = '(\\d{1,2}).[–](?:\\d{1,2}).(\\d{1,2}).(\\d{4})';
@@ -250,6 +255,8 @@ const wDatePatterns: LangRegExp = {
   P: new RegExp(wDatePatternP, 'giu'),
   PGW: new RegExp(wDatePatternE, 'giu'),
   S: new RegExp(wDatePatternS, 'giu'),
+  SV: new RegExp(wDatePatternX, 'giu'),
+  SW: new RegExp(wDatePatternE, 'giu'),
   T: new RegExp(wDatePatternT, 'giu'),
   TPO: new RegExp(wDatePatternTPO, 'giu'),
   TG: new RegExp(wDatePatternE, 'giu'),
@@ -288,10 +295,14 @@ const wParsingE = (groups: string[]): WDateParsingResult => {
     month = groups[1];
     date = groups[2];
     year = groups[3];
-  } else {
+  } else if (groups[4]) {
     month = groups[4];
     date = groups[5];
     year = groups[6];
+  } else {
+    month = groups[7];
+    date = groups[8];
+    year = groups[9];
   }
 
   return [year, month, date];
@@ -341,6 +352,7 @@ const wDateParsing: WDateParsing = {
   J: wParsingJ,
   KO: wParsingJ,
   PGW: wParsingE,
+  SW: wParsingE,
   TG: wParsingE,
   TW: wParsingTW,
 };

src/common/parsing_rules.ts

@@ -29,6 +29,7 @@ export const extractSourceEnhanced = (src: string, lang: string) => {
 
   const firstPattern: LangRegExp = {
     common: firstPatternCommon,
+    SW: firstPatternTW,
     TW: firstPatternTW,
   };
 

src/locales/languages.ts

@@ -26,4 +26,6 @@ export default [
   { locale: 'sv-SE', code: 'Z' },
   { locale: 'nl-NL', code: 'O' },
   { locale: 'wes-PGW', code: 'PGW' },
+  { locale: 'sl-SI', code: 'SV' },
+  { locale: 'sw-KE', code: 'SW' },
 ];

src/locales/sl-SI/text.json

@@ -1,15 +1,15 @@
 {
-	"januaryVariations": "januar",
-	"februaryVariations": "februar",
-	"marchVariations": "marec",
-	"aprilVariations": "april",
-	"mayVariations": "maj",
-	"juneVariations": "junij",
-	"julyVariations": "julij",
-	"augustVariations": "avgust",
-	"septemberVariations": "september",
-	"octoberVariations": "oktober",
-	"novemberVariations": "november",
-	"decemberVariations": "december",
-	"partMinutesSeparatorVariations": "min."
+  "januaryVariations": "januar|januarja",
+  "februaryVariations": "februar|februarja",
+  "marchVariations": "marec|marca",
+  "aprilVariations": "april|aprila",
+  "mayVariations": "maj|maja",
+  "juneVariations": "junij|junija",
+  "julyVariations": "julij|julija",
+  "augustVariations": "avgust|avgusta",
+  "septemberVariations": "september|septembra",
+  "octoberVariations": "oktober|oktobra",
+  "novemberVariations": "november|novembra",
+  "decemberVariations": "december|decembra",
+  "partMinutesSeparatorVariations": "min."
 }

src/node/utils.node.ts

@@ -16,6 +16,8 @@ import O from '../locales/nl-NL/text.json' assert { type: 'json' };
 import P from '../locales/pl-PL/text.json' assert { type: 'json' };
 import PGW from '../locales/wes-PGW/text.json' assert { type: 'json' };
 import S from '../locales/es-ES/text.json' assert { type: 'json' };
+import SV from '../locales/sl-SI/text.json' assert { type: 'json' };
+import SW from '../locales/sw-KE/text.json' assert { type: 'json' };
 import T from '../locales/pt-BR/text.json' assert { type: 'json' };
 import TG from '../locales/tl-PH/text.json' assert { type: 'json' };
 import TK from '../locales/tr-TR/text.json' assert { type: 'json' };
@@ -34,7 +36,37 @@ declare global {
 }
 
 global.jw_epub_parser = {
-  languages: { CH, CHS, E, F, FI, I, J, K, KO, M, MG, O, P, PGW, S, T, TG, TK, TND, TNK, TPO, TTM, TW, U, VZ, X, Z },
+  languages: {
+    CH,
+    CHS,
+    E,
+    F,
+    FI,
+    I,
+    J,
+    K,
+    KO,
+    M,
+    MG,
+    O,
+    P,
+    PGW,
+    S,
+    SV,
+    SW,
+    T,
+    TG,
+    TK,
+    TND,
+    TNK,
+    TPO,
+    TTM,
+    TW,
+    U,
+    VZ,
+    X,
+    Z,
+  },
   path: path,
   readFile: readFile,
 };

test/enhancedParsing/list.json

@@ -19,5 +19,8 @@
   { "language": "VZ", "issue": "202411" },
   { "language": "X", "issue": "202411" },
   { "language": "O", "issue": "202411" },
-  { "language": "PGW", "issue": "202411" }
+  { "language": "Z", "issue": "202411" },
+  { "language": "PGW", "issue": "202411" },
+  { "language": "SV", "issue": "202411" },
+  { "language": "SW", "issue": "202411" }
 ]