feat(module): add support for korean language

sws2apps · Oct 4, 2024 · a007c88 · a007c88
1 parent e8b8401
commit a007c88
Show file tree

Hide file tree

Showing 9 changed files with 354 additions and 3 deletions.
diff --git a/.vscode/extensions.json b/.vscode/extensions.json
@@ -0,0 +1,5 @@
+{
+  "recommendations": [
+    "esbenp.prettier-vscode"
+  ]
+}
diff --git a/README.md b/README.md
@@ -118,7 +118,8 @@ French
 German
 Italian
 Japanese
-Madagascar Sign Language, Malagasy
+Korean
+Malagasy
 Polish, Portuguese Brazil
 Romanian, Russian
 Spanish

diff --git a/src/browser/utils.browser.ts b/src/browser/utils.browser.ts
@@ -7,6 +7,7 @@ import F from '../locales/fr-FR/text.json';
 import I from '../locales/it-IT/text.json';
 import J from '../locales/ja-JP/text.json';
 import K from '../locales/uk-UA/text.json';
+import KO from '../locales/ko-KR/text.json';
 import M from '../locales/ro-RO/text.json';
 import MG from '../locales/mg-MG/text.json';
 import P from '../locales/pl-PL/text.json';
@@ -30,6 +31,6 @@ declare global {
 }
 
 window.jw_epub_parser = {
-  languages: { CH, CHS, E, F, I, J, K, M, MG, P, S, T, TG, TK, TND, TNK, TPO, TTM, TW, U, VZ, X },
+  languages: { CH, CHS, E, F, I, J, K, KO, M, MG, P, S, T, TG, TK, TND, TNK, TPO, TTM, TW, U, VZ, X },
   path: path,
 };
diff --git a/src/common/date_parser.ts b/src/common/date_parser.ts
@@ -28,6 +28,12 @@ option2 = `(\\d{1,2})月(\\d{1,2})日`;
 option3 = `(?:\\d{4})年(\\d{1,2})月(\\d{1,2})日`;
 const mwbDatePatternJ = `${option1}|${option2}|${option3}`;
 
+// date like 1-) 11월 4-10일; or 2) 11월 25일–12월 1일; or 3) 2024년 12월 30일–2025년 1월 5일
+option1 = `(\\d{1,2})월 (\\d{1,2})[-–](?:\\d{1,2})일`;
+option2 = `(\\d{1,2})월 (\\d{1,2})일`;
+option3 = `(?:\\d{4})년 (\\d{1,2})월 (\\d{1,2})일`;
+const mwbDatePatternKO = `${option1}|${option2}|${option3}`;
+
 // date like 1-) 4-10 listopada; or 2) 25 listopada do 1 grudnia; or 3) 30 grudnia 2024 do 5 stycznia 2025
 option1 = `(\\d{1,2})[-–](?:\\d{1,2}) (${wordWithDiacritics})`;
 option2 = `(\\d{1,2}) (${wordWithDiacritics}) do (?:\\d{1,2}) (?:${wordWithDiacritics})`;
@@ -64,6 +70,7 @@ const mwbDatePatterns: LangRegExp = {
   CHS: new RegExp(mwbDatePatternJ, 'giu'),
   E: new RegExp(mwbDatePatternE, 'giu'),
   J: new RegExp(mwbDatePatternJ, 'giu'),
+  KO: new RegExp(mwbDatePatternKO, 'giu'),
   P: new RegExp(mwbDatePatternP, 'giu'),
   S: new RegExp(mwbDatePatternS, 'giu'),
   T: new RegExp(mwbDatePatternT, 'giu'),
@@ -116,6 +123,7 @@ const mwbDateParsing: MWBDateParsing = {
   CHS: mwbParsingE,
   E: mwbParsingE,
   J: mwbParsingE,
+  KO: mwbParsingE,
   TG: mwbParsingE,
   TW: mwbParsingE,
 };
@@ -178,6 +186,11 @@ option1 = `(\\d{4})年(?:nián)?(\\d{1,2})月(?:yuè)?(\\d{1,2})[-–～](\\d{1,
 option2 = `(\\d{4})年(?:nián)?(\\d{1,2})月(?:yuè)?(\\d{1,2})日`;
 const wDatePatternJ = `${option1}|${option2}`;
 
+// date like 1-) 2024년 10월 7-13일; or 2) 2024년 10월 28일–2024년 11월 3일
+option1 = `(\\d{4})년 (\\d{1,2})월 (\\d{1,2})[-–](\\d{1,2})일`;
+option2 = `(\\d{4})년 (\\d{1,2})월 (\\d{1,2})일`;
+const wDatePatternKO = `${option1}|${option2}`;
+
 // date like 1-) Artykuł do studium w tygodniu od 14 do 20 października 2024 roku; or 2) Artykuł do studium w tygodniu od 28 października do 3 listopada 2024 roku; or 3) Artykuł do studium w tygodniu od 30 grudnia 2024 roku do 5 stycznia 2025 roku.
 option1 = `(\\d{1,2}) do (?:\\d{1,2})? (${wordWithDiacritics}) (\\d{4})`;
 option2 = `(\\d{1,2}) (${wordWithDiacritics}) do (?:\\d{1,2}) (?:${wordWithDiacritics}) (\\d{4})`;
@@ -223,6 +236,7 @@ const wDatePatterns: LangRegExp = {
   CHS: new RegExp(wDatePatternJ, 'giu'),
   E: new RegExp(wDatePatternE, 'giu'),
   J: new RegExp(wDatePatternJ, 'giu'),
+  KO: new RegExp(wDatePatternKO, 'giu'),
   P: new RegExp(wDatePatternP, 'giu'),
   S: new RegExp(wDatePatternS, 'giu'),
   T: new RegExp(wDatePatternT, 'giu'),
@@ -314,6 +328,7 @@ const wDateParsing: WDateParsing = {
   CHS: wParsingJ,
   E: wParsingE,
   J: wParsingJ,
+  KO: wParsingJ,
   TG: wParsingE,
   TW: wParsingTW,
 };

diff --git a/src/locales/languages.ts b/src/locales/languages.ts
@@ -4,6 +4,7 @@ export default [
 	{ locale: 'it-IT', code: 'I' },
 	{ locale: 'ja-JP', code: 'J' },
 	{ locale: 'uk-UA', code: 'K' },
+	{ locale: 'ko-KR', code: 'KO' },
 	{ locale: 'ro-RO', code: 'M' },
 	{ locale: 'mg-MG', code: 'MG' },
 	{ locale: 'pl-PL', code: 'P' },

diff --git a/src/node/utils.node.ts b/src/node/utils.node.ts
@@ -8,6 +8,7 @@ import F from '../locales/fr-FR/text.json' assert { type: 'json' };
 import I from '../locales/it-IT/text.json' assert { type: 'json' };
 import J from '../locales/ja-JP/text.json' assert { type: 'json' };
 import K from '../locales/uk-UA/text.json' assert { type: 'json' };
+import KO from '../locales/ko-KR/text.json' assert { type: 'json' };
 import M from '../locales/ro-RO/text.json' assert { type: 'json' };
 import MG from '../locales/mg-MG/text.json' assert { type: 'json' };
 import P from '../locales/pl-PL/text.json' assert { type: 'json' };
@@ -29,7 +30,7 @@ declare global {
 }
 
 global.jw_epub_parser = {
-  languages: { CH, CHS, E, F, I, J, K, M, MG, P, S, T, TG, TK, TND, TNK, TPO, TTM, TW, U, VZ, X },
+  languages: { CH, CHS, E, F, I, J, K, KO, M, MG, P, S, T, TG, TK, TND, TNK, TPO, TTM, TW, U, VZ, X },
   path: path,
   readFile: readFile,
 };
diff --git a/test/enhancedParsing/list.json b/test/enhancedParsing/list.json
@@ -6,6 +6,7 @@
   { "language": "I", "issue": "202411" },
   { "language": "J", "issue": "202411" },
   { "language": "K", "issue": "202411" },
+  { "language": "KO", "issue": "202411" },
   { "language": "M", "issue": "202411" },
   { "language": "MG", "issue": "202411" },
   { "language": "S", "issue": "202411" },