feat(locales): support haitian creole

README.md

@@ -114,9 +114,10 @@ Currently, we only support enhanced parsing for the following languages:
 ```bash
 Chinese Mandarin (Simplified), Chinese Mandarin (Traditional)
 Dutch
-Enlish
+English
 French
 German
+Haitian Creole
 Italian
 Japanese
 Korean

crowdin.yml

@@ -1,4 +1,4 @@
-'pull_request_title': 'feat(localize): updated translation from Crowdin'
+'pull_request_title': 'fix(locales): updated translation from Crowdin'
 'pull_request_labels': ['crowdin']
 
 files:

src/browser/utils.browser.ts

@@ -2,6 +2,7 @@ import * as path from 'path-browserify';
 
 import CH from '../locales/cmn-Hant/text.json';
 import CHS from '../locales/ch-CHS/text.json';
+import CR from '../locales/ht-HT/text.json';
 import E from '../locales/en/text.json';
 import F from '../locales/fr-FR/text.json';
 import FI from '../locales/fi-FI/text.json';
@@ -15,6 +16,7 @@ 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 ST from '../locales/et-EE/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';
@@ -40,6 +42,7 @@ window.jw_epub_parser = {
   languages: {
     CH,
     CHS,
+    CR,
     E,
     F,
     FI,
@@ -53,6 +56,7 @@ window.jw_epub_parser = {
     P,
     PGW,
     S,
+    ST,
     SV,
     SW,
     T,

src/common/date_parser.ts

@@ -1,6 +1,7 @@
 import { JWEPUBParserError } from '../classes/error.js';
 import { WDateParsing, WDateParsingResult, LangRegExp, MWBDateParsingResult, MWBDateParsing } from '../types/index.js';
 import { getMonthNames } from './language_rules.js';
+import overrides from './override.js';
 
 const dateRangeSeparator = `\\s? bis |[-–—]| do | — | – \\s?`;
 const wordWithDiacritics = `\\p{L}+|\\p{L}+\\p{M}*`;
@@ -75,6 +76,7 @@ const mwbDatePatterns: LangRegExp = {
   P: new RegExp(mwbDatePatternP, 'giu'),
   PGW: new RegExp(mwbDatePatternE, 'giu'),
   S: new RegExp(mwbDatePatternS, 'giu'),
+  ST: new RegExp(mwbDatePatternX, 'giu'),
   SV: new RegExp(mwbDatePatternX, 'giu'),
   SW: new RegExp(mwbDatePatternE, 'giu'),
   T: new RegExp(mwbDatePatternT, 'giu'),
@@ -255,6 +257,7 @@ const wDatePatterns: LangRegExp = {
   P: new RegExp(wDatePatternP, 'giu'),
   PGW: new RegExp(wDatePatternE, 'giu'),
   S: new RegExp(wDatePatternS, 'giu'),
+  ST: new RegExp(wDatePatternX, 'giu'),
   SV: new RegExp(wDatePatternX, 'giu'),
   SW: new RegExp(wDatePatternE, 'giu'),
   T: new RegExp(wDatePatternT, 'giu'),
@@ -340,22 +343,34 @@ const wDateParsing: WDateParsing = {
 // #endregion
 
 export const extractWTStudyDate = (src: string, lang: string) => {
-  const srcClean = src
+  src = src
     .trim()
     .replace('  ', ' ')
     .replace('​', '')
     .replace('⁠', '')
     .replace(/\u200F/g, '');
 
+  let finalSrc = src;
+
+  const overrideLang = overrides[lang];
+
+  if (overrideLang) {
+    const overrideSrc = overrideLang[src];
+
+    if (overrideSrc) {
+      finalSrc = overrideSrc;
+    }
+  }
+
   const datePattern = wDatePatterns[lang] || wDatePatterns.common;
 
-  const match = srcClean.match(datePattern);
+  const match = finalSrc.match(datePattern);
 
   if (!match) {
-    throw new JWEPUBParserError('wtstudy', `Parsing failed for Watchtower Study Date. The input was: ${src}`);
+    throw new JWEPUBParserError('wtstudy', `Parsing failed for Watchtower Study Date. The input was: ${finalSrc}`);
   }
 
-  const groups = Array.from(datePattern.exec(srcClean)!);
+  const groups = Array.from(datePattern.exec(finalSrc)!);
 
   const parseDataFunc = wDateParsing[lang] || wDateParsing.common;
 

src/common/override.ts

@@ -0,0 +1,13 @@
+import { Override } from '../types';
+
+const overrides: Override = {
+  PGW: {
+    'Fine-Fine Things Wey You See for Bible': 'Fine-Fine Things Wey You See for Bible: (10 min.)',
+  },
+  ST: {
+    'Seda artiklit uuritakse vahemikus 28. oktoobrist 3.novembrini 2024.':
+      'Seda artiklit uuritakse vahemikus 28. oktoobrist 3. novembrini 2024.',
+  },
+};
+
+export default overrides;

src/common/parsing_rules.ts

@@ -1,6 +1,7 @@
 import { JWEPUBParserError } from '../classes/error.js';
 import { LangRegExp } from '../types/index.js';
 import { getPartMinutesSeparatorVariations } from './language_rules.js';
+import overrides from './override.js';
 
 export const extractSongNumber = (src: string) => {
   const parseNum = src.match(/(\d+)/);
@@ -19,8 +20,16 @@ export const extractSongNumber = (src: string) => {
 export const extractSourceEnhanced = (src: string, lang: string) => {
   const variations = getPartMinutesSeparatorVariations(lang);
 
-  if (src === 'Fine-Fine Things Wey You See for Bible' && lang === 'PGW') {
-    src = 'Fine-Fine Things Wey You See for Bible: (10 min.)';
+  let finalSrc = src;
+
+  const overrideLang = overrides[lang];
+
+  if (overrideLang) {
+    const overrideSrc = overrideLang[src];
+
+    if (overrideSrc) {
+      finalSrc = overrideSrc;
+    }
   }
 
   // separate minutes from title
@@ -45,13 +54,13 @@ export const extractSourceEnhanced = (src: string, lang: string) => {
 
   const langPattern = firstPattern[lang] || firstPattern.common;
 
-  const matchFirstPattern = src.match(langPattern);
+  const matchFirstPattern = finalSrc.match(langPattern);
 
   if (!matchFirstPattern) {
-    throw new JWEPUBParserError('jw-epub-parser', `Parsing failed. The input was: ${src}`);
+    throw new JWEPUBParserError('jw-epub-parser', `Parsing failed. The input was: ${finalSrc}`);
   }
 
-  const groupsFirstPattern = Array.from(langPattern.exec(src)!);
+  const groupsFirstPattern = Array.from(langPattern.exec(finalSrc)!);
 
   const fulltitle = groupsFirstPattern.at(1)!.trim();
   const time = +groupsFirstPattern.at(2)!.trim();

src/locales/et-EE/text.json

@@ -1,15 +1,15 @@
 {
-	"januaryVariations": "Jaanuar|jaanuarini|jaanuarist",
-	"februaryVariations": "Veebruar|veebruarini|veebruarist",
-	"marchVariations": "Märts|märtsini|märtsist",
-	"aprilVariations": "Aprill|aprillini|aprillist",
-	"mayVariations": "Mai|maini|maist",
-	"juneVariations": "Juuni|juunini",
-	"julyVariations": "Juuli|juulini|juulist",
-	"augustVariations": "August|augustini|augustist",
-	"septemberVariations": "September|septembrini|septembrist",
-	"octoberVariations": "Oktoober|oktoobrini|oktoobrist",
-	"novemberVariations": "November|novembrini|novembrist",
-	"decemberVariations": "Detsember|detsembrini|detsembrist",
-	"partMinutesSeparatorVariations": "min"
+  "januaryVariations": "Jaanuar|jaanuarini|jaanuarist",
+  "februaryVariations": "Veebruar|veebruarini|veebruarist",
+  "marchVariations": "Märts|märtsini|märtsist",
+  "aprilVariations": "Aprill|aprillini|aprillist",
+  "mayVariations": "Mai|maini|maist",
+  "juneVariations": "Juuni|juunini",
+  "julyVariations": "Juuli|juulini|juulist",
+  "augustVariations": "August|augustini|augustist",
+  "septemberVariations": "September|septembrini|septembrist",
+  "octoberVariations": "Oktoober|oktoobrini|oktoobrist",
+  "novemberVariations": "November|novembrini|novembrist",
+  "decemberVariations": "Detsember|detsembrini|detsembrist",
+  "partMinutesSeparatorVariations": "min"
 }

src/locales/languages.ts

@@ -28,4 +28,6 @@ export default [
   { locale: 'wes-PGW', code: 'PGW' },
   { locale: 'sl-SI', code: 'SV' },
   { locale: 'sw-KE', code: 'SW' },
+  { locale: 'et-EE', code: 'ST' },
+  { locale: 'ht-HT', code: 'CR' },
 ];

src/node/utils.node.ts

@@ -3,6 +3,7 @@ import { readFile } from 'fs/promises';
 
 import CH from '../locales/cmn-Hant/text.json' assert { type: 'json' };
 import CHS from '../locales/ch-CHS/text.json' assert { type: 'json' };
+import CR from '../locales/ht-HT/text.json' assert { type: 'json' };
 import E from '../locales/en/text.json' assert { type: 'json' };
 import F from '../locales/fr-FR/text.json' assert { type: 'json' };
 import FI from '../locales/fi-FI/text.json' assert { type: 'json' };
@@ -16,6 +17,7 @@ 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 ST from '../locales/et-EE/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' };
@@ -39,6 +41,7 @@ global.jw_epub_parser = {
   languages: {
     CH,
     CHS,
+    CR,
     E,
     F,
     FI,
@@ -52,6 +55,7 @@ global.jw_epub_parser = {
     P,
     PGW,
     S,
+    ST,
     SV,
     SW,
     T,

src/types/index.ts

@@ -81,3 +81,9 @@ export type MWBDateParsingResult = [month: string, date: string];
 export type MWBDateParsing = {
   [lang: string]: (groups: string[]) => MWBDateParsingResult;
 };
+
+export type Override = {
+  [language: string]: {
+    [src: string]: string;
+  };
+};

test/enhancedParsing/list.json

@@ -22,5 +22,6 @@
   { "language": "Z", "issue": "202411" },
   { "language": "PGW", "issue": "202411" },
   { "language": "SV", "issue": "202411" },
-  { "language": "SW", "issue": "202411" }
+  { "language": "SW", "issue": "202411" },
+  { "language": "CR", "issue": "202411" }
 ]