Merge pull request #25 from amerharb/fix-medial-yeh

Fix medial yeh
Seen-Arabic · Jan 18, 2024 · addf8c4 · addf8c4
2 parents 9434361 + b5dfa9c
commit addf8c4
Show file tree

Hide file tree

Showing 5 changed files with 119 additions and 25 deletions.
diff --git a/src/constants/arabic-letters.ts b/src/constants/arabic-letters.ts
@@ -5,10 +5,10 @@ export const ARABIC_DOTLESS_DICT: { [key: string]: string } = {
 	آ: 'ا',
 	ٱ: 'ا',
 	ء: '',
-	ب: 'ٮ',
-	پ: 'ٮ',
-	ت: 'ٮ',
-	ث: 'ٮ',
+	ب: '\u066E', // Dotless Beh in all four contextual forms.
+	پ: '\u066E',
+	ت: '\u066E',
+	ث: '\u066E',
 	ج: 'ح',
 	چ: 'ح',
 	خ: 'ح',
@@ -33,14 +33,14 @@ export const ARABIC_DOTLESS_DICT: { [key: string]: string } = {
 	گ: 'ک',
 	ل: 'ل',
 	م: 'م',
-	ن: 'ں',
+	ن: '\u06BA', // Dotless Noon in all four contextual forms, however most fonts shows dots in initial and medial forms
 	ه: 'ه',
 	و: 'و',
 	ؤ: 'و',
 	ة: 'ه',
-	ى: 'ى',
-	ي: 'ى',
-	ئ: 'ى',
+	ى: '\u0649', // Dotless Yeh in all four contextual forms, however some fonts shows final form in all cases
+	ي: '\u0649',
+	ئ: '\u0649',
 };
 
 export const STANDARD_LETTERS: string[] = [

diff --git a/src/index.ts b/src/index.ts
@@ -1,9 +1,12 @@
 import * as constants from './constants';
 import * as script from './scripts';
 import * as utils from './utils';
+import { OldArabicOptions } from './options';
 
 export const ArabicServices = {
 	constants,
 	utils,
 	...script,
 };
+
+export type { OldArabicOptions };
diff --git a/src/options.ts b/src/options.ts
@@ -0,0 +1,16 @@
+export type OldArabicOptions = {
+	replaceMidNoonWithBah?: boolean;
+	replaceMidYahWithBah?: boolean;
+};
+
+const defaultOldArabicOptions: OldArabicOptions = {
+	replaceMidNoonWithBah: true,
+	replaceMidYahWithBah: false,
+};
+
+export function fillDefaultOptions(options: OldArabicOptions): OldArabicOptions {
+	return {
+		...defaultOldArabicOptions,
+		...options,
+	};
+}
diff --git a/src/scripts/scripts.ts b/src/scripts/scripts.ts
@@ -11,6 +11,7 @@ import {
 	YAA,
 } from '../constants/arabic-letters';
 import { setCharAt, similarityScore } from '../utils';
+import { fillDefaultOptions, type OldArabicOptions } from '../options';
 
 /**
  * Remove all tashkeel from text
@@ -30,32 +31,40 @@ export function removeTashkeel(text: string): string {
 /**
  * Remove all dots & tashkeel from text
  * @param sentence string to convert to old arabic
+ * @param option
  * @returns string in old arabic
  * @example
  *   Input: "الخَيْلُ وَاللّيْلُ وَالبَيْداءُ تَعرِفُني"
  *   Output: "الحىل واللىل والٮىدا ٮعرڡٮى"
  */
-export function toOldArabic(sentence: string): string {
+export function toOldArabic(sentence: string, option: OldArabicOptions = {}): string {
+	const { replaceMidNoonWithBah, replaceMidYahWithBah } = fillDefaultOptions(option);
 	sentence = removeTashkeel(sentence.trim());
 	let newSentence = '';
 	for (let letter = 0; letter < sentence.length; letter++) {
 		// if letter is not Arabic letter => append to newSentence
 		if (!ARABIC_DOTLESS_DICT.hasOwnProperty(sentence[letter])) {
 			newSentence += sentence[letter];
 		} else {
-			// letter is Arabic letter => replace it with its corresponding dotless letter
-			newSentence += ARABIC_DOTLESS_DICT[sentence[letter]];
-			// Handle 'ن' Issue
-			if (sentence[letter] == 'ن') {
-				const nextLetter = letter + 1;
-				// if 'ن' is not last character replace it with 'ب' corresponding dotless letter => 'ٮ'
-				if (nextLetter < sentence.length) {
-					let temp = newSentence.substring(0, newSentence.length - 1);
-					if (ARABIC_DOTLESS_DICT.hasOwnProperty(sentence[nextLetter]) || sentence[nextLetter] == 'ـ') {
-						temp += ARABIC_DOTLESS_DICT['ب'];
-						newSentence = temp;
-					}
-				}
+			if (
+				// Handle 'ن' Issue
+				replaceMidNoonWithBah &&
+				sentence[letter] == 'ن' &&
+				letter + 1 < sentence.length &&
+				(ARABIC_DOTLESS_DICT.hasOwnProperty(sentence[letter + 1]) || sentence[letter + 1] == 'ـ')
+			) {
+				newSentence += ARABIC_DOTLESS_DICT['ب'];
+			} else if (
+				// Handle 'ي' Issue
+				replaceMidYahWithBah &&
+				sentence[letter] == 'ي' &&
+				letter + 1 < sentence.length &&
+				(ARABIC_DOTLESS_DICT.hasOwnProperty(sentence[letter + 1]) || sentence[letter + 1] == 'ـ')
+			) {
+				newSentence += ARABIC_DOTLESS_DICT['ب'];
+			} else {
+				// if letter is Arabic letter => append corresponding dotless letter to newSentence
+				newSentence += ARABIC_DOTLESS_DICT[sentence[letter]];
 			}
 		}
 	}

diff --git a/tests/scripts.test.ts b/tests/scripts.test.ts
@@ -28,6 +28,14 @@ describe('#toOldArabic()', () => {
 			const expected = `ٮٮں`;
 			expect(actual).toBe(expected);
 		});
+		it('should convert ن to ں in all case', () => {
+			const text = `ننن`;
+			const actual = ArabicServices.toOldArabic(text, { replaceMidNoonWithBah: false });
+			const expected = `ںںں`;
+			const expectedUnicode = '\u06BA\u06BA\u06BA';
+			expect(actual).toBe(expected);
+			expect(actual).toBe(expectedUnicode);
+		});
 	});
 
 	describe('Test Tatweel with ن letter', () => {
@@ -37,6 +45,52 @@ describe('#toOldArabic()', () => {
 			const expected = `ٮــٮــں`;
 			expect(actual).toBe(expected);
 		});
+		it('should convert ن to ں last letter in all cases', () => {
+			const text = `نــنــن`;
+			const actual = ArabicServices.toOldArabic(text, { replaceMidNoonWithBah: false });
+			const expected = `ںــںــں`;
+			const expectedUnicode = '\u06BA\u0640\u0640\u06BA\u0640\u0640\u06BA';
+			expect(actual).toBe(expected);
+			expect(actual).toBe(expectedUnicode);
+		});
+	});
+
+	describe('Test ي letter', () => {
+		it('should convert ي to ى last letter, else convert it to ٮ', () => {
+			const text = `ييي`;
+			const actual = ArabicServices.toOldArabic(text, { replaceMidYahWithBah : true });
+			const expected = `ٮٮى`;
+			const expectedUnicode = '\u066E\u066E\u0649';
+			expect(actual).toBe(expected);
+			expect(actual).toBe(expectedUnicode);
+		});
+		it('should convert ي to ى in all case (default)', () => {
+			const text = `ييي`;
+			const actual = ArabicServices.toOldArabic(text, { replaceMidNoonWithBah: false });
+			const expected = `ىىى`;
+			const expectedUnicode = '\u0649\u0649\u0649';
+			expect(actual).toBe(expected);
+			expect(actual).toBe(expectedUnicode);
+		});
+	});
+
+	describe('Test Tatweel with ي letter', () => {
+		it('should convert ي to ى last letter, else convert it to ٮ', () => {
+			const text = `يــيــي`;
+			const actual = ArabicServices.toOldArabic(text, { replaceMidYahWithBah : true });
+			const expected = `ٮــٮــى`;
+			const expectedUnicode = '\u066E\u0640\u0640\u066E\u0640\u0640\u0649';
+			expect(actual).toBe(expected);
+			expect(actual).toBe(expectedUnicode);
+		});
+		it('should convert ي to ں last letter in all cases (default)', () => {
+			const text = `يــيــي`;
+			const actual = ArabicServices.toOldArabic(text);
+			const expected = `ىــىــى`;
+			const expectedUnicode = '\u0649\u0640\u0640\u0649\u0640\u0640\u0649';
+			expect(actual).toBe(expected);
+			expect(actual).toBe(expectedUnicode);
+		});
 	});
 
 	describe('Test with Poetry Text', () => {
@@ -46,13 +100,25 @@ describe('#toOldArabic()', () => {
 			const expected = `الحىـل واللىـل والٮىـدا ٮعرڡٮى والسىڡ والرمح والٯرطاس والٯلـم`;
 			expect(actual).toBe(expected);
 		});
+		it('should remove all dots & tashkeel from Poetry text, with option replace Yeh with dotless Bah', () => {
+			const text = `الخَيْـلُ وَاللّيْـلُ وَالبَيْـداءُ تَعرِفُني وَالسّيفُ وَالرّمحُ والقرْطاسُ وَالقَلَـمُ`;
+			const actual = ArabicServices.toOldArabic(text, { replaceMidYahWithBah: true });
+			const expected = `الحٮـل واللٮـل والٮٮـدا ٮعرڡٮى والسٮڡ والرمح والٯرطاس والٯلـم`;
+			expect(actual).toBe(expected);
+		});
 	});
 
 	describe('Test with Quran Text', () => {
-		it('should remove tashkeel from Quran text', () => {
+		it('should remove all dots & tashkeel from Quran text', () => {
 			const text = `وَقَالُواْ ٱلۡحَمۡدُ لِلَّهِ ٱلَّذِيٓ أَذۡهَبَ عَنَّا ٱلۡحَزَنَۖ إِنَّ رَبَّنَا لَغَفُورٞ شَكُورٌ`;
-			const actual = ArabicServices.removeTashkeel(text);
-			const expected = `وقالوا الحمد لله الذي أذهب عنا الحزن إن ربنا لغفور شكور`;
+			const actual = ArabicServices.toOldArabic(text);
+			const expected = `وٯالوا الحمد لله الدى ادهٮ عٮا الحرں اں رٮٮا لعڡور سکور`;
+			expect(actual).toBe(expected);
+		});
+		it('should remove all dots & tashkeel from Quran text', () => {
+			const text = `وَقَالُواْ ٱلۡحَمۡدُ لِلَّهِ ٱلَّذِيٓ أَذۡهَبَ عَنَّا ٱلۡحَزَنَۖ إِنَّ رَبَّنَا لَغَفُورٞ شَكُورٌ`;
+			const actual = ArabicServices.toOldArabic(text, { replaceMidNoonWithBah: false });
+			const expected = `وٯالوا الحمد لله الدى ادهٮ عںا الحرں اں رٮںا لعڡور سکور`;
 			expect(actual).toBe(expected);
 		});
 	});