added word to letters method

Seen-Arabic · Nov 26, 2023 · 20bc9b6 · 20bc9b6
1 parent f43e5b9
commit 20bc9b6
Show file tree

Hide file tree

Showing 3 changed files with 95 additions and 0 deletions.
diff --git a/src/main/java/io/github/seen_arabic/arabic_services/ArabicServices.java b/src/main/java/io/github/seen_arabic/arabic_services/ArabicServices.java
@@ -88,6 +88,31 @@ public static String tashfeer(String text) {
         return newSentence.toString().trim();
     }
 
+    public static String wordToLetters(String word) {
+        StringBuilder newWord = new StringBuilder();
+
+        // Loop through each character in the input word
+        for (int i = 0; i < word.length(); i++) {
+            String letter = word.charAt(i) + "";
+
+            // Check if the current letter has a pronunciation in PRONOUNCED_LETTERS
+            if (Data.PRONOUNCED_LETTERS.containsKey(letter)) {
+                newWord.append(Data.PRONOUNCED_LETTERS.get(letter));
+
+                // Add a space after the pronounced letter unless it's the last letter in the
+                // word
+                if (i != word.length() - 1) {
+                    newWord.append(' ');
+                }
+            } else {
+                // If the letter is not in PRONOUNCED_LETTERS, keep it unchanged
+                newWord.append(letter);
+            }
+        }
+
+        return newWord.toString().trim();
+    }
+
     private static String handleNoonIssue(String text) {
         String arabicLetters = String.join("", Data.LETTERS_DICT.keySet()) + "ـ";
         String regex = Data.NOON + "(" + "?=[^" + arabicLetters + "]" + ")|" + Data.NOON + "\\z";

diff --git a/src/main/java/io/github/seen_arabic/arabic_services/Data.java b/src/main/java/io/github/seen_arabic/arabic_services/Data.java
@@ -15,6 +15,7 @@ class Data {
 
     static final Map<String, String> LETTERS_DICT;
     static final Map<Character, String[]> LETTERS_TASHFEER_REPLACEMENT_DICT;
+    static final Map<String, String> PRONOUNCED_LETTERS;
 
     static final String[] TASHKEEL = {
             "\u0600",
@@ -193,6 +194,43 @@ class Data {
         LETTERS_TASHFEER_REPLACEMENT_DICT.put('و', new String[] { "ۅ", "ۆ", "ۇ", "ۈ", "ۏ", "ۉ", "ۋ" });
         LETTERS_TASHFEER_REPLACEMENT_DICT.put('ي', new String[] { "ۍ", "ێ", "ې", "ے", "ی۪" });
 
+        PRONOUNCED_LETTERS = new HashMap<>();
+        PRONOUNCED_LETTERS.put("ا", "ألف");
+        PRONOUNCED_LETTERS.put("إ", "ألف_مكسورة");
+        PRONOUNCED_LETTERS.put("أ", "ألف");
+        PRONOUNCED_LETTERS.put("آ", "ألف_مد");
+        PRONOUNCED_LETTERS.put("ء", "همزة"); // Not in STANDARD_LETTERS[]
+        PRONOUNCED_LETTERS.put("ب", "باء");
+        PRONOUNCED_LETTERS.put("ت", "تاء");
+        PRONOUNCED_LETTERS.put("ث", "ثاء");
+        PRONOUNCED_LETTERS.put("ج", "جيم");
+        PRONOUNCED_LETTERS.put("ح", "حاء");
+        PRONOUNCED_LETTERS.put("خ", "خاء");
+        PRONOUNCED_LETTERS.put("د", "دال");
+        PRONOUNCED_LETTERS.put("ذ", "ذال");
+        PRONOUNCED_LETTERS.put("ر", "راء");
+        PRONOUNCED_LETTERS.put("ز", "زاي");
+        PRONOUNCED_LETTERS.put("س", "سين");
+        PRONOUNCED_LETTERS.put("ش", "شين");
+        PRONOUNCED_LETTERS.put("ص", "صاد");
+        PRONOUNCED_LETTERS.put("ض", "ضاد");
+        PRONOUNCED_LETTERS.put("ط", "طاء");
+        PRONOUNCED_LETTERS.put("ظ", "ظاء");
+        PRONOUNCED_LETTERS.put("ع", "عين");
+        PRONOUNCED_LETTERS.put("غ", "غين");
+        PRONOUNCED_LETTERS.put("ف", "فاء");
+        PRONOUNCED_LETTERS.put("ق", "قاف");
+        PRONOUNCED_LETTERS.put("ك", "كاف");
+        PRONOUNCED_LETTERS.put("ل", "لام");
+        PRONOUNCED_LETTERS.put("م", "ميم");
+        PRONOUNCED_LETTERS.put("ن", "نون");
+        PRONOUNCED_LETTERS.put("ه", "هاء");
+        PRONOUNCED_LETTERS.put("و", "واو");
+        PRONOUNCED_LETTERS.put("ؤ", "همزة_متوسطة_مضمومة");
+        PRONOUNCED_LETTERS.put("ى", "ألف_لينة");
+        PRONOUNCED_LETTERS.put("ي", "ياء");
+        PRONOUNCED_LETTERS.put("ئ", "همزة_متوسطة_مكسورة");
+        PRONOUNCED_LETTERS.put("ة", "تاء_مربوطة"); // Not in STANDARD_LETTERS[]
     }
 
     private Data() {

diff --git a/src/test/java/io/github/seen_arabic/arabic_services/ArabicServicesTest.java b/src/test/java/io/github/seen_arabic/arabic_services/ArabicServicesTest.java
@@ -92,4 +92,36 @@ public void testTashfeer() {
         assertNotEquals(result, inputWord);
     }
 
+    @Test
+    public void testWordToLetters() {
+        itShouldReturnAStringWithPronouncedArabicLetters();
+        itShouldHandleEmptyInput();
+        itShouldHandleInputWithNoPronouncedArabicLetters();
+        itShouldHandleInputWithSpaces();
+    }
+
+    private void itShouldReturnAStringWithPronouncedArabicLetters() {
+        String input = "هذه جملة عربية";
+        String result = ArabicServices.wordToLetters(input);
+        assertNotNull(result);
+        assertNotEquals(input, result);
+    }
+
+    private void itShouldHandleEmptyInput() {
+        String input = "";
+        String result = ArabicServices.wordToLetters(input);
+        assertEquals("", result);
+    }
+
+    private void itShouldHandleInputWithNoPronouncedArabicLetters() {
+        String input = "12345 not in Arabic letters";
+        String result = ArabicServices.wordToLetters(input);
+        assertEquals(input, result);
+    }
+
+    private void itShouldHandleInputWithSpaces() {
+        String input = "هذه جملة اخرى";
+        String result = ArabicServices.wordToLetters(input);
+        assertEquals("هاء ذال هاء  جيم ميم لام تاء_مربوطة  ألف خاء راء ألف_لينة", result);
+    }
 }