Add Arabic pronunciation processing

2026-09-13 21:44:42 +02:00
parent 49a03a0700
commit f5ab97e6a0
4 changed files with 182 additions and 1 deletions
@@ -0,0 +1,104 @@
+import Foundation
+
+public struct ArabicPronunciationProcessor: Sendable {
+    public init() {}
+
+    public func process(
+        lyrics: String,
+        settings: ArabicPronunciationSettings
+    ) -> ArabicPronunciationProcessingResult {
+        guard settings.isEnabled else {
+            return ArabicPronunciationProcessingResult(
+                text: lyrics,
+                notes: [.processingDisabled]
+            )
+        }
+
+        var processedText = lyrics.precomposedStringWithCanonicalMapping
+        var notes = processingNotes(for: processedText, settings: settings)
+
+        if settings.tanweenPolicy == .removeWhenUnwanted {
+            processedText = removeTanween(from: processedText)
+        }
+
+        if settings.tanweenPolicy == .addWhenPronunciationRequires,
+           containsArabicLetter(in: processedText),
+           !containsTanween(in: processedText) {
+            notes.append(.tanweenAdditionNeedsReview)
+        }
+
+        return ArabicPronunciationProcessingResult(
+            text: processedText,
+            notes: notes
+        )
+    }
+
+    private func processingNotes(
+        for text: String,
+        settings: ArabicPronunciationSettings
+    ) -> [ArabicPronunciationProcessingNote] {
+        guard containsArabicLetter(in: text) else {
+            return []
+        }
+
+        switch settings.diacritizationPolicy {
+        case .unspecified:
+            return [.diacritizationPolicyUnspecified]
+        case .pronunciationTargeted where !containsDiacritics(in: text):
+            return [.pronunciationTargetedDiacriticsNeedReview]
+        case .fullTashkeel where !containsDiacritics(in: text):
+            return [.fullTashkeelNeedsReview]
+        default:
+            return []
+        }
+    }
+
+    private func removeTanween(from text: String) -> String {
+        String(text.unicodeScalars.filter { scalar in
+            !isTanween(scalar)
+        })
+    }
+
+    private func containsArabicLetter(in text: String) -> Bool {
+        text.unicodeScalars.contains { scalar in
+            let isArabicBlock = (0x0600...0x06FF).contains(scalar.value)
+                || (0x0750...0x077F).contains(scalar.value)
+                || (0x08A0...0x08FF).contains(scalar.value)
+            return isArabicBlock && CharacterSet.letters.contains(scalar)
+        }
+    }
+
+    private func containsDiacritics(in text: String) -> Bool {
+        text.unicodeScalars.contains(where: isArabicDiacritic)
+    }
+
+    private func containsTanween(in text: String) -> Bool {
+        text.unicodeScalars.contains(where: isTanween)
+    }
+
+    private func isArabicDiacritic(_ scalar: Unicode.Scalar) -> Bool {
+        (0x064B...0x065F).contains(scalar.value) || scalar.value == 0x0670
+    }
+
+    private func isTanween(_ scalar: Unicode.Scalar) -> Bool {
+        (0x064B...0x064D).contains(scalar.value)
+    }
+}
+
+public struct ArabicPronunciationProcessingResult: Equatable, Sendable {
+    public var text: String
+    public var notes: [ArabicPronunciationProcessingNote]
+
+    public init(text: String, notes: [ArabicPronunciationProcessingNote] = []) {
+        self.text = text
+        self.notes = notes
+    }
+}
+
+public enum ArabicPronunciationProcessingNote: Equatable, Sendable {
+    case processingDisabled
+    case diacritizationPolicyUnspecified
+    case pronunciationTargetedDiacriticsNeedReview
+    case fullTashkeelNeedsReview
+    case tanweenAdditionNeedsReview
+}