Skip to content

REFACTORING_PLAN_PHASE1.md

Latest commit

 

History

History
127 lines (105 loc) · 3.77 KB

REFACTORING_PLAN_PHASE1.md

File metadata and controls

127 lines (105 loc) · 3.77 KB

FASE 1 - CRITICAL FIXES CHECKLIST

Task 1.1: iOS 26 Dependencies Removal (4-6h)

Files da modificare:

1. MirrorSmokerStopper/AI/CoachLLM.swift

  • Rimuovere @available(iOS 26, *)
  • Sostituire con implementazione rule-based
  • Mantenere stessa interfaccia pubblica
  • Test: Verificare che compili su iOS 17
// PRIMA (FAKE):
@available(iOS 26, *)
class CoachLLM {
    // Fake Foundation Models implementation
}

// DOPO (REAL):
class CoachLLM {
    // Rule-based implementation che funziona davvero
    private let templates: [String: [String]] = [
        "craving": [
            "Take 5 deep breaths. This craving will pass in 3-5 minutes.",
            "Remember why you decided to quit. You're stronger than this urge.",
            "Drink a glass of water and do 10 jumping jacks."
        ],
        "stress": [
            "Try the 4-7-8 breathing technique: inhale 4, hold 7, exhale 8.",
            "Take a 2-minute walk, even if it's just around the room.",
            "Listen to your favorite calming music for a few minutes."
        ]
    ]
}

2. MirrorSmokerStopper/AI/AICoachManager.swift

  • Rimuovere @available(iOS 26, *)
  • Implementare fallback per tutte le funzioni
  • Sostituire chiamate fake con logica reale
  • Test: Verificare funzionalità su dispositivo iOS 17

3. MirrorSmokerStopper/Views/AICoachTestView.swift

  • Rimuovere @available(iOS 26, *)
  • Sostituire con test view funzionante
  • Rimuovere riferimenti a Foundation Models
  • Test: View si apre senza errori

4. MirrorSmokerStopper/AI/QuitPlanOptimizer.swift

// Rimuovere:
// This would use Apple Intelligence Foundation Models
// Would be enhanced with Apple Intelligence

// Sostituire con:
private func generateOptimizedRecommendation(
    profile: UserProfile,
    behavioralData: BehavioralAnalysisResult
) -> QuitPlanRecommendation {
    // Algoritmo matematico reale basato su:
    // - Dependency level
    // - Historical success rates
    // - Behavioral patterns
    // - Medical recommendations
}

Validation Checklist:

  • App compila senza warning iOS 26
  • Nessun crash su iPhone fisico iOS 17/18
  • AI Coach mostra contenuti reali
  • Test su simulatore iOS 17.0

Task 1.2: Marketing Metadata Cleanup (2-3h)

Files da correggere:

1. fastlane/metadata/en-US/description.txt

- • World-class AI Coach leveraging Apple Intelligence
+ • Intelligent AI Coach with personalized recommendations
- • 100% LOCAL PROCESSING with Apple Intelligence  
+ • 100% LOCAL PROCESSING with advanced algorithms
- • AI-optimized quit plans powered by Apple Intelligence
+ • AI-optimized quit plans powered by behavioral analysis

2. fastlane/metadata/*/promotional_text.txt (tutte le lingue)

- Revolutionary AI Coach + Apple Intelligence integration
+ Revolutionary AI Coach with advanced behavioral analysis

3. fastlane/metadata/*/keywords.txt

- Apple Intelligence,quit smoking,iOS,AI coach
+ AI coach,quit smoking,iOS,behavioral analysis,health

Validation:

  • Nessun riferimento a "Apple Intelligence"
  • Nessun riferimento a "Foundation Models"
  • Claims sono veritieri e implementati
  • Test: Preview in fastlane looks correct

Task 1.3: Core ML References Cleanup (1-2h)

MirrorSmokerStopper/AI/CoachEngine.swift

// PRIMA:
// Core ML model for risk classification (placeholder until actual model is trained)
private var model: MLModel?

// DOPO:
// Statistical risk classification based on behavioral patterns
private let riskAnalyzer = StatisticalRiskAnalyzer()

Validation:

  • Nessun riferimento a modelli ML inesistenti
  • Risk classification funziona con algoritmi statistici
  • Performance accettabile (<100ms per prediction)