config.yaml

# Configuración de Aliah-Plus

# Umbrales de similitud facial
similarity:
  secure_match: 0.85      # >85% = Match Seguro
  probable_match: 0.72    # 72-85% = Coincidencia Probable
  discard_below: 0.72     # <72% = Descartado
  default_threshold: 0.75 # Umbral por defecto

# Motores de búsqueda
search_engines:
  - name: yandex
    enabled: true
    priority: 1
    max_results: 50
    timeout: 30
    
  - name: bing
    enabled: true
    priority: 2
    max_results: 50
    timeout: 30
    
  - name: pimeyes
    enabled: true
    priority: 3
    max_results: 50
    timeout: 60
    stealth_mode: true
    requires_ocr: true

# OCR Configuration
ocr:
  enabled: true
  gpu: true
  confidence_threshold: 0.60
  languages: ['en', 'es']
  preprocessing_methods: 7  # Número de métodos de preprocesamiento
  
  # Dominios conocidos a buscar
  known_platforms:
    - onlyfans
    - fansly
    - patreon
    - instagram
    - twitter
    - tiktok
    - chaturbate

# Cross-referencing
cross_referencing:
  enabled: true
  min_sources_for_correlation: 2
  domain_similarity_threshold: 0.85
  deduplicate_results: true
  
# Modelos de ML
models:
  face_detection: mtcnn  # Opciones: mtcnn, mediapipe, retinaface
  face_recognition: ArcFace  # Opciones: ArcFace, Facenet512, VGG-Face
  ocr: easyocr  # Opciones: easyocr, pytesseract
  
  # Configuración de ArcFace
  arcface:
    embedding_size: 512
    
# Vector Database
vector_db:
  type: qdrant  # Opciones: qdrant, chromadb, memory
  host: localhost
  port: 6333
  collection_name: aliah_faces
  distance_metric: cosine
  cache_ttl: 86400  # 24 horas en segundos
  
# Web Scraping
scraping:
  max_results_per_engine: 50
  timeout: 30
  retry_attempts: 3
  retry_delay: 2  # segundos
  
  # Stealth mode
  stealth_mode: true
  headless: true
  
  # Proxies
  use_proxies: false
  proxy_rotation: false
  proxy_list: []
  
  # Rate limiting
  rate_limit:
    requests_per_minute: 30
    delay_between_requests: 2  # segundos
    
  # User agents
  rotate_user_agents: true
  custom_user_agents: []

# API Configuration
api:
  title: "Aliah-Plus API"
  version: "1.0.0"
  description: "Sistema Avanzado de Re-Identificación Facial"
  
  # Rate limiting
  rate_limit:
    enabled: true
    requests_per_minute: 10
    burst: 20
    
  # CORS
  cors:
    allow_origins: ["*"]
    allow_methods: ["*"]
    allow_headers: ["*"]
    
  # File upload
  max_file_size: 10485760  # 10MB en bytes
  allowed_extensions: ['.jpg', '.jpeg', '.png', '.webp']

# Logging
logging:
  level: INFO  # Opciones: DEBUG, INFO, WARNING, ERROR
  format: "{time:YYYY-MM-DD HH:mm:ss} | {level} | {message}"
  rotation: "100 MB"
  retention: "30 days"
  log_file: "logs/aliah_plus_{time}.log"

# Performance
performance:
  # Parallel processing
  max_workers: 4
  concurrent_searches: 3
  
  # Cache
  enable_cache: true
  cache_size: 1000  # Número de búsquedas en cache
  
  # Timeouts
  global_timeout: 120  # segundos
  embedding_timeout: 10
  scraping_timeout: 60

# Security
security:
  # Sanitización
  sanitize_inputs: true
  validate_file_types: true
  check_file_headers: true
  
  # Privacy
  store_images: false
  anonymize_ips: true
  log_user_data: false
  
  # API Keys (para futuras integraciones)
  require_api_key: false
  api_keys: []

# Features toggles
features:
  ocr_extraction: true
  cross_referencing: true
  stealth_scraping: true
  vector_db_storage: true
  similarity_validation: true
  multi_engine_search: true