# OMOP Pipeline Configuration

# Database Configuration
database:
  host: localhost
  port: 5432
  database: omop_cdm
  user: dom
  password: loli
  pool_size: 10
  max_overflow: 20
  pool_timeout: 30
  pool_recycle: 3600

# ETL Configuration
etl:
  batch_size: 1000
  num_workers: 8
  max_retries: 3
  retry_delay: 5  # seconds
  checkpoint_interval: 10000  # records

# Mapping Configuration
mapping:
  cache_size: 10000
  use_custom_mappings: true
  unmapped_concept_id: 0

# Validation Configuration
validation:
  min_completeness: 0.95
  max_error_rate: 0.05
  check_referential_integrity: true
  check_date_consistency: true
  check_value_ranges: true

# Logging Configuration
logging:
  level: INFO
  file: logs/omop_pipeline.log
  max_bytes: 10485760  # 10MB
  backup_count: 5
  format: "%(asctime)s - %(name)s - %(levelname)s - %(message)s"

# Performance Configuration
performance:
  enable_parallel_processing: true
  monitor_memory: true
  memory_threshold: 0.8  # 80% of available memory
  circuit_breaker_threshold: 0.5  # 50% error rate
  circuit_breaker_window: 100  # records

# Schema Configuration
schema:
  omop_schema: omop
  staging_schema: staging
  audit_schema: audit
  create_indexes: true
  create_constraints: true