# 胡汉三千年项目 - 数据导入配置
# 符号数据导入器配置文件

# 数据库配置
database:
  # 数据库文件路径
  path: "symbols.db"
  # 数据库连接超时（秒）
  timeout: 30
  # 是否启用WAL模式（提高并发性能）
  wal_mode: true

# 数据验证配置
validation:
  # 必填字段检查
  required_fields: ["symbol_id", "symbol_form", "symbol_name"]
  # 符号ID格式正则表达式
  symbol_id_pattern: "^[A-Za-z0-9_]+$"
  # 允许的阴阳属性值
  valid_yin_yang: ["yin", "yang", "neutral"]
  # 允许的刻法类型
  valid_engraving: ["yin_engraving", "yang_engraving", "mixed"]
  # 置信度范围
  confidence_range:
    min: 1
    max: 100

# 导入配置
import:
  # 批量导入时的批次大小
  batch_size: 100
  # 是否在导入时显示进度
  show_progress: true
  # 导入失败时的处理方式
  on_error: "continue"  # 可选值：stop, continue, skip
  # 重复记录的处理方式
  duplicate_handling: "update"  # 可选值：skip, update, error

# 文件格式配置
file_formats:
  csv:
    # 默认分隔符
    delimiter: ","
    # 支持的编码格式
    encodings: ["utf-8", "gbk", "gb2312", "latin1"]
    # 是否跳过空行
    skip_empty_lines: true
    # 是否跳过标题行
    skip_header: false
  
  json:
    # JSON根节点名称（如果数据在特定节点下）
    root_node: "symbols"
    # 是否验证JSON格式
    validate_json: true

# 日志配置
logging:
  # 日志级别
  level: "INFO"  # DEBUG, INFO, WARNING, ERROR
  # 日志文件路径
  file_path: "import_log.log"
  # 是否在控制台显示日志
  console_output: true
  # 日志格式
  format: "%(asctime)s - %(levelname)s - %(message)s"

# 示例数据配置
sample_data:
  # 是否自动生成示例数据
  auto_generate: true
  # 示例符号数量
  symbol_count: 10
  # 示例关联数量
  link_count: 5
  # 示例传播路径数量
  path_count: 3

# 高级配置
advanced:
  # 是否启用数据缓存
  enable_cache: true
  # 缓存大小（MB）
  cache_size: 100
  # 是否启用并行处理
  parallel_processing: false
  # 并行工作线程数
  max_workers: 4
  # 数据库连接池大小
  connection_pool_size: 10

# 数据源配置
data_sources:
  # 支持的源类型
  supported_types: ["csv", "json", "excel", "database", "api"]
  
  # CSV源配置
  csv:
    # 默认文件扩展名
    extensions: [".csv", ".txt"]
    # 最大文件大小（MB）
    max_file_size: 100
    # 是否自动检测分隔符
    auto_detect_delimiter: true
  
  # JSON源配置
  json:
    extensions: [".json", ".jsonl"]
    max_file_size: 50
    # 是否支持JSON Lines格式
    support_jsonl: true
  
  # Excel源配置
  excel:
    extensions: [".xlsx", ".xls"]
    max_file_size: 50
    # 默认工作表名称
    default_sheet: "Sheet1"

# 字段映射配置
field_mappings:
  # 符号表字段映射
  symbols:
    # 标准字段名 -> 可能的数据源字段名
    symbol_id: ["id", "symbol_id", "symbolID", "标识符"]
    symbol_form: ["form", "symbol_form", "symbolForm", "符号形态"]
    symbol_name: ["name", "symbol_name", "symbolName", "符号名称"]
    yin_yang_attribute: ["yin_yang", "attribute", "阴阳属性", "阴阳"]
    engraving_type: ["engraving", "type", "刻法类型", "刻法"]
    origin_civilization: ["civilization", "origin", "起源文明", "文明"]
    origin_period: ["period", "era", "起源时期", "时期"]
    geographical_context: ["geography", "context", "地理背景", "地理"]
    functional_context: ["function", "purpose", "功能背景", "功能"]
    phonetic_context: ["phonetic", "sound", "语音背景", "语音"]
    semantic_context: ["semantic", "meaning", "语义背景", "语义"]
    description: ["desc", "description", "描述", "说明"]
  
  # 关联表字段映射
  cross_civilization_links:
    source_symbol_id: ["source_id", "from", "源符号", "起点"]
    target_symbol_id: ["target_id", "to", "目标符号", "终点"]
    link_type: ["type", "relation_type", "关联类型", "类型"]
    confidence_level: ["confidence", "level", "置信度", "可信度"]
    evidence_description: ["evidence", "proof", "证据描述", "证据"]
    source_reference: ["reference", "source", "来源参考", "参考"]
  
  # 传播路径表字段映射
  symbol_transmission_paths:
    path_name: ["name", "path", "路径名称", "名称"]
    start_civilization: ["start", "origin", "起点文明", "起点"]
    end_civilization: ["end", "destination", "终点文明", "终点"]
    transmission_period: ["period", "time", "传播时期", "时期"]
    transmission_route: ["route", "path", "传播路线", "路线"]
    supporting_evidence: ["evidence", "support", "支持证据", "证据"]

# 错误处理配置
error_handling:
  # 验证错误处理
  validation_errors:
    # 是否记录验证错误
    log_errors: true
    # 是否生成错误报告
    generate_report: true
    # 错误报告格式
    report_format: "html"  # html, json, csv
  
  # 导入错误处理
  import_errors:
    # 是否继续导入其他记录
    continue_on_error: true
    # 是否记录错误详情
    log_details: true
    # 错误记录保存路径
    error_log_path: "import_errors.log"
  
  # 数据库错误处理
  database_errors:
    # 是否自动重试
    auto_retry: true
    # 最大重试次数
    max_retries: 3
    # 重试间隔（秒）
    retry_interval: 5

# 性能优化配置
performance:
  # 内存使用限制（MB）
  memory_limit: 512
  # 是否启用内存优化
  memory_optimization: true
  # 是否启用延迟加载
  lazy_loading: true
  # 查询超时时间（秒）
  query_timeout: 60

# 安全配置
security:
  # 是否验证文件类型
  validate_file_type: true
  # 允许的文件类型
  allowed_file_types: ["csv", "json", "xlsx", "xls"]
  # 是否检查文件大小
  check_file_size: true
  # 最大允许文件大小（MB）
  max_file_size: 100
  # 是否扫描恶意内容
  scan_malicious_content: false

# 备份配置
backup:
  # 是否自动备份
  auto_backup: true
  # 备份文件路径
  backup_path: "backups"
  # 备份保留天数
  retention_days: 30
  # 备份频率
  backup_frequency: "daily"  # daily, weekly, monthly

# 监控配置
monitoring:
  # 是否启用性能监控
  enable_monitoring: true
  # 监控指标
  metrics: ["import_speed", "memory_usage", "error_rate", "success_rate"]
  # 监控数据保存路径
  metrics_path: "metrics"
  # 监控数据保留天数
  metrics_retention: 7