config-driven-data-ingestion 1.0.0

A comprehensive, config-driven data ingestion library for Python

Config-Driven Data Loading Framework

Framework that eliminates repetitive data loading code by using configuration files. Instead of writing custom code for each data source, we configure once and reuse everywhere.

Framework Architecture

┌─────────────────┐    ┌──────────────────┐    ┌─────────────────┐
│   DATA SOURCES  │    │   ORCHESTRATOR   │    │  LiqForecast    │
│                 │    │                  │    │    Schema       │
│ • CSV Files     │────│ Config Reader    │────│                 │
│ • Excel Files   │    │ Data Processor   │    │ • forecast_data │
│ • REST APIs     │    │ Column Mapper    │    │ • market_trends │
│ • JSON Files    │    │ Database Writer  │    │ • risk_metrics  │
└─────────────────┘    └──────────────────┘    └─────────────────┘

Core Components

• DataSourceFactory: Creates appropriate loaders based on configuration type
• DataOrchestrator: Manages the entire pipeline with error handling and retry logic [One Time Implementation]
• DataProcessor: Handles transformations and column mapping
• DatabaseWriter: Executes batch operations to Database schema tables

---

Sample Configuration Files

Data Sources Configuration (data-sources.yml)

# Data Sources Configuration for LiqForecast Schema
data_sources:
  market_data_csv:
    type: "csv"
    source:
      file_path: "/data/market/daily_rates.csv"
      delimiter: ","
      header: true
      encoding: "UTF-8"
    target:
      schema: "MarketData"
      table: "market_trends"
      batch_size: 500
    column_mapping:
      - source: "date" → target: "trade_date"
      - source: "currency_pair" → target: "currency"
      - source: "rate" → target: "exchange_rate"
      - source: "volume" → target: "trading_volume"
  risk_metrics_excel:
    type: "excel"
    source:
      file_path: "/data/risk/monthly_risk.xlsx"
      sheet_name: "RiskData"
      skip_rows: 1
    target:
      schema: "RiskMetrics"
      table: "risk_metrics"
      batch_size: 200
    column_mapping:
      - source: "Portfolio ID" → target: "portfolio_id"
      - source: "VaR 95%" → target: "var_95"
      - source: "Expected Shortfall" → target: "expected_shortfall"
      - source: "Liquidity Score" → target: "liquidity_score"
    validation:
      required_columns: ["Portfolio ID", "VaR 95%"]
      data_quality_checks: true
  rest_api_data:
    type: "rest_api"
    source:
      url: "https://api.provider.com/v1/liq"
      method: "GET"
      headers:
        Authorization: "Bearer ${API_TOKEN}"
        Content-Type: "application/json"
      timeout: 30
      retry_attempts: 3
    target:
      schema: "APIData"
      table: "forecast_data"
      batch_size: 1000
    column_mapping:
      - source: "id" → target: "id"
      - source: "assetClass" → target: "asset_class"
      - source: "predictedLiquidity" → target: "predicted_liquidity"
      - source: "confidenceLevel" → target: "confidence_level"
      - source: "forecastDate" → target: "forecast_date"
  config_json:
    type: "json"
    source:
      file_path: "/config/portfolio_settings.json"
      json_path: "$.portfolios[*]"
    target:
      schema: "ConfigData"
      table: "portfolio_config"
      batch_size: 100
    column_mapping:
      - source: "id" → target: "portfolio_id"
      - source: "name" → target: "portfolio_name"
      - source: "riskProfile" → target: "risk_profile"
      - source: "liquidityThreshold" → target: "liquidity_threshold"

# Global Settings
global_settings:
  error_handling:
    continue_on_error: true
    error_threshold: 10
    notification_email: "[e-mail]"
  data_quality:
    enable_validation: true
    null_value_handling: "skip"
    duplicate_handling: "ignore"
  performance:
    connection_pool_size: 10
    query_timeout: 300
    memory_limit: "2GB"

High Level Architecture Diagram

Class Diagram

Sequence Diagram

Key Developer Benefits

• Code Reusability: Write once, configure multiple times - no duplicate data loading logic
• Maintenance Reduction: Single codebase handles all data sources through configuration
• Easy Onboarding: New data sources added via YAML files, not code changes
• Error Handling: Built-in retry logic and comprehensive error reporting
• Performance: Batch processing and connection pooling optimize database operations