Output Settings

Output settings control file formats and organization.

Configuration

output:
  format: csv
  compression: none
  compression_level: 6

  files:
    journal_entries: true
    acdoca: true
    master_data: true
    documents: true
    subledgers: true
    trial_balances: true
    labels: true
    controls: true

Format

Output file format selection.

output:
  format: csv        # CSV format (default)
  format: json       # JSON format
  format: jsonl      # Newline-delimited JSON
  format: parquet    # Apache Parquet columnar
  format: sap        # SAP S/4HANA table format
  format: oracle     # Oracle EBS GL tables
  format: netsuite   # NetSuite journal entries

CSV Format

Standard comma-separated values:

document_id,posting_date,company_code,account,debit,credit
abc-123,2024-01-15,1000,1100,"1000.00","0.00"
abc-123,2024-01-15,1000,4000,"0.00","1000.00"

Characteristics:

• UTF-8 encoding
• Header row included
• Quoted strings when needed
• Decimals as strings

JSON Format

Structured JSON with nested objects:

[
  {
    "header": {
      "document_id": "abc-123",
      "posting_date": "2024-01-15",
      "company_code": "1000"
    },
    "lines": [
      {"account": "1100", "debit": "1000.00", "credit": "0.00"},
      {"account": "4000", "debit": "0.00", "credit": "1000.00"}
    ]
  }
]

Parquet Format

Apache Parquet columnar format for analytics:

output:
  format: parquet
  compression: snappy     # snappy (default), gzip, zstd

Parquet files are self-describing with embedded schema and support columnar compression. Ideal for Spark, DuckDB, Polars, pandas, and cloud data warehouses.

ERP Formats

Export in native ERP table schemas for load testing and integration validation:

# SAP S/4HANA
output:
  format: sap
  sap:
    tables: [bkpf, bseg, acdoca, lfa1, kna1, mara, csks, cepc]
    client: "100"

# Oracle EBS
output:
  format: oracle
  oracle:
    ledger_id: 1

# NetSuite
output:
  format: netsuite
  netsuite:
    subsidiary_id: 1
    include_custom_fields: true

See ERP Output Formats for full field mappings.

Streaming Mode

Enable streaming output for memory-efficient generation of large datasets:

output:
  format: csv           # Any format
  streaming: true       # Enable streaming mode

See Streaming Output for details.

Compression

File compression options.

output:
  compression: none     # No compression
  compression: gzip     # Gzip compression (.gz)
  compression: zstd     # Zstandard compression (.zst)

Compression Level

When compression is enabled:

output:
  compression: gzip
  compression_level: 6    # 1-9, higher = smaller + slower

Compression Comparison

File Selection

Control which files are generated:

output:
  files:
    # Core transaction data
    journal_entries: true    # journal_entries.csv
    acdoca: true             # acdoca.csv (SAP format)

    # Master data
    master_data: true        # vendors.csv, customers.csv, etc.

    # Document flow
    documents: true          # purchase_orders.csv, invoices.csv, etc.

    # Subsidiary ledgers
    subledgers: true         # ar_open_items.csv, ap_open_items.csv, etc.

    # Period close
    trial_balances: true     # trial_balances/*.csv

    # ML labels
    labels: true             # anomaly_labels.csv, fraud_labels.csv

    # Controls
    controls: true           # internal_controls.csv, sod_rules.csv

Output Directory Structure

With all files enabled:

output/
├── master_data/
│   ├── chart_of_accounts.csv
│   ├── vendors.csv
│   ├── customers.csv
│   ├── materials.csv
│   ├── fixed_assets.csv
│   └── employees.csv
├── transactions/
│   ├── journal_entries.csv
│   └── acdoca.csv
├── documents/
│   ├── purchase_orders.csv
│   ├── goods_receipts.csv
│   ├── vendor_invoices.csv
│   ├── payments.csv
│   ├── sales_orders.csv
│   ├── deliveries.csv
│   ├── customer_invoices.csv
│   └── customer_receipts.csv
├── subledgers/
│   ├── ar_open_items.csv
│   ├── ar_aging.csv
│   ├── ap_open_items.csv
│   ├── ap_aging.csv
│   ├── fa_register.csv
│   ├── fa_depreciation.csv
│   ├── inventory_positions.csv
│   └── inventory_movements.csv
├── period_close/
│   └── trial_balances/
│       ├── 2024_01.csv
│       ├── 2024_02.csv
│       └── ...
├── consolidation/
│   ├── eliminations.csv
│   └── currency_translation.csv
├── fx/
│   ├── daily_rates.csv
│   └── period_rates.csv
├── graphs/                      # If graph_export enabled
│   ├── pytorch_geometric/
│   └── neo4j/
├── labels/
│   ├── anomaly_labels.csv
│   └── fraud_labels.csv
└── controls/
    ├── internal_controls.csv
    ├── control_mappings.csv
    └── sod_rules.csv

Examples

Development (Fast)

output:
  format: csv
  compression: none
  files:
    journal_entries: true
    master_data: true
    labels: true

Production (Compact)

output:
  format: csv
  compression: zstd
  compression_level: 6
  files:
    journal_entries: true
    acdoca: true
    master_data: true
    documents: true
    subledgers: true
    trial_balances: true
    labels: true
    controls: true

ML Training Focus

output:
  format: csv
  compression: gzip
  files:
    journal_entries: true
    labels: true                 # Important for supervised learning
    master_data: true            # For feature engineering

SAP Integration

output:
  format: csv
  compression: none
  files:
    journal_entries: false
    acdoca: true                 # SAP ACDOCA format
    master_data: true
    documents: true

Validation

See Also

• Output Formats
• Graph Export
• datasynth-output Crate