TestDataX

This command-line interface application enables quick and customizable test data generation across various formats. It uses Mimesis for synthetic data, offers flexible schema configurations, and simplifies output to multiple database dialects or file types. Users can define precise parameters for data volume, types, and constraints for each target data set.

Requirements

• Python 3.11+

Quick Start

# Install from PyPI
pip install testdatax

# Generate sample data
testdatax --rows 1000 --format json --output data.json

Features

• Generate realistic test data with Mimesis
• Support for multiple output formats (CSV, JSON, SQL, etc.)
• Customizable schema definitions
• Configurable data generation parameters
• CLI tool for easy test data generation

Supported Formats

• JSON
• CSV
• ORC
• Parquet
• MySQL
• MSSQL
• Oracle

CLI Usage

testdatax -o <output_file> -f <format> -s <schema_file> -r <num_rows> -p <provider> [-d]

Options:

• -o, --output: Output file path (table_name for sql exports)
• -f, --format: Output format (csv, json, orc, parquet, mysql, mssql, oracle)
• -r, --rows: Number of rows to generate (default: 10)
• -s, --schema: Path to schema file
• -p, --provider: Data provider; only mimesis is supported (default: mimesis)
• --seed: Seed for reproducible output (optional)
• --null-rate: Default NULL probability (0-1) for nullable fields - default: 0.1
• -d, --debug: Enable debug output

Reproducibility: passing --seed makes generation deterministic — the same schema, row count, provider and seed produce identical output every run, which is ideal for stable test fixtures.

Usage Examples

Generate 10 rows of CSV data:

testdatax -o users.csv -f csv -s schema.json -r 10

Generate 1000 rows of Parquet data with debug output:

testdatax -o large_dataset.parquet -f parquet -s users_schema.json -r 1000 -d

Generate 1000 rows of Parquet data using Mimesis provider:

testdatax -o large_dataset.parquet -f parquet -s users_schema.json -r 1000 -p mimesis

Generate JSON data with default row count (10):

testdatax -o data.json -f json -s schema.json

Generate ORC file with specific schema:

testdatax -o analytics.orc -f orc -s analytics_schema.json -r 100

Generate MySQL with default row count (1000), table_name as 'default':

testdatax -o default.sql -f mysql -r 1000

Generate MSSQL with default row count (1000), table_name as 'mstest':

testdatax -o mstest.sql -f mssql -r 1000

Generate Oracle with default row count (1000), table_name as 'oracle':

testdatax -o oracle.sql -f oracle -r 1000

Each command consists of:

• -o, --output: Specify the output file path and name
• -f, --format: Output format (csv, json, orc, parquet, mysql, mssql, oracle)
• -s, --schema: Path to your schema definition file
• -r, --rows: Number of rows to generate (optional, defaults to 10)
• -p, --provider: Data provider; only mimesis is supported (default: mimesis)
• -d, --debug: Enable debug logging (optional)

Schema Example

{
  "username": {
    "type": "string",
    "provider_field": "name"
  },
  "date_joined": {
    "type": "datetime"
  },
  "date": {
    "type": "date"
  },
  "age": {
    "type": "integer",
    "min": 18,
    "max": 99
  },
  "is_active": {
    "type": "boolean"
  },
  "float": {
    "type": "float"
  },
  "uuid": {
    "type": "uuid"
  },
  "status": {
    "type": "enum",
    "values": ["active", "inactive", "pending"]
  }
}

Schema Configuration

The schema file defines the structure and constraints of your generated data. Each field in the schema can have the following properties:

Basic Field Properties

• type: (required) The data type of the field
• nullable: (optional) Boolean to allow null values (default: false)
• unique: (optional) Boolean to ensure unique values (default: false)

Type-Specific Properties

String Fields

{
  "username": {
    "type": "string",
    "min_length": 5,
    "max_length": 20,
    "provider_field": "user_name"  // Use provider-specific field to generate realistic data
  },
  "description": {
    "type": "text",
    "min_length": 100,
    "max_length": 500
  }
}

Numeric Fields

{
  "age": {
    "type": "integer",
    "min": 18,
    "max": 99
  },
  "score": {
    "type": "float",
    "min": 0.0,
    "max": 100.0,
    "precision": 2
  }
}

Date and Time Fields

{
  "created_at": {
    "type": "datetime",
    "start_date": "2020-01-01",
    "end_date": "2023-12-31"
  },
  "birth_date": {
    "type": "date",
    "format": "%Y-%m-%d"
  }
}

Note: start_date/end_date bound the generated range (inclusive). format applies a strftime pattern to date/datetime values in the CSV and JSON outputs only; the SQL, Parquet and ORC exporters keep native date types and ignore format.

Enum Fields

{
  "status": {
    "type": "enum",
    "values": ["pending", "active", "suspended"],
    "weights": [0.2, 0.7, 0.1]  // Optional probability weights
  }
}

Using Mimesis provider fields

Specify Mimesis-backed generators with provider_field:

{
  "name": {
    "type": "string",
    "provider_field": "name"
  },
  "email": {
    "type": "string",
    "provider_field": "email"
  },
  "address": {
    "type": "string",
    "provider_field": "address"
  },
  "company": {
    "type": "string",
    "provider_field": "company"
  }
}

Complete Example

{
  "user_id": {
    "type": "uuid",
    "unique": true
  },
  "username": {
    "type": "string",
    "provider_field": "user_name",
    "unique": true
  },
  "email": {
    "type": "string",
    "provider_field": "email",
    "unique": true
  },
  "age": {
    "type": "integer",
    "min": 18,
    "max": 99
  },
  "status": {
    "type": "enum",
    "values": ["active", "inactive"],
    "weights": [0.8, 0.2]
  },
  "created_at": {
    "type": "datetime",
    "start_date": "2020-01-01",
    "end_date": "2023-12-31"
  },
  "is_verified": {
    "type": "boolean",
    "nullable": true
  }
}

Data provider

TestDataX generates synthetic values using Mimesis. The CLI accepts -p mimesis (default); other values are rejected.

Migration from older schemas

• Prefer the JSON key provider_field for Mimesis field names.
• The legacy key faker is still accepted as a deprecated alias: it maps to the same string Mimesis value_provider name (the Faker library is not used). Rename to provider_field when updating schemas.

Supported Data Types

• string
• text
• integer
• bigint
• float
• decimal
• boolean
• date
• datetime
• blob
• uuid
• enum

Database Type Mappings

License

This project is licensed under the MIT License - see the LICENSE file for details.