Nesting directories

Nesting models in directories

As your collection of datagen models grows, you can organize them into nested directory structures. This allows you to:

• Namespace your models - Avoid naming conflicts by organizing models in separate directories
• Group related models - Structure models by service, domain, feature, or any criteria that suits your needs
• Maintain clarity - Keep large codebases organized and easier to navigate

You are free to organize your directories however you wish. For example, you might structure them by:

• Services or microservices
• Database schemas
• Business domains
• Features or modules

Example structure

Consider organizing models by service, where each service has its own set of tables:

models/
├── serviceA/
│   ├── tableA.dg
│   ├── tableB.dg
│   └── tableC.dg
└── serviceB/
    ├── tableA.dg
    ├── tableD.dg
    └── tableE.dg

Referencing nested models

When referencing models in nested directories, use the fully qualified path:

self.datagen.<directory>.<model>().<field>()

For the structure above:

• To refer to tableB in serviceA: self.datagen.serviceA.tableB().<column>()
• To refer to tableD in serviceB: self.datagen.serviceB.tableD().<column>()

Note

Notice that both serviceA and serviceB have a model named tableA. This is perfectly fine because they are accessed via their fully qualified paths, so there’s no confusion about which model is being referenced.

All paths are relative to the root models directory you provide to datagen.

Running datagen with nested directories

Important

When running datagen with nested directory structures, you must provide the root directory as input, not individual subdirectories.

For example, if your models are organized as shown above:

# Correct - provide the root directory
datagen -d ./models -o output

# Incorrect - providing subdirectories will cause errors
datagen -d ./models/serviceA -o output
datagen -d ./models/serviceB -o output

Providing individual subdirectories like serviceA or serviceB will cause data generation errors, especially when models reference each other across directories using self.datagen paths.