fauxgen

A factory method generator that streamlines test data creation by automating field value generation.

Overview

Testing classes with numerous fields often requires extensive boilerplate code. fauxgen addresses this challenge by automatically generating factory methods from class definitions, significantly reducing test code complexity.

Currently supports pandera.DataFrameModel with plans for future expansions.

Quick Start

uv add fauxgen
uv run fauxgen gen --module-dir <your_module>

Usage Example

Traditional Approach (Without fauxgen)

Testing DataFrame models traditionally requires explicit specification of all fields, even when testing a single validation:

import pandera as pa
from pandera.typing import Series

class UserSchema(pa.DataFrameModel):
    id: Series[int] = pa.Field(ge=1)
    age: Series[int] = pa.Field(ge=0, le=150)
    name: Series[str] = pa.Field()
    email: Series[str] = pa.Field(nullable=True)
    active: Series[bool] = pa.Field()

def test_user_registration():
    # Forced to specify every field when testing age validation
    df_user = pd.DataFrame([
        {
            "id": 1,            # Unrelated to test
            "age": 151,         # Actual test target
            "name": "test",     # Unrelated to test
            "email": "test@example.com",  # Unrelated to test
            "active": True,     # Unrelated to test
        },
    ]).pipe(DataFrame[UserSchema])
    # Test age validation...

Simplified Testing (With fauxgen)

fauxgen generates factory methods that enable focused testing by automatically handling irrelevant fields:

from .factories import user_schema_record  # Generated by fauxgen

def test_user_registration():
    # Focus solely on the field under test
    # Other fields are automatically populated with valid values
    df_user = pd.DataFrame([
        user_schema_record(age=151), # Set specific age for test
    ]).pipe(DataFrame[UserSchema])  # Test age validation...

def test_user_email_optional():
    # Effortlessly test specific scenarios
    # without worrying about irrelevant fields
    df_user = pd.DataFrame([
        user_schema_record(email=None)  # Set email to None for test
    ]).pipe(DataFrame[UserSchema])  # Test nullable email...

How It Works

Input: pandera.DataFrameModel Definition

import pandera as pa
from pandera.typing import Series

class TestSchema(pa.DataFrameModel):
    int_col: Series[int] = pa.Field(ge=0.1, le=10.0)
    float_col_only_ge: Series[float] = pa.Field(ge=10)
    float_col_only_le: Series[float] = pa.Field(le=10)
    pa_bool_col: Series[pa.Bool] = pa.Field(nullable=True)

Output: Generated Factory Method

import fauxgen as f

class TestSchemaRecord(TypedDict):
    int_col: int
    float_col_only_ge: float
    float_col_only_le: float
    pa_bool_col: bool | None

def test_schema_record(
    *,
    int_col: int | f.Unset = f.UNSET,
    float_col_only_ge: float | f.Unset = f.UNSET,
    float_col_only_le: float | f.Unset = f.UNSET,
    pa_bool_col: bool | None | f.Unset = f.UNSET,
    seed_: int | None = None,
) -> TestSchemaRecord:
    return {
        "int_col": f.Unset.unwrap_or_else(int_col, lambda: f.gen_int(ge=0.1, le=10, seed=seed_)),
        "float_col_only_ge": f.Unset.unwrap_or_else(float_col_only_ge, lambda: f.gen_float(ge=10, le=110, seed=seed_)),
        "float_col_only_le": f.Unset.unwrap_or_else(float_col_only_le, lambda: f.gen_float(ge=-90, le=10, seed=seed_)),
        "pa_bool_col": f.Unset.unwrap_or_else(pa_bool_col, lambda: f.gen_bool(seed=seed_)),
    }

Key Features

1. Enhanced Type Safety:

   • Leverages TypedDict for comprehensive type checking
   • Provides full IDE support with type hints and autocompletion

2. Intelligent Field Generation:

   • Selectively override specific fields while auto-generating others
   • Maintains data integrity through validation-aware value generation

3. Validation-Aware Generation:

   • Automatically respects field constraints (ge, le, etc.)
   • Properly handles optional fields with nullable support