A simple and easy to use Data Validation library for Python.

Validoopsie

Validoopsie is a remarkably lightweight and user-friendly data validation library for Python. It’s designed to help you easily declare classes and chain validations together, in a style reminiscent of popular DataFrame libraries. This makes it a familiar and intuitive tool for developers who regularly work with dataframes.

Thanks to the excellent work by Narwhals, Validoopsie incorporates the "Bring Your Own DataFrame" (BYOD) concept. This flexibility allows you to use any DataFrame that Narwhals supports for your data validation tasks. To explore the full range of supported DataFrames, you can visit this link.

The syntax of Validoopsie has been thoughtfully crafted to ensure ease of use. Every validation function is encapsulated in its own method, which can be seamlessly linked together. This method-specific design prioritizes simplicity and readability, freeing you from the need to adapt to a new API each time you switch libraries. It allows you to focus on maintaining clean and understandable code.

Validoopsie draws significant inspiration from the Great Expectations library. It strives to distill the data validation process into something straightforward and efficient. Whether you're checking data integrity or ensuring compliance with data standards, Validoopsie provides a streamlined yet powerful solution to make these tasks accessible and straightforward.

Table of Contents

1. Installation
2. Getting Started
3. Version
4. Dunder Methods
5. Validate Method
6. Development
7. License

Installation

• pip

  pip install Validoopsie

• uv

  uv add Validoopsie

Getting Started

• 📖 Documentation
• 🚨 Impact Levels in Validoopsie
• 🎯 Threshold Levels in Validoopsie
• 🛠️ Contribution Guidelines
• ✨ Contributing a new Validation to Validoopsie
• 🧑‍💻 Develop your own custom validation
• 🗂️ Validation Catalog

Validoopsie is incredibly easy to use, so much so that you could do it half-asleep. The simplicity of the library is enhanced by the BYOD (Bring Your Own DataFrame) concept, where you merely need to utilize the Validate class and chain your desired validations together. This approach ensures that you can get started with minimal effort and without any unnecessary complexity.

import pandas as pd

from validoopsie import Validate

p_df = pd.DataFrame(
    {
        "name": ["John", "Doe", "Jane"],
        "target_name": ["John", "Doe", "Jane"],
        "last_name": ["Smith", "Smith", "Smith"],
        "age": [25, 30, 35],
    },
)

# `vd` stands for Validate Data
vd = Validate(p_df)
vd.EqualityValidation.PairColumnEquality(
    column="name",
    target_column="age",
    impact="high",
).UniqueValidation.ColumnUniqueValuesToBeInList(
    column="last_name",
    values=["Smith"],
).ValuesValidation.ColumnValuesToBeBetween(
    column="age",
    min_value=20,
    max_value=40,
)

vd.results

OUTPUT:

{
  "Summary": {
    "passed": false,
    "validations": [
      "PairColumnEquality_name",
      "ColumnUniqueValuesToBeInList_last_name",
      "ColumnValuesToBeBetween_age"
    ],
    "failed_validation": ["PairColumnEquality_name"]
  },
  "PairColumnEquality_name": {
    "validation": "PairColumnEquality",
    "impact": "high",
    "timestamp": "2025-10-07T11:25:04.213211+02:00",
    "column": "name",
    "result": {
      "status": "Fail",
      "threshold_pass": false,
      "message": "The column 'name' is not equal to the column'age'.",
      "failing_items": [
        "Doe - column name - column age - 30",
        "Jane - column name - column age - 35",
        "John - column name - column age - 25"
      ],
      "failed_number": 3,
      "frame_row_number": 3,
      "threshold": 0.0,
      "failed_percentage": 1.0
    }
  },
  "ColumnUniqueValuesToBeInList_last_name": {
    "validation": "ColumnUniqueValuesToBeInList",
    "impact": "low",
    "timestamp": "2025-10-07T11:25:04.216417+02:00",
    "column": "last_name",
    "result": {
      "status": "Success",
      "threshold_pass": true,
      "message": "All items passed the validation.",
      "frame_row_number": 3,
      "threshold": 0.0
    }
  },
  "ColumnValuesToBeBetween_age": {
    "validation": "ColumnValuesToBeBetween",
    "impact": "low",
    "timestamp": "2025-10-07T11:25:04.217300+02:00",
    "column": "age",
    "result": {
      "status": "Success",
      "threshold_pass": true,
      "message": "All items passed the validation.",
      "frame_row_number": 3,
      "threshold": 0.0
    }
  }
}

You can also display the validation results in a formatted table using the display_summary method, which provides a clean and readable view of your validation results:

vd.display_summary()

Display a full detailed table with all available information

vd.display_summary(information="full")

Customize the table format (supports tabulate formatting options)

vd.display_summary(tablefmt="pipe", maxcolwidths=20)

The display_summary method supports two information levels:

• "short" (default): Shows key metrics like timestamp, impact, status, validation type, column, threshold, and failure details
• "full": Shows all available validation and result fields

You can also customize the table appearance using any tabulate formatting options such as tablefmt for different table styles (e.g., "github", "grid", "pipe", "html") and maxcolwidths to control column width.

Version

You can check the installed version of Validoopsie at any time:

from validoopsie import __version__

print(__version__)

Dunder Methods

The Validate class supports several Python dunder methods for convenience:

• repr(vd) — returns a summary string with row count and number of validations:

  >>> repr(vd)
  'Validate(rows=3, validations=3)'
  

• str(vd) — returns a human-readable status string:

  >>> str(vd)
  'Validate: 3 validation(s), passed=False'
  

• len(vd) — returns the number of validations:

  >>> len(vd)
  3
  

Validate Method

To ensure that all your validations have been correctly executed and to handle any potential errors that may arise during the validation process, you can use the validate method. However, it's important to note that errors will only be raised if the impact level is set to high. Without this setting, potential issues may not trigger an error message.

NOTE: Raised error is a custom ValueError.

import pandas as pd

from validoopsie import Validate

p_df = pd.DataFrame(
    {
        "name": ["John", "Doe", "Jane"],
        "target_name": ["John", "Doe", "Jane"],
        "last_name": ["Smith", "Smith", "Smith"],
        "age": [25, 30, 35],
    },
)

# `vd` stands for Validate Data
vd = Validate(p_df)
vd.EqualityValidation.PairColumnEquality(
    column="name",
    target_column="age",
    impact="high",
).UniqueValidation.ColumnUniqueValuesToBeInList(
    column="last_name",
    values=["Smith"],
).ValuesValidation.ColumnValuesToBeBetween(
    column="age",
    min_value=20,
    max_value=40,
).validate()

Thanks to loguru output will provide a very condenced information on validations and their status in a colorful way.

Development

Validoopsie includes a Makefile to simplify development tasks:

# Install dependencies
make setup

# Run linters (mypy, ruff)
make lint

# Run tests (includes doctests, stubtest)
make test

# Run both lint and test
make all

For more information on development, check the contribution guidelines.

License

MIT © Validoopsie

Original Creator - Akmal Soliev