mini-expectations 0.1.0

A tiny, code-first data quality expectations library for pandas DataFrames.

mini_expectations

A tiny, code-first data quality library for pandas DataFrames.

It aims to provide a small, readable alternative to Great Expectations with:

• Pandas-only focus
• Code-first API (no YAML, no configs)
• Fail-fast behavior via a single ExpectationFailed exception
• Simple chaining from DataFrame → column → expectation

Installation

Install from your local checkout:

pip install -e .

You will also need pandas in your environment.

Basic usage

import pandas as pd
from mini_expectations import expect, ExpectationFailed

df = pd.DataFrame(
    {
        "id": [1, 2, 3],
        "amount": [10.0, 20.5, 0.1],
        "email": ["[e-mail]", "[e-mail]", "[e-mail]"],
        "status": ["active", "active", "inactive"],
    }
)

# DataFrame-level checks
expect(df).to_have_columns(["id", "amount", "email"])
expect(df).to_have_no_nulls()
expect(df).to_have_unique_column("id")

# Column-level checks
expect(df).column("amount").to_be_positive()
expect(df).column("amount").to_be_between(0, 100)
expect(df).column("email").to_match_regex(r".+@.+")
expect(df).column("status").to_be_in_set(["active", "inactive"])
expect(df).column("id").to_have_no_nulls()

# Row count checks
expect(df).row_count().to_be_between(1, 1_000_000)

All expectations return self so you can chain them:

expect(df) \
    .to_have_columns(["id", "amount", "email"]) \
    .to_have_no_nulls() \
    .to_have_unique_column("id")

When an expectation fails, ExpectationFailed is raised with a human-readable message that describes what was expected and what was actually observed.

API overview

All entry points live under a single function:

• expect(df) → returns a DataFrameExpectations object.

DataFrame-level expectations:

• to_have_columns(columns): assert that all listed columns exist.
• to_have_no_nulls(subset=None): assert that there are no nulls in the DataFrame (or in a subset of columns).
• to_have_unique_column(column): assert that a column exists, has no nulls, and contains only unique values.
• row_count(): returns a RowCountExpectation wrapper for the number of rows.

Column navigation and expectations:

• column(name): navigate to a ColumnExpectations object.
• to_be_positive(strictly=True): assert that numeric values are > 0 (or >= 0 when strictly=False).
• to_match_regex(pattern): assert that all non-null stringified values match a regex pattern.
• to_be_unique(): assert that non-null values are unique.
• to_be_between(min_value, max_value, inclusive=True): assert that non-null numeric values lie in a given range.
• to_be_in_set(allowed_values): assert that non-null values come from a fixed set.
• to_have_no_nulls(): assert that a column has no null values.

Row count expectations:

• RowCountExpectation.to_be_between(min_value, max_value): assert that the number of rows lies between two bounds (inclusive).

Error handling

All failed expectations raise:

• ExpectationFailed: a subclass of AssertionError with clear, descriptive messages.

You can choose to let expectations fail fast, or catch the exception to aggregate or log errors:

try:
    expect(df).column("amount").to_be_positive()
except ExpectationFailed as exc:
    print("Data quality issue:", exc)