Filters

The Filters library provides an easy and readable way to create complex data validation and processing pipelines, including:

• Validating complex JSON structures in API requests or config files.
• Parsing timestamps and converting to UTC.
• Converting Unicode strings to NFC, normalising line endings and removing unprintable characters.
• Decoding Base64, including URL-safe variants.

And much more!

The output from one filter can be piped into the input of another, enabling you to chain filters together to quickly and easily create complex data schemas and pipelines.

Philosophy

Filters applies the UNIX philosophy to data validation: do one thing well, and compose small tools together.

Each filter performs a single, focused task. Chain them using the | operator to build sophisticated validation pipelines that are easy to read and maintain.

Type-safe: Full type hint support for IDE autocomplete and static analysis.

Opinionated: Makes deliberate choices to handle common issues automatically (Unicode normalisation, UTC conversion, etc.) so you write less boilerplate.

Quick Start

Install via pip:

pip install phx-filters

Create a validation schema:

import filters as f
from decimal import Decimal

# Define your schema
schema = f.FilterRunner(
    f.FilterMapper({
        "lat": f.Required | f.Decimal | f.Min(Decimal(-90)) | f.Max(Decimal(90)),
        "lon": f.Required | f.Decimal | f.Min(Decimal(-180)) | f.Max(Decimal(180)),
        "name": f.Required | f.Unicode | f.Strip,
    })
)

# Validate data
result = schema.apply({"lat": "42.36", "lon": "-71.06", "name": "  Boston  "})

if result.is_valid():
    clean_data = result.value
    # clean_data = {
    #     "lat": Decimal("42.36"),
    #     "lon": Decimal("-71.06"),
    #     "name": "Boston"
    # }
else:
    errors = result.error_messages
    # errors = {"lat": ["Decimal value is too small (minimum is -90)."]}

FilterRunner provides a familiar interface similar to Django forms, making it easy to integrate into web applications.

Examples

Validate API Request Data

When building APIs, you need to validate request payloads and handle errors gracefully. FilterRunner makes this straightforward:

from decimal import Decimal
import filters as f

# Define validation for a user registration endpoint
user_schema = f.FilterRunner(
    f.FilterMapper(
        {
            "email": f.Required | f.Unicode | f.Strip | f.MaxLength(254),
            "age": f.Required | f.Int | f.Min(13) | f.Max(120),
            "timezone": f.Decimal | f.Min(Decimal("-15")) | f.Max(Decimal("15")),
        },
        allow_extra_keys=False,
    )
)

# Validate incoming data
result = user_schema.apply(request_data)

if result.is_valid():
    # Save to database
    user = User.create(**result.value)
else:
    # Return validation errors to client
    return {"errors": result.error_messages}, 400

Parse Complex JSON Structures

Filters excels at validating nested data structures with complex constraints:

schema = f.FilterRunner(
    f.JsonDecode |
    f.FilterMapper(
        {
            "birthday": f.Date,
            "gender": f.CaseFold | f.Choice(choices={"f", "m", "n"}),
            "utcOffset": (
                f.Decimal |
                f.Min(Decimal("-15")) |
                f.Max(Decimal("15")) |
                f.Round(to_nearest="0.25")
            ),
        },
        allow_extra_keys=False,
        allow_missing_keys=False,
    )
)

result = schema.apply('{"birthday":"1879-03-14", "gender":"M", "utcOffset":"1"}')

Process Lists of Data

Use FilterRepeater to apply validation to every item in a collection:

# Clean a list of user-generated strings
schema = f.FilterRunner(
    f.FilterRepeater(f.Unicode | f.Strip | f.MaxLength(100))
)

result = schema.apply([
    "  some text  ",
    b"\xe2\x99\xaa unicode bytes ",
    "another string",
])

For more examples and detailed documentation, visit https://filters.readthedocs.io/

Features

• Composable: Chain filters using the | operator
• Type-safe: Full type hint support for IDE autocomplete and mypy
• Familiar API: FilterRunner provides Django-form-like interface
• Extensible: Create custom filters by extending BaseFilter
• Battle-tested: Used in production applications for years
• Well-documented: Comprehensive docs at https://filters.readthedocs.io/

Requirements

Filters is known to be compatible with the following Python versions:

• 3.14
• 3.13
• 3.12

Note

I'm only one person, so to keep from getting overwhelmed, I'm only committing to supporting the 3 most recent versions of Python.

Installation

Install the latest stable version via pip:

pip install phx-filters

Important

Make sure to install phx-filters, not filters. I created the latter at a previous job years ago, and after I left they never touched that project again and stopped responding to my emails — so in the end I had to fork it 🤷

Extensions

The following extensions are available:

• Django Filters: Adds filters designed to work with Django applications. To install:

  pip install phx-filters[django]
  

• ISO Filters: Adds filters for interpreting standard codes and identifiers. To install:

  pip install phx-filters[iso]
  

Tip

To install multiple extensions, separate them with commas, e.g.:

pip install phx-filters[django,iso]

Maintainers

To install the distribution for local development, some additional setup is required:

1. Install uv (only needs to be done once).

2. Run the following command to install additional dependencies:

   uv sync --group=dev
   

3. Activate pre-commit hook:

   uv run autohooks activate --mode=pythonpath
   

Running Unit Tests and Type Checker

Run the tests for all supported versions of Python using tox:

uv run tox -p

Note

The first time this runs, it will take awhile, as mypy needs to build up its cache. Subsequent runs should be much faster.

If you just want to run unit tests in the current virtualenv (using pytest):

uv run pytest

If you just want to run type checking in the current virtualenv (using mypy):

uv run mypyc src test

Documentation

To build the documentation locally:

1. Switch to the docs directory:

   cd docs
   

2. Build the documentation:

   uv run make html
   

Releases

Steps to build releases are based on Packaging Python Projects Tutorial.

Important

Make sure to build releases off of the main branch, and check that all changes from develop have been merged before creating the release!

1. Build the Project

1. Delete artefacts from previous builds, if applicable:

   rm dist/*
   

2. Run the build:

   uv build
   

3. The build artefacts will be located in the dist directory at the top level of the project.

2. Upload to PyPI

1. Create a PyPI API token (you only have to do this once).

2. Increment the version number in pyproject.toml.

3. Upload build artefacts to PyPI:

   uv publish
   

3. Create GitHub Release

1. Create a tag and push to GitHub:

   git tag <version>
   git push <version>
   

   <version> must match the updated version number in pyproject.toml.

2. Go to the Releases page for the repo.

3. Click Draft a new release.

4. Select the tag that you created in step 1.

5. Specify the title of the release (e.g., Filters v1.2.3).

6. Write a description for the release. Make sure to include: - Credit for code contributed by community members. - Significant functionality that was added/changed/removed. - Any backwards-incompatible changes and/or migration instructions. - SHA256 hashes of the build artefacts.

7. GPG-sign the description for the release (ASCII-armoured).

8. Attach the build artefacts to the release.

9. Click Publish release.