retrie

What it is and what it does

Retrie solves the performance problem of matching or replacing large sets of strings using naive regex unions. Instead of compiling a pattern like `(?:abc|abs|foo)` which becomes slow as the word list grows, it builds a Trie data structure that produces a more efficient pattern like `(?:ab[cs]|foo)`. The package provides three main classes—Trie (the underlying structure), Blacklist (filter out unwanted strings), Whitelist (keep only allowed strings), and Replacer (perform bulk find-and-replace)—each with options to match whole words or substrings.

The implementation is pure Python with minimal dependencies (only typing and cached-property), making it portable and easy to integrate. It supports both Python 2.7 and modern Python versions, and has been actively maintained since its 2020 release. The API is straightforward: instantiate a class with a word list or mapping, optionally configure matching behavior, and call methods like `filter()`, `cleanse_text()`, or `replace()` on your input.

Use it for:

• Filter spam or profanity from user-generated text by maintaining a blacklist of forbidden terms.
• Extract only whitelisted keywords or entities from documents for data cleaning pipelines.
• Perform bulk find-and-replace operations (e.g., synonym substitution, URL rewriting) in a single pass.
• Build content moderation systems that need to match many patterns efficiently without regex compilation overhead.
• Normalize or standardize text by replacing multiple variant spellings or abbreviations with canonical forms.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Builds efficient Trie-based regex patterns for fast matching and replacing strings in bulk, avoiding the performance penalty of simple regex unions on large word sets.

Yes. Retrie is a focused, well-maintained tool that solves a real performance problem for bulk string matching and replacement. Its low install friction, permissive MIT license, and broad Python version support make it a safe dependency. Use it when you need to filter or replace large word sets and regex performance matters; skip it if you only have a handful of patterns to match.

Install

retrie on PyPI

pip install retrie

uv add retrie

poetry add retrie

Installing retrie

Before you install

Low friction: pure Python, no compiled dependencies, and actively maintained with a recent commit on 2026-08-01. Supports Python 2.7 through 3.12, though reliance on the unmaintained cached-property backport for older Python versions may warrant attention in long-term projects.

License in practice

MIT license permits commercial and private use with minimal restrictions—include a copy of the license and you are free to modify and distribute.

Quickstart

pip install retrie

from retrie.retrie import Blacklist
blacklist = Blacklist(["abc", "foo"], match_substrings=False)
blacklist.cleanse_text("good abc foobar")

Verify before relying

• Whether the Trie structure provides measurable performance gains for specific word-set sizes or text lengths.
• How the package handles Unicode edge cases or non-ASCII character matching in practice.

Package facts

Evidence: retrie-0.3.1-py2.py3-none-any.whl

Keywords: pure-Python, regex, trie, regex-trie, blacklist, whitelist, re, search, replace

Tags

More Python Modules packages