doorplate

Parse US addresses into components, with USPS Publication 28 and RESO output, a match key, a URL-safe slug, and diagnostics. A Rust library with Python bindings (PyO3, abi3 wheels for Python 3.10 and later); no runtime dependencies and no data to download.

This library formats and parses; it does not validate that an address exists or can receive mail.

Alpha release (0.5.0). Expect rough edges. The match key is versioned (k4:) and will probably get new versions as the parser improves, so store the version prefix with your keys and plan to re-key. k4 is new in 0.5.0 and has only been checked on the project's own data (37 million rows of public records), so a k5 may follow. Accuracy and speed are measurements on a few datasets, not guarantees.

Help wanted: messy addresses. If it parses your addresses badly, please open an issue with a sample. I want to fix them. Private data can be anonymized first (how).

Quickstart

pip install doorplate

import doorplate as dp

a = dp.parse("123 north main street apt 4, Miami FL 33101")
a.street_name          # "MAIN"
a.unit_number          # "4"
a.place                # {"name": "MIAMI", "state": "FL", "county": "DADE", ...}
a.warnings             # []
a.to_usps()["line1"]   # "123 N MAIN ST APT 4"
a.to_display()["line1"]  # "123 N Main St Apt 4" (to_long(): "123 North Main Street Apartment 4"); display only, not Pub 28
a.key()                # "k4:S|123|N|MAIN|ST||FL|33101"
a.slug()               # "123-n-main-st-apt-4-miami-fl-33101"
a.to_reso()            # RESO Data Dictionary 2.0 fields
a.to_dict()            # everything above as plain Python values, and provenance (the library version and the state data version; also `a.library_version`, `a.state_data_version`)

dp.parse_fields(line1="THE COVE", line2="1111 ASBURY WAY",
                city="Orlando", state="FL", zip="32801")
dp.parse_many(["1 Main St, Miami FL", "PO Box 5, Tampa FL"])  # GIL released
dp.parse_fields_many(line1=["1 Main St", "PO Box 5"], city=["Miami", "Tampa"])
dp.parse("281 EVELYN DR DOTHAN", default_state="AL").city     # "DOTHAN"

• Unusable input (no address text at all) raises ParseError, a ValueError; in parse_many that element is None. Anything else comes back with warnings (what each one tells you, and how often it is wrong).
• default_state is a lookup hint for input with no state; it never fills state. A value that is not a US state, DC or territory raises ValueError.

Install

pip install doorplate                  # the parser
pip install "doorplate[pandas]"        # plus the pandas helper
pip install "doorplate[polars]"        # plus the Polars helper
pip install "doorplate[duckdb]"        # plus the DuckDB functions (DuckDB and PyArrow)

Python 3.10 or later. Wheels are built for Linux (x86_64, aarch64), macOS (universal2) and Windows (x64), and tested on Python 3.10 and 3.13. To build from source (Rust 1.83 or later):

pip install maturin
cd crates/doorplate-py
maturin develop --release      # builds the extension into your virtualenv
pytest tests

The API

The calls are parse, parse_fields, parse_many, parse_fields_many, Parser, tag, parse_tokens, load_layer, Anonymizer and address_type; a ParsedAddress is read-only and typed (py.typed), with the components, warnings, unparsed, and key(), slug(), to_usps(), to_display(), to_reso() and to_dict(). The tour is the Python guide; every function and class is in the Python reference.

Your data, by shape

A column of addresses in pandas, Polars or DuckDB, split columns, mixed shapes, a CSV file: see pandas, Polars and DuckDB (one call per shape, with timings), or the examples in the repository.

Batches, pandas, Polars and DuckDB

parse_many(texts) parses a list on every core with the GIL released (about 3x a loop of parse), and the helpers doorplate.pandas.parse_column, doorplate.polars.parse_column and doorplate.duckdb build on it:

from doorplate.polars import parse_column   # or doorplate.pandas

out = df.hstack(parse_column(df["address"]))                    # Polars
out = df.join(parse_column(df["address"], default_state="FL"))  # pandas

The columns, DuckDB in SQL, doing it by hand and the timings are in pandas, Polars and DuckDB.

Custom layers

Your data spells TRAIL as TRAYL, or lives in a place the shipped list lacks? doorplate.load_layer("DIR") adds your own spellings, words, places and ZIPs on top of the shipped tables (custom_layer= on every call). A layer can change a key, so every parse reports it. The files, the rules, a demo layer, a template and a prompt for an LLM are in Custom layers.

Sharing messy data

doorplate.Anonymizer(seed=None).anonymize("123 N Zephyrmill Ave Apt 4, Miami, FL 33101") gives a copy that keeps the shape and loses the names and numbers, even when the text does not parse. It is obfuscation, not anonymity: read the output before you share it. The rules, the limits and what it can leak are in Anonymizing data.

Coming from usaddress

For a script that calls usaddress.parse and usaddress.tag, change one import (label compatibility, not identical output):

import doorplate.compat as usaddress    # was: import usaddress

The same calls under doorplate's own names are parse_tokens(s) and tag(s). The mapping and the differences are in Coming from usaddress.

Accuracy and speed

The suite is 19,586 labeled addresses that neither library was made from; it also holds both libraries' known blind spots on purpose, so read its natural stratum as the estimate for ordinary input. Speed is on 4 CPUs, 8 GB. These are measurements on a few data sets, not promises for yours. Method and every other figure: accuracy and speed and size; how to rerun them: running the benchmarks.

Disclaimer

doorplate is provided "as is", without warranty of any kind, under the MIT or Apache-2.0 license at your option; the license text is what governs. In plain terms: use it at your own risk. Nothing here guarantees that a parse, a warning, a match key or a figure is correct, complete or fit for your purpose. Accuracy and speed numbers are measurements on particular datasets, not promises for yours. doorplate parses and formats addresses; it does not validate that an address exists or can receive mail, and Pub 28 formatting is not certified validation. Check its output before you rely on it for anything with legal, financial, safety or delivery consequences.

doorplate is an independent project. It is not affiliated with, endorsed by or certified by the US Postal Service, RESO, DataMade (the authors of usaddress) or any data source it cites; those names are used only to say what the output follows or is compared with.

More

• Documentation site: the Python guide, reference, pandas, Polars and DuckDB, the key, warnings and how it works.
• Project README: what it does, the Rust library, the command-line tool.
• Examples: a notebook tour and a pipeline for each DataFrame library.

Licensed under either of MIT or Apache-2.0 at your option.