Python package resources: read data from a ZIP-backed package

Operation contract

The fixture creates an owned package archive containing region labels, imports that package from the ZIP and reads its data with files(). The resource object is traversable but not necessarily a pathlib.Path. The result is text with an explicit encoding. The archive and interpreter path change are owned and cleaned up at the end.

Failure and ownership boundary

Do not form paths from arbitrary received resource names. APIs that require a real file can use as_file inside its context and must not retain a temporary extraction path afterward. Package data also has to be included in the distribution build. Python pyproject.toml: build a wheel and inspect its metadata and Python zipapp: build an executable archive without bundling an interpreter cover those packaging boundaries.

Working program

from importlib.resources import files
import importlib
from pathlib import Path
import sys
import tempfile
from zipfile import ZipFile

with tempfile.TemporaryDirectory() as directory:
    archive = Path(directory) / "receipt_regions.zip"
    with ZipFile(archive, "w") as package:
        package.writestr("receipt_regions/__init__.py", "")
        package.writestr("receipt_regions/regions.txt", "DEL\nBOM\n")
    sys.path.insert(0, str(archive))
    try:
        importlib.import_module("receipt_regions")
        text = files("receipt_regions").joinpath("regions.txt").read_text(encoding="utf-8")
        print(text.splitlines())
    finally:
        sys.path.remove(str(archive))
        sys.modules.pop("receipt_regions", None)

Output

['DEL', 'BOM']

Costs and limits

Reading n resource bytes allocates the decoded text and may decompress archive data. A tiny package does not establish safe budgets for arbitrary compressed input. This archive is generated locally, not received from an untrusted upload.

Common Mistakes

• A package resource is not always a filesystem Path.
• Keep a temporary extracted path within its resource context.

Connected lessons

Python pyproject.toml: build a wheel and inspect its metadata, Python zipapp: build an executable archive without bundling an interpreter, Python pathlib files: specify encoding and close the resource owner.

Check the next state boundary

Python wheel installation: test the built artifact outside its source directory.