Repositories
From 2023 to 2026 the code of every file ever published to PyPI lived in 272 git repositories on GitHub, named pypi-mirror-1 to pypi-mirror-272.
They belonged to the PyPI Data project that ran on this domain (see About). Git stores each unique file once, so the whole of PyPI fit into about 440 GB. Each repository held the projects released in one span of time, and this page listed their sizes and date ranges.
The repositories were removed in 2026 together with the original pypi-data GitHub organisation, and no public copy is known. With Tom Forbes' agreement, the pypi-data name on GitHub now holds the code of this site's scanner.