ReproCheck 0.11.0b1

ReproCheck 0.11.0b1

This is the first public beta of ReproCheck.

ReproCheck audits a Python project for reproducibility problems that are cheap to
fix and expensive to discover later: an untracked data path, a floating action
tag, a dependency pinned in one file and missing from another. It reads the
project, writes a report, and can apply one safe fix with an audit trail.

What it does

• Deterministic static audit. One rule per problem, each with a stable id
  (RC002, RC140, ...), a severity, a confidence and the evidence it fired on.
  The same tree always produces the same findings: no model, no heuristics that
  change between runs, no network calls.
• Python and dependency consistency checks. A single version declaration
  where there should be one, dependencies declared in one place and used in
  another, unpinned or conflicting requirements across pyproject.toml,
  requirements*.txt and lock files.
• GitHub Actions reference checks. Floating action tags, and a version
  declared by one file and required by another.
• Reproduction attempt in an isolated working copy. reprocheck reproduce
  builds an environment for the project in a temporary workspace. The analysed
  project is never modified, and a fresh baseline is compared against the saved
  one so any difference is visible.
• Optional runtime checks. --runtime-checks imports the project and runs
  pytest collection. It never executes test functions.
• Human-readable reports. JSON for machines, Markdown for people, plus a
  terminal summary.
• Baselines and diffs. Save a report as a reference and compare the next one
  against it, so "what changed since the last run" is an answer, not a diff of
  two JSON files by hand.
• Deterministic fix suggestions. The suggestion engine reads the same facts
  as the checks, so a proposal can never contain a pattern the check did not
  prove was missing.
• Safe fix application and rollback. --apply writes a backup and a JSON
  audit record for every change; --rollback restores the original bytes.

Install

pip install reprocheck==0.11.0b1

Requires Python 3.11 or newer. One runtime dependency, packaging.

reprocheck scan .
reprocheck suggest .

Beta limitations

Please read these before trusting a report.

• environment.yml and Conda are not analysed. Dependencies are read from
  pyproject.toml and requirements*.txt.
• CI that does not run on GitHub Actions is not analysed.
• setup.py and setup.cfg have limited coverage.
• A project whose version is derived from .git can fail to reproduce from a
  copy, because the copy is not a repository.
• A virtual environment is not a security sandbox. reproduce runs the
  analysed project's build backend, and --runtime-checks imports its code,
  with your privileges. See SECURITY.md
  before pointing it at code you do not trust, and use a container or a VM.
• Only RC140 has an applicable fix at the moment, one suggestion at a time.
  There is no "apply all" and no interactive prompt.
• PASS means only that the checks which ran succeeded. Anything under
  Not verified — the full test suite, external datasets, remote URLs, index
  state — remains unknown.
• ReproCheck reports facts about a repository. It is not proof that a
  scientific study is reproducible, and it does not attempt to establish that.

Verified

• Python 3.11.16, 3.12.3 and 3.13.15: full suite green locally, and green in
  GitHub Actions on the same matrix.
• Wheel and sdist installed and smoke-tested in clean environments, from this
  release and from the TestPyPI index.
• pint, astropy and napari produce identical findings from an installed
  release, a locally built wheel and the published artifact.

Assets

reprocheck-0.11.0b1-py3-none-any.whl	171.713 bytes, sha256 3186b8a2…
reprocheck-0.11.0b1.tar.gz	272.613 bytes, sha256 190cd8e7…

The sha256 digests of these files are identical to the ones computed locally
before upload, on both PyPI and TestPyPI.

Reporting

Bugs and feature requests: https://github.com/RafaelMoreiraDev/ReproCheck/issues
Support scope: SUPPORT.md
Security: SECURITY.md