This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
- ALL tests MUST pass for code to be considered complete and working
- Never describe code as "working as expected" if there are ANY failing tests
- Even if specific feature tests pass, failing tests elsewhere indicate broken functionality
- Changes that break existing tests must be fixed before considering implementation complete
- A successful implementation must pass linting, type checking, AND all existing tests
libvcs is a lite, typed Python tool for:
- Detecting and parsing URLs for Git, Mercurial, and Subversion repositories
- Providing command abstractions for git, hg, and svn
- Synchronizing repositories locally
- Creating pytest fixtures for testing with temporary repositories
The library powers vcspull, a tool for managing and synchronizing multiple git, svn, and mercurial repositories.
This project uses:
- Python 3.9+
- uv for dependency management
- ruff for linting and formatting
- mypy for type checking
- pytest for testing
# Install dependencies
uv pip install --editable .
uv pip sync
# Install with development dependencies
uv pip install --editable . -G dev# Run all tests
just test
# or directly with pytest
uv run pytest
# Run a single test file
uv run pytest tests/sync/test_git.py
# Run a specific test
uv run pytest tests/sync/test_git.py::test_remotes
# Run tests with test watcher
just start
# or
uv run ptw .# Run ruff for linting
just ruff
# or directly
uv run ruff check .
# Format code with ruff
just ruff-format
# or directly
uv run ruff format .
# Run ruff linting with auto-fixes
uv run ruff check . --fix --show-fixes
# Run mypy for type checking
just mypy
# or directly
uv run mypy src tests
# Watch mode for linting (using entr)
just watch-ruff
just watch-mypyFollow this workflow for code changes:
- Format First:
uv run ruff format . - Run Tests:
uv run pytest - Run Linting:
uv run ruff check . --fix --show-fixes - Check Types:
uv run mypy - Verify Tests Again:
uv run pytest
# Build documentation
just build-docs
# Start documentation server with auto-reload
just start-docs
# Update documentation CSS/JS
just design-docslibvcs is organized into three main modules:
-
URL Detection and Parsing (
libvcs.url)- Base URL classes in
url/base.py - VCS-specific implementations in
url/git.py,url/hg.py, andurl/svn.py - URL registry in
url/registry.py - Constants in
url/constants.py
- Base URL classes in
-
Command Abstraction (
libvcs.cmd)- Command classes for git, hg, and svn in
cmd/git.py,cmd/hg.py, andcmd/svn.py - Built on top of Python's subprocess module (via
_internal/subprocess.py)
- Command classes for git, hg, and svn in
-
Repository Synchronization (
libvcs.sync)- Base sync classes in
sync/base.py - VCS-specific sync implementations in
sync/git.py,sync/hg.py, andsync/svn.py
- Base sync classes in
-
Internal Utilities (
libvcs._internal)- Subprocess wrappers in
_internal/subprocess.py - Data structures in
_internal/dataclasses.pyand_internal/query_list.py - Runtime helpers in
_internal/run.pyand_internal/shortcuts.py
- Subprocess wrappers in
-
pytest Plugin (
libvcs.pytest_plugin)- Provides fixtures for creating temporary repositories for testing
libvcs uses pytest for testing with many custom fixtures. The pytest plugin (pytest_plugin.py) defines fixtures for creating temporary repositories for testing. These include:
create_git_remote_repo: Creates a git repository for testingcreate_hg_remote_repo: Creates a Mercurial repository for testingcreate_svn_remote_repo: Creates a Subversion repository for testinggit_repo,svn_repo,hg_repo: Pre-made repository instancesset_home,gitconfig,hgconfig,git_commit_envvars: Environment fixtures
These fixtures handle setup and teardown automatically, creating isolated test environments.
For running tests with actual VCS commands, tests will be skipped if the corresponding VCS binary is not installed.
def test_repo_sync(git_repo):
# git_repo is already a GitSync instance with a clean repository
# Use it directly in your tests
assert git_repo.get_revision() == "initial"Use typing.NamedTuple for parameterized tests:
class RepoFixture(t.NamedTuple):
test_id: str # For test naming
repo_args: dict[str, t.Any]
expected_result: str
@pytest.mark.parametrize(
list(RepoFixture._fields),
REPO_FIXTURES,
ids=[test.test_id for test in REPO_FIXTURES],
)
def test_sync(
# Parameters and fixtures...
):
# Test implementation- Use namespace imports:
import enuminstead offrom enum import Enum - For typing, use
import typing as tand access via namespace:t.NamedTuple, etc. - Use
from __future__ import annotationsat the top of all Python files
Follow Python community conventions (Django, pytest, Sphinx patterns):
Method naming:
- Use
get_*prefix for methods that perform I/O or subprocess calls (e.g.,get_remotes(),get_revision()) - Use
is_*prefix for boolean checks (e.g.,is_valid()) - Use
has_*prefix for existence checks (e.g.,has_remote())
Parameter naming:
- Use descriptive names instead of underscore-prefixed built-in shadows
- BAD:
_all,_type,_list(cryptic, non-standard) - GOOD:
all_remotes,include_all,file_type,path_list(self-documenting)
Examples:
# BAD - cryptic underscore prefix
def fetch(_all: bool = False): ...
def rev_list(_all: bool = False): ...
# GOOD - descriptive parameter names
def fetch(all_remotes: bool = False): ...
def rev_list(include_all: bool = False): ...
# BAD - inconsistent getter naming
def remotes(): ... # No prefix
def get_revision(): ... # Has prefix
# GOOD - consistent getter naming for subprocess calls
def get_remotes(): ...
def get_revision(): ...Rationale: Major Python projects (Django, pytest, Sphinx) don't use _all style prefixes. They either use the built-in name directly as a keyword-only argument, or use descriptive alternatives. Underscore prefixes are reserved for private/internal parameters only.
Follow NumPy docstring style for all functions and methods:
"""Short description of the function or class.
Detailed description using reStructuredText format.
Parameters
----------
param1 : type
Description of param1
param2 : type
Description of param2
Returns
-------
type
Description of return value
"""All functions and methods MUST have working doctests. Doctests serve as both documentation and tests.
CRITICAL RULES:
- Doctests MUST actually execute - never comment out
asyncio.run()or similar calls - Doctests MUST NOT be converted to
.. code-block::as a workaround (code-blocks don't run) - If you cannot create a working doctest, STOP and ask for help
Available tools for doctests:
doctest_namespacefixtures:tmp_path,asyncio,create_git_remote_repo,create_hg_remote_repo,create_svn_remote_repo,example_git_repo- Ellipsis for variable output:
# doctest: +ELLIPSIS - Update
pytest_plugin.pyto add new fixtures todoctest_namespace
# doctest: +SKIP is NOT permitted - it's just another workaround that doesn't test anything. If a VCS binary might not be installed, pytest already handles skipping via skip_if_binaries_missing. Use the fixtures properly.
Async doctest pattern:
>>> async def example():
... result = await some_async_function()
... return result
>>> asyncio.run(example())
'expected output'Using fixtures in doctests:
>>> git = Git(path=tmp_path) # tmp_path from doctest_namespace
>>> git.run(['status'])
'...'When output varies, use ellipsis:
>>> git.clone(url=f'file://{create_git_remote_repo()}') # doctest: +ELLIPSIS
'Cloning into ...'Format commit messages as:
Component/File(commit-type[Subcomponent/method]): Concise description
why: Explanation of necessity or impact.
what:
- Specific technical changes made
- Focused on a single topic
Common commit types:
- feat: New features or enhancements
- fix: Bug fixes
- refactor: Code restructuring without functional change
- docs: Documentation updates
- chore: Maintenance (dependencies, tooling, config)
- test: Test-related updates
- style: Code style and formatting
Example:
url/git(feat[GitURL]): Add support for custom SSH port syntax
why: Enable parsing of Git URLs with custom SSH ports
what:
- Add port capture to SCP_REGEX pattern
- Update GitURL.to_url() to include port if specified
- Add tests for the new functionality
For multi-line commits, use heredoc to preserve formatting:
git commit -m "$(cat <<'EOF'
feat(Component[method]) add feature description
why: Explanation of the change.
what:
- First change
- Second change
EOF
)"Maintain CHANGES (changelog) and MIGRATION (upgrade guide) for all user-facing changes.
File structure:
CHANGES: Organized by version with sections in this order of precedence:### Breaking changes- API changes that require user action### New features- New functionality### Bug fixes- Corrections to existing behavior### Documentation- Doc-only changes### Developmentor### Internal- Tooling, CI, refactoring
MIGRATION: Detailed migration instructions with before/after examples
Maintenance-only releases: For releases with no user-facing changes (only internal/development work), use:
_Maintenance only, no bug fixes, or new features_PR references - where to put them:
- DO: Put PR number in section headers or at end of bullet items in the files
- DON'T: Put PR number in commit message titles (causes linkback notification noise in the PR)
For larger changes with dedicated sections:
#### API Naming Consistency (#507)
Renamed parameters and methods...For smaller changes in a list:
### Bug fixes
- Fix argument expansion in `rev_list` (#455)
- Remove unused command: `Svn.mergelist` (#450)Commit messages should NOT include PR numbers:
# GOOD - no PR in commit message
CHANGES(docs): Document breaking API changes for 0.39.x
# BAD - PR in commit message creates noise
CHANGES(docs): Document breaking API changes for 0.39.x (#507)The PR reference in the file content creates a clean linkback when the PR merges, while keeping commit messages focused and avoiding duplicate notifications.
When stuck in debugging loops:
- Pause and acknowledge the loop
- Minimize to MVP: Remove all debugging cruft and experimental code
- Document the issue comprehensively for a fresh approach
- Format for portability (using quadruple backticks)