pip

pip is the de facto standard tool for installing and managing Python packages, maintained by the Python Packaging Authority (PyPA) and bundled with every modern CPython installation. It resolves version constraints across a dependency graph, downloads wheels or source distributions from PyPI (or any configured index), and handles the full install/upgrade/uninstall lifecycle from the command line.

Unlike most entries in this directory, pip isn’t imported into application code — it’s the tool developers run (pip install, pip freeze, pip list) to manage the packages their code depends on, making it one of the most-invoked pieces of software in the Python ecosystem.

What You Get

• The pip install, pip uninstall, pip freeze, pip list, and pip download command-line subcommands
• A dependency resolver that computes a consistent set of package versions across a project’s declared and transitive requirements
• Support for installing from PyPI, private indexes, VCS URLs (git/hg/svn), local paths, and requirements files
• Wheel building and caching so repeated installs of the same package/version don’t re-download or re-compile
• A vendored copy of key dependencies (_vendor/) so pip can bootstrap itself without depending on anything else being installed first

Common Use Cases

• Installing project dependencies from a requirements.txt or pyproject.toml in local development, CI, or a Docker build
• Freezing an environment’s exact installed versions for reproducible builds (pip freeze > requirements.txt)
• Installing a package directly from a git branch or local editable path during development (pip install -e .)
• Managing per-project isolated environments in combination with venv or virtualenv

Under The Hood

Architecture - The CLI entry point (src/pip/_internal/cli) dispatches to per-command modules under commands/; installation itself flows through resolution/ (dependency version solving), index/ (querying PyPI’s simple/JSON APIs), distributions/ and wheel_builder.py (unpacking sdists/wheels), and operations/ (the actual filesystem install step) — a clean separation between “decide what to install” and “do the install.”

Tech Stack - Pure Python, deliberately vendoring its own copies of dependencies like packaging, certifi, urllib3, and resolvelib under _vendor/ so that pip — which is often the first thing run in a fresh environment — never has a chicken-and-egg dependency problem.

Code Quality - An extensive tests/ tree split into unit/, functional/, and typing/ suites, run against multiple Python versions in CI; docs/ includes both user-facing docs and a news/ directory of per-PR changelog fragments, reflecting PyPA’s mature release process.

API Design - As a CLI rather than an importable library, its “API” is the command-line surface (pip install, pip freeze, etc.) plus a stable --format=json output for some subcommands; PyPA explicitly discourages importing pip._internal as a Python API, keeping the internal module structure free to change between releases.