pygit — Build Your Own Git

A functional subset of Git built from scratch — same object model, same content addressing, same DAG structure. Blobs, trees, and commits are stored as zlib-compressed SHA-1-addressed objects, just like real Git.

Supported Commands

Command	Description
pygit init	Initialize .pygit/ repository
pygit hash-object <file>	Store file as blob, print SHA-1
pygit cat-file <sha>	Read and decode a stored object
pygit add <file>	Stage file (update index)
pygit write-tree	Write index snapshot as tree object
pygit commit-tree <tree> -m <msg>	Create commit object
pygit commit -m <msg>	Stage all + commit (porcelain)
pygit log	Walk commit DAG, print history
pygit checkout <sha>	Restore working tree from commit
pygit status	Compare index vs working tree
pygit diff	Show unstaged changes

Quick Start

python3 pygit.py init
echo "hello" > hello.txt
python3 pygit.py add hello.txt
python3 pygit.py commit -m "Initial commit"
python3 pygit.py log

commit a3f1c8...
Author: pygit
Date:   2024-01-01

    Initial commit

Object Model

Blob   — raw file content (zlib-compressed, SHA-1 addressed)
Tree   — directory snapshot: list of (mode, name, sha) entries
Commit — tree SHA + parent SHA + author + message
          ↓
          .pygit/objects/ab/cdef123...  (2-char dir / 38-char file)

What I Learned

• How Git's content-addressable storage works: SHA-1 hash of "type len\0content" determines the file path
• Why commits form a DAG (directed acyclic graph): each commit stores its parent's SHA, enabling history traversal
• How git add works: the index is a snapshot of staged file paths → blob SHAs, not file contents
• The difference between plumbing (hash-object, write-tree) and porcelain (commit, log) commands

---

Built as part of the Build Your Own X curriculum.