feat/m1-object-store-skeleton

minigit/cli.py

@@ -13,6 +13,7 @@
 
 from minigit import __version__
 from minigit.errors import MiniGitError
+from minigit.objects import register_subcommands
 
 
 def _register_commands(subparsers) -> None:
@@ -27,6 +28,8 @@ def register_subcommands(subparsers):
             parser.set_defaults(handler=cmd_add)
     """
 
+    register_subcommands(subparsers)
+
 
 def build_parser() -> argparse.ArgumentParser:
     parser = argparse.ArgumentParser(prog="minigit", description="A version control system.")

minigit/objects.py

@@ -8,3 +8,75 @@
 
 Build the `ObjectStore` class here, per the interface contract.
 """
+
+import zlib
+from pathlib import Path
+
+from minigit.errors import ObjectNotFoundError
+
+
+class ObjectStore:
+    root: Path
+    objects_dir: Path
+    _fake_store: dict[str, tuple[str, bytes]]
+
+    def __init__(self, repo_path=".") -> None:
+        """
+        Initialize the object store.
+        """
+        self.root = Path(repo_path)
+        self.objects_dir = self.root / ".minigit" / "objects"
+        self._fake_store = {}
+
+    def hash_object(self, data: bytes, obj_type: str) -> str:
+        """
+        Return the SHA-1 hash of the object, given its data and type.
+        """
+        # placeholder - real SHA-1 of "<type> <len>\0<content>" lands Week 2
+        return f"{zlib.crc32(obj_type.encode() + data):040x}"
+
+    def write_object(self, data: bytes, obj_type: str) -> str:
+        """
+        Writes the object's hash into self._fake_store and returns the hash.
+        Allow duplicates to be written.
+        """
+        obj_hash = self.hash_object(data, obj_type)
+        self._fake_store[obj_hash] = (obj_type, data)
+        return obj_hash
+
+    def read_object(self, hash: str) -> tuple[str, bytes]:
+        """
+        Reads the object from self._fake_store and returns a tuple of (type, data).
+        Raise ObjectNotFoundError(hash) if the object is not found.
+        """
+        if hash not in self._fake_store:
+            raise ObjectNotFoundError(hash)
+
+        return self._fake_store[hash]
+
+
+_cli_store = ObjectStore()
+
+
+def run_hash_object(args) -> int:
+    with open(args.path, "rb") as f:
+        data = f.read()
+    obj_hash = _cli_store.write_object(data, "blob")
+    print(obj_hash)
+    return 0
+
+
+def run_cat_file(args) -> int:
+    _, obj_data = _cli_store.read_object(args.hash)
+    print(obj_data.decode("utf-8", errors="replace"), end="")
+    return 0
+
+
+def register_subcommands(subparsers):
+    hash_parser = subparsers.add_parser("hash-object")
+    hash_parser.add_argument("path")
+    hash_parser.set_defaults(handler=run_hash_object)
+
+    cat_parser = subparsers.add_parser("cat-file")
+    cat_parser.add_argument("hash")
+    cat_parser.set_defaults(handler=run_cat_file)

tests/test_objects.py

@@ -0,0 +1,98 @@
+"""Object tests: Tests for module 1 - object storage."""
+
+from pathlib import Path
+
+import pytest
+
+from minigit.cli import main
+from minigit.errors import ObjectNotFoundError
+from minigit.objects import ObjectStore
+
+
+def test_round_trip(tmp_path: Path) -> None:
+    """
+    Test that we can write an object and then read it back.
+    """
+
+    store = ObjectStore(tmp_path)
+    obj_hash = store.write_object(b"hi", "blob")
+    assert store.read_object(obj_hash) == ("blob", b"hi")
+
+
+def test_identical_objects_same_hash(tmp_path: Path):
+    """
+    Test that writing the same object twice returns the same hash.
+    """
+    store = ObjectStore(tmp_path)
+
+    hash1 = store.hash_object(b"hi", "blob")
+    hash2 = store.hash_object(b"hi", "blob")
+
+    assert hash1 == hash2
+
+
+def test_different_objects_different_hashes(tmp_path: Path):
+    """
+    Test that writing different objects returns different hashes.
+    """
+    store = ObjectStore(tmp_path)
+
+    hash1 = store.hash_object(b"hi", "blob")
+    hash2 = store.hash_object(b"hello", "blob")
+
+    assert hash1 != hash2
+
+
+def test_idempotent_write(tmp_path: Path):
+    """
+    Test that writing the same object twice returns the same hash and does not raise an error.
+    """
+    store = ObjectStore(tmp_path)
+
+    hash1 = store.write_object(b"hi", "blob")
+    hash2 = store.write_object(b"hi", "blob")
+
+    assert hash1 == hash2
+    assert store.read_object(hash1) == ("blob", b"hi")
+
+
+def test_unknown_hash_raises(tmp_path: Path):
+    """
+    Test that reading an unknown hash raises ObjectNotFoundError.
+    """
+    store = ObjectStore(tmp_path)
+
+    with pytest.raises(ObjectNotFoundError):
+        store.read_object("does-not-exist")
+
+
+def test_hash_object_cli_print(tmp_path, capsys):
+    """
+    Test that the hash-object CLI command prints the correct hash.
+    """
+
+    test_file = tmp_path / "test.txt"
+    test_file.write_bytes(b"hello minigit")
+
+    assert main(["hash-object", str(test_file)]) == 0
+
+    captured = capsys.readouterr()
+    obj_hash = captured.out.strip()
+
+    assert len(obj_hash) == 40  # current placeholder hash has length 40, like SHA-1
+
+
+def test_cat_file_cli_print(tmp_path, capsys):
+    """
+    Test that the cat-file CLI command prints the correct object data.
+    """
+
+    test_file = tmp_path / "test.txt"
+    test_file.write_bytes(b"hello minigit")
+
+    assert main(["hash-object", str(test_file)]) == 0
+    obj_hash = capsys.readouterr().out.strip()
+
+    assert main(["cat-file", obj_hash]) == 0
+    captured = capsys.readouterr()
+    assert captured.out == "hello minigit"