introduce repo concept, make plumbing CLI available as dvc-data command

We need to introduce repo concept so that the plumbing CLI command
is compatible with `dvc init` created repositories.

Author: skshetry

noxfile.py

@@ -28,7 +28,7 @@ def lint(session: nox.Session) -> None:
     args = *(session.posargs or ("--show-diff-on-failure",)), "--all-files"
     session.run("pre-commit", "run", *args)
     session.run("python", "-m", "mypy")
-    session.run("python", "-m", "pylint", "src", "cli.py")
+    session.run("python", "-m", "pylint", "src")
 
 
 @nox.session

pyproject.toml

@@ -62,7 +62,7 @@ warn_no_return = true
 warn_redundant_casts = true
 warn_unreachable = true
 ignore_missing_imports = true
-files = ["src", "tests", "cli.py"]
+files = ["src", "tests"]
 
 [tool.pylint.message_control]
 enable = ["c-extension-no-member", "no-else-return"]

setup.cfg

@@ -32,6 +32,10 @@ install_requires=
     nanotime>=0.5.2
 
 [options.extras_require]
+cli =
+    typer>=0.4,<0.5
+all =
+    %(cli)s
 tests =
     pytest==7.1.2
     pytest-sugar==0.9.4
@@ -45,8 +49,12 @@ tests =
     universal-pathlib==0.0.19
     s3fs[boto3]>=2022.02.0; python_version < '3.11'
 dev =
+    %(all)s
     %(tests)s
-    typer>=0.4,<0.5
+
+[options.entry_points]
+console_scripts =
+    dvc-data = dvc_data.__main__:app
 
 [options.packages.find]
 exclude =
@@ -67,4 +75,4 @@ select = B,C,E,F,W,T4,B902,T,P
 show_source = true
 count = true
 per-file-ignores =
-    cli.py:B008
+    src/dvc_data/cli.py:B008

src/dvc_data/__main__.py

@@ -0,0 +1,17 @@
+try:
+    from .cli import app
+except ImportError:  # pragma: no cover
+
+    def app():  # type: ignore[misc]
+        print(
+            "dvc-data could not run because the required "
+            "dependencies are not installed.\n"
+            "Please install it with: pip install 'dvc-data[cli]'"
+        )
+        return 1
+
+
+if __name__ == "__main__":
+    import sys
+
+    sys.exit(app())

cli.py renamed to src/dvc_data/cli.py

@@ -1,5 +1,3 @@
-#! /usr/bin/env python3
-
 import enum
 import hashlib
 import json
@@ -28,6 +26,7 @@
 from dvc_data.hashfile.hash_info import HashInfo
 from dvc_data.hashfile.state import State
 from dvc_data.objects.tree import Tree, merge
+from dvc_data.repo import NotARepo, Repo
 from dvc_data.stage import stage as _stage
 from dvc_data.transfer import transfer as _transfer
 
@@ -59,9 +58,6 @@
     "LinkEnum", {lt: lt for lt in ["reflink", "hardlink", "symlink", "copy"]}
 )
 SIZE_HELP = "Human readable size, eg: '1kb', '100Mb', '10GB' etc"
-ODB_PATH = typer.Option(
-    ".dvc/cache", help="Path to the root of the odb", envvar="ODB_PATH"
-)
 
 
 class Application(typer.Typer):
@@ -138,21 +134,27 @@ def from_shortoid(odb: HashFileDB, oid: str) -> str:
         raise typer.Exit(1) from exc
 
 
-def get_odb(path, **config):
-    state = State(root_dir=os.getcwd(), tmp_dir=os.path.join(path, "tmp"))
-    return HashFileDB(LocalFileSystem(), path, state=state, **config)
+def get_odb(**config):
+    try:
+        repo = Repo.discover()
+    except NotARepo as exc:
+        typer.echo(exc, err=True)
+        raise typer.Abort(1)
+
+    state = State(root_dir=repo.root, tmp_dir=repo.tmp_dir)
+    return HashFileDB(repo.fs, repo.object_dir, state=state, **config)
 
 
 @app.command(help="Oid to path")
-def o2p(oid: str = typer.Argument(..., allow_dash=True), db: str = ODB_PATH):
-    odb = get_odb(db)
+def o2p(oid: str = typer.Argument(..., allow_dash=True)):
+    odb = get_odb()
     path = odb.oid_to_path(from_shortoid(odb, oid))
     typer.echo(path)
 
 
 @app.command(help="Path to Oid")
-def p2o(path: Path = typer.Argument(..., allow_dash=True), db: str = ODB_PATH):
-    odb = get_odb(db)
+def p2o(path: Path = typer.Argument(..., allow_dash=True)):
+    odb = get_odb()
     fs_path = relpath(path)
     if fs_path == "-":
         fs_path = sys.stdin.read().strip()
@@ -164,10 +166,9 @@ def p2o(path: Path = typer.Argument(..., allow_dash=True), db: str = ODB_PATH):
 @app.command(help="Provide content of the objects")
 def cat(
     oid: str = typer.Argument(..., allow_dash=True),
-    db: str = ODB_PATH,
     check: bool = typer.Option(False, "--check", "-c"),
 ):
-    odb = get_odb(db)
+    odb = get_odb()
     oid = from_shortoid(odb, oid)
     if check:
         try:
@@ -184,11 +185,10 @@ def cat(
 @app.command(help="Stage and optionally write object to the database")
 def stage(
     path: Path = dir_file_type,
-    db: str = ODB_PATH,
     write: bool = typer.Option(False, "--write", "-w"),
     shallow: bool = False,
 ):
-    odb = get_odb(db)
+    odb = get_odb()
     fs_path = relpath(path)
 
     fs = odb.fs
@@ -210,8 +210,8 @@ def stage(
 
 @app.command("ls", help="List objects in a tree")
 @app.command("ls-tree", help="List objects in a tree")
-def ls(oid: str = typer.Argument(..., allow_dash=True), db: str = ODB_PATH):
-    odb = get_odb(db)
+def ls(oid: str = typer.Argument(..., allow_dash=True)):
+    odb = get_odb()
     oid = from_shortoid(odb, oid)
     try:
         tree = Tree.load(odb, HashInfo("md5", oid))
@@ -224,8 +224,8 @@ def ls(oid: str = typer.Argument(..., allow_dash=True), db: str = ODB_PATH):
 
 
 @app.command(help="Verify objects in the database")
-def fsck(db: str = ODB_PATH):
-    odb = get_odb(db)
+def fsck():
+    odb = get_odb()
     ret = 0
     for oid in odb.all():
         try:
@@ -237,10 +237,8 @@ def fsck(db: str = ODB_PATH):
 
 
 @app.command(help="Diff two objects in the database")
-def diff(
-    short_oid1, short_oid2: str, db: str = ODB_PATH, unchanged: bool = False
-):
-    odb = get_odb(db)
+def diff(short_oid1, short_oid2: str, unchanged: bool = False):
+    odb = get_odb()
     obj1 = odb.get(from_shortoid(odb, short_oid1))
     obj2 = odb.get(from_shortoid(odb, short_oid2))
     d = _diff(load(odb, obj1.hash_info), load(odb, obj2.hash_info), odb)
@@ -271,8 +269,8 @@ def _prepare_info(entry):
 
 
 @app.command(help="Merge two trees and optionally write to the database.")
-def merge_tree(oid1: str, oid2: str, db: str = ODB_PATH, force: bool = False):
-    odb = get_odb(db)
+def merge_tree(oid1: str, oid2: str, force: bool = False):
+    odb = get_odb()
     oid1 = from_shortoid(odb, oid1)
     oid2 = from_shortoid(odb, oid2)
     obj1 = load(odb, odb.get(oid1).hash_info)
@@ -300,7 +298,7 @@ def merge_tree(oid1: str, oid2: str, db: str = ODB_PATH, force: bool = False):
 
 
 @app.command()
-def update_tree(oid: str, patch_file: Path = file_type, db: str = ODB_PATH):
+def update_tree(oid: str, patch_file: Path = file_type):
     """Update tree contents virtually with a patch file in json format.
 
     Example patch file for reference:
@@ -316,7 +314,7 @@ def update_tree(oid: str, patch_file: Path = file_type, db: str = ODB_PATH):
 
     Example: ./cli.py update-tree f23d4 patch.json
     """
-    odb = get_odb(db)
+    odb = get_odb()
     oid = from_shortoid(odb, oid)
     obj = load(odb, odb.get(oid).hash_info)
     assert isinstance(obj, Tree)
@@ -374,9 +372,8 @@ def checkout(
     type: List[LinkEnum] = typer.Option(  # pylint: disable=redefined-builtin
         ["copy"]
     ),
-    db: str = ODB_PATH,
 ):
-    odb = get_odb(db, type=[t.value for t in type])
+    odb = get_odb(type=[t.value for t in type])
     oid = from_shortoid(odb, oid)
     obj = load(odb, odb.get(oid).hash_info)
     with Tqdm(total=len(obj), desc="Checking out", unit="obj") as pbar:

src/dvc_data/repo.py

@@ -0,0 +1,55 @@
+import os
+
+from dvc_objects.fs import LocalFileSystem
+from dvc_objects.fs.base import FileSystem
+
+
+class NotARepo(Exception):
+    pass
+
+
+localfs = LocalFileSystem()
+
+
+class Repo:
+    def __init__(self, root: str = "", fs: FileSystem = None) -> None:
+        fs = fs or localfs
+        root = root or fs.path.getcwd()
+        control_dir: str = os.getenv("DVC_DIR") or fs.path.join(root, ".dvc")
+
+        if not fs.isdir(control_dir):
+            raise NotARepo(f"{root} is not a data repo.")
+
+        self.fs = fs or localfs
+        self.root = root
+        self._control_dir = control_dir
+        self._tmp_dir: str = fs.path.join(self._control_dir, "tmp")
+        self._object_dir: str = fs.path.join(self._control_dir, "cache")
+
+    @classmethod
+    def discover(
+        cls,
+        start: str = ".",
+        fs: FileSystem = None,
+    ) -> "Repo":
+        remaining = True
+        fs = fs or localfs
+        path = start = fs.path.abspath(start)
+        while remaining:
+            try:
+                return cls(path, fs)
+            except NotARepo:
+                path, remaining = fs.path.split(path)
+        raise NotARepo(f"No data repository was found at {start}")
+
+    @property
+    def control_dir(self):
+        return self._control_dir
+
+    @property
+    def tmp_dir(self):
+        return self._tmp_dir
+
+    @property
+    def object_dir(self):
+        return self._object_dir