Merge pull request #1138 from dandi/digest-empty-zarr

Make internal Zarr-checksumming tree types support empty Zarrs

dandi/cli/tests/test_digest.py

@@ -1,3 +1,4 @@
+import os
 from pathlib import Path
 
 from click.testing import CliRunner
@@ -52,3 +53,12 @@ def test_digest_zarr():
         r = runner.invoke(digest, ["--digest", "zarr-checksum", "sample.zarr"])
         assert r.exit_code == 0
         assert r.output == "sample.zarr: 4313ab36412db2981c3ed391b38604d6-5--1516\n"
+
+
+def test_digest_empty_zarr(tmp_path: Path) -> None:
+    runner = CliRunner()
+    with runner.isolated_filesystem():
+        os.mkdir("empty.zarr")
+        r = runner.invoke(digest, ["--digest", "zarr-checksum", "empty.zarr"])
+        assert r.exit_code == 0
+        assert r.output == "empty.zarr: 481a2f77ab786a0f45aafd5db0971caa-0--0\n"

dandi/support/digests.py

@@ -15,7 +15,7 @@
 import hashlib
 import logging
 import os.path
-from pathlib import Path
+from pathlib import Path, PurePath
 from typing import Dict, List, Optional, Tuple, Union, cast
 
 from dandischema.digests.dandietag import DandiETag
@@ -123,10 +123,10 @@ def digest_file(f: Path) -> Tuple[Path, str, int]:
             dgst = md5file_nocache(f)
         return (f, dgst, os.path.getsize(f))
 
-    zcc = ZCDirectory()
+    zcc = ZCTree()
     for p, digest, size in threaded_walk(path, digest_file):
         zcc.add(p.relative_to(path), digest, size)
-    return zcc.get_digest_size()[0]
+    return zcc.get_digest()
 
 
 @dataclass
@@ -169,10 +169,22 @@ def get_digest_size(self) -> Tuple[str, int]:
             size += sz
         return (cast(str, get_checksum(files, dirs)), size)
 
-    def add(self, path: Path, digest: str, size: int) -> None:
+
+@dataclass
+class ZCTree:
+    """
+    Tree root node used for building an in-memory tree of Zarr entries and
+    their digests when calculating a complete Zarr checksum
+
+    :meta private:
+    """
+
+    tree: ZCDirectory = field(init=False, default_factory=ZCDirectory)
+
+    def add(self, path: PurePath, digest: str, size: int) -> None:
         *dirs, name = path.parts
         parts = []
-        d = self
+        d = self.tree
         for dirname in dirs:
             parts.append(dirname)
             e = d.children.setdefault(dirname, ZCDirectory())
@@ -185,6 +197,14 @@ def add(self, path: Path, digest: str, size: int) -> None:
         assert name not in d.children, f"File {pstr} encountered twice"
         d.children[name] = ZCFile(digest=digest, size=size)
 
+    def get_digest(self) -> str:
+        if self.tree.children:
+            return self.tree.get_digest_size()[0]
+        else:
+            # get_checksum() refuses to operate on empty directories, so we
+            # return the checksum for an empty Zarr ourselves:
+            return "481a2f77ab786a0f45aafd5db0971caa-0--0"
+
 
 def md5file_nocache(filepath: Union[str, Path]) -> str:
     """

dandi/support/tests/test_digests.py

@@ -9,7 +9,6 @@
 
 from pathlib import Path
 
-import pytest
 from pytest_mock import MockerFixture
 
 from .. import digests
@@ -79,9 +78,7 @@ def test_get_zarr_checksum(mocker: MockerFixture, tmp_path: Path) -> None:
     assert get_zarr_checksum(tmp_path) == "25627e0fc7c609d10100d020f7782a25-8--197"
     assert get_zarr_checksum(sub1) == "64af93ad7f8d471c00044d1ddbd4c0ba-4--97"
 
-    with pytest.raises(ValueError) as excinfo:
-        get_zarr_checksum(empty)
-    assert str(excinfo.value) == "Cannot compute a Zarr checksum for an empty directory"
+    assert get_zarr_checksum(empty) == "481a2f77ab786a0f45aafd5db0971caa-0--0"
 
     spy = mocker.spy(digests, "md5file_nocache")
     assert (