Add new zarr checksum format to model

dandischema/digests/zarr.py

@@ -9,6 +9,7 @@
 
 """Passed to the json() method of pydantic models for serialization."""
 ENCODING_KWARGS = {"separators": (",", ":")}
+ZARR_CHECKSUM_PATTERN = "([0-9a-f]{32})-([0-9]+)--([0-9]+)"
 
 
 def generate_directory_digest(md5: str, file_count: int, size: int):
@@ -18,7 +19,7 @@ def generate_directory_digest(md5: str, file_count: int, size: int):
 
 def parse_directory_digest(digest: str):
     """Parse a directory digest into its constituent parts"""
-    match = re.match("([0-9a-f]{32})-([0-9]+)--([0-9]+)", digest)
+    match = re.match(ZARR_CHECKSUM_PATTERN, digest)
     if match is None:
         raise ValueError(f"Cannot parse directory digest {digest}")
     return match.group(1), int(match.group(2)), int(match.group(3))

dandischema/models.py

@@ -22,6 +22,7 @@
 
 from .consts import DANDI_SCHEMA_VERSION
 from .digests.dandietag import DandiETag
+from .digests.zarr import ZARR_CHECKSUM_PATTERN, parse_directory_digest
 from .model_types import (
     AccessTypeDict,
     AgeReferenceTypeDict,
@@ -1124,11 +1125,17 @@ def digest_check(cls, v, values, **kwargs):
             if v.get(DigestType.dandi_etag):
                 raise ValueError("Digest cannot have both etag and zarr checksums.")
             digest = v.get(DigestType.dandi_zarr_checksum)
-            if not re.fullmatch(MD5_PATTERN, digest):
+            if not re.fullmatch(ZARR_CHECKSUM_PATTERN, digest):
                 raise ValueError(
                     f"Digest must have an appropriate dandi-zarr-checksum value. "
                     f"Got {digest}"
                 )
+            _checksum, _file_count, zarr_size = parse_directory_digest(digest)
+            content_size = values.get("contentSize")
+            if content_size != zarr_size:
+                raise ValueError(
+                    f"contentSize {content_size} is not equal to the checksum size {zarr_size}."
+                )
         else:
             if DigestType.dandi_etag not in v:
                 raise ValueError("A non-zarr asset must have a dandi-etag.")

dandischema/tests/test_models.py

@@ -118,7 +118,7 @@ def test_asset_digest():
         for el in exc.value.errors()
     )
 
-    digest = 33 * "a"
+    digest = 32 * "a"
     digest_model = {models.DigestType.dandi_zarr_checksum: digest}
     with pytest.raises(pydantic.ValidationError) as exc:
         models.BareAsset(
@@ -133,14 +133,23 @@ def test_asset_digest():
             for val in set([el["msg"] for el in exc.value.errors()])
         ]
     )
-    digest = 32 * "a"
+    digest = f"{32 * 'a'}-1--42"
     digest_model = {models.DigestType.dandi_zarr_checksum: digest}
-    models.BareAsset(
-        contentSize=100,
-        encodingFormat="application/x-zarr",
-        digest=digest_model,
-        path="/",
+    with pytest.raises(pydantic.ValidationError) as exc:
+        models.BareAsset(
+            contentSize=100,
+            encodingFormat="application/x-zarr",
+            digest=digest_model,
+            path="/",
+        )
+    assert any(
+        [
+            "contentSize 100 is not equal to the checksum size 42." in val
+            for val in set([el["msg"] for el in exc.value.errors()])
+        ]
     )
+    digest = f"{32 * 'a'}-1--100"
+    digest_model = {models.DigestType.dandi_zarr_checksum: digest}
     with pytest.raises(pydantic.ValidationError) as exc:
         models.PublishedAsset(
             contentSize=100,
@@ -357,7 +366,7 @@ def test_autogenerated_titles():
 
 
 def test_dantimeta_1():
-    """ checking basic metadata for publishing"""
+    """checking basic metadata for publishing"""
     # meta data without doi, datePublished and publishedBy
     meta_dict = {
         "identifier": "DANDI:999999",