Merge pull request #317 from dandi/relax-on-default

Relax various patterns defined in `dandischema.models` when DANDI Schema is not set to a vendor

Author: yarikoptic

dandischema/conf.py

@@ -14,11 +14,16 @@
 _MODELS_MODULE_NAME = "dandischema.models"
 """The full import name of the module containing the DANDI Pydantic models"""
 
-_UNVENDORED_ID_PATTERN = r"[A-Z][-A-Z]*"
-_UNVENDORED_DOI_PREFIX_PATTERN = r"10\.\d{4,}"
+UNVENDORED_ID_PATTERN = r"[A-Z][-A-Z]*"
+UNVENDORED_DOI_PREFIX_PATTERN = r"10\.\d{4,}"
 
 logger = logging.getLogger(__name__)
 
+DEFAULT_INSTANCE_NAME = "DANDI-ADHOC"
+"""
+The default name of the DANDI instance
+"""
+
 
 class SpdxLicenseListInfo(BaseModel):
     """
@@ -74,14 +79,12 @@ class Config(BaseSettings):
     model_config = SettingsConfigDict(env_prefix="dandi_")
 
     instance_name: Annotated[
-        str, StringConstraints(pattern=rf"^{_UNVENDORED_ID_PATTERN}$")
-    ] = "DANDI-ADHOC"
+        str, StringConstraints(pattern=rf"^{UNVENDORED_ID_PATTERN}$")
+    ] = DEFAULT_INSTANCE_NAME
     """Name of the DANDI instance"""
 
     doi_prefix: Optional[
-        Annotated[
-            str, StringConstraints(pattern=rf"^{_UNVENDORED_DOI_PREFIX_PATTERN}$")
-        ]
+        Annotated[str, StringConstraints(pattern=rf"^{UNVENDORED_DOI_PREFIX_PATTERN}$")]
     ] = None
     """
     The DOI prefix at DataCite

dandischema/models.py

@@ -37,7 +37,12 @@
 from pydantic_core import CoreSchema
 from zarr_checksum.checksum import InvalidZarrChecksum, ZarrDirectoryDigest
 
-from dandischema.conf import get_instance_config
+from dandischema.conf import (
+    DEFAULT_INSTANCE_NAME,
+    UNVENDORED_DOI_PREFIX_PATTERN,
+    UNVENDORED_ID_PATTERN,
+    get_instance_config,
+)
 
 from .consts import DANDI_SCHEMA_VERSION
 from .digests.dandietag import DandiETag
@@ -48,13 +53,17 @@
 _INSTANCE_CONFIG = get_instance_config()
 
 # Regex pattern for the prefix of identifiers
-ID_PATTERN = _INSTANCE_CONFIG.instance_name
+ID_PATTERN = (
+    _INSTANCE_CONFIG.instance_name
+    if _INSTANCE_CONFIG.instance_name != DEFAULT_INSTANCE_NAME
+    else UNVENDORED_ID_PATTERN
+)
 
 # The pattern that a DOI prefix of a dandiset must conform to
 DOI_PREFIX_PATTERN = (
     re.escape(_INSTANCE_CONFIG.doi_prefix)
     if _INSTANCE_CONFIG.doi_prefix is not None
-    else None
+    else UNVENDORED_DOI_PREFIX_PATTERN
 )
 
 # Use DJANGO_DANDI_WEB_APP_URL to point to a specific deployment.
@@ -74,10 +83,13 @@
 )
 ASSET_UUID_PATTERN = r"^dandiasset:" + UUID_PATTERN
 VERSION_PATTERN = r"\d{6}/\d+\.\d+\.\d+"
+_INNER_DANDI_DOI_PATTERN = (
+    rf"{DOI_PREFIX_PATTERN}/{ID_PATTERN.lower()}\.{VERSION_PATTERN}"
+)
 DANDI_DOI_PATTERN = (
-    rf"^{DOI_PREFIX_PATTERN}/{ID_PATTERN.lower()}\.{VERSION_PATTERN}$"
-    if DOI_PREFIX_PATTERN is not None
-    else None
+    rf"^{_INNER_DANDI_DOI_PATTERN}$"
+    if _INSTANCE_CONFIG.doi_prefix is not None
+    else rf"^({_INNER_DANDI_DOI_PATTERN}|)$"  # This matches an empty string as well
 )
 DANDI_PUBID_PATTERN = rf"^{ID_PATTERN}:{VERSION_PATTERN}$"
 DANDI_NSKEY = "dandi"  # Namespace for DANDI ontology
@@ -1879,14 +1891,11 @@ class Publishable(DandiBaseModel):
 
 _doi_field_kwargs: dict[str, Any] = {
     "title": "DOI",
+    "pattern": DANDI_DOI_PATTERN,
     "json_schema_extra": {"readOnly": True, "nskey": DANDI_NSKEY},
 }
-if DANDI_DOI_PATTERN is not None:
-    _doi_field_kwargs["pattern"] = DANDI_DOI_PATTERN
-else:
+if _INSTANCE_CONFIG.doi_prefix is None:
     _doi_field_kwargs["default"] = ""
-    # restricting the value to empty string to indicate that there is no DOI
-    _doi_field_kwargs["pattern"] = r"^$"
 
 
 class PublishedDandiset(Dandiset, Publishable):

dandischema/tests/test_models.py

@@ -793,17 +793,19 @@ def _get_field_pattern(
         # Without any environment variables set. dandischema is unvendorized.
         (
             {},
-            "DANDI-ADHOC",
-            None,
+            r"[A-Z][-A-Z]*",
+            r"10\.\d{4,}",
             {
                 "dandiset_id": "DANDI-ADHOC:001350/draft",
                 "dandiset_identifier": "DANDI-ADHOC:001350",
                 "published_dandiset_id": "DANDI-ADHOC:001350/0.250511.1527",
+                "published_dandiset_doi": "",
             },
             {
                 "dandiset_id": "45:001350/draft",  # Invalid id prefix
                 "dandiset_identifier": "DANDI-ADHOC:001350",
                 "published_dandiset_id": "DANDI-ADHOC:001350/0.250511.1527",
+                "published_dandiset_doi": "",
             },
         ),
         (
@@ -832,34 +834,39 @@ def _get_field_pattern(
                 "instance_name": "DANDI",
             },
             "DANDI",
-            None,
+            r"10\.\d{4,}",
             {
                 "dandiset_id": "DANDI:001425/draft",
                 "dandiset_identifier": "DANDI:001425",
                 "published_dandiset_id": "DANDI:001425/0.250514.0602",
+                "published_dandiset_doi": "10.48324/dandi.001425/0.250514.0602",
             },
             {
                 "dandiset_id": "DANDI:001425/draft",
                 "dandiset_identifier": "DANDI:001425",
                 # Not matching the `ID_PATTERN` regex
                 "published_dandiset_id": "DANDI3:001425/0.250514.0602",
+                "published_dandiset_doi": "10.48324/dandi.001425/0.250514.0602",
             },
         ),
         # === EMBER DANDI instance test cases ===
         # Without any environment variables set. dandischema is unvendorized.
         (
             {},
-            "DANDI-ADHOC",
-            None,
+            r"[A-Z][-A-Z]*",
+            r"10\.\d{4,}",
             {
                 "dandiset_id": "DANDI-ADHOC:000005/draft",
-                "dandiset_identifier": "DANDI-ADHOC:000005",
+                "dandiset_identifier": "ABC:000005",
                 "published_dandiset_id": "DANDI-ADHOC:000005/0.250404.1839",
+                "published_dandiset_doi": "10.60533/ember-dandi.000005/0.250404.1839",
             },
             {
                 "dandiset_id": "DANDI-ADHOC:000005/draft",
-                "dandiset_identifier": "-DANDI-ADHOC:000005",  # Invalid id prefix
+                "dandiset_identifier": "ABC:000005",
                 "published_dandiset_id": "DANDI-ADHOC:000005/0.250404.1839",
+                # Invalid registrant code in the DOI prefix
+                "published_dandiset_doi": "10.605/ember-dandi.000005/0.250404.1839",
             },
         ),
         (
@@ -889,7 +896,7 @@ def _get_field_pattern(
 def test_vendorization(
     clear_dandischema_modules_and_set_env_vars: None,
     exp_id_pattern: str,
-    exp_doi_prefix_pattern: Optional[str],
+    exp_doi_prefix_pattern: str,
     # Fields that are valid for the vendorization
     valid_vendored_fields: dict[str, str],
     # Fields that are invalid for the vendorization
@@ -915,12 +922,11 @@ class VendoredFieldModel(BaseModel):
         published_dandiset_id: str = Field(
             pattern=_get_field_pattern("id", models_.PublishedDandiset)
         )
-        if exp_doi_prefix_pattern is not None:
-            published_dandiset_doi: str = Field(
-                pattern=_get_field_pattern("doi", models_.PublishedDandiset)
-            )
+        published_dandiset_doi: str = Field(
+            pattern=_get_field_pattern("doi", models_.PublishedDandiset)
+        )
 
-        model_config = ConfigDict(strict=True, extra="forbid")
+        model_config = ConfigDict(strict=True)
 
     # Validate the valid vendored fields against the vendored patterns
     VendoredFieldModel.model_validate(valid_vendored_fields)