tol-sdk 1.8.8__tar.gz → 1.8.9__tar.gz

{tol_sdk-1.8.8/src/tol_sdk.egg-info → tol_sdk-1.8.9}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: tol-sdk
-Version: 1.8.8
+Version: 1.8.9
 Summary: SDK for interaction with ToL, Sanger and external services
 Author-email: ToL Platforms Team <tol-platforms@sanger.ac.uk>
 License: MIT

{tol_sdk-1.8.8 → tol_sdk-1.8.9}/pyproject.toml

@@ -8,7 +8,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "tol-sdk"
-version = "1.8.8"
+version = "1.8.9"
 description = "SDK for interaction with ToL, Sanger and external services"
 readme = "README.md"
 authors = [

{tol_sdk-1.8.8 → tol_sdk-1.8.9}/src/tol/ena/ena_datasource.py

@@ -11,6 +11,7 @@ from cachetools.func import ttl_cache
 
 import requests
 from requests.auth import HTTPBasicAuth
+from requests.exceptions import HTTPError
 
 
 from .client import EnaApiClient
@@ -118,9 +119,13 @@ class EnaDataSource(
         if object_type == 'submittable_taxon':
             ena_response = []
             for object_id in object_ids:
-                response = client.get_detail(object_type, [object_id])
-                if response and isinstance(response, list):
-                    ena_response.extend(response)
+                try:
+                    response = client.get_detail(object_type, [object_id])
+                    if response and isinstance(response, list):
+                        ena_response.extend(response)
+                except HTTPError as http_error:
+                    if http_error.response.status_code != 400:
+                        raise
         else:
             ena_response = client.get_detail(object_type, object_ids)
         # For a checklist we need to convert into a list of dicts

{tol_sdk-1.8.8 → tol_sdk-1.8.9}/src/tol/validators/__init__.py

@@ -19,8 +19,9 @@ from .types import TypesValidator  # noqa
 from .unique_values import UniqueValuesValidator  # noqa
 from .unique_whole_organisms import UniqueWholeOrganismsValidator  # noqa
 from .interfaces import Condition  # noqa
-from .min_one_valid_value import MinOneValidValueValidator   # noqa
+from .min_one_valid_value import MinOneValidValueValidator  # noqa
 from .value_check import ValueCheckValidator  # noqa
 from .branching import BranchingValidator  # noqa
 from .unique_value_check import UniqueValueCheckValidator  # noqa
-from .date_sorting import DateSortingValidator # noqa
+from .date_sorting import DateSortingValidator  # noqa
+from .taxon_matches_goat import TaxonMatchesGoatValidator  # noqa

{tol_sdk-1.8.8 → tol_sdk-1.8.9}/src/tol/validators/allowed_keys.py

@@ -30,6 +30,7 @@ class AllowedKeysValidator(Validator):
         **kwargs
     ) -> None:
 
+        del kwargs
         super().__init__()
         self.__config = config
 

tol_sdk-1.8.9/src/tol/validators/taxon_matches_goat.py

@@ -0,0 +1,86 @@
+# SPDX-FileCopyrightText: 2026 Genome Research Ltd.
+#
+# SPDX-License-Identifier: MIT
+
+from dataclasses import dataclass
+
+from tol.core import DataObject, Validator
+from tol.sources.goat import GoatDataSource, goat
+
+
+class TaxonMatchesGoatValidator(Validator):
+    """
+    Validates a stream of `DataObject` instances, checking whether its Taxonomy information
+    matches that in GoaT
+    """
+    @dataclass(slots=True, frozen=True, kw_only=True)
+    class Config:
+        species_field: str | None = None
+        genus_field: str | None = None
+        family_field: str | None = None
+        superfamily_field: str | None = None
+        phylum_field: str | None = None
+        kingdom_field: str | None = None
+        superkingdom_field: str | None = None
+        domain_field: str | None = None
+
+    __slots__ = ['__config', '__goat_datasource', '_cached_taxa']
+    __config: Config
+    __goat_datasource: GoatDataSource
+    _cached_taxa: dict[str, DataObject]
+
+    def __init__(self, config: Config) -> None:
+        super().__init__()
+        self.__config = config
+        self.__goat_datasource = goat()
+        self._cached_taxa = {}
+
+    def _validate_data_object(self, obj: DataObject) -> None:
+        taxon_id = obj.get_field_by_name('TAXON_ID')
+
+        # Check whether we already have the information for this id in the cache.
+        # If we don't, fetch it from GoaT and add it to the cache
+        taxon: DataObject | None
+        if taxon_id in self._cached_taxa:
+            taxon = self._cached_taxa[taxon_id]
+        else:
+            taxon = self.__goat_datasource.get_one('taxon', taxon_id)
+
+            # Add this taxon to cached taxa.
+            # Error if GoaT has no taxon with this id
+            if taxon is not None:
+                self._cached_taxa[taxon_id] = taxon
+            else:
+                self.add_error(
+                    object_id=obj.id,
+                    detail=f'Invalid Taxon ID: {taxon_id}',
+                    field='taxon_id'
+                )
+
+                # We can't validate a taxon that doesn't exist, so after this error move on
+                # to the next DataObject
+                return
+
+        # Check that each associated taxonomy rank for this taxon matches those in GoaT
+        taxonomic_ranks = ('species', 'genus', 'family', 'superfamily',
+                           'phylum', 'kingdom', 'superkingdom', 'domain')
+        for rank in taxonomic_ranks:
+            # From the rank to check, get the name of its field in the data object we're validating
+            # from the validator config. If this field name is `None`, then this taxonomic rank
+            # isn't being checked (likely because the data object does not have this field)
+            field_name: str | None = getattr(self.__config, f'{rank}_field')
+            if field_name is None:
+                continue
+
+            # Fetch the values of these taxonomic ranks
+            value_in_data_object = obj.get_field_by_name(field_name)
+            value_in_goat = taxon.get_field_by_name(f'{rank}.scientific_name')
+
+            # Ensure the value in the data object matches the one in GoaT
+            if value_in_data_object != value_in_goat:
+                self.add_warning(
+                    object_id=obj.id,
+                    detail=(f'Value for {field_name} ({value_in_data_object}) '
+                            f'does not match the value in GoaT ({value_in_goat})'),
+                    field=field_name,
+                )

{tol_sdk-1.8.8 → tol_sdk-1.8.9/src/tol_sdk.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: tol-sdk
-Version: 1.8.8
+Version: 1.8.9
 Summary: SDK for interaction with ToL, Sanger and external services
 Author-email: ToL Platforms Team <tol-platforms@sanger.ac.uk>
 License: MIT

{tol_sdk-1.8.8 → tol_sdk-1.8.9}/src/tol_sdk.egg-info/SOURCES.txt

@@ -356,6 +356,7 @@ src/tol/validators/regex.py
 src/tol/validators/regex_by_value.py
 src/tol/validators/specimens_have_same_taxon.py
 src/tol/validators/sts_fields.py
+src/tol/validators/taxon_matches_goat.py
 src/tol/validators/tolid.py
 src/tol/validators/types.py
 src/tol/validators/unique_value_check.py