tol-sdk 1.8.8__py3-none-any.whl → 1.8.9__py3-none-any.whl

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/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/validators/allowed_keys.py

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

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.dist-info → tol_sdk-1.8.9.dist-info}/METADATA

@@ -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.dist-info → tol_sdk-1.8.9.dist-info}/RECORD

@@ -142,7 +142,7 @@ tol/eln/sanitise.py,sha256=fMj-VrQTnw4zn2X0wnjWQAI8gWAa8RYqNuv23LXQssI,406
 tol/ena/__init__.py,sha256=T3TCqaHpgi2Uk2PjPGu60GaG2V8cTrHJlVLtZfLFhTQ,174
 tol/ena/client.py,sha256=ldmm7Z9_auQf1zVWjsFLXYgbKvGtSHTsr88YO3zfv2Y,6731
 tol/ena/converter.py,sha256=nxbo4IFzzOvKNMq3Aeiw5iDqVWvY33nTngLppjHAoGY,1144
-tol/ena/ena_datasource.py,sha256=jEvyUaH4pfFxmdtn6O_PwOdjPz6u80uAT3SLlR2f5nM,8968
+tol/ena/ena_datasource.py,sha256=QRj4pXFeAaC0qPdmtPCrCD3DKVsXhu1jYZ97P3AHlnA,9184
 tol/ena/ena_methods.py,sha256=jgpLssZq-F-vgkO-fYu4jrXenmNkdFpFKAY3VKp5HHE,9209
 tol/ena/factory.py,sha256=3IJCmGLo59PWbGmNqmKho5WYG330OjL8SoZYstIwHt4,3423
 tol/ena/filter.py,sha256=UzOx5ivXvA0TY2QuNzFmS-zDPVNnaAx07DMVkAwVsAE,3370
@@ -335,8 +335,8 @@ tol/treeval/treeval_datasource.py,sha256=GzY6JwH67b5QdV-UVdCFJfgGAIuZ96J2nl53YxZ
 tol/utils/__init__.py,sha256=764-Na1OaNGUDWpMIu51ZtXG7n_nB5MccUFK6LmkWRI,138
 tol/utils/csv.py,sha256=mihww25fSn72c4h-RFeqD_pFIG6KHZP4v1_C0rx81ws,421
 tol/utils/s3.py,sha256=aoYCwJ-qcMqFrpxmViFqPa0O1jgp0phtztO3-0CSNjw,491
-tol/validators/__init__.py,sha256=_ETv6oGQ2bTH_6-foYFy9T5wP5OG3cl96zEjvrIS7zk,1399
-tol/validators/allowed_keys.py,sha256=RJcHBiguL84B8hjSRaXLNES21yZqaKFwJNp2Tz9zvh0,1506
+tol/validators/__init__.py,sha256=sF9i4rxi-NsJkqLvvxj2k_I_QPRgkhQZc9ErSqT7quk,1465
+tol/validators/allowed_keys.py,sha256=eLZnr6DVqK6ru6b-T-1hvzSdAxlWZkTNZuohrbO0vS8,1525
 tol/validators/allowed_values.py,sha256=-Yy3Sqo1WYacGKlot_dn3M2o7Oj5MXOioJrJmrWCCxs,1536
 tol/validators/allowed_values_from_datasource.py,sha256=9cVwllBbzfCls8UsojazfCInt9_AakA0_H9pBO1wSL4,3173
 tol/validators/assert_on_condition.py,sha256=eBGgSVfIQ6e45SheM-ZDg7daXJjyZxRVS5L8AWvbXag,2027
@@ -351,6 +351,7 @@ tol/validators/regex.py,sha256=dLAi_vQt9_DsT6wQZmbYC7X5-Wp15l0leUE6XkPaItg,2602
 tol/validators/regex_by_value.py,sha256=XM5EnT4vgD17rfpR3bUE9I56IemSw26BI9MZtMakd4E,2582
 tol/validators/specimens_have_same_taxon.py,sha256=BaJcZ38ZprPcuGTIorSxxC9uGN0_lj6HS6B54EObcuY,2183
 tol/validators/sts_fields.py,sha256=aYbzy15btEg4-ocDT1qrspe7-atoWRrOJ_KmuPU6J14,8936
+tol/validators/taxon_matches_goat.py,sha256=-5UJus3WZOc_ji5Kmat43nI6Auhr924BVi-igEnFP6Q,3482
 tol/validators/tolid.py,sha256=VOb6lNFz11H_0KaWX8_nvsw8xJEa6KrjB0p-5lkcqog,3885
 tol/validators/types.py,sha256=jMVpckRp8RS93f7usf58YH_K-5rKWgZIYs7bO9dHhQc,2914
 tol/validators/unique_value_check.py,sha256=sFvDooYkKeORvULGEOTsgIcxlbe0AXDWxY3Gbr3j0KI,1282
@@ -359,9 +360,9 @@ tol/validators/unique_whole_organisms.py,sha256=RdqA1GzIf3LTdrmNGGdxv0aW2udDY2P9
 tol/validators/value_check.py,sha256=DdNx_B1gns01zgBg5N6Bwia46Aukw6MAteM-M37Kv1k,1122
 tol/validators/interfaces/__init__.py,sha256=jtOxnwnwqV_29xjmmMcS_kvlt-pQiWwQYJn2YRP07_w,172
 tol/validators/interfaces/condition_evaluator.py,sha256=nj8Cb8hi47OBy6OVNfeLhF-Pjwtr8MiOSymYL6hfVes,3766
-tol_sdk-1.8.8.dist-info/licenses/LICENSE,sha256=RF9Jacy-9BpUAQQ20INhTgtaNBkmdTolYCHtrrkM2-8,1077
-tol_sdk-1.8.8.dist-info/METADATA,sha256=iyFf12GxWPRsRVPQgR5mrGzLDhAH-OrJ58-4T6CROLA,3142
-tol_sdk-1.8.8.dist-info/WHEEL,sha256=qELbo2s1Yzl39ZmrAibXA2jjPLUYfnVhUNTlyF1rq0Y,92
-tol_sdk-1.8.8.dist-info/entry_points.txt,sha256=jH3HfTwxjzog7E3lq8CKpUWGIRY9FSXbyL6CpUmv6D0,36
-tol_sdk-1.8.8.dist-info/top_level.txt,sha256=PwKMQLphyZNvagBoriVbl8uwHXQl8IC1niawVG0iXMM,10
-tol_sdk-1.8.8.dist-info/RECORD,,
+tol_sdk-1.8.9.dist-info/licenses/LICENSE,sha256=RF9Jacy-9BpUAQQ20INhTgtaNBkmdTolYCHtrrkM2-8,1077
+tol_sdk-1.8.9.dist-info/METADATA,sha256=MgDxVlsVbpKw9s2vFEdeeTUYpv9hdL48ETxeGTbGF-Y,3142
+tol_sdk-1.8.9.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
+tol_sdk-1.8.9.dist-info/entry_points.txt,sha256=jH3HfTwxjzog7E3lq8CKpUWGIRY9FSXbyL6CpUmv6D0,36
+tol_sdk-1.8.9.dist-info/top_level.txt,sha256=PwKMQLphyZNvagBoriVbl8uwHXQl8IC1niawVG0iXMM,10
+tol_sdk-1.8.9.dist-info/RECORD,,

{tol_sdk-1.8.8.dist-info → tol_sdk-1.8.9.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (80.10.1)
+Generator: setuptools (80.10.2)
 Root-Is-Purelib: true
 Tag: py3-none-any