Merge pull request #4482 from broadinstitute/pdo-missing-project-error

Pdo missing project error

seqr/views/apis/data_manager_api.py

@@ -468,7 +468,10 @@ def get_loaded_projects(request, sample_type, dataset_type):
     projects = get_internal_projects().filter(is_demo=False)
     project_samples = None
     if AirtableSession.is_airtable_enabled():
-        project_samples = _fetch_airtable_loadable_project_samples(request.user, dataset_type, sample_type)
+        try:
+            project_samples = _fetch_airtable_loadable_project_samples(request.user, dataset_type, sample_type)
+        except ValueError as e:
+            return create_json_response({'error': str(e)}, status=400)
         projects = projects.filter(guid__in=project_samples.keys())
     if dataset_type == Sample.DATASET_TYPE_VARIANT_CALLS:
         exclude_sample_type = Sample.SAMPLE_TYPE_WES if sample_type == Sample.SAMPLE_TYPE_WGS else Sample.SAMPLE_TYPE_WGS
@@ -587,9 +590,13 @@ def _get_valid_project_samples(project_samples, dataset_type, sample_type, user)
 
     missing_family_samples = defaultdict(list)
     for project, individuals_by_sample_id in missing_project_samples.items():
-        loaded_samples = {sample['sample_id'] for sample in _get_dataset_type_samples_for_matched_pdos(
-            user, dataset_type, sample_type, AVAILABLE_PDO_STATUSES, project_guid=project,
-        )}
+        try:
+            loaded_samples = {sample['sample_id'] for sample in _get_dataset_type_samples_for_matched_pdos(
+                user, dataset_type, sample_type, AVAILABLE_PDO_STATUSES, project_guid=project,
+            )}
+        except ValueError as e:
+            errors.append(str(e))
+            continue
         for sample_id, individual in individuals_by_sample_id.items():
             if sample_id in loaded_samples:
                 individual_ids.append(individual['id'])

seqr/views/utils/airtable_utils.py

@@ -4,7 +4,6 @@
 from django.core.exceptions import PermissionDenied
 
 from seqr.utils.logging_utils import SeqrLogger
-from seqr.utils.middleware import ErrorsWarningsException
 from seqr.views.utils.terra_api_utils import is_cloud_authenticated
 
 from settings import AIRTABLE_API_KEY, AIRTABLE_URL, BASE_URL
@@ -171,11 +170,15 @@ def get_samples_for_matched_pdos(self, pdo_statuses, pdo_fields=None, project_gu
         invalid_pdo_samples = []
         for sample in sample_records.values():
             sample_id = sample.get('SeqrCollaboratorSampleID') or sample['CollaboratorSampleID']
-            project_guids = [
-                re.match(f'{BASE_URL}project/([^/]+)/project_page', url).group(1) for url in sample['SeqrProject']
+            project_matches = [
+                re.match(f'{BASE_URL}project/([^/]+)/project_page', url)
+                for url in sample.get('SeqrProject', []) if url
             ]
-            if len(project_guids) > 1 and len(project_guids) < len(sample['PDOStatus']):
+            if any(pm is None for pm in project_matches) or (1 < len(project_matches) < len(sample['PDOStatus'])):
                 invalid_pdo_samples.append(sample_id)
+                continue
+
+            project_guids = [match.group(1) for match in project_matches]
             pdos = [{
                 'project_guid': project_guids[i] if len(project_guids) > 1 else project_guids[0],
                 **{field: sample[field][i] for field in pdo_fields}
@@ -189,8 +192,6 @@ def get_samples_for_matched_pdos(self, pdo_statuses, pdo_fields=None, project_gu
 
         if invalid_pdo_samples:
             samples = ', '.join(sorted(invalid_pdo_samples))
-            raise ErrorsWarningsException([
-                f'The following samples are associated with misconfigured PDOs in Airtable: {samples}'
-            ])
+            raise ValueError(f'The following samples are associated with misconfigured PDOs in Airtable: {samples}')
 
         return {record_id: sample for record_id, sample in sample_records.items() if sample['pdos']}