fix issues in fix_file_counts_cr command

src/metax_api/management/commands/fix_file_counts_cr.py

@@ -0,0 +1,19 @@
+import logging
+
+from django.core.management.base import BaseCommand
+
+from metax_api.models import CatalogRecord
+
+logger = logging.getLogger(__name__)
+
+class Command(BaseCommand):
+    def handle(self, *args, **options):
+        CRS = CatalogRecord.objects.all()
+        crs_sum = CRS.count()
+        logger.info(f"fix_file_counts command found {crs_sum} catalog records with file_count=0 and byte_size=0")
+        i = 1
+        for catalog_record in CRS:
+            logger.info(f"Calculating {i}/{crs_sum} {catalog_record.identifier} ")
+            catalog_record.calculate_directory_byte_sizes_and_file_counts()
+            i += 1
+        logger.info(f"fix_file_counts command executed successfully")

src/metax_api/models/catalog_record.py

@@ -2097,9 +2097,13 @@ def calculate_directory_byte_sizes_and_file_counts(self):
         if self.research_dataset.get('directories', None):
             dir_identifiers = [d['identifier'] for d in self.research_dataset['directories']]
 
+        file_dir_identifiers = []
         if self.research_dataset.get('files', None):
-            file_dir_identifiers = [File.objects.get(identifier=f['identifier']).parent_directory.identifier
-                for f in self.research_dataset['files']]
+            try:
+                file_dir_identifiers = [File.objects.get(identifier=f['identifier']).parent_directory.identifier
+                    for f in self.research_dataset['files']]
+            except Exception as e:
+                _logger.error(e)
 
         if not dir_identifiers and not file_dir_identifiers:
             return
@@ -2108,6 +2112,9 @@ def calculate_directory_byte_sizes_and_file_counts(self):
 
         highest_level_dirs_by_project = self._get_top_level_parent_dirs_by_project(dir_identifiers)
 
+        if len(highest_level_dirs_by_project) == 0:
+            return
+
         directory_data = {}
 
         for project_identifier, dir_paths in highest_level_dirs_by_project.items():