Added edition deletion from ES

NYPL · Dec 18, 2024 · 54811be · 54811be
1 parent 6f7ee0b
commit 54811be
Show file tree

Hide file tree

Showing 2 changed files with 15 additions and 12 deletions.
diff --git a/scripts/__init__.py b/scripts/__init__.py
@@ -14,3 +14,4 @@
 from .countCABooks import main as countCA
 from .nyplLoginFlags import main as nyplFlags
 from .deleteUMPManifestLinks import main as deleteUMPManifests
+from .deleteProblemWorks import main as deleteWorks
diff --git a/services/sources/publisher_backlist_service.py b/services/sources/publisher_backlist_service.py
@@ -11,7 +11,7 @@
 from managers import S3Manager, WebpubManifest
 from .source_service import SourceService
 from managers import DBManager, ElasticsearchManager
-from elasticsearch_dsl import Search
+from elasticsearch_dsl import Search, Q
 
 logger = create_log(__name__)
 
@@ -50,7 +50,7 @@ def delete_records(
     def delete_manifest(self, record_metadata_dict):
         self.db_manager.createSession()
         try:
-            record = self.session.query(Record).filter(Record.source_id == record_metadata_dict['DRB Record_ID']).first()
+            record = self.db_manager.session.query(Record).filter(Record.source_id == record_metadata_dict['DRB Record_ID']).first()
             if record:
                 key_name = self.get_metadata_file_name(record, record_metadata_dict)
                 self.s3_manager.s3Client.delete_object(Bucket= self.s3_bucket, Key= key_name)
@@ -68,7 +68,7 @@ def delete_work(self, record_metadata_dict):
                 record_uuid_str = str(record.uuid)
                 edition =  self.db_manager.session.query(Edition).filter(Edition.dcdw_uuids.contains([record_uuid_str])).first()
                 work = self.db_manager.session.query(Work).filter(Work.id == edition.work_id).first()
-                if self.checkAllEditionsRelatedToRecord(record_uuid_str, work) == True:
+                if self.has_only_one_edition(work):
                     work_uuid_str = str(work.uuid)
                     es_work_resp = Search(index=os.environ['ELASTICSEARCH_INDEX']).query('match', uuid=work_uuid_str)
                     self.db_manager.session.query(Work).filter(Work.id == edition.work_id).delete()
@@ -81,20 +81,22 @@ def delete_work(self, record_metadata_dict):
         finally:
             self.db_manager.session.close()
 
-    def checkAllEditionsRelatedToRecord(self, record_uuid_str, work):
-        for edition in self.db_manager.session.query(Edition).filter(Edition.work_id == work.id).all():
-            if record_uuid_str not in edition.dcdw_uuids:
-                return False
+    def has_only_one_edition(self, work):
+        if len(work.editions) > 1:
+            return False
         return True
 
     def delete_pub_backlist_edition_only(self, record_uuid_str, work):
-        for edition in self.db_manager.session.query(Edition) \
+        edition = self.db_manager.session.query(Edition) \
             .filter(Edition.work_id == work.id) \
             .filter(Edition.dcdw_uuids.contains([record_uuid_str])) \
-            .all():
-                self.db_manager.session.delete(edition)
-                # es_edition_resp = Search(index=os.environ['ELASTICSEARCH_INDEX']).query('match', uuid=uuid)
-                # es_edition_esp.delete()
+            .first()
+        self.db_manager.session.delete(edition)
+        es_work_resp = Search(index=os.environ['ELASTICSEARCH_INDEX']).query('match', uuid=str(work.uuid))
+        for work_hit in es_work_resp:
+            for edition_hit in work_hit:
+                edition_es_response = Search(index=os.environ['ELASTICSEARCH_INDEX']).query('nested', path='editions', query=Q('match', **{'editions.edition_id': edition_hit['edition_id']}))
+                edition_es_response.delete()
 
     def get_metadata_file_name(self, record, record_metadata_dict):
         key_format = f"{self.prefix}{record.source}"