Add scheduled task to delete stale document stubs.

mayan/apps/documents/apps.py

@@ -56,7 +56,10 @@ from .links import (
     link_document_version_download, link_document_version_list,
     link_document_version_revert, link_trash_can_empty
 )
-from .literals import CHECK_DELETE_PERIOD_INTERVAL, CHECK_TRASH_PERIOD_INTERVAL
+from .literals import (
+    CHECK_DELETE_PERIOD_INTERVAL, CHECK_TRASH_PERIOD_INTERVAL,
+    DELETE_STALE_STUBS_INTERVAL
+)
 from .models import (
     DeletedDocument, Document, DocumentPage, DocumentType,
     DocumentTypeFilename, DocumentVersion
@@ -183,6 +186,10 @@ class DocumentsApp(MayanAppConfig):
                     'task': 'documents.tasks.task_check_trash_periods',
                     'schedule': timedelta(seconds=CHECK_TRASH_PERIOD_INTERVAL),
                 },
+                'task_delete_stubs': {
+                    'task': 'documents.tasks.task_delete_stubs',
+                    'schedule': timedelta(seconds=DELETE_STALE_STUBS_INTERVAL),
+                },
             }
         )
 
@@ -208,6 +215,9 @@ class DocumentsApp(MayanAppConfig):
                 'documents.tasks.task_check_trash_periods': {
                     'queue': 'documents_periodic'
                 },
+                'documents.tasks.task_delete_stubs': {
+                    'queue': 'documents_periodic'
+                },
                 'documents.tasks.task_clear_image_cache': {
                     'queue': 'tools'
                 },

mayan/apps/documents/literals.py

@@ -5,10 +5,12 @@ from django.utils.translation import ugettext_lazy as _
 CACHE_PATH = 'document_cache/'
 CHECK_DELETE_PERIOD_INTERVAL = 60
 CHECK_TRASH_PERIOD_INTERVAL = 60
+DELETE_STALE_STUBS_INTERVAL = 60
 DEFAULT_DELETE_PERIOD = 30
 DEFAULT_DELETE_TIME_UNIT = 'days'
 DEFAULT_ZIP_FILENAME = 'document_bundle.zip'
 DOCUMENT_IMAGE_TASK_TIMEOUT = 20
+STUB_EXPIRATION_INTERVAL = 60 * 60 * 24  # 24 hours
 UPDATE_PAGE_COUNT_RETRY_DELAY = 10
 UPLOAD_NEW_VERSION_RETRY_DELAY = 10
 NEW_DOCUMENT_RETRY_DELAY = 10

mayan/apps/documents/tasks.py

@@ -13,7 +13,7 @@ from common.models import SharedUploadedFile
 
 from .literals import (
     UPDATE_PAGE_COUNT_RETRY_DELAY, UPLOAD_NEW_VERSION_RETRY_DELAY,
-    NEW_DOCUMENT_RETRY_DELAY
+    NEW_DOCUMENT_RETRY_DELAY, STUB_EXPIRATION_INTERVAL
 )
 from .models import (
     DeletedDocument, Document, DocumentPage, DocumentType, DocumentVersion
@@ -41,6 +41,7 @@ def task_check_delete_periods():
                 document_type, delta
             )
             for document in DeletedDocument.objects.filter(document_type=document_type):
+                # TODO: Don't iterate, filter documents by expiration
                 if now() > document.deleted_date_time + delta:
                     logger.info(
                         'Document "%s" with id: %d, trashed on: %s, exceded '
@@ -75,6 +76,7 @@ def task_check_trash_periods():
                 document_type, delta
             )
             for document in Document.objects.filter(document_type=document_type):
+                # TODO: Don't iterate, filter documents by expiration
                 if now() > document.date_added + delta:
                     logger.info(
                         'Document "%s" with id: %d, added on: %s, exceded '
@@ -98,6 +100,16 @@ def task_clear_image_cache():
     logger.info('Finished document cache invalidation')
 
 
+@app.task(ignore_result=True)
+def task_delete_stubs():
+    logger.info('Executing')
+
+    for stale_stub_document in Document.objects.filter(is_stub=True, date_added__lt=now() - timedelta(seconds=STUB_EXPIRATION_INTERVAL)):
+        stale_stub_document.delete(trash=False)
+
+    logger.info('Finshed')
+
+
 @app.task(compression='zlib')
 def task_get_document_page_image(document_page_id, *args, **kwargs):
     document_page = DocumentPage.objects.get(pk=document_page_id)