Finish document upload task refactor. Increase failure tolerance to database Operational Errors.

This commit is contained in:
Roberto Rosario
2015-07-15 04:36:23 -04:00
parent 0b1230f214
commit 91a9b3b045
7 changed files with 141 additions and 113 deletions

View File

@@ -1,14 +1,17 @@
import logging
from django.contrib.auth.models import User
from django.db import OperationalError
from django.core.files import File
from django.db import OperationalError, transaction
from django.utils.translation import ugettext_lazy as _
from mayan.celery import app
from common.compressed_files import CompressedFile, NotACompressedFile
from common.models import SharedUploadedFile
from converter.models import Transformation
from documents.models import DocumentType
from documents.models import Document, DocumentType
from documents.settings import setting_language
from metadata.api import save_metadata_list
from .literals import DEFAULT_SOURCE_TASK_RETRY_DELAY
@@ -31,23 +34,36 @@ def task_check_interval_source(source_id):
@app.task(bind=True, default_retry_delay=DEFAULT_SOURCE_TASK_RETRY_DELAY, ignore_result=True)
def task_post_document_version_upload(self, source_id, document_version_id):
def task_upload_document(self, source_id, document_type_id, shared_uploaded_file_id, description=None, label=None, language=None, metadata_dict_list=None, user_id=None):
try:
document_type = DocumentType.objects.get(pk=document_type_id)
source = Source.objects.get_subclass(pk=source_id)
document_version = DocumentVersion.objects.get(pk=document_version_id)
shared_upload = SharedUploadedFile.objects.get(pk=shared_uploaded_file_id)
if user_id:
user = User.objects.get(pk=user_id)
else:
user = None
with shared_upload.open() as file_object:
source.upload_document(file_object=file_object, document_type=document_type, description=description, label=label, language=language, metadata_dict_list=metadata_dict_list, user=user)
Transformation.objects.copy(source=Source.objects.get_subclass(pk=source_id), targets=document_version.pages.all())
except OperationalError as exception:
logger.warning('Operational error during post source document upload processing: %s. Retrying.', exception)
logger.warning('Operational exception while trying to create new document "%s" from source id %d; %s. Retying.', label or shared_upload.filename, source_id, exception)
raise self.retry(exc=exception)
else:
try:
shared_upload.delete()
except OperationalError as exception:
logger.warning('Operational error during attempt to delete shared upload file: %s; %s. Retrying.', shared_upload, exception)
@app.task(bind=True, default_retry_delay=DEFAULT_SOURCE_TASK_RETRY_DELAY, ignore_result=True)
def task_upload_document(self, document_type_id, shared_uploaded_file_id, label=None, language=None, user_id=None, description=None, metadata_dict_list=None):
def task_source_handle_upload(self, document_type_id, shared_uploaded_file_id, source_id, description=None, expand=False, label=None, language=None, metadata_dict_list=None, skip_list=None, user_id=None):
try:
shared_uploaded_file = SharedUploadedFile.objects.get(pk=shared_uploaded_file_id)
source = Source.objects.get_subclass(pk=source_id)
document_type = DocumentType.objects.get(pk=document_type_id)
source = Source.objects.get_subclass(pk=source_id)
shared_upload = SharedUploadedFile.objects.get(pk=shared_uploaded_file_id)
if user_id:
user = User.objects.get(pk=user_id)
@@ -55,62 +71,60 @@ def task_upload_document(self, document_type_id, shared_uploaded_file_id, label=
user = None
if not label:
label = shared_uploaded_file.filename
label = shared_upload.filename
with transaction.atomic():
document = DocumentVersion.objects.create(document_type=document_type)
document_version = document.new_document(
file_object=file_object, label=label, description=description,
language=language, _user=user
)
if metadata_dict_list:
save_metadata_list(metadata_dict_list, document, create=True)
task_post_source_document_version_upload.delay(source_id=source_id, document_version_id=document_version.pk, metadata_dict_list=metadata_dict_list)
except OperationalError as exception:
logger.warning('Operational error during attempt to handle source upload: %s. Retrying.', exception)
raise self.retry(exc=exception)
try:
shared_uploaded_file.delete()
except OperationalError as exception:
logger.warning('Operational error during attempt to delete shared upload file: %s; %s. Retrying.', shared_uploaded_file, exception)
@app.task(bind=True, default_retry_delay=DEFAULT_SOURCE_TASK_RETRY_DELAY, ignore_result=True)
def task_source_handle_upload(self, label, document_type_id, shared_uploaded_file_id, source_id, description=None, expand=False, language=None, metadata_dict_list=None, user_id=None):
try:
shared_uploaded_file = SharedUploadedFile.objects.get(pk=shared_uploaded_file_id)
source = Source.objects.get_subclass(pk=source_id)
document_type = DocumentType.objects.get(pk=document_type_id)
if user_id:
user = User.objects.get(pk=user_id)
else:
user = None
if not label:
label = shared_uploaded_file.filename
except OperationalError as exception:
logger.warning('Operational error during attempt to load data to handle source upload: %s. Retrying.', exception)
raise self.retry(exc=exception)
with shared_uploaded_file.open() as file_object:
source.handle_upload(description=description, document_type=document_type, expand=expand, file_object=file_object, label=label, language=language, metadata_dict_list=metadata_dict_list, user=user)
kwargs = {
'description': description, 'document_type_id': document_type.pk,
'label': label, 'language': language,
'metadata_dict_list': metadata_dict_list,
'source_id': source_id, 'user_id': user_id
}
try:
shared_uploaded_file.delete()
except OperationalError as exception:
logger.warning('Operational error during attempt to delete shared upload file: %s; %s. Retrying.', shared_uploaded_file, exception)
if not skip_list:
skip_list = []
# TODO: Report/record how was file uploaded
# if result['is_compressed'] is None:
# messages.success(request, _('File uploaded successfully.'))
with shared_upload.open() as file_object:
if expand:
try:
compressed_file = CompressedFile(file_object)
for compressed_file_child in compressed_file.children():
# TODO: find way to uniquely identify child files
# Use filename in the mean time.
if unicode(compressed_file_child) not in skip_list:
kwargs.update({'label': unicode(compressed_file_child)})
# if result['is_compressed'] is True:
# messages.success(request, _('File uncompressed successfully and uploaded as individual files.'))
try:
child_shared_uploaded_file = SharedUploadedFile.objects.create(file=File(compressed_file_child))
except OperationalError as exception:
logger.warning('Operational error while preparing to upload child document: %s. Rescheduling.', exception)
# if result['is_compressed'] is False:
# messages.warning(request, _('File was not a compressed file, uploaded as it was.'))
task_source_handle_upload.delay(
document_type_id=document_type_id,
shared_uploaded_file_id=shared_uploaded_file_id,
source_id=source_id, description=description,
expand=expand, label=label,
language=language,
metadata_dict_list=metadata_dict_list,
skip_list=skip_list, user_id=user_id
)
return
else:
skip_list.append(unicode(compressed_file_child))
task_upload_document.delay(shared_uploaded_file_id=child_shared_uploaded_file.pk, **kwargs)
finally:
compressed_file_child.close()
compressed_file_child.close()
try:
shared_upload.delete()
except OperationalError as exception:
logger.warning('Operational error during attempt to delete shared upload file: %s; %s. Retrying.', shared_uploaded_file, exception)
except NotACompressedFile:
logging.debug('Exception: NotACompressedFile')
task_upload_document.delay(shared_uploaded_file_id=shared_upload.pk, **kwargs)
else:
task_upload_document.delay(shared_uploaded_file_id=shared_upload.pk, **kwargs)