def upload_file(self, file_object, filename=None, use_file_name=False, document_type=None, expand=False, metadata_dict_list=None, user=None, document=None, new_version_data=None, command_line=False): is_compressed = None if expand: try: cf = CompressedFile(file_object) count = 1 for fp in cf.children(): if command_line: print 'Uploading file #%d: %s' % (count, fp) self.upload_single_file(file_object=fp, filename=None, document_type=document_type, metadata_dict_list=metadata_dict_list, user=user) fp.close() count += 1 except NotACompressedFile: is_compressed = False logging.debug('Exception: NotACompressedFile') if command_line: raise self.upload_single_file(file_object=file_object, filename=filename, document_type=document_type, metadata_dict_list=metadata_dict_list, user=user) else: is_compressed = True else: self.upload_single_file(file_object, filename, use_file_name, document_type, metadata_dict_list, user, document, new_version_data) file_object.close() return {'is_compressed': is_compressed}
def handle_upload(self, file_object, description=None, document_type=None, expand=False, label=None, language=None, metadata_dict_list=None, metadata_dictionary=None, user=None): if not document_type: document_type = self.document_type kwargs = { 'description': description, 'document_type': document_type, 'label': label, 'language': language, 'metadata_dict_list': metadata_dict_list, 'metadata_dictionary': metadata_dictionary, 'user': user } if expand: try: compressed_file = CompressedFile(file_object) for compressed_file_child in compressed_file.children(): kwargs.update({'label': unicode(compressed_file_child)}) self.upload_document( file_object=File(compressed_file_child), **kwargs ) compressed_file_child.close() except NotACompressedFile: logging.debug('Exception: NotACompressedFile') self.upload_document(file_object=file_object, **kwargs) else: self.upload_document(file_object=file_object, **kwargs)
def handle_upload(self, file_object, description=None, document_type=None, expand=False, label=None, language=None, metadata_dict_list=None, metadata_dictionary=None, user=None): if not document_type: document_type = self.document_type kwargs = { 'description': description, 'document_type': document_type, 'label': label, 'language': language, 'metadata_dict_list': metadata_dict_list, 'metadata_dictionary': metadata_dictionary, 'user': user } if expand: try: compressed_file = CompressedFile(file_object) for compressed_file_child in compressed_file.children(): kwargs.update({'label': force_text(compressed_file_child)}) self.upload_document( file_object=File(compressed_file_child), **kwargs ) compressed_file_child.close() except NotACompressedFile: logging.debug('Exception: NotACompressedFile') self.upload_document(file_object=file_object, **kwargs) else: self.upload_document(file_object=file_object, **kwargs)
def handle_upload(self, file_object, description=None, document_type=None, expand=False, label=None, language=None, metadata_dict_list=None, metadata_dictionary=None, tag_ids=None, user=None): """ Handle an upload request from a file object which may be an individual document or a compressed file containing multiple documents. """ if not document_type: document_type = self.document_type kwargs = { 'description': description, 'document_type': document_type, 'label': label, 'language': language, 'metadata_dict_list': metadata_dict_list, 'metadata_dictionary': metadata_dictionary, 'tag_ids': tag_ids, 'user': user } if expand: try: compressed_file = CompressedFile(file_object) for compressed_file_child in compressed_file.children(): kwargs.update({'label': force_text(compressed_file_child)}) self.upload_document( file_object=File(compressed_file_child), **kwargs ) compressed_file_child.close() except NotACompressedFile: logging.debug('Exception: NotACompressedFile') self.upload_document(file_object=file_object, **kwargs) else: self.upload_document(file_object=file_object, **kwargs)
def handle_upload(self, file_object, description=None, document_type=None, expand=False, label=None, language=None, user=None): """ Handle an upload request from a file object which may be an individual document or a compressed file containing multiple documents. """ documents = [] if not document_type: document_type = self.document_type kwargs = { 'description': description, 'document_type': document_type, 'label': label, 'language': language, 'user': user } if expand: try: compressed_file = CompressedFile(file_object) for compressed_file_child in compressed_file.children(): kwargs.update({'label': force_text(compressed_file_child)}) documents.append( self.upload_document( file_object=File(compressed_file_child), **kwargs)) compressed_file_child.close() except NotACompressedFile: logging.debug('Exception: NotACompressedFile') documents.append( self.upload_document(file_object=file_object, **kwargs)) else: documents.append( self.upload_document(file_object=file_object, **kwargs)) # Return a list of newly created documents. Used by the email source # to assign the from and subject metadata values. return documents
def new_document(self, document_type, file_object, label=None, command_line=False, description=None, expand=False, language=None, user=None): versions_created = [] if expand: try: compressed_file = CompressedFile(file_object) count = 1 for compressed_file_child in compressed_file.children(): if command_line: print 'Uploading file #%d: %s' % (count, compressed_file_child) versions_created.append(self.upload_single_document(document_type=document_type, file_object=compressed_file_child, description=description, label=unicode(compressed_file_child), language=language or LANGUAGE, user=user)) compressed_file_child.close() count += 1 except NotACompressedFile: logging.debug('Exception: NotACompressedFile') if command_line: raise versions_created.append(self.upload_single_document(document_type=document_type, file_object=file_object, description=description, label=label, language=language or LANGUAGE, user=user)) else: versions_created.append(self.upload_single_document(document_type=document_type, file_object=file_object, description=description, label=label, language=language or LANGUAGE, user=user)) return versions_created
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, tag_ids=None, user_id=None): SharedUploadedFile = apps.get_model(app_label='common', model_name='SharedUploadedFile') DocumentType = apps.get_model(app_label='documents', model_name='DocumentType') try: document_type = DocumentType.objects.get(pk=document_type_id) shared_upload = SharedUploadedFile.objects.get( pk=shared_uploaded_file_id) if not label: label = shared_upload.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) kwargs = { 'description': description, 'document_type_id': document_type.pk, 'label': label, 'language': language, 'metadata_dict_list': metadata_dict_list, 'source_id': source_id, 'tag_ids': tag_ids, 'user_id': user_id } if not skip_list: skip_list = [] 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 force_text(compressed_file_child) not in skip_list: kwargs.update( {'label': force_text(compressed_file_child)}) 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) 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, tag_ids=tag_ids, user_id=user_id) return else: skip_list.append(force_text(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_upload, 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)
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, tag_ids=None, user_id=None): SharedUploadedFile = apps.get_model( app_label='common', model_name='SharedUploadedFile' ) DocumentType = apps.get_model( app_label='documents', model_name='DocumentType' ) try: document_type = DocumentType.objects.get(pk=document_type_id) shared_upload = SharedUploadedFile.objects.get( pk=shared_uploaded_file_id ) if not label: label = shared_upload.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) kwargs = { 'description': description, 'document_type_id': document_type.pk, 'label': label, 'language': language, 'metadata_dict_list': metadata_dict_list, 'source_id': source_id, 'tag_ids': tag_ids, 'user_id': user_id } if not skip_list: skip_list = [] 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 force_text(compressed_file_child) not in skip_list: kwargs.update( {'label': force_text(compressed_file_child)} ) 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 ) 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, tag_ids=tag_ids, user_id=user_id ) return else: skip_list.append(force_text(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_upload, 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 )
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: document_type = DocumentType.objects.get(pk=document_type_id) shared_upload = SharedUploadedFile.objects.get(pk=shared_uploaded_file_id) if not label: label = shared_upload.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) 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, } if not skip_list: skip_list = [] 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)}) 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, ) 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_upload, 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)