示例#1
0
    def test__add_to_bulk(self):
        retry_queue = Queue()
        old_date_modified = datetime.datetime.utcnow().isoformat()
        queue_resource_item = {
            'doc_type': 'Tender',
            'id': uuid.uuid4().hex,
            'dateModified': datetime.datetime.utcnow().isoformat()
        }
        resource_item_doc_dict = {
            'doc_type': 'Tender',
            '_rev': '1-' + uuid.uuid4().hex,
            'id': queue_resource_item['id'],
            'dateModified': queue_resource_item['dateModified']
        }
        resource_item_dict = {
            'doc_type': 'Tender',
            'id': queue_resource_item['id'],
            'dateModified': queue_resource_item['dateModified']
        }
        worker = ResourceItemWorker(config_dict=self.worker_config,
                                    log_dict=self.log_dict,
                                    retry_resource_items_queue=retry_queue)
        worker.db = MagicMock()

        # Successfull adding to bulk
        start_length = len(worker.bulk)
        worker._add_to_bulk(resource_item_dict, queue_resource_item,
                            resource_item_doc_dict)
        end_length = len(worker.bulk)
        self.assertGreater(end_length, start_length)

        # Update exist doc in bulk
        start_length = len(worker.bulk)
        new_resource_item_dict = deepcopy(resource_item_dict)
        new_resource_item_dict['dateModified'] = datetime.datetime.utcnow(
        ).isoformat()
        worker._add_to_bulk(new_resource_item_dict, queue_resource_item,
                            resource_item_doc_dict)
        end_length = len(worker.bulk)
        self.assertEqual(start_length, end_length)

        # Ignored dublicate in bulk
        start_length = end_length
        worker._add_to_bulk(
            {
                'doc_type': 'Tender',
                'id': queue_resource_item['id'],
                '_id': queue_resource_item['id'],
                'dateModified': old_date_modified
            }, queue_resource_item, resource_item_dict)
        end_length = len(worker.bulk)
        self.assertEqual(start_length, end_length)
        del worker
示例#2
0
    def test__save_bulk_docs(self):
        self.worker_config['bulk_save_limit'] = 3
        retry_queue = Queue()
        worker = ResourceItemWorker(config_dict=self.worker_config,
                                    retry_resource_items_queue=retry_queue)
        doc_id_1 = uuid.uuid4().hex
        doc_id_2 = uuid.uuid4().hex
        doc_id_3 = uuid.uuid4().hex
        doc_id_4 = uuid.uuid4().hex
        date_modified = datetime.datetime.utcnow().isoformat()
        worker.bulk = {
            doc_id_1: {'id': doc_id_1, 'dateModified': date_modified},
            doc_id_2: {'id': doc_id_2, 'dateModified': date_modified},
            doc_id_3: {'id': doc_id_3, 'dateModified': date_modified},
            doc_id_4: {'id': doc_id_4, 'dateModified': date_modified}
        }
        update_return_value = [
            (True, doc_id_1, '1-' + uuid.uuid4().hex),
            (True, doc_id_2, '2-' + uuid.uuid4().hex),
            (False, doc_id_3, Exception(u'New doc with oldest dateModified.')),
            (False, doc_id_4, Exception(u'Document update conflict.'))
        ]
        worker.db = MagicMock()
        worker.db.update.return_value = update_return_value

        # Test success response from couchdb
        worker._save_bulk_docs()
        sleep(0.1)
        self.assertEqual(len(worker.bulk), 0)
        self.assertEqual(worker.retry_resource_items_queue.qsize(), 1)

        # Test failed response from couchdb
        worker.db.update.side_effect = Exception('Some exceptions')
        worker.bulk = {
            doc_id_1: {'id': doc_id_1, 'dateModified': date_modified},
            doc_id_2: {'id': doc_id_2, 'dateModified': date_modified},
            doc_id_3: {'id': doc_id_3, 'dateModified': date_modified},
            doc_id_4: {'id': doc_id_4, 'dateModified': date_modified}
        }
        worker._save_bulk_docs()
        sleep(0.2)
        self.assertEqual(worker.retry_resource_items_queue.qsize(), 5)
        self.assertEqual(len(worker.bulk), 0)

        worker.config['historical'] = True
        worker.bulk = {
            doc_id_1: {'id': doc_id_1, 'rev': randint(10, 99)}
        }
        worker._save_bulk_docs()
        sleep(0.2)
        self.assertEqual(worker.retry_resource_items_queue.qsize(), 6)
        self.assertEqual(len(worker.bulk), 0)
        worker.config['historical'] = False
    def test__add_to_bulk(self):
        retry_queue = PriorityQueue()
        old_date_modified = datetime.datetime.utcnow().isoformat()
        resource_item_id = uuid.uuid4().hex
        priority = 1

        local_resource_item = {
            'doc_type': 'Tender',
            '_rev': '1-' + uuid.uuid4().hex,
            'id': resource_item_id,
            'dateModified': old_date_modified
        }
        new_date_modified = datetime.datetime.utcnow().isoformat()
        public_resource_item = {
            'id': resource_item_id,
            'dateModified': new_date_modified
        }
        worker = ResourceItemWorker(config_dict=self.worker_config,
                                    retry_resource_items_queue=retry_queue)
        worker.db = MagicMock()

        # Successfull adding to bulk
        start_length = len(worker.bulk)
        worker._add_to_bulk(local_resource_item, public_resource_item,
                            priority)
        end_length = len(worker.bulk)
        self.assertGreater(end_length, start_length)

        # Update exist doc in bulk
        start_length = len(worker.bulk)
        new_public_resource_item = deepcopy(public_resource_item)
        new_public_resource_item['dateModified'] =\
            datetime.datetime.utcnow().isoformat()
        worker._add_to_bulk(local_resource_item, new_public_resource_item,
                            priority)
        end_length = len(worker.bulk)
        self.assertEqual(start_length, end_length)

        # Ignored dublicate in bulk
        start_length = end_length
        worker._add_to_bulk(
            local_resource_item, {
                'doc_type': 'Tender',
                'id': local_resource_item['id'],
                '_id': local_resource_item['id'],
                'dateModified': old_date_modified
            }, priority)
        end_length = len(worker.bulk)
        self.assertEqual(start_length, end_length)
        del worker
示例#4
0
    def test__save_bulk_docs(self):
        self.worker_config['bulk_save_limit'] = 3
        retry_queue = Queue()
        worker = ResourceItemWorker(config_dict=self.worker_config,
                                    log_dict=self.log_dict,
                                    retry_resource_items_queue=retry_queue)
        doc_id_1 = uuid.uuid4().hex
        doc_id_2 = uuid.uuid4().hex
        doc_id_3 = uuid.uuid4().hex
        doc_id_4 = uuid.uuid4().hex
        date_modified = datetime.datetime.utcnow().isoformat()
        worker.bulk = {
            doc_id_1: {
                'id': doc_id_1,
                'dateModified': date_modified
            },
            doc_id_2: {
                'id': doc_id_2,
                'dateModified': date_modified
            },
            doc_id_3: {
                'id': doc_id_3,
                'dateModified': date_modified
            },
            doc_id_4: {
                'id': doc_id_4,
                'dateModified': date_modified
            }
        }
        update_return_value = [
            (True, doc_id_1, '1-' + uuid.uuid4().hex),
            (True, doc_id_2, '2-' + uuid.uuid4().hex),
            (False, doc_id_3, Exception(u'New doc with oldest dateModified.')),
            (False, doc_id_4, Exception(u'Document update conflict.'))
        ]
        worker.db = MagicMock()
        worker.db.update.return_value = update_return_value

        self.assertEqual(worker.log_dict['update_documents'], 0)
        self.assertEqual(worker.log_dict['save_documents'], 0)
        self.assertEqual(worker.log_dict['skiped'], 0)
        self.assertEqual(worker.log_dict['add_to_retry'], 0)

        # Test success response from couchdb
        worker._save_bulk_docs()
        self.assertEqual(worker.log_dict['update_documents'], 1)
        self.assertEqual(worker.log_dict['save_documents'], 1)
        self.assertEqual(worker.log_dict['skiped'], 1)
        self.assertEqual(worker.log_dict['add_to_retry'], 1)

        # Test failed response from couchdb
        worker.db.update.side_effect = Exception('Some exceptions')
        worker.bulk = {
            doc_id_1: {
                'id': doc_id_1,
                'dateModified': date_modified
            },
            doc_id_2: {
                'id': doc_id_2,
                'dateModified': date_modified
            },
            doc_id_3: {
                'id': doc_id_3,
                'dateModified': date_modified
            },
            doc_id_4: {
                'id': doc_id_4,
                'dateModified': date_modified
            }
        }
        worker._save_bulk_docs()
        self.assertEqual(worker.log_dict['update_documents'], 1)
        self.assertEqual(worker.log_dict['save_documents'], 1)
        self.assertEqual(worker.log_dict['skiped'], 1)
        self.assertEqual(worker.log_dict['add_to_retry'], 5)