class BaseSaver(object):
# 文档字符串
'''
BaseSaver class allows users to save all infos data fetched from website.
:Usage:
'''
# 数据存储器的静态成员定义
SAVE_MODES = ('mongodb', 'neo4j', 'mysql')
# 初始化方法：
def __init__(self, save_mode="neo4j"):
# 文档字符串
'''
Initialize an instance of BaseSaver.
:Args:
- save_mode : a str of database to save data in.
'''
# 方法实现
if save_mode not in self.SAVE_MODES:
raise RuntimeError('存储模式指定有误，请输入mongodb、neo4j或者mysql')
self.save_mode = save_mode
if self.save_mode == 'mongodb':
# mongodb initialize
print('>>>> we are in mongodb.')
self.connector = MongoClient(
**MONGO_CONF)[MONGO_CONF.get('authSource')]
elif self.save_mode == 'neo4j':
# neo4j initialize
print('>>>> we are in neo4j.')
self.connector = Graph(**NEO_CONF)
else:
# mysql initialize
print('>>>> we are in mysql.')
self.connector = pymysql.connect(**SQL_CONF)
self.cursor = self.connector.cursor()
sql = RESORT_SQL.format(table_name)
print(sql)
self.cursor.execute(sql)
self.connector.commit()
# 数据存储方法：
def data_save(self, file_name):
# 文档字符串
'''
Saves spider fetched data into different databases.
Wipes out the old data and saves the new fetched ones.
:Args:
- file_name : a str of file name to fetch data from.
'''
# 方法实现
# 此处可以拓展成任意文件类型，其他文件类型的数据转换成json再写即可
file_path = os.path.join(save_path, file_name + '.json')
if not os.access(file_path, os.F_OK):
raise RuntimeError(f'数据文件{file_path}不存在，请检查数据！')
with open(file_path, 'r', encoding='utf-8') as file:
self.json_data = json.load(file, encoding='utf-8')
if self.save_mode == 'mongodb':
print('>>> we are saving to mongodb.')
# 删除原始数据
self.connector.drop_collection(collection)
# 保存新数据
self.connector[collection].insert_many(self.json_data)
elif self.save_mode == 'neo4j':
print('>>> we are saving to neo4j.')
# 删除原始数据, 一定要小心使用
self.graph_cleaner()
# 保存新数据
self.graph_builder()
else:
print('>>> we are saving to mysql.')
# 删除原始数据，一定要小心使用
self.cursor.execute(f"DELETE FROM {table_name}")
# 准备sql语句
data_key = self.json_data[0].keys()
sql_key = ','.join(data_key)
sql_value = ', '.join([f'%({key})s' for key in data_key])
# 保存新数据
sql = '''
INSERT INTO {0}({1})
VALUES ({2});
'''.format(table_name, sql_key, sql_value)
print(sql)
self.cursor.executemany(sql, self.json_data)
self.connector.commit()
# 知识图谱删除方法：
def graph_cleaner(self):
pass
# 知识图谱生成方法：
def graph_builder(self):
pass
# 数据存储器退出方法：
def __del__(self):
# 文档字符串
'''
The deconstructor of BaseSaver class.
Deconstructs an instance of BaseSaver, closes Databases.
'''
# 方法实现
print(f'>>>> closing {self.save_mode}.')
if self.save_mode == 'mongodb':
self.connector.client.close()
elif self.save_mode == 'mysql':
self.connector.close()
