python sqlalchemy结果集转JSON

class AlchemyEncoder(json.JSONEncoder):
    def default(self, obj):
        if isinstance(obj.__class__, DeclarativeMeta):
            # an SQLAlchemy class
            fields = {}
            for field in [x for x in dir(obj) if not x.startswith('_') and x != 'metadata']:
                data = obj.__getattribute__(field)
                try:
                    json.dumps(data)     # this will fail on non-encodable values, like other classes
                    fields[field] = data
                except TypeError:    # 添加了对datetime的处理
                    # print(type(data),data)
                    if isinstance(data, datetime.datetime):
                        fields[field] = data.strftime("%Y-%m-%d %H:%M:%S.%f")[:-3] #SQLserver数据库中毫秒是3位，日期格式;2015-05-12 11:13:58.543
                    elif isinstance(data, datetime.date):
                        fields[field] = datadata.strftime("%Y-%m-%d")
                    elif isinstance(data, decimal.Decimal):
                        fields[field]= float(data)                        
                    else:
                        fields[field] = AlchemyEncoder.default(self, data) #如果是自定义类，递归调用解析JSON，这个是对象映射关系表 也加入到JSON
            # a json-encodable dict
            return fields

        return json.JSONEncoder.default(self, obj)

使用：   rts = json.dumps(objects, cls=AlchemyEncoder,ensure_ascii=False)# 加上ensure_ascii=False 是解决中文乱码问题

以上代码是根据网上找的一个类修改的，我修改了通过实体关系递归引用实体，格式化日期。

个人感觉是最简单的sqlalchemy转JSON 的方案了.