如何使用Python读取Hive数据库？-Python教程-PHP中文网

实际业务读取hive数据库的代码

代码说明和领悟

后续附上修改成mysql的一个例子代码

首页

后端开发

Python教程

如何使用Python读取Hive数据库？

PHPz

May 09, 2023 pm 04:28 PM

python hive

实际业务读取hive数据库的代码

import logging
import pandas as pd
from impala.dbapi import connect
import sqlalchemy
from sqlalchemy.orm import sessionmaker
import os
import time
import os
import datetime
from dateutil.relativedelta import relativedelta
from typing import Dict, List
import logging
import threading
import pandas as pd
import pickle
class HiveHelper(object):
    def __init__(
        self,
        host=&#39;10.2.32.22&#39;,
        port=21051,
        database=&#39;ur_ai_dw&#39;,
        auth_mechanism=&#39;LDAP&#39;,
        user=&#39;urbi&#39;,
        password=&#39;Ur#730xd&#39;,
        logger:logging.Logger=None
        ):
        self.host = host
        self.port = port
        self.database = database
        self.auth_mechanism = auth_mechanism
        self.user = user
        self.password = password
        self.logger = logger
        self.impala_conn = None
        self.conn = None
        self.cursor = None
        self.engine = None
        self.session = None
    def create_table_code(self, file_name):
        &#39;&#39;&#39;创建表类代码&#39;&#39;&#39;
        os.system(f&#39;sqlacodegen {self.connection_str} > {file_name}&#39;)
        return self.conn
    def get_conn(self):
        &#39;&#39;&#39;创建连接或获取连接&#39;&#39;&#39;
        if self.conn is None:
            engine = self.get_engine()
            self.conn = engine.connect()
        return self.conn
    def get_impala_conn(self):
        &#39;&#39;&#39;创建连接或获取连接&#39;&#39;&#39;
        if self.impala_conn is None:
            self.impala_conn = connect(
                host=self.host,
                port=self.port,
                database=self.database,
                auth_mechanism=self.auth_mechanism,
                user=self.user,
                password=self.password
                )
        return self.impala_conn
    def get_engine(self):
        &#39;&#39;&#39;创建连接或获取连接&#39;&#39;&#39;
        if self.engine is None:
            self.engine = sqlalchemy.create_engine(&#39;impala://&#39;, creator=self.get_impala_conn)
        return self.engine
    def get_cursor(self):
        &#39;&#39;&#39;创建连接或获取连接&#39;&#39;&#39;
        if self.cursor is None:
            self.cursor = self.conn.cursor()
        return self.cursor
    def get_session(self) -> sessionmaker:
        &#39;&#39;&#39;创建连接或获取连接&#39;&#39;&#39;
        if self.session is None:
            engine = self.get_engine()
            Session = sessionmaker(bind=engine)
            self.session = Session()
        return self.session
    def close_conn(self):
        &#39;&#39;&#39;关闭连接&#39;&#39;&#39;
        if self.conn is not None:
            self.conn.close()
            self.conn = None
        self.dispose_engine()
        self.close_impala_conn()
    def close_impala_conn(self):
        &#39;&#39;&#39;关闭impala连接&#39;&#39;&#39;
        if self.impala_conn is not None:
            self.impala_conn.close()
            self.impala_conn = None
    def close_session(self):
        &#39;&#39;&#39;关闭连接&#39;&#39;&#39;
        if self.session is not None:
            self.session.close()
            self.session = None
        self.dispose_engine()
    def dispose_engine(self):
        &#39;&#39;&#39;释放engine&#39;&#39;&#39;
        if self.engine is not None:
            # self.engine.dispose(close=False)
            self.engine.dispose()
            self.engine = None
    def close_cursor(self):
        &#39;&#39;&#39;关闭cursor&#39;&#39;&#39;
        if self.cursor is not None:
            self.cursor.close()
            self.cursor = None
    def get_data(self, sql, auto_close=True) -> pd.DataFrame:
        &#39;&#39;&#39;查询数据&#39;&#39;&#39;
        conn = self.get_conn()
        data = None
        try:
            # 异常重试3次
            for i in range(3):
                try:
                    data = pd.read_sql(sql, conn)
                    break
                except Exception as ex:
                    if i == 2:
                        raise ex # 往外抛出异常
                    time.sleep(60) # 一分钟后重试
        except Exception as ex:
            self.logger.exception(ex)
            raise ex # 往外抛出异常
        finally:
            if auto_close:
                self.close_conn()
        return data
pass
class VarsHelper():
    def __init__(self, save_dir, auto_save=True):
        self.save_dir = save_dir
        self.auto_save = auto_save
        self.values = {}
        if not os.path.exists(os.path.dirname(self.save_dir)):
            os.makedirs(os.path.dirname(self.save_dir))
        if os.path.exists(self.save_dir):
            with open(self.save_dir, &#39;rb&#39;) as f:
                self.values = pickle.load(f)
                f.close()
    def set_value(self, key, value):
        self.values[key] = value
        if self.auto_save:
            self.save_file()
    def get_value(self, key):
        return self.values[key]
    def has_key(self, key):
        return key in self.values.keys()
    def save_file(self):
        with open(self.save_dir, &#39;wb&#39;) as f:
            pickle.dump(self.values, f)
            f.close()
pass
class GlobalShareArgs():
    args = {
        "debug": False
    }
    def get_args():
        return GlobalShareArgs.args
    def set_args(args):
        GlobalShareArgs.args = args
    def set_args_value(key, value):
        GlobalShareArgs.args[key] = value
    def get_args_value(key, default_value=None):
        return GlobalShareArgs.args.get(key, default_value)
    def contain_key(key):
        return key in GlobalShareArgs.args.keys()
    def update(args):
        GlobalShareArgs.args.update(args)
pass
class ShareArgs():
    args = {
        "labels_dir":"./hjx/shop_group/month_w_amt/data/labels", # 标签目录
        "labels_output_dir":"./hjx/shop_group/month_w_amt/data/labels_output", # 聚类导出标签目录
        "common_datas_dir":"./hjx/data", # 共用数据目录。ur_bi_dw的公共
        "only_predict": False, # 只识别，不训练
        "delete_model": True, # 先删除模型，仅在训练时使用
        "export_excel": False, # 导出excel
        "classes": 12, # 聚类数
        "batch_size": 16,
        "hidden_size": 32,
        "max_nrof_epochs": 100,
        "learning_rate": 0.0005,
        "loss_type": "categorical_crossentropy",
        "avg_model_num": 10,
        "steps_per_epoch": 4.0, # 4.0
        "lr_callback_patience": 4, 
        "lr_callback_cooldown": 1,
        "early_stopping_callback_patience": 6,
        "get_data": True,
    }
    def get_args():
        return ShareArgs.args
    def set_args(args):
        ShareArgs.args = args
    def set_args_value(key, value):
        ShareArgs.args[key] = value
    def get_args_value(key, default_value=None):
        return ShareArgs.args.get(key, default_value)
    def contain_key(key):
        return key in ShareArgs.args.keys()
    def update(args):
        ShareArgs.args.update(args)
pass
class UrBiGetDatasBase():
    # 线程锁列表，同保存路径共用锁
    lock_dict:Dict[str, threading.Lock] = {}
    # 时间列表，用于判断是否超时
    time_dict:Dict[str, datetime.datetime] = {}
    # 用于记录是否需要更新超时时间
    get_data_timeout_dict:Dict[str, bool] = {}
    def __init__(
        self,
        host=&#39;10.2.32.22&#39;,
        port=21051,
        database=&#39;ur_ai_dw&#39;,
        auth_mechanism=&#39;LDAP&#39;,
        user=&#39;urbi&#39;,
        password=&#39;Ur#730xd&#39;,
        save_dir=None,
        logger:logging.Logger=None,
        ):
        self.save_dir = save_dir
        self.logger = logger
        self.db_helper = HiveHelper(
            host=host,
            port=port,
            database=database,
            auth_mechanism=auth_mechanism,
            user=user,
            password=password,
            logger=logger
            )
        # 创建子目录
        if self.save_dir is not None and not os.path.exists(self.save_dir):
            os.makedirs(self.save_dir)
        self.vars_helper = None
        if GlobalShareArgs.get_args_value(&#39;debug&#39;):
            self.vars_helper = VarsHelper(&#39;./hjx/data/vars/UrBiGetDatas&#39;) 
    def close(self):
        &#39;&#39;&#39;关闭连接&#39;&#39;&#39;
        self.db_helper.close_conn()
    def get_last_time(self, key_name) -> bool:
        &#39;&#39;&#39;获取是否超时&#39;&#39;&#39;
        # 转静态路径，确保唯一性
        key_name = os.path.abspath(key_name)
        if self.vars_helper is not None and self.vars_helper.has_key(&#39;UrBiGetDatasBase.time_list&#39;):
            UrBiGetDatasBase.time_dict = self.vars_helper.get_value(&#39;UrBiGetDatasBase.time_list&#39;)
        timeout = 12 # 12小时
        if GlobalShareArgs.get_args_value(&#39;debug&#39;):
            timeout = 24 # 24小时
        get_data_timeout = False
        if key_name not in UrBiGetDatasBase.time_dict.keys() or (datetime.datetime.today() - UrBiGetDatasBase.time_dict[key_name]).total_seconds()>(timeout*60*60):
            self.logger.info(&#39;超时%d小时，重新查数据：%s&#39;, timeout, key_name)
            # UrBiGetDatasBase.time_list[key_name] = datetime.datetime.today()
            get_data_timeout = True
        else:
            self.logger.info(&#39;未超时%d小时，跳过查数据：%s&#39;, timeout, key_name)
        # if self.vars_helper is not None :
        #     self.vars_helper.set_value(&#39;UrBiGetDatasBase.time_list&#39;, UrBiGetDatasBase.time_list)
        UrBiGetDatasBase.get_data_timeout_dict[key_name] = get_data_timeout
        return get_data_timeout
    def save_last_time(self, key_name):
        &#39;&#39;&#39;更新状态超时&#39;&#39;&#39;
        # 转静态路径，确保唯一性
        key_name = os.path.abspath(key_name)
        if UrBiGetDatasBase.get_data_timeout_dict[key_name]:
            UrBiGetDatasBase.time_dict[key_name] = datetime.datetime.today()
        if self.vars_helper is not None :
            UrBiGetDatasBase.time_dict[key_name] = datetime.datetime.today()
            self.vars_helper.set_value(&#39;UrBiGetDatasBase.time_list&#39;, UrBiGetDatasBase.time_dict)
    def get_lock(self, key_name) -> threading.Lock:
        &#39;&#39;&#39;获取锁&#39;&#39;&#39;
        # 转静态路径，确保唯一性
        key_name = os.path.abspath(key_name)
        if key_name not in UrBiGetDatasBase.lock_dict.keys():
            UrBiGetDatasBase.lock_dict[key_name] = threading.Lock()
        return UrBiGetDatasBase.lock_dict[key_name]
    def get_data_of_date(
        self,
        save_dir,
        sql,
        sort_columns:List[str],
        del_index_list=[-1], # 删除最后下标
        start_date = datetime.datetime(2017, 1, 1), # 开始时间
        offset = relativedelta(months=3), # 时间间隔
        date_format_fun = lambda d: &#39;%04d%02d01&#39; % (d.year, d.month), # 查询语句中替代时间参数的格式化
        filename_format_fun = lambda d: &#39;%04d%02d.csv&#39; % (d.year, d.month), # 查询语句中替代时间参数的格式化
        stop_date = &#39;20700101&#39;, # 超过时间则停止
        data_format_fun = None, # 格式化数据
        ):
        &#39;&#39;&#39;分时间增量读取数据&#39;&#39;&#39;
        # 创建文件夹
        if not os.path.exists(save_dir):
            os.makedirs(save_dir)
        else:
            #删除最后一个文件
            file_list = os.listdir(save_dir)
            if len(file_list)>0:
                file_list.sort()
                for del_index in del_index_list:
                    os.remove(os.path.join(save_dir,file_list[del_index]))
                    print(&#39;删除最后一个文件：&#39;, file_list[del_index])
        select_index = -1
        # start_date = datetime.datetime(2017, 1, 1)
        while True:
            end_date = start_date + offset
            start_date_str = date_format_fun(start_date)
            end_date_str = date_format_fun(end_date)
            self.logger.info(&#39;date: %s-%s&#39;, start_date_str, end_date_str)
            file_path = os.path.join(save_dir, filename_format_fun(start_date))
            # self.logger.info(&#39;file_path: %s&#39;, file_path)
            if not os.path.exists(file_path):
                data:pd.DataFrame = self.db_helper.get_data(sql % (start_date_str, end_date_str))
                if data is None:
                    break
                self.logger.info(&#39;data: %d&#39;, len(data))
                # self.logger.info(&#39;data: %d&#39;, data.columns)
                if len(data)>0:
                    select_index+=1
                    if data_format_fun is not None:
                        data = data_format_fun(data)
                    # 排序
                    data = data.sort_values(sort_columns)
                    data.to_csv(file_path)
                elif select_index!=-1:
                    break
                elif stop_date < start_date_str:
                    raise Exception("读取数据异常，时间超出最大值！")
            start_date = end_date
pass
class UrBiGetDatas(UrBiGetDatasBase):
    def __init__(
        self,
        host=&#39;10.2.32.22&#39;,
        port=21051,
        database=&#39;ur_ai_dw&#39;,
        auth_mechanism=&#39;LDAP&#39;,
        user=&#39;urbi&#39;,
        password=&#39;Ur#730xd&#39;,
        save_dir=&#39;./hjx/data/ur_bi_dw_data&#39;,
        logger:logging.Logger=None
        ):
        self.save_dir = save_dir
        self.logger = logger
        super().__init__(
            host=host,
            port=port,
            database=database,
            auth_mechanism=auth_mechanism,
            user=user,
            password=password,
            save_dir=save_dir,
            logger=logger
            )
    def get_dim_date(self):
        &#39;&#39;&#39;日期数据&#39;&#39;&#39;
        file_path = os.path.join(self.save_dir,&#39;ur_bi_dw.dim_date.csv&#39;)
        now_lock = self.get_lock(file_path)
        now_lock.acquire() # 加锁
        try:
            # 设置超时4小时才重新查数据
            if not self.get_last_time(file_path):
                return
            sql = &#39;SELECT * FROM ur_bi_dw.dim_date&#39;
            data:pd.DataFrame = self.db_helper.get_data(sql)
            columns = list(data.columns)
            columns = {c:&#39;dim_date.&#39;+c for c in columns}
            data = data.rename(columns=columns)
            data = data.sort_values([&#39;dim_date.date_key&#39;])
            data.to_csv(file_path)
            # 更新超时时间
            self.save_last_time(file_path)
        except Exception as ex:
            self.logger.exception(ex)
            raise ex # 往外抛出异常
        finally:
            now_lock.release() # 释放锁
    def get_dim_shop(self):
        &#39;&#39;&#39;店铺数据&#39;&#39;&#39;
        file_path = os.path.join(self.save_dir,&#39;ur_bi_dw.dim_shop.csv&#39;)
        now_lock = self.get_lock(file_path)
        now_lock.acquire() # 加锁
        try:
            # 设置超时4小时才重新查数据
            if not self.get_last_time(file_path):
                return
            sql = &#39;SELECT * FROM ur_bi_dw.dim_shop&#39;
            data:pd.DataFrame = self.db_helper.get_data(sql)
            columns = list(data.columns)
            columns = {c:&#39;dim_shop.&#39;+c for c in columns}
            data = data.rename(columns=columns)
            data = data.sort_values([&#39;dim_shop.shop_no&#39;])
            data.to_csv(file_path)
            # 更新超时时间
            self.save_last_time(file_path)
        except Exception as ex:
            self.logger.exception(ex)
            raise ex # 往外抛出异常
        finally:
            now_lock.release() # 释放锁
    def get_dim_vip(self):
        &#39;&#39;&#39;会员数据&#39;&#39;&#39;
        sub_dir = os.path.join(self.save_dir,&#39;vip_no&#39;)
        now_lock = self.get_lock(sub_dir)
        now_lock.acquire() # 加锁
        try:
            # 设置超时4小时才重新查数据
            if not self.get_last_time(sub_dir):
                return
            sql = &#39;&#39;&#39;SELECT dv.*, dd.date_key, dd.date_name2 
            FROM ur_bi_dw.dim_vip as dv
            INNER JOIN ur_bi_dw.dim_date as dd
            ON dv.card_create_date=dd.date_name2 
            where dd.date_key >= %s
            and dd.date_key < %s&#39;&#39;&#39;
            # data:pd.DataFrame = self.db_helper.get_data(sql)
            sort_columns = [&#39;dv.vip_no&#39;]
            # TODO:
            self.get_data_of_date(
                save_dir=sub_dir,
                sql=sql,
                sort_columns=sort_columns,
                start_date=datetime.datetime(2017, 1, 1), # 开始时间
                offset=relativedelta(years=1)
            )
            # 更新超时时间
            self.save_last_time(sub_dir)
        except Exception as ex:
            self.logger.exception(ex)
            raise ex # 往外抛出异常
        finally:
            now_lock.release() # 释放锁
    def get_weather(self):
        &#39;&#39;&#39;天气数据&#39;&#39;&#39;
        sub_dir = os.path.join(self.save_dir,&#39;weather&#39;)
        now_lock = self.get_lock(sub_dir)
        now_lock.acquire() # 加锁
        try:
            # 设置超时4小时才重新查数据
            if not self.get_last_time(sub_dir):
                return
            sql = """
            select weather.* from ur_bi_ods.ods_base_weather_data_1200 as weather
            where weather.date_key>=%s and weather.date_key<%s
            """
            sort_columns = [&#39;weather.date_key&#39;,&#39;weather.areaid&#39;]
            def data_format_fun(data):
                columns = list(data.columns)
                columns = {c:&#39;weather.&#39;+c for c in columns}
                data = data.rename(columns=columns)
                return data
            self.get_data_of_date(
                save_dir=sub_dir,
                sql=sql,
                sort_columns=sort_columns,
                del_index_list=[-2, -1], # 删除最后下标
                data_format_fun=data_format_fun,
            )
            # 更新超时时间
            self.save_last_time(sub_dir)
        except Exception as ex:
            self.logger.exception(ex)
            raise ex # 往外抛出异常
        finally:
            now_lock.release() # 释放锁
    def get_weather_city(self):
        &#39;&#39;&#39;天气城市数据&#39;&#39;&#39;
        file_path = os.path.join(self.save_dir,&#39;ur_bi_dw.weather_city.csv&#39;)
        now_lock = self.get_lock(file_path)
        now_lock.acquire() # 加锁
        try:
            # 设置超时4小时才重新查数据
            if not self.get_last_time(file_path):
                return
            sql = &#39;SELECT * FROM ur_bi_dw.dim_weather_city as weather_city&#39;
            data:pd.DataFrame = self.db_helper.get_data(sql)
            columns = list(data.columns)
            columns = {c:&#39;weather_city.&#39;+c for c in columns}
            data = data.rename(columns=columns)
            data.to_csv(file_path)
            # 更新超时时间
            self.save_last_time(file_path)
        except Exception as ex:
            self.logger.exception(ex)
            raise ex # 往外抛出异常
        finally:
            now_lock.release() # 释放锁
    def get_dim_goods(self):
        &#39;&#39;&#39;货品数据&#39;&#39;&#39;
        file_path = os.path.join(self.save_dir,&#39;ur_bi_dw.dim_goods.csv&#39;)
        now_lock = self.get_lock(file_path)
        now_lock.acquire() # 加锁
        try:
            # 设置超时4小时才重新查数据
            if not self.get_last_time(file_path):
                return
            sql = &#39;SELECT * FROM ur_bi_dw.dim_goods&#39;
            data:pd.DataFrame = self.db_helper.get_data(sql)
            columns = list(data.columns)
            columns = {c:&#39;dim_goods.&#39;+c for c in columns}
            data = data.rename(columns=columns)
            data.to_csv(file_path)
            # 更新超时时间
            self.save_last_time(file_path)
        except Exception as ex:
            self.logger.exception(ex)
            raise ex # 往外抛出异常
        finally:
            now_lock.release() # 释放锁
    def get_dim_goods_market_shop_date(self):
        &#39;&#39;&#39;店铺商品生命周期数据&#39;&#39;&#39;
        file_path = os.path.join(self.save_dir,&#39;ur_bi_dw.dim_goods_market_shop_date.csv&#39;)
        now_lock = self.get_lock(file_path)
        now_lock.acquire() # 加锁
        try:
            # 设置超时4小时才重新查数据
            if not self.get_last_time(file_path):
                return
            # sql = &#39;SELECT * FROM ur_bi_dw.dim_goods_market_shop_date as goods_shop_date&#39;
            sql = &#39;&#39;&#39;
            select shop_no, sku_no, shop_market_date, lifecycle_end_date, lifecycle_days
            FROM ur_bi_dw.dim_goods_market_shop_date
            where lifecycle_end_date is not null
            &#39;&#39;&#39;
            data:pd.DataFrame = self.db_helper.get_data(sql)
            columns = list(data.columns)
            columns = {c:c.replace(&#39;lifecycle_end_date.&#39;,&#39;&#39;) for c in columns}
            data = data.rename(columns=columns)
            data = data.sort_values([&#39;shop_market_date&#39;])
            data.to_csv(file_path, index=False)
            # 更新超时时间
            self.save_last_time(file_path)
        except Exception as ex:
            self.logger.exception(ex)
            raise ex # 往外抛出异常
        finally:
            now_lock.release() # 释放锁
    def get_dim_goods_market_date(self):
        &#39;&#39;&#39;全国商品生命周期数据&#39;&#39;&#39;
        file_path = os.path.join(self.save_dir,&#39;ur_bi_dw.dim_goods_market_date.csv&#39;)
        now_lock = self.get_lock(file_path)
        now_lock.acquire() # 加锁
        try:
            # 设置超时4小时才重新查数据
            if not self.get_last_time(file_path):
                return
            sql = &#39;&#39;&#39;
            select * FROM ur_bi_dw.dim_goods_market_date
            &#39;&#39;&#39;
            data:pd.DataFrame = self.db_helper.get_data(sql)
            columns = list(data.columns)
            columns = {c:&#39;dim_goods_market_date.&#39;+c for c in columns}
            data = data.rename(columns=columns)
            data = data.sort_values([&#39;dim_goods_market_date.sku_no&#39;])
            data.to_csv(file_path, index=False)
            # 更新超时时间
            self.save_last_time(file_path)
        except Exception as ex:
            self.logger.exception(ex)
            raise ex # 往外抛出异常
        finally:
            now_lock.release() # 释放锁
    def get_dim_goods_color_dev_sizes(self):
        &#39;&#39;&#39;商品开发码数数据&#39;&#39;&#39;
        file_path = os.path.join(self.save_dir,&#39;dim_goods_color_dev_sizes.csv&#39;)
        now_lock = self.get_lock(file_path)
        now_lock.acquire() # 加锁
        try:
            # 设置超时4小时才重新查数据
            if not self.get_last_time(file_path):
                return
            # sql = &#39;SELECT * FROM ur_bi_dw.dim_goods_market_shop_date as goods_shop_date&#39;
            sql = &#39;SELECT * FROM ur_bi_dm.dim_goods_color_dev_sizes&#39;
            data:pd.DataFrame = self.db_helper.get_data(sql)
            columns = list(data.columns)
            columns = {c:c.replace(&#39;dim_goods_color_dev_sizes.&#39;,&#39;&#39;) for c in columns}
            data = data.rename(columns=columns)
            data.to_csv(file_path, index=False)
            # 更新超时时间
            self.save_last_time(file_path)
        except Exception as ex:
            self.logger.exception(ex)
            raise ex # 往外抛出异常
        finally:
            now_lock.release() # 释放锁
    def get_dwd_daily_sales_size(self):
        &#39;&#39;&#39;实际销售金额&#39;&#39;&#39;
        sub_dir = os.path.join(self.save_dir,&#39;dwd_daily_sales_size_all&#39;)
        now_lock = self.get_lock(sub_dir)
        now_lock.acquire() # 加锁
        try:
            # 设置超时4小时才重新查数据
            if not self.get_last_time(sub_dir):
                return
            sql = """
            select shop_no,sku_no,date_key,`size`,
                sum(tag_price) as `tag_price`,
                sum(sales_qty) as `sales_qty`,
                sum(sales_tag_amt) as `sales_tag_amt`,
                sum(sales_amt) as `sales_amt`,
                count(0) as `sales_count`
            from ur_bi_dw.dwd_daily_sales_size as sales
            where sales.date_key>=%s and sales.date_key<%s
                and sales.currency_code=&#39;CNY&#39;
            group by shop_no,sku_no,date_key,`size`
            """
            sort_columns = [&#39;date_key&#39;,&#39;shop_no&#39;,&#39;sku_no&#39;]
            self.get_data_of_date(
                save_dir=sub_dir,
                sql=sql,
                sort_columns=sort_columns,
                start_date=datetime.datetime(2017, 1, 1), # 开始时间
            )
            # 更新超时时间
            self.save_last_time(sub_dir)
        except Exception as ex:
            self.logger.exception(ex)
            raise ex # 往外抛出异常
        finally:
            now_lock.release() # 释放锁
    def get_dwd_daily_delivery_size(self):
        &#39;&#39;&#39;实际配货金额&#39;&#39;&#39;
        sub_dir = os.path.join(self.save_dir,&#39;dwd_daily_delivery_size_all&#39;)
        now_lock = self.get_lock(sub_dir)
        now_lock.acquire() # 加锁
        try:
            # 设置超时4小时才重新查数据
            if not self.get_last_time(sub_dir):
                return
            sql = """
            select shop_no,sku_no,date_key,`size`,
                sum(delivery.shop_distr_received_qty) as `shop_distr_received_qty`,
                sum(delivery.shop_distr_received_amt) as `shop_distr_received_amt`,
                sum(delivery.online_distr_received_qty) as `online_distr_received_qty`,
                sum(delivery.online_distr_received_amt) as `online_distr_received_amt`,
                sum(delivery.pr_received_qty) as `pr_received_qty`,
                count(0) as `delivery_count`
            from ur_bi_dw.dwd_daily_delivery_size as delivery
            where delivery.date_key>=%s and delivery.date_key<%s
                and delivery.currency_code=&#39;CNY&#39;
            group by shop_no,sku_no,date_key,`size`
            """
            sort_columns = [&#39;date_key&#39;,&#39;shop_no&#39;,&#39;sku_no&#39;]
            self.get_data_of_date(
                save_dir=sub_dir,
                sql=sql,
                sort_columns=sort_columns,
                start_date=datetime.datetime(2017, 1, 1), # 开始时间
            )
            # 更新超时时间
            self.save_last_time(sub_dir)
        except Exception as ex:
            self.logger.exception(ex)
            raise ex # 往外抛出异常
        finally:
            now_lock.release() # 释放锁
    def get_v_last_nation_sales_status(self):
        &#39;&#39;&#39;商品畅滞销数据&#39;&#39;&#39;
        file_path = os.path.join(self.save_dir,&#39;v_last_nation_sales_status.csv&#39;)
        now_lock = self.get_lock(file_path)
        now_lock.acquire() # 加锁
        try:
            # 设置超时4小时才重新查数据
            if not self.get_last_time(file_path):
                return
            sql = &#39;SELECT * FROM ur_bi_dw.v_last_nation_sales_status&#39;
            data:pd.DataFrame = self.db_helper.get_data(sql)
            columns = list(data.columns)
            columns = {c:c.replace(&#39;v_last_nation_sales_status.&#39;,&#39;&#39;) for c in columns}
            data = data.rename(columns=columns)
            data.to_csv(file_path, index=False)
            # 更新超时时间
            self.save_last_time(file_path)
        except Exception as ex:
            self.logger.exception(ex)
            raise ex # 往外抛出异常
        finally:
            now_lock.release() # 释放锁
    def get_dwd_daily_finacial_goods(self):
        &#39;&#39;&#39;商品成本价数据&#39;&#39;&#39;
        file_path = os.path.join(self.save_dir,&#39;dwd_daily_finacial_goods.csv&#39;)
        now_lock = self.get_lock(file_path)
        now_lock.acquire() # 加锁
        try:
            # 设置超时4小时才重新查数据
            if not self.get_last_time(file_path):
                return
            sql = """
            select t1.sku_no,t1.`size`,t1.cost_tax_incl from ur_bi_dw.dwd_daily_finacial_goods as t1
            inner join (
                select sku_no,`size`,max(date_key) as date_key
                from ur_bi_dw.dwd_daily_finacial_goods
                where currency_code=&#39;CNY&#39; and country_code=&#39;CN&#39;
                group by sku_no,`size`
            ) as t2
            on t2.sku_no=t1.sku_no
                and t2.`size`=t1.`size`
                and t2.date_key=t1.date_key
            where t1.currency_code=&#39;CNY&#39; and t1.country_code=&#39;CN&#39;
            """
            data:pd.DataFrame = self.db_helper.get_data(sql)
            columns = list(data.columns)
            columns = {c:c.replace(&#39;t1.&#39;,&#39;&#39;) for c in columns}
            data = data.rename(columns=columns)
            data.to_csv(file_path, index=False)
            # 更新超时时间
            self.save_last_time(file_path)
        except Exception as ex:
            self.logger.exception(ex)
            raise ex # 往外抛出异常
        finally:
            now_lock.release() # 释放锁
    def get_dim_size_group(self):
        &#39;&#39;&#39;尺码映射数据&#39;&#39;&#39;
        file_path = os.path.join(self.save_dir,&#39;dim_size_group.csv&#39;)
        now_lock = self.get_lock(file_path)
        now_lock.acquire() # 加锁
        try:
            # 设置超时4小时才重新查数据
            if not self.get_last_time(file_path):
                return
            sql = """select * from ur_bi_dw.dim_size_group"""
            data:pd.DataFrame = self.db_helper.get_data(sql)
            columns = list(data.columns)
            columns = {c:c.replace(&#39;dim_size_group.&#39;,&#39;&#39;) for c in columns}
            data = data.rename(columns=columns)
            data.to_csv(file_path, index=False)
            # 更新超时时间
            self.save_last_time(file_path)
        except Exception as ex:
            self.logger.exception(ex)
            raise ex # 往外抛出异常
        finally:
            now_lock.release() # 释放锁
pass
def get_common_datas(
    host=&#39;10.2.32.22&#39;,
    port=21051,
    database=&#39;ur_ai_dw&#39;,
    auth_mechanism=&#39;LDAP&#39;,
    user=&#39;urbi&#39;,
    password=&#39;Ur#730xd&#39;,
    logger:logging.Logger=None):
    # 共用文件
    common_datas_dir = ShareArgs.get_args_value(&#39;common_datas_dir&#39;)
    common_ur_bi_dir = os.path.join(common_datas_dir, &#39;ur_bi_data&#39;)
    ur_bi_get_datas = UrBiGetDatas(
        host=host,
        port=port,
        database=database,
        auth_mechanism=auth_mechanism,
        user=user,
        password=password,
        save_dir=common_ur_bi_dir,
        logger=logger
    )
    try:
        logger.info(&#39;正在查询日期数据...&#39;)
        ur_bi_get_datas.get_dim_date()
        logger.info(&#39;查询日期数据完成！&#39;)
        logger.info(&#39;正在查询店铺数据...&#39;)
        ur_bi_get_datas.get_dim_shop()
        logger.info(&#39;查询店铺数据完成！&#39;)
        logger.info(&#39;正在查询天气数据...&#39;)
        ur_bi_get_datas.get_weather()
        logger.info(&#39;查询天气数据完成！&#39;)
        logger.info(&#39;正在查询天气城市数据...&#39;)
        ur_bi_get_datas.get_weather_city()
        logger.info(&#39;查询天气城市数据完成！&#39;)
        logger.info(&#39;正在查询货品数据...&#39;)
        ur_bi_get_datas.get_dim_goods()
        logger.info(&#39;查询货品数据完成！&#39;)
        logger.info(&#39;正在查询实际销量数据...&#39;)
        ur_bi_get_datas.get_dwd_daily_sales_size()
        logger.info(&#39;查询实际销量数据完成！&#39;)
    except Exception as ex:
        logger.exception(ex)
        raise ex # 往外抛出异常
    finally:
        ur_bi_get_datas.close()
pass
class CustomUrBiGetDatas(UrBiGetDatasBase):
    def __init__(
        self,
        host=&#39;10.2.32.22&#39;,
        port=21051,
        database=&#39;ur_ai_dw&#39;,
        auth_mechanism=&#39;LDAP&#39;,
        user=&#39;urbi&#39;,
        password=&#39;Ur#730xd&#39;,
        save_dir=&#39;./hjx/data/ur_bi_data&#39;,
        logger:logging.Logger=None
        ):
        self.save_dir = save_dir
        self.logger = logger
        super().__init__(
            host=host,
            port=port,
            database=database,
            auth_mechanism=auth_mechanism,
            user=user,
            password=password,
            save_dir=save_dir,
            logger=logger
            )
    def get_sales_goal_amt(self):
        &#39;&#39;&#39;销售目标金额&#39;&#39;&#39;
        file_path = os.path.join(self.save_dir,&#39;month_of_year_sales_goal_amt.csv&#39;)
        now_lock = self.get_lock(file_path)
        now_lock.acquire() # 加锁
        try:
            # 设置超时4小时才重新查数据
            if not self.get_last_time(file_path):
                return
            sql = &#39;&#39;&#39;
            select sales_goal.shop_no,
                if(sales_goal.serial=&#39;Y&#39;,&#39;W&#39;,sales_goal.serial) as `sales_goal.serial`,
                dates.month_of_year,
                sum(sales_goal.sales_goal_amt) as sales_goal_amt
            from ur_bi_dw.dwd_sales_goal_west as sales_goal
            inner join ur_bi_dw.dim_date as dates
                on sales_goal.date_key = dates.date_key
            group by sales_goal.shop_no,
                if(sales_goal.serial=&#39;Y&#39;,&#39;W&#39;,sales_goal.serial),
                dates.month_of_year
            &#39;&#39;&#39;
            data:pd.DataFrame = self.db_helper.get_data(sql)
            data = data.rename(columns={
                &#39;shop_no&#39;:&#39;sales_goal.shop_no&#39;,
                &#39;serial&#39;:&#39;sales_goal.serial&#39;,
                &#39;month_of_year&#39;:&#39;dates.month_of_year&#39;,
            })
            # 排序
            data = data.sort_values([&#39;sales_goal.shop_no&#39;,&#39;sales_goal.serial&#39;,&#39;dates.month_of_year&#39;])
            data.to_csv(file_path)
            # 更新超时时间
            self.save_last_time(file_path)
        except Exception as ex:
            self.logger.exception(ex)
            raise ex # 往外抛出异常
        finally:
            now_lock.release() # 释放锁
    def get_shop_serial_area(self):
        &#39;&#39;&#39;店-系列面积&#39;&#39;&#39;
        file_path = os.path.join(self.save_dir,&#39;shop_serial_area.csv&#39;)
        now_lock = self.get_lock(file_path)
        now_lock.acquire() # 加锁
        try:
            # 设置超时4小时才重新查数据
            if not self.get_last_time(file_path):
                return
            sql = &#39;&#39;&#39;
            select shop_serial_area.shop_no,
                if(shop_serial_area.serial=&#39;Y&#39;,&#39;W&#39;,shop_serial_area.serial) as `shop_serial_area.serial`,
                shop_serial_area.month_of_year,
                sum(shop_serial_area.area) as `shop_serial_area.area`
            from ur_bi_dw.dwd_shop_serial_area as shop_serial_area
            where shop_serial_area.area is not null
            group by shop_serial_area.shop_no,if(shop_serial_area.serial=&#39;Y&#39;,&#39;W&#39;,shop_serial_area.serial),shop_serial_area.month_of_year
            &#39;&#39;&#39;
            data:pd.DataFrame = self.db_helper.get_data(sql)
            data = data.rename(columns={
                &#39;shop_no&#39;:&#39;shop_serial_area.shop_no&#39;,
                &#39;serial&#39;:&#39;shop_serial_area.serial&#39;,
                &#39;month_of_year&#39;:&#39;shop_serial_area.month_of_year&#39;,
                &#39;area&#39;:&#39;shop_serial_area.area&#39;,
            })
            # 排序
            data = data.sort_values([&#39;shop_serial_area.shop_no&#39;,&#39;shop_serial_area.serial&#39;,&#39;shop_serial_area.month_of_year&#39;])
            data.to_csv(file_path)
            # 更新超时时间
            self.save_last_time(file_path)
        except Exception as ex:
            self.logger.exception(ex)
            raise ex # 往外抛出异常
        finally:
            now_lock.release() # 释放锁
pass
def get_datas(
    host=&#39;10.2.32.22&#39;,
    port=21051,
    database=&#39;ur_ai_dw&#39;,
    auth_mechanism=&#39;LDAP&#39;,
    user=&#39;urbi&#39;,
    password=&#39;Ur#730xd&#39;,
    save_dir=&#39;./data/sales_forecast/ur_bi_dw_data&#39;,
    logger:logging.Logger=None):
    ur_bi_get_datas = CustomUrBiGetDatas(
        host=host,
        port=port,
        database=database,
        auth_mechanism=auth_mechanism,
        user=user,
        password=password,
        save_dir=save_dir,
        logger=logger
    )
    try:
        # 店，系列，品类，年月，销售目标金额
        logger.info(&#39;正在查询年月销售目标金额数据...&#39;)
        ur_bi_get_datas.get_sales_goal_amt()
        logger.info(&#39;查询年月销售目标金额数据完成！&#39;)
    except Exception as ex:
        logger.exception(ex)
        raise ex # 往外抛出异常
    finally:
        ur_bi_get_datas.close()
pass
def getdata_ur_bi_dw(
    host=&#39;10.2.32.22&#39;,
    port=21051,
    database=&#39;ur_ai_dw&#39;,
    auth_mechanism=&#39;LDAP&#39;,
    user=&#39;urbi&#39;,
    password=&#39;Ur#730xd&#39;,
    save_dir=&#39;./data/sales_forecast/ur_bi_dw_data&#39;,
    logger=None
):
    get_common_datas(
        host=host,
        port=port,
        database=database,
        auth_mechanism=auth_mechanism,
        user=user,
        password=password,
        logger=logger
    )
    get_datas(
        host=host,
        port=port,
        database=database,
        auth_mechanism=auth_mechanism,
        user=user,
        password=password,
        save_dir=save_dir,
        logger=logger
    )
pass
# 代码入口
# getdata_ur_bi_dw(
#     host=ur_bi_dw_host,
#     port=ur_bi_dw_port,
#     database=ur_bi_dw_database,
#     auth_mechanism=ur_bi_dw_auth_mechanism,
#     user=ur_bi_dw_user,
#     password=ur_bi_dw_password,
#     save_dir=ur_bi_dw_save_dir,
#     logger=logger
#     )

登录后复制

代码说明和领悟

每个类的具体作用说明，代码需要根据下面的文字说明进行“食用”：

（第一层）HiveHelper完成了连接数据库、关闭数据库连接、生成事务、执行、引擎、连接等功能

VarsHelper提供了一个简单的持久化功能，可以将对象以文件的形式存放在磁盘上。并提供设置值、获取值、判断值是否存在的方法

GlobalShareArgs提供了一个字典，并且提供了获取字典、设置字典、设置字典键值对、设置字典键的值、判断键是否在字典中、更新字典等方法

ShareArgs跟GlobalShareArgs类似，只是一开始字典的初始化的键值对比较多

（第二层）UrBiGetDataBase类，提供了线程锁字典、时间字典、超时判断字典，都是类变量；使用了HiveHelper类，但注意，不是继承。在具体的sql读数时，提供了线程固定和时间判断

（第三层）UrBiGetDatas类，获取hive数据库那边的日期数据、店铺数据、会员数据、天气数据、天气城市数据、商品数据、店铺生命周期数据、全国商品生命周期数据、商品开发码数数据、实际销售金额、实际配货金额、商品畅滞销数据、商品成本价数据、尺码映射数据等。

（第四层）get_common_data函数，使用URBiGetData类读取日期、店铺、天气、天气城市、货品、实际销量数据，并缓存到文件夹./yongjian/data/ur_bi_data下面

CustomUrBiGetData类，继承了UrBiGetDatasBase类，读取销售目标金额、点系列面积数据。

（这个也是第四层）get_datas函数，通过CustomUrBiGetData类，读取年月销售目标金额。

总的函数：（这个是总的调用入口函数）get_data_ur_bi_dw函数，调用了get_common_data和get_datas函数进行读取数据，然后将数据保存到某个文件夹目录下面。

举一反三，如果你不是hive数据库，你可以将第一层这个底层更换成mysql。主页有解释如果进行更换。第二层不需要改变，第三层就是你想要进行读取的数据表，不同的数据库你想要读取的数据表也不同，所以sql需要你在这里写，套用里面的方法即可，基本上就是修改sql就好了。

这种方法的好处在于，数据不会重复读取，并且读取的数据都可以得到高效的使用。

后续附上修改成mysql的一个例子代码

import logging
import pandas as pd
from impala.dbapi import connect
import sqlalchemy
from sqlalchemy.orm import sessionmaker
import os
import time
import os
import datetime
from dateutil.relativedelta import relativedelta
from typing import Dict, List
import logging
import threading
import pandas as pd
import pickle
class MySqlHelper(object):
    def __init__(
        self,
        host=&#39;192.168.15.144&#39;,
        port=3306,
        database=&#39;test_ims&#39;,
        user=&#39;spkjz_writer&#39;,
        password=&#39;7cmoP3QDtueVJQj2q4Az&#39;,
        logger:logging.Logger=None
        ):
        self.host = host
        self.port = port
        self.database = database
        self.user = user
        self.password = password
        self.logger = logger
        self.connection_str = &#39;mysql+pymysql://%s:%s@%s:%d/%s&#39; %(
            self.user, self.password, self.host, self.port, self.database
        )
        self.conn = None
        self.cursor = None
        self.engine = None
        self.session = None
    def create_table_code(self, file_name):
        &#39;&#39;&#39;创建表类代码&#39;&#39;&#39;
        os.system(f&#39;sqlacodegen {self.connection_str} > {file_name}&#39;)
        return self.conn
    def get_conn(self):
        &#39;&#39;&#39;创建连接或获取连接&#39;&#39;&#39;
        if self.conn is None:
            engine = self.get_engine()
            self.conn = engine.connect()
        return self.conn
    def get_engine(self):
        &#39;&#39;&#39;创建连接或获取连接&#39;&#39;&#39;
        if self.engine is None:
            self.engine = sqlalchemy.create_engine(self.connection_str)
        return self.engine
    def get_cursor(self):
        &#39;&#39;&#39;创建连接或获取连接&#39;&#39;&#39;
        if self.cursor is None:
            self.cursor = self.conn.cursor()
        return self.cursor
    def get_session(self) -> sessionmaker:
        &#39;&#39;&#39;创建连接或获取连接&#39;&#39;&#39;
        if self.session is None:
            engine = self.get_engine()
            Session = sessionmaker(bind=engine)
            self.session = Session()
        return self.session
    def close_conn(self):
        &#39;&#39;&#39;关闭连接&#39;&#39;&#39;
        if self.conn is not None:
            self.conn.close()
            self.conn = None
        self.dispose_engine()
    def close_session(self):
        &#39;&#39;&#39;关闭连接&#39;&#39;&#39;
        if self.session is not None:
            self.session.close()
            self.session = None
        self.dispose_engine()
    def dispose_engine(self):
        &#39;&#39;&#39;释放engine&#39;&#39;&#39;
        if self.engine is not None:
            # self.engine.dispose(close=False)
            self.engine.dispose()
            self.engine = None
    def close_cursor(self):
        &#39;&#39;&#39;关闭cursor&#39;&#39;&#39;
        if self.cursor is not None:
            self.cursor.close()
            self.cursor = None
    def get_data(self, sql, auto_close=True) -> pd.DataFrame:
        &#39;&#39;&#39;查询数据&#39;&#39;&#39;
        conn = self.get_conn()
        data = None
        try:
            # 异常重试3次
            for i in range(3):
                try:
                    data = pd.read_sql(sql, conn)
                    break
                except Exception as ex:
                    if i == 2:
                        raise ex # 往外抛出异常
                    time.sleep(60) # 一分钟后重试
        except Exception as ex:
            self.logger.exception(ex)
            raise ex # 往外抛出异常
        finally:
            if auto_close:
                self.close_conn()
        return data
pass
class VarsHelper():
    def __init__(self, save_dir, auto_save=True):
        self.save_dir = save_dir
        self.auto_save = auto_save
        self.values = {}
        if not os.path.exists(os.path.dirname(self.save_dir)):
            os.makedirs(os.path.dirname(self.save_dir))
        if os.path.exists(self.save_dir):
            with open(self.save_dir, &#39;rb&#39;) as f:
                self.values = pickle.load(f)
                f.close()
    def set_value(self, key, value):
        self.values[key] = value
        if self.auto_save:
            self.save_file()
    def get_value(self, key):
        return self.values[key]
    def has_key(self, key):
        return key in self.values.keys()
    def save_file(self):
        with open(self.save_dir, &#39;wb&#39;) as f:
            pickle.dump(self.values, f)
            f.close()
pass
class GlobalShareArgs():
    args = {
        "debug": False
    }
    def get_args():
        return GlobalShareArgs.args
    def set_args(args):
        GlobalShareArgs.args = args
    def set_args_value(key, value):
        GlobalShareArgs.args[key] = value
    def get_args_value(key, default_value=None):
        return GlobalShareArgs.args.get(key, default_value)
    def contain_key(key):
        return key in GlobalShareArgs.args.keys()
    def update(args):
        GlobalShareArgs.args.update(args)
pass
class ShareArgs():
    args = {
        "labels_dir":"./hjx/shop_group/month_w_amt/data/labels", # 标签目录
        "labels_output_dir":"./hjx/shop_group/month_w_amt/data/labels_output", # 聚类导出标签目录
        "common_datas_dir":"./hjx/data", # 共用数据目录。ur_bi_dw的公共
        "only_predict": False, # 只识别，不训练
        "delete_model": True, # 先删除模型，仅在训练时使用
        "export_excel": False, # 导出excel
        "classes": 12, # 聚类数
        "batch_size": 16,
        "hidden_size": 32,
        "max_nrof_epochs": 100,
        "learning_rate": 0.0005,
        "loss_type": "categorical_crossentropy",
        "avg_model_num": 10,
        "steps_per_epoch": 4.0, # 4.0
        "lr_callback_patience": 4, 
        "lr_callback_cooldown": 1,
        "early_stopping_callback_patience": 6,
        "get_data": True,
    }
    def get_args():
        return ShareArgs.args
    def set_args(args):
        ShareArgs.args = args
    def set_args_value(key, value):
        ShareArgs.args[key] = value
    def get_args_value(key, default_value=None):
        return ShareArgs.args.get(key, default_value)
    def contain_key(key):
        return key in ShareArgs.args.keys()
    def update(args):
        ShareArgs.args.update(args)
pass
class IMSGetDatasBase():
    # 线程锁列表，同保存路径共用锁
    lock_dict:Dict[str, threading.Lock] = {}
    # 时间列表，用于判断是否超时
    time_dict:Dict[str, datetime.datetime] = {}
    # 用于记录是否需要更新超时时间
    get_data_timeout_dict:Dict[str, bool] = {}
    def __init__(
        self,
        host=&#39;192.168.15.144&#39;,
        port=3306,
        database=&#39;test_ims&#39;,
        user=&#39;spkjz_writer&#39;,
        password=&#39;Ur#7cmoP3QDtueVJQj2q4Az&#39;,
        save_dir=None,
        logger:logging.Logger=None,
        ):
        self.save_dir = save_dir
        self.logger = logger
        self.db_helper = MySqlHelper(
            host=host,
            port=port,
            database=database,
            user=user,
            password=password,
            logger=logger
            )
        # 创建子目录
        if self.save_dir is not None and not os.path.exists(self.save_dir):
            os.makedirs(self.save_dir)
        self.vars_helper = None
        if GlobalShareArgs.get_args_value(&#39;debug&#39;):
            self.vars_helper = VarsHelper(&#39;./hjx/data/vars/IMSGetDatas&#39;) # 把超时时间保存到文件，注释该行即可停掉，只用于调试
    def close(self):
        &#39;&#39;&#39;关闭连接&#39;&#39;&#39;
        self.db_helper.close_conn()
    def get_last_time(self, key_name) -> bool:
        &#39;&#39;&#39;获取是否超时&#39;&#39;&#39;
        # 转静态路径，确保唯一性
        key_name = os.path.abspath(key_name)
        if self.vars_helper is not None and self.vars_helper.has_key(&#39;IMSGetDatasBase.time_list&#39;):
            IMSGetDatasBase.time_dict = self.vars_helper.get_value(&#39;IMSGetDatasBase.time_list&#39;)
        timeout = 12 # 12小时
        if GlobalShareArgs.get_args_value(&#39;debug&#39;):
            timeout = 24 # 24小时
        get_data_timeout = False
        if key_name not in IMSGetDatasBase.time_dict.keys() or (datetime.datetime.today() - IMSGetDatasBase.time_dict[key_name]).total_seconds()>(4*60*60):
            self.logger.info(&#39;超时%d小时，重新查数据：%s&#39;, timeout, key_name)
            # IMSGetDatasBase.time_list[key_name] = datetime.datetime.today()
            get_data_timeout = True
        else:
            self.logger.info(&#39;未超时%d小时，跳过查数据：%s&#39;, timeout, key_name)
        # if self.vars_helper is not None :
        #     self.vars_helper.set_value(&#39;IMSGetDatasBase.time_list&#39;, IMSGetDatasBase.time_list)
        IMSGetDatasBase.get_data_timeout_dict[key_name] = get_data_timeout
        return get_data_timeout
    def save_last_time(self, key_name):
        &#39;&#39;&#39;更新状态超时&#39;&#39;&#39;
        # 转静态路径，确保唯一性
        key_name = os.path.abspath(key_name)
        if IMSGetDatasBase.get_data_timeout_dict[key_name]:
            IMSGetDatasBase.time_dict[key_name] = datetime.datetime.today()
        if self.vars_helper is not None :
            IMSGetDatasBase.time_dict[key_name] = datetime.datetime.today()
            self.vars_helper.set_value(&#39;IMSGetDatasBase.time_list&#39;, IMSGetDatasBase.time_dict)
    def get_lock(self, key_name) -> threading.Lock:
        &#39;&#39;&#39;获取锁&#39;&#39;&#39;
        # 转静态路径，确保唯一性
        key_name = os.path.abspath(key_name)
        if key_name not in IMSGetDatasBase.lock_dict.keys():
            IMSGetDatasBase.lock_dict[key_name] = threading.Lock()
        return IMSGetDatasBase.lock_dict[key_name]
    def get_data_of_date(
        self,
        save_dir,
        sql,
        sort_columns:List[str],
        del_index_list=[-1], # 删除最后下标
        start_date = datetime.datetime(2017, 1, 1), # 开始时间
        offset = relativedelta(months=3), # 时间间隔
        date_format_fun = lambda d: &#39;%04d%02d01&#39; % (d.year, d.month), # 查询语句中替代时间参数的格式化
        filename_format_fun = lambda d: &#39;%04d%02d.csv&#39; % (d.year, d.month), # 查询语句中替代时间参数的格式化
        stop_date = &#39;20700101&#39;, # 超过时间则停止
        ):
        &#39;&#39;&#39;分时间增量读取数据&#39;&#39;&#39;
        # 创建文件夹
        if not os.path.exists(save_dir):
            os.makedirs(save_dir)
        else:
            #删除最后一个文件
            file_list = os.listdir(save_dir)
            if len(file_list)>0:
                file_list.sort()
                for del_index in del_index_list:
                    os.remove(os.path.join(save_dir,file_list[del_index]))
                    print(&#39;删除最后一个文件：&#39;, file_list[del_index])
        select_index = -1
        # start_date = datetime.datetime(2017, 1, 1)
        while True:
            end_date = start_date + offset
            start_date_str = date_format_fun(start_date)
            end_date_str = date_format_fun(end_date)
            self.logger.info(&#39;date: %s-%s&#39;, start_date_str, end_date_str)
            file_path = os.path.join(save_dir, filename_format_fun(start_date))
            # self.logger.info(&#39;file_path: %s&#39;, file_path)
            if not os.path.exists(file_path):
                data:pd.DataFrame = self.db_helper.get_data(sql % (start_date_str, end_date_str))
                if data is None:
                    break
                self.logger.info(&#39;data: %d&#39;, len(data))
                # self.logger.info(&#39;data: %d&#39;, data.columns)
                if len(data)>0:
                    select_index+=1
                    # 排序
                    data = data.sort_values(sort_columns)
                    data.to_csv(file_path)
                elif select_index!=-1:
                    break
                elif stop_date < start_date_str:
                    raise Exception("读取数据异常，时间超出最大值！")
            start_date = end_date
pass
class CustomIMSGetDatas(IMSGetDatasBase):
    def __init__(
        self,
        host=&#39;192.168.13.134&#39;,
        port=4000,
        database=&#39;test_ims&#39;,
        user=&#39;root&#39;,
        password=&#39;rootimmsadmin&#39;,
        save_dir=&#39;./hjx/data/export_ims_data&#39;,
        logger:logging.Logger=None
        ):
        self.save_dir = save_dir
        self.logger = logger
        super().__init__(
            host=host,
            port=port,
            database=database,
            user=user,
            password=password,
            save_dir=save_dir,
            logger=logger
            )
    def get_ims_w_amt_pro(self):
        &#39;&#39;&#39;年月系列占比数据&#39;&#39;&#39;
        file_path = os.path.join(self.save_dir,&#39;ims_w_amt_pro.csv&#39;)
        now_lock = self.get_lock(file_path)
        now_lock.acquire() # 加锁
        try:
            # 设置超时4小时才重新查数据
            # if not self.get_last_time(file_path):
            #     return
            sql = &#39;SELECT * FROM ims_w_amt_pro&#39;
            data:pd.DataFrame = self.db_helper.get_data(sql)
            data = data.rename(columns={
                &#39;serial_forecast_proportion&#39;: &#39;forecast_proportion&#39;,
            })
            data.to_csv(file_path)
            # # 更新超时时间
            # self.save_last_time(file_path)
        except Exception as ex:
            self.logger.exception(ex)
            raise ex # 往外抛出异常
        finally:
            now_lock.release() # 释放锁
pass
def get_datas(
    host=&#39;192.168.13.134&#39;,
    port=4000,
    database=&#39;test_ims&#39;,
    user=&#39;root&#39;,
    password=&#39;rootimmsadmin&#39;,
    save_dir=&#39;./hjx/data/export_ims_data&#39;,
    logger:logging.Logger=None
    ):
    ur_bi_get_datas = CustomIMSGetDatas(
        host=host,
        port=port,
        database=database,
        user=user,
        password=password,
        save_dir=save_dir,
        logger=logger
    )
    try:
        # 年月系列占比数据
        logger.info(&#39;正在查询年月系列占比数据...&#39;)
        ur_bi_get_datas.get_ims_w_amt_pro()
        logger.info(&#39;查询年月系列占比数据完成！&#39;)
    except Exception as ex:
        logger.exception(ex)
        raise ex # 往外抛出异常
    finally:
        ur_bi_get_datas.close()
pass
def getdata_export_ims(
    host=&#39;192.168.13.134&#39;,
    port=4000,
    database=&#39;test_ims&#39;,
    user=&#39;root&#39;,
    password=&#39;rootimmsadmin&#39;,
    save_dir=&#39;./hjx/data/export_ims_data&#39;,
    logger:logging.Logger=None
    ):
    get_datas(
        host=host,
        port=port,
        database=database,
        user=user,
        password=password,
        save_dir=save_dir,
        logger=logger
    )
pass

登录后复制

以上是如何使用Python读取Hive数据库？的详细内容。更多信息请关注PHP中文网其他相关文章！

本站声明

本文内容由网友自发贡献，版权归原作者所有，本站不承担相应法律责任。如您发现有涉嫌抄袭侵权的内容，请联系admin@php.cn