学习大数据DAY59 全量抽取和增量抽取实战

目录

需求流程：

需求分析与规范

作业

作业2

需求流程：

全量抽取 增量抽取 - DataX Kettle Sqoop ...
场景: 业务部门同事或者甲方的工作人员给我们的部门经理和你提出了新的需 求
流程: 联系 => 开会讨论 => 确认需求 => 落地
需求文档( 具体需要的东西)
原型文档 (报表的原型 纸笔/画图工具)
第一张汇总报表需要的指标 - 决策报表 汇总表 每次计算只有一天的记录 -
大 BOSS:
- 全部会员数 新增会员数
- 有效会员数 有效会员占比
- 流失会员数: 倒推一年含一年无消费记录的会员
- 净增有效会员数
- 会员消费级别分类人数 (A >=2000 B >=1000 < 2000 C >=500 <1000 D >=100
<500 E <100)
- 会员消费总额
- 会员消费总次数
- 会员消费单价
- 流失会员数
- 新增流失会员
- 60 天会员复购率
- 180 天会员复购率
- 365 天会员复购率
第二张报表用于市场营销 - 明细报表, 普通报表 - 市场部同事
- 筛选大于 30 笔的会员或者消费总金额大于 1500 的会员作为目标用户 用于电
话营销
- 字段: 姓名 手机号 消费总额 消费次数 城市 门店 付款偏好 (手机 刷卡
现金..) 关注的疾病
- 该会员最近 3 个月的月消费订单数和额度 m1_total m1_sale m2_total
m2_sale
第三张报表用于市场营销 - 2022.1-2023.12 每个月消费前 20 的会员名单
24X20=480 条 - 市场部经理
- T+1(月) yyyy-mm 月份
- 会员姓名 联系方式... 消费级别分类, 最近 30 天消费订单数和总额
- 该会员当月前 30 天消费, 60 天消费, 90 天消费 (困难点)
- 报表排序方式: 默认按消费总额倒序 / 按消费次数倒序
- 报表默认显示 2021 年 1 月份的数据, 可选 2 年内任何一个月的数据查看

需求分析与规范

经理整理出一个类似宽表文档的东西 - 方便后续的明细查询和指标计算
它决定了我们需要抽取哪些表
crm.user_base_info_his 客户信息表
erp.u_memcard_reg 会员卡信息表
erp.u_sale_m 订单表
1900W 数据
erp.u_sale_pay 订单支付表
1200W erp.c_memcard_class_group 会员分组表
erp.u_memcard_reg_c 疾病关注表
his.chronic_patient_info_new 检测表
erp.c_org_busi 门店表
# 额外的从文件处理的码值表
erp.c_code_value
# 7 个全量
系统名前缀 _ 表名 _(full|inc)
crm.user_base_info_his
全量 => ods_lijinquan.crm_user_base_info_his_full
erp.u_memcard_reg
全量=> ods_lijinquan.erp_u_memcard_reg_full
erp.c_memcard_class_group
全量 => ods_lijinquan.erp_c_memcard_class_group_full
erp.u_memcard_reg_c
全量=>ods_lijinquan.erp_u_memcard_reg_c_full
his.chronic_patient_info_new
全量 => ods_lijinquan.his_chronic_patient_info_new_full
erp.c_org_busi 全量 => ods_lijinquan.erp_c_org_busi_full
erp.c_code_value
全量文件处理
=>ods_lijinquan.c_code_value_full
# 增量
erp.u_sale_m 先 做 全 量 ( 一 次 性 ) 再 做 增 量 ( 每 天 执 行 )
=> ods_lijinquan.erp_u_sale_m_inc
erp.u_sale_pay 同上 增量 => ods_lijinquan.erp_u_sale_pay_inc

作业

完成 7 张全量表的抽取, 部署到调度平台 7 个调度任务
所有表最后都要跟源表的总数进行对比 需要一致
全量表处理
升级辅助脚本
- 自动读取表的字段信息, 自动生成 datax json 文件 和 Hive
建表文件
full.py: 
#!/bin/python3
import pymysql
import sys
# 自动写 datax 的 json 文件
if len(sys.argv)!=3:
print("使用方法为:python3 full.py 数据库名 表名")
sys.exit()
sys_name=sys.argv[1]
table_name=sys.argv[2]
# datax_json=f"{sys_name}.{table_name}_full.json"
db=pymysql.connect(
host='zhiyun.pub',
port=23306,
user='zhiyun',
password='zhiyun',
database='information_schema'
)
cursor=db.cursor()
cursor.execute(f"select column_name,data_type from
information_schema.columns where table_schema='{sys_name}' and
table_name='{table_name}'")
data=cursor.fetchall()
fileds=[]for field in data:
field_name = field[0]
field_type = field[1]
#转换成 hive 类型
field_hive_type="string"
if field_type=="int" or field_type=="tinyint" or
field_type=="bigint":
field_hive_type="int"
if field_type=="float" or field_type=="double":
field_hive_type="float"
fileds.append([field_name,field_hive_type])
db.close()
print("=============== 配置 datax ===============")
file_path=f"/zhiyun/shihaihong/jobs/{sys_name}_{table_name}_fu
ll.json"
template_path="/zhiyun/shihaihong/jobs/template.json"
with open(template_path,"r",encoding="utf-8") as f:
template_content=f.read()
new_content=template_content.replace("#sys_name#",sys_name)
new_content=new_content.replace("#table_name#",table_name)
#列的替换
lines=[]
for filed in fileds:
line='
{"name":"'+filed[0]+'
","type":"'+filed[1]+'"},'
lines.append(line)
columns="\n".join(lines)
columns=columns.strip(",")
new_content=new_content.replace("\"#columns#\"",columns)
#写入到新的配置
with open(file_path,"w",encoding="utf-8") as ff:
ff.write(new_content)
ff.close()
f.close()
print("datax 文件配置成功")
print("=============== 配置 hive ===============")file_path=f"/zhiyun/shihaihong/sql/{sys_name}_{table_name}_ful
l.sql"
template_path="/zhiyun/shihaihong/sql/template.sql"
with open(template_path,"r",encoding="utf-8") as f:
template_content=f.read()
new_content=template_content.replace("#sys_name#",sys_name)
new_content=new_content.replace("#table_name#",table_name)
#列的替换
lines=[]
for filed in fileds:
line=f"
{filed[0]} {filed[1]},"
lines.append(line)
columns="\n".join(lines)
columns=columns.strip(",")
new_content=new_content.replace("#columns#",columns)
#写入到新的配置
with open(file_path,"w",encoding="utf-8") as ff:
ff.write(new_content)
ff.close()
print("hive 建表文件生成成功")

json 模板：

template.json:

{
"job": {
"content": [
{
"reader": {
"name": "mysqlreader",
"parameter": {
"column": ["*"],
"connection": [
{
"jdbcUrl": ["jdbc:mysql://zhiyun.pub:233
06/crm?useSSL=false"
],
"table": [
"#table_name#"
]
}
],
"password": "zhiyun",
"username": "zhiyun"
}
},
"writer": {
"name": "hdfswriter",
"parameter": {
"column": [
"#column#"
],
"defaultFS": "hdfs://cdh02:8020",
"fieldDelimiter": "\t",
"fileName":
"#sys_name#_#table_name#_full.data",
"fileType": "orc",
"path":
"/zhiyun/shihaihong/ods/#sys_name#_#table_name#_full",
"writeMode": "truncate"
}
}
}
],
"setting": {
"speed": {
"channel": "3"
}
}
}
}
sql 模板文件： 
create database if not exists ods_shihaihong location
"/zhiyun/shihaihong/ods";-- 增量表
create external table if not exists
ods_shihaihong.#table_name#_full(
#columns#
)
row format delimited fields terminated by "\t"
lines terminated by "\n"
stored as orc
location "/zhiyun/shihaihong/ods/#sys_name#_#table_name#_full";
测试：
python3 python/full.py crm user_base_info_his
生成 json 和 sql 文件：
根据 json 和 sql 文件写出获得数据的 shell 脚本：
crm_user_base_info_his_full.sh ： 
#!/bin/bash
echo "生成全量配置文件"mkdir -p /zhiyun/shihaihong/jobs
echo '{
"job": {
"content": [
{
"reader": {
"name": "mysqlreader",
"parameter": {
"column": ["*"],
"connection": [
{
"jdbcUrl": [
"jdbc:mysql://zhiyun.pub:233
06/crm?useSSL=false"
],
"table": [
"user_base_info_his"
]
}
],
"password": "zhiyun",
"username": "zhiyun"
}
},
"writer": {
"name": "hdfswriter",
"parameter": {
"column": [
{"name":
"id","type":"int"},
{"name":"user_id","type":"string"},
{"name":"user_type","type":"string"}
,
{"name":"source","type":"string"},
{"name":"erp_code","type":"string"}
,
{"name":"active_time","type":"strin
g"},
{"name":"name","type":"string"},
{"name":"sex","type":"string"},
{"name":"education","type":"string"}
,{"name":"job","type":"string"},
{"name":"email","type":"string"},
{"name":"wechat","type":"string"},
{"name":"webo","type":"string"},
{"name":"birthday","type":"string"}
,
{"name":"age","type":"int"},
{"name":"id_card_no","type":"string
"},
{"name":"social_insurance_no","type
":"string"},
{"name":"address","type":"string"},
{"name":"last_subscribe_time","type
":"int"}
],
"defaultFS": "hdfs://cdh02:8020",
"fieldDelimiter": "\t",
"fileName":
"crm_user_base_info_his_full.data",
"fileType": "orc",
"path":
"/zhiyun/shihaihong/ods/crm_user_base_info_his_full",
"writeMode": "truncate"
}
}
}
],
"setting": {
"speed": {
"channel": "3"
}
}
}
}' > /zhiyun/shihaihong/jobs/crm_user_base_info_his_full.json
echo "开始抽取"
hadoop fs -mkdir -p
/zhiyun/shihaihong/ods/crm_user_base_info_his_full
python /opt/datax/bin/datax.py
/zhiyun/shihaihong/jobs/crm_user_base_info_his_full.json
echo "hive 建表"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e 'create database if not exists ods_shihaihong location
"/zhiyun/shihaihong/ods";
-- 增量表
create external table if not exists
ods_shihaihong.user_base_info_his_full(
id int,
user_id string,
user_type string,
source string,
erp_code string,
active_time string,
name string,
sex string,
education string,
job string,
email string,
wechat string,
webo string,
birthday string,
age int,
id_card_no string,
social_insurance_no string,
address string,
last_subscribe_time int
)
row format delimited fields terminated by "\t"
lines terminated by "\n"
stored as orc
location "/zhiyun/shihaihong/ods/crm_user_base_info_his_full";
'
# echo "加载数据"
# beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "
# load data inpath
\"/zhiyun/shihaihong/tmp/crm_user_base_info_his_full/*\"
overwrite into table ods_shihaihong.crm_user_base_info_his_full
partition(createtime='$day');
# "
echo "验证数据"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "
select count(1) from ods_shihaihong.user_base_info_his_full;
"echo "抽取完成"
运行测试：
在本人数据库检查：
在生产调度中心设置任务：
在 GLUE IDE 插入 sh 文件后，执行一次：
运行成功。
其它六张表用同样的方式操作即可。
其它六张表的 shell 脚本： erp_u_memcard_reg_full.sh： 
#!/bin/bash
echo "生成全量配置文件"
mkdir -p /zhiyun/shihaihong/jobs
echo '{
"job": {
"content": [
{
"reader": {
"name": "mysqlreader",
"parameter": {
"column": ["*"],
"connection": [
{
"jdbcUrl": [
"jdbc:mysql://zhiyun.pub:233
06/erp?useSSL=false"
],
"table": [
"u_memcard_reg"
]
}
],
"password": "zhiyun",
"username": "zhiyun"
}
},
"writer": {
"name": "hdfswriter",
"parameter": {
"column": [
{"name":
"id","type":"int"},
{"name":"memcardno","type":"string"}
,
{"name":"busno","type":"string"},
{"name":"introducer","type":"string
"},
{"name":"cardtype","type":"int"},
{"name":"cardlevel","type":"int"},{"name":"cardpass","type":"string"}
,
{"name":"cardstatus","type":"int"},
{"name":"saleamount","type":"string
"},
{"name":"realamount","type":"string
"},
{"name":"puramount","type":"string"}
,
{"name":"integral","type":"string"}
,
{"name":"integrala","type":"string"}
,
{"name":"integralflag","type":"int"}
,
{"name":"cardholder","type":"string
"},
{"name":"cardaddress","type":"strin
g"},
{"name":"sex","type":"string"},
{"name":"tel","type":"string"},
{"name":"handset","type":"string"},
{"name":"fax","type":"string"},
{"name":"createuser","type":"string
"},
{"name":"createtime","type":"string
"},
{"name":"tstatus","type":"int"},
{"name":"notes","type":"string"},
{"name":"stamp","type":"string"},
{"name":"idcard","type":"string"},
{"name":"birthday","type":"string"}
,
{"name":"allowintegral","type":"int
"},
{"name":"apptype","type":"string"},
{"name":"applytime","type":"string"}
,
{"name":"invalidate","type":"string
"},
{"name":"lastdate","type":"string"}
,
{"name":"bak1","type":"string"},{"name":"scrm_userid","type":"strin
g"}
],
"defaultFS": "hdfs://cdh02:8020",
"fieldDelimiter": "\t",
"fileName":
"erp_u_memcard_reg_full.data",
"fileType": "orc",
"path":
"/zhiyun/shihaihong/ods/erp_u_memcard_reg_full",
"writeMode": "truncate"
}
}
}
],
"setting": {
"speed": {
"channel": "3"
}
}
}
}' > /zhiyun/shihaihong/jobs/erp_u_memcard_reg_full.json
echo "开始抽取"
hadoop fs -mkdir -p /zhiyun/shihaihong/ods/erp_u_memcard_reg_full
python /opt/datax/bin/datax.py
/zhiyun/shihaihong/jobs/erp_u_memcard_reg_full.json
echo "hive 建表"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e '
create database if not exists ods_shihaihong location
"/zhiyun/shihaihong/ods";
-- 增量表
create external table if not exists
ods_shihaihong.u_memcard_reg_full(
id int,
memcardno string,
busno string,
introducer string,
cardtype int,
cardlevel int,
cardpass string,
cardstatus int,saleamount string,
realamount string,
puramount string,
integral string,
integrala string,
integralflag int,
cardholder string,
cardaddress string,
sex string,
tel string,
handset string,
fax string,
createuser string,
createtime string,
tstatus int,
notes string,
stamp string,
idcard string,
birthday string,
allowintegral int,
apptype string,
applytime string,
invalidate string,
lastdate string,
bak1 string,
scrm_userid string
)
row format delimited fields terminated by "\t"
lines terminated by "\n"
stored as orc
location "/zhiyun/shihaihong/ods/erp_u_memcard_reg_full";
'
# echo "加载数据"
# beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "
# load data inpath
\"/zhiyun/shihaihong/tmp/erp_u_memcard_reg_full/*\" overwrite
into table ods_shihaihong.erp_u_memcard_reg_full
partition(createtime='$day');
# "
echo "验证数据"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "select count(1) from ods_shihaihong.u_memcard_reg_full;
"
echo "抽取完成"
erp.c_memcard_class_group: 
#!/bin/bash
echo "生成全量配置文件"
mkdir -p /zhiyun/shihaihong/jobs
echo '{
"job": {
"content": [
{
"reader": {
"name": "mysqlreader",
"parameter": {
"column": ["*"],
"connection": [
{
"jdbcUrl": [
"jdbc:mysql://zhiyun.pub:233
06/erp?useSSL=false"
],
"table": [
"c_memcard_class_group"
]
}
],
"password": "zhiyun",
"username": "zhiyun"
}
},
"writer": {
"name": "hdfswriter",
"parameter": {
"column": [
{"name":
"createtime","type":"string"},{"name":"createuser","type":"string
"},
{"name":"groupid","type":"int"},
{"name":"groupname","type":"string"}
,
{"name":"notes","type":"string"},
{"name":"stamp","type":"int"}
],
"defaultFS": "hdfs://cdh02:8020",
"fieldDelimiter": "\t",
"fileName":
"erp_c_memcard_class_group_full.data",
"fileType": "orc",
"path":
"/zhiyun/shihaihong/ods/erp_c_memcard_class_group_full",
"writeMode": "truncate"
}
}
}
],
"setting": {
"speed": {
"channel": "3"
}
}
}
}' > /zhiyun/shihaihong/jobs/erp_c_memcard_class_group_full.json
echo "开始抽取"
hadoop fs -mkdir -p
/zhiyun/shihaihong/ods/erp_c_memcard_class_group_full
python /opt/datax/bin/datax.py
/zhiyun/shihaihong/jobs/erp_c_memcard_class_group_full.json
echo "hive 建表"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e '
create database if not exists ods_shihaihong location
"/zhiyun/shihaihong/ods";
-- 增量表
create external table if not exists
ods_shihaihong.c_memcard_class_group_full(
createtime string,
createuser string,groupid int,
groupname string,
notes string,
stamp int
)
row format delimited fields terminated by "\t"
lines terminated by "\n"
stored as orc
location
"/zhiyun/shihaihong/ods/erp_c_memcard_class_group_full";
'
# echo "加载数据"
# beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "
# load data inpath
\"/zhiyun/shihaihong/tmp/erp_c_memcard_class_group_full/*\"
overwrite into table
ods_shihaihong.erp_c_memcard_class_group_full
partition(createtime='$day');
# "
echo "验证数据"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "
select count(1) from ods_shihaihong.c_memcard_class_group_full;
"
echo "抽取完成"
erp.u_memcard_reg_c:
#!/bin/bash
echo "生成全量配置文件"
mkdir -p /zhiyun/shihaihong/jobs
echo '{
"job": {
"content": [
{
"reader": {
"name": "mysqlreader","parameter": {
"column": ["*"],
"connection": [
{
"jdbcUrl": [
"jdbc:mysql://zhiyun.pub:233
06/erp?useSSL=false"
],
"table": [
"u_memcard_reg_c"
]
}
],
"password": "zhiyun",
"username": "zhiyun"
}
},
"writer": {
"name": "hdfswriter",
"parameter": {
"column": [
{"name":
"id","type":"int"},
{"name":"memcardno","type":"string"}
,
{"name":"sickness","type":"string"}
,
{"name":"status","type":"string"}
],
"defaultFS": "hdfs://cdh02:8020",
"fieldDelimiter": "\t",
"fileName":
"erp_u_memcard_reg_c_full.data",
"fileType": "orc",
"path":
"/zhiyun/shihaihong/ods/erp_u_memcard_reg_c_full",
"writeMode": "truncate"
}
}
}
],
"setting": {
"speed": {"channel": "3"
}
}
}
}' > /zhiyun/shihaihong/jobs/erp_u_memcard_reg_c_full.json
echo "开始抽取"
hadoop fs -mkdir -p
/zhiyun/shihaihong/ods/erp_u_memcard_reg_c_full
python /opt/datax/bin/datax.py
/zhiyun/shihaihong/jobs/erp_u_memcard_reg_c_full.json
echo "hive 建表"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e '
create database if not exists ods_shihaihong location
"/zhiyun/shihaihong/ods";
-- 增量表
create external table if not exists
ods_shihaihong.u_memcard_reg_c_full(
id int,
memcardno string,
sickness string,
status string
)
row format delimited fields terminated by "\t"
lines terminated by "\n"
stored as orc
location "/zhiyun/shihaihong/ods/erp_u_memcard_reg_c_full";
'
# echo "加载数据"
# beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "
# load data inpath
\"/zhiyun/shihaihong/tmp/erp_u_memcard_reg_c_full/*\" overwrite
into table ods_shihaihong.erp_u_memcard_reg_c_full
partition(createtime='$day');
# "
echo "验证数据"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "
select count(1) from ods_shihaihong.u_memcard_reg_c_full;
"echo "抽取完成"
his.chronic_patient_info_new: 
#!/bin/bash
echo "生成全量配置文件"
mkdir -p /zhiyun/shihaihong/jobs
echo '{
"job": {
"content": [
{
"reader": {
"name": "mysqlreader",
"parameter": {
"column": ["*"],
"connection": [
{
"jdbcUrl": [
"jdbc:mysql://zhiyun.pub:233
06/his?useSSL=false"
],
"table": [
"chronic_patient_info_new"
]
}
],
"password": "zhiyun",
"username": "zhiyun"
}
},
"writer": {
"name": "hdfswriter",
"parameter": {
"column": [
{"name":
"id","type":"int"},{"name":"member_id","type":"string"}
,
{"name":"erp_code","type":"string"}
,
{"name":"extend","type":"string"},
{"name":"detect_time","type":"strin
g"},
{"name":"bec_chr_mbr_date","type":"
string"}
],
"defaultFS": "hdfs://cdh02:8020",
"fieldDelimiter": "\t",
"fileName":
"his_chronic_patient_info_new_full.data",
"fileType": "orc",
"path":
"/zhiyun/shihaihong/ods/his_chronic_patient_info_new_full",
"writeMode": "truncate"
}
}
}
],
"setting": {
"speed": {
"channel": "3"
}
}
}
}' >
/zhiyun/shihaihong/jobs/his_chronic_patient_info_new_full.json
echo "开始抽取"
hadoop fs -mkdir -p
/zhiyun/shihaihong/ods/his_chronic_patient_info_new_full
python /opt/datax/bin/datax.py
/zhiyun/shihaihong/jobs/his_chronic_patient_info_new_full.json
echo "hive 建表"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e '
create database if not exists ods_shihaihong location
"/zhiyun/shihaihong/ods";
-- 增量表create external table if not exists
ods_shihaihong.chronic_patient_info_new_full(
id int,
member_id string,
erp_code string,
extend string,
detect_time string,
bec_chr_mbr_date string
)
row format delimited fields terminated by "\t"
lines terminated by "\n"
stored as orc
location
"/zhiyun/shihaihong/ods/his_chronic_patient_info_new_full";
'
# echo "加载数据"
# beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "
# load data inpath
\"/zhiyun/shihaihong/tmp/his_chronic_patient_info_new_full/*\"
overwrite into table
ods_shihaihong.his_chronic_patient_info_new_full
partition(createtime='$day');
# "
echo "验证数据"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "
select count(1) from
ods_shihaihong.chronic_patient_info_new_full;
"
echo "抽取完成"
erp.c_org_busi : 
#!/bin/bash
echo "生成全量配置文件"
mkdir -p /zhiyun/shihaihong/jobsecho '{
"job": {
"content": [
{
"reader": {
"name": "mysqlreader",
"parameter": {
"column": ["*"],
"connection": [
{
"jdbcUrl": [
"jdbc:mysql://zhiyun.pub:233
06/erp?useSSL=false"
],
"table": [
"c_org_busi"
]
}
],
"password": "zhiyun",
"username": "zhiyun"
}
},
"writer": {
"name": "hdfswriter",
"parameter": {
"column": [
{"name":
"id","type":"int"},
{"name":"busno","type":"string"},
{"name":"orgname","type":"string"},
{"name":"orgsubno","type":"string"}
,
{"name":"orgtype","type":"string"},
{"name":"salegroup","type":"string"}
,
{"name":"org_tran_code","type":"str
ing"},
{"name":"accno","type":"string"},
{"name":"sendtype","type":"string"}
,
{"name":"sendday","type":"string"},{"name":"maxday","type":"string"},
{"name":"minday","type":"string"},
{"name":"notes","type":"string"},
{"name":"stamp","type":"string"},
{"name":"status","type":"string"},
{"name":"customid","type":"string"}
,
{"name":"whl_vendorno","type":"stri
ng"},
{"name":"whlgroup","type":"string"}
,
{"name":"rate","type":"string"},
{"name":"creditamt","type":"string"}
,
{"name":"creditday","type":"string"}
,
{"name":"peoples","type":"string"},
{"name":"area","type":"string"},
{"name":"abc","type":"string"},
{"name":"address","type":"string"},
{"name":"tel","type":"string"},
{"name":"principal","type":"string"}
,
{"name":"identity_card","type":"str
ing"},
{"name":"mobil","type":"string"},
{"name":"corporation","type":"strin
g"},
{"name":"saler","type":"string"},
{"name":"createtime","type":"string
"},
{"name":"bank","type":"string"},
{"name":"bankno","type":"string"},
{"name":"bak1","type":"string"},
{"name":"bak2","type":"string"},
{"name":"a_bak1","type":"string"},
{"name":"aa_bak1","type":"string"},
{"name":"b_bak1","type":"string"},
{"name":"bb_bak1","type":"string"},
{"name":"y_bak1","type":"string"},
{"name":"t_bak1","type":"string"},
{"name":"ym_bak1","type":"string"},
{"name":"tm_bak1","type":"string"},{"name":"supervise_code","type":"st
ring"},
{"name":"monthrent","type":"string"}
,
{"name":"wms_warehid","type":"strin
g"},
{"name":"settlement_cycle","type":"
string"},
{"name":"apply_cycle","type":"strin
g"},
{"name":"applydate","type":"string"}
,
{"name":"accounttype","type":"strin
g"},
{"name":"applydate_last","type":"st
ring"},
{"name":"paymode","type":"string"},
{"name":"yaolian_flag","type":"stri
ng"},
{"name":"org_longitude","type":"str
ing"},
{"name":"org_latitude","type":"stri
ng"},
{"name":"org_province","type":"stri
ng"},
{"name":"org_city","type":"string"}
,
{"name":"org_area","type":"string"}
,
{"name":"business_time","type":"str
ing"},
{"name":"yaolian_group","type":"str
ing"},
{"name":"pacard_storeid","type":"st
ring"},
{"name":"opening_time","type":"stri
ng"},
{"name":"ret_ent_id","type":"string
"},
{"name":"ent_id","type":"string"}
],
"defaultFS": "hdfs://cdh02:8020",
"fieldDelimiter": "\t","fileName": "erp_c_org_busi_full.data",
"fileType": "orc",
"path":
"/zhiyun/shihaihong/ods/erp_c_org_busi_full",
"writeMode": "truncate"
}
}
}
],
"setting": {
"speed": {
"channel": "3"
}
}
}
}' > /zhiyun/shihaihong/jobs/erp_c_org_busi_full.json
echo "开始抽取"
hadoop fs -mkdir -p /zhiyun/shihaihong/ods/erp_c_org_busi_full
python /opt/datax/bin/datax.py
/zhiyun/shihaihong/jobs/erp_c_org_busi_full.json
echo "hive 建表"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e '
create database if not exists ods_shihaihong location
"/zhiyun/shihaihong/ods";
-- 增量表
create external table if not exists
ods_shihaihong.c_org_busi_full(
id int,
busno string,
orgname string,
orgsubno string,
orgtype string,
salegroup string,
org_tran_code string,
accno string,
sendtype string,
sendday string,
maxday string,
minday string,
notes string,
stamp string,status string,
customid string,
whl_vendorno string,
whlgroup string,
rate string,
creditamt string,
creditday string,
peoples string,
area string,
abc string,
address string,
tel string,
principal string,
identity_card string,
mobil string,
corporation string,
saler string,
createtime string,
bank string,
bankno string,
bak1 string,
bak2 string,
a_bak1 string,
aa_bak1 string,
b_bak1 string,
bb_bak1 string,
y_bak1 string,
t_bak1 string,
ym_bak1 string,
tm_bak1 string,
supervise_code string,
monthrent string,
wms_warehid string,
settlement_cycle string,
apply_cycle string,
applydate string,
accounttype string,
applydate_last string,
paymode string,
yaolian_flag string,
org_longitude string,
org_latitude string,
org_province string,org_city string,
org_area string,
business_time string,
yaolian_group string,
pacard_storeid string,
opening_time string,
ret_ent_id string,
ent_id string
)
row format delimited fields terminated by "\t"
lines terminated by "\n"
stored as orc
location "/zhiyun/shihaihong/ods/erp_c_org_busi_full";
'
# echo "加载数据"
# beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "
# load data inpath
\"/zhiyun/shihaihong/tmp/erp_c_org_busi_full/*\" overwrite into
table ods_shihaihong.erp_c_org_busi_full
partition(createtime='$day');
# "
echo "验证数据"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "
select count(1) from ods_shihaihong.c_org_busi_full;
"
echo "抽取完成"
erp.c_code_value: 
#!/bin/bash
echo "生成全量配置文件"
mkdir -p /zhiyun/shihaihong/jobs
echo '{
"job": {"content": [
{
"reader": {
"name": "mysqlreader",
"parameter": {
"column": ["*"],
"connection": [
{
"jdbcUrl": [
"jdbc:mysql://zhiyun.pub:233
06/erp?useSSL=false"
],
"table": [
"c_code_value"
]
}
],
"password": "zhiyun",
"username": "zhiyun"
}
},
"writer": {
"name": "hdfswriter",
"parameter": {
"column": [
{"name":
"id","type":"int"},
{"name":"cat_name","type":"string"}
,
{"name":"cat_code","type":"string"}
,
{"name":"val_name","type":"string"}
,
{"name":"var_desc","type":"string"}
],
"defaultFS": "hdfs://cdh02:8020",
"fieldDelimiter": "\t",
"fileName":
"erp_c_code_value_full.data",
"fileType": "orc",
"path":
"/zhiyun/shihaihong/ods/erp_c_code_value_full",
"writeMode": "truncate"}
}
}
],
"setting": {
"speed": {
"channel": "3"
}
}
}
}' > /zhiyun/shihaihong/jobs/erp_c_code_value_full.json
echo "开始抽取"
hadoop fs -mkdir -p /zhiyun/shihaihong/ods/erp_c_code_value_full
python /opt/datax/bin/datax.py
/zhiyun/shihaihong/jobs/erp_c_code_value_full.json
echo "hive 建表"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e '
create database if not exists ods_shihaihong location
"/zhiyun/shihaihong/ods";
-- 增量表
create external table if not exists
ods_shihaihong.c_code_value_full(
id int,
cat_name string,
cat_code string,
val_name string,
var_desc string
)
row format delimited fields terminated by "\t"
lines terminated by "\n"
stored as orc
location "/zhiyun/shihaihong/ods/erp_c_code_value_full";
'
# echo "加载数据"
# beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "
# load data inpath
\"/zhiyun/shihaihong/tmp/erp_c_code_value_full/*\" overwrite
into table ods_shihaihong.erp_c_code_value_full
partition(createtime='$day');
# "echo "验证数据"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "
select count(1) from ods_shihaihong.c_code_value_full;
"
echo "抽取完成"
任务调度：
erp.u_memcard_reg:
erp.c_memcard_class_group:
erp.u_memcard_reg_c:
his.chronic_patient_info_new:
erp.c_org_busi :
erp.c_code_value :
将码值表文件上传到 data 中，用 python 写一个数据清洗的脚本： 
#!/bin/python3
import os
import pandas as pd
from openpyxl import load_workbook
ss=''
lst=[]#使用 pandas 的 read_excel 函数读取指定路径的 Excel 文件。
sheet_name=None 表示读取文件中的所有工作表，而 header=2 表示数据的表
头位于第 3 行（索引从 0 开始）
dfs = pd.read_excel('/zhiyun/shihaihong/data/12.码值
表.xlsx',sheet_name=None,header=2)
dir=list(dfs.keys())
#获取 xlsx 文件数据
for i in range(len(dir)):
if i>1:
#获取 A2 行数据
wb = load_workbook(filename='/zhiyun/shihaihong/data/12.
码值表.xlsx')
str_head = wb[dir[i]]['A2'].value
data=dfs[dir[i]]
#获取其它行数据
lst1=[]
for i in data.columns:
for j in range(len(data)):
if data[i][j] != 'NaN':
lst1.append(str(data[i][j]))
n=int(len(lst1)/2)
for i in range(n):
ss=f"{str_head.split('-')[0]}|{str_head.split('-')[
1]}|{lst1[i]}|{lst1[i+n]}"
lst.append(ss)
print("写入数据到 data")
template_path = "/zhiyun/shihaihong/data/code_value.txt"
with open(template_path,"w",encoding="utf-8") as f:
content="\n".join(lst)
f.write(content)
f.close
print("上传 data 文件 到 hdfs")
os.system(f"hdfs dfs -mkdir -p /zhiyun/shihaihong/filetxt/")
os.system(f"hdfs dfs -put {template_path}
/zhiyun/shihaihong/filetxt/")
#!/bin/bash
# 作用: 完成从编写配置文件到验证数据的整个过程
# 需要在任何节点都可以执行
# 创建本人文件夹
mkdir -p /zhiyun/shihaihong/data /zhiyun/shihaihong/jobs
/zhiyun/shihaihong/python /zhiyun/shihaihong/shell
/zhiyun/shihaihong/sql
echo "hive 建表"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e'
create database if not exists ods_shihaihong location
"/zhiyun/shihaihong/ods";
create external table if not exists
ods_shihaihong.c_code_value_full(
cat_name string,
cat_code string,
val_name string,
var_desc string
)
row format delimited fields terminated by "|"
lines terminated by "\n"
stored as textfile
location "/zhiyun/shihaihong/filetxt";'
echo "hive 建表完成"
echo "验证数据"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "
select count(1) from ods_shihaihong.c_code_value_full;
"
echo "验证完成"

执行后，用 shell 脚本抽取
 

任务调度：

作业2

完成 2 张增量表的处理 历史数据调度任务 + 增量调度任务 4 个调度任务
所有表最后都要跟源表的总数进行对比 需要一致
抽取的增量表：
erp.u_sale_m 先 做 全 量 ( 一 次 性 ) 再 做 增 量 ( 每 天 执 行 ) =>
ods_lijinquan.erp_u_sale_m_inc
erp.u_sale_pay 同上 增量 => ods_lijinquan.erp_u_sale_pay_inc
首次抽取为全量抽取：
erp_u_sale_m_full.sh: 
#!/bin/bash
echo "生成全量配置文件"
mkdir -p /zhiyun/shihaihong/jobs
echo '{
"job": {
"content": [
{
"reader": {
"name": "mysqlreader",
"parameter": {
"column": ["*"],
"connection": [
{
"jdbcUrl": [
"jdbc:mysql://zhiyun.pub:233
06/erp?useSSL=false"
],
"table": [
"u_sale_m"]
}
],
"password": "zhiyun",
"username": "zhiyun"
}
},
"writer": {
"name": "hdfswriter",
"parameter": {
"column": [
{"name":"id","type":"int"},
{"name":"saleno","type":"string"},
{"name":"busno","type":"string"},
{"name":"posno","type":"string"},
{"name":"extno","type":"string"},
{"name":"extsource","type":"string"}
,
{"name":"o2o_trade_from","type":"st
ring"},
{"name":"channel","type":"int"},
{"name":"starttime","type":"string"}
,
{"name":"finaltime","type":"string"}
,
{"name":"payee","type":"string"},
{"name":"discounter","type":"string
"},
{"name":"crediter","type":"string"}
,
{"name":"returner","type":"string"}
,
{"name":"warranter1","type":"string
"},
{"name":"warranter2","type":"string
"},
{"name":"stdsum","type":"string"},
{"name":"netsum","type":"string"},
{"name":"loss","type":"string"},
{"name":"discount","type":"float"},
{"name":"member","type":"string"},
{"name":"precash","type":"string"},
{"name":"stamp","type":"string"},{"name":"shiftid","type":"string"},
{"name":"shiftdate","type":"string"}
,
{"name":"yb_saleno","type":"string"
}
],
"defaultFS": "hdfs://cdh02:8020",
"fieldDelimiter": "\t",
"fileName": "erp_u_sale_m_full.data",
"fileType": "orc",
"path":
"/zhiyun/shihaihong/ods/erp_u_sale_m_full",
"writeMode": "truncate"
}
}
}
],
"setting": {
"speed": {
"channel": "3"
}
}
}
}' > /zhiyun/shihaihong/jobs/erp_u_sale_m_full.json
echo "开始抽取"
hadoop fs -mkdir -p /zhiyun/shihaihong/ods/erp_u_sale_m_full
python /opt/datax/bin/datax.py
/zhiyun/shihaihong/jobs/erp_u_sale_m_full.json
echo "hive 建表"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e '
create database if not exists ods_shihaihong location
"/zhiyun/shihaihong/ods";
-- 增量表
create external table if not exists ods_shihaihong.u_sale_m_full(
id int,
saleno string,
busno string,
posno string,
extno string,
extsource string,
o2o_trade_from string,channel int,
starttime string,
finaltime string,
payee string,
discounter string,
crediter string,
returner string,
warranter1 string,
warranter2 string,
stdsum string,
netsum string,
loss string,
discount float,
member string,
precash string,
stamp string,
shiftid string,
shiftdate string,
yb_saleno string
)
row format delimited fields terminated by "\t"
lines terminated by "\n"
stored as orc
location "/zhiyun/shihaihong/ods/erp_u_sale_m_full";
'
# echo "加载数据"
# beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "
# load data inpath \"/zhiyun/shihaihong/tmp/erp_u_sale_m_full/*\"
overwrite into table ods_shihaihong.erp_u_sale_m_full
partition(createtime='$day');
# "
echo "验证数据"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "
select count(1) from ods_shihaihong.u_sale_m_full;
"
echo "抽取完成"

写入任务调度平台：

将 sh 文件内容复制入 GLUE IDE 中，执行一次：
后续用增量抽取： 
#!/bin/bash
day=$(date -d "yesterday" +%Y-%m-%d)
if [ $1 != "" ]; thenday=$(date -d "$1 -1 day" +%Y-%m-%d);
fi;
echo "抽取的日期为 $day"
echo "生成增量配置文件"
mkdir -p /zhiyun/shihaihong/jobs
echo '
{
"job": {
"content": [
{
"reader": {
"name": "mysqlreader",
"parameter": {
"connection": [
{
"jdbcUrl": [
"jdbc:mysql://zhiyun.pub:233
06/erp?useSSL=false"
],
"querySql": [
"select * from u_sale_m where
stamp between '\'"$day" 00:00:00\'' and '\'"$day" 23:59:59\'' and
id>0"
]
}
],
"password": "zhiyun",
"username": "zhiyun"
}
},
"writer": {
"name": "hdfswriter",
"parameter": {
"column": [
{"name":"id","type":"int"},
{"name":"saleno","type":"string"},
{"name":"busno","type":"string"},
{"name":"posno","type":"string"},
{"name":"extno","type":"string"},{"name":"extsource","type":"string"}
,
{"name":"o2o_trade_from","type":"str
ing"},
{"name":"channel","type":"int"},
{"name":"starttime","type":"string"}
,
{"name":"finaltime","type":"string"}
,
{"name":"payee","type":"string"},
{"name":"discounter","type":"string"}
,
{"name":"crediter","type":"string"},
{"name":"returner","type":"string"},
{"name":"warranter1","type":"string"}
,
{"name":"warranter2","type":"string"}
,
{"name":"stdsum","type":"string"},
{"name":"netsum","type":"string"},
{"name":"loss","type":"string"},
{"name":"discount","type":"float"},
{"name":"member","type":"string"},
{"name":"precash","type":"string"},
{"name":"stamp","type":"string"},
{"name":"shiftid","type":"string"},
{"name":"shiftdate","type":"string"}
,
{"name":"yb_saleno","type":"string"}
],
"defaultFS": "hdfs://cdh02:8020",
"fieldDelimiter": "\t",
"fileName": "erp_u_sale_m_inc.data",
"fileType": "orc",
"path":
"/zhiyun/shihaihong/tmp/erp_u_sale_m_inc",
"writeMode": "truncate"
}
}
}
],
"setting": {
"speed": {"channel": 2
}
}
}
}' > /zhiyun/shihaihong/jobs/erp_u_sale_m_inc.json
echo "开始抽取"
hadoop fs -mkdir -p /zhiyun/shihaihong/tmp/erp_u_sale_m_inc
python /opt/datax/bin/datax.py
/zhiyun/shihaihong/jobs/erp_u_sale_m_inc.json
echo "hive 建表"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e '
create database if not exists ods_shihaihong location
"/zhiyun/shihaihong/ods";
create external table if not exists
ods_shihaihong.erp_u_sale_m_inc(
id int,
saleno string,
busno string,
posno string,
extno string,
extsource string,
o2o_trade_from string,
channel int,
starttime string,
finaltime string,
payee string,
discounter string,
crediter string,
returner string,
warranter1 string,
warranter2 string,
stdsum string,
netsum string,
loss string,
discount float,
member string,
precash string,
shiftid string,
shiftdate string,
yb_saleno string) partitioned by (stamp string)
row format delimited fields terminated by "\t"
lines terminated by "\n"
stored as orc
location "/zhiyun/shihaihong/ods/erp_u_sale_m_inc";
'
echo "加载数据"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "
load data inpath '/zhiyun/shihaihong/tmp/erp_u_sale_m_inc/*'
overwrite into table ods_shihaihong.erp_u_sale_m_inc
partition(stamp='"$day"');
"
echo "验证数据"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "
show partitions ods_shihaihong.erp_u_sale_m_inc;
select count(1) from ods_shihaihong.erp_u_sale_m_inc where stamp
= '"$day"';
select * from ods_shihaihong.erp_u_sale_m_inc where stamp =
'"$day"' limit 5;
"
echo "抽取完成"
任务调度：
执行一次，输入参数：
erp_u_sale_pay_inc ：
全量抽取： 
#!/bin/bash
echo "生成全量配置文件"
mkdir -p /zhiyun/shihaihong/jobsecho '{
"job": {
"content": [
{
"reader": {
"name": "mysqlreader",
"parameter": {
"column": ["*"],
"connection": [
{
"jdbcUrl": [
"jdbc:mysql://zhiyun.pub:233
06/erp?useSSL=false"
],
"table": [
"u_sale_pay"
]
}
],
"password": "zhiyun",
"username": "zhiyun"
}
},
"writer": {
"name": "hdfswriter",
"parameter": {
"column": [
{"name":
"id","type":"int"},
{"name":"saleno","type":"string"},
{"name":"cardno","type":"string"},
{"name":"netsum","type":"string"},
{"name":"paytype","type":"string"},
{"name":"bak1","type":"string"}
],
"defaultFS": "hdfs://cdh02:8020",
"fieldDelimiter": "\t",
"fileName": "erp_u_sale_pay_full.data",
"fileType": "orc",
"path":
"/zhiyun/shihaihong/ods/erp_u_sale_pay_full",
"writeMode": "truncate"}
}
}
],
"setting": {
"speed": {
"channel": "3"
}
}
}
}' > /zhiyun/shihaihong/jobs/erp_u_sale_pay_full.json
echo "开始抽取"
hadoop fs -mkdir -p /zhiyun/shihaihong/ods/erp_u_sale_pay_full
python /opt/datax/bin/datax.py
/zhiyun/shihaihong/jobs/erp_u_sale_pay_full.json
echo "hive 建表"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e '
create database if not exists ods_shihaihong location
"/zhiyun/shihaihong/ods";
-- 增量表
create external table if not exists
ods_shihaihong.u_sale_pay_full(
id int,
saleno string,
cardno string,
netsum string,
paytype string,
bak1 string
)
row format delimited fields terminated by "\t"
lines terminated by "\n"
stored as orc
location "/zhiyun/shihaihong/ods/erp_u_sale_pay_full";
'
# echo "加载数据"
# beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "
# load data inpath
\"/zhiyun/shihaihong/tmp/erp_u_sale_pay_full/*\" overwrite into
table ods_shihaihong.erp_u_sale_pay_full
partition(createtime='$day');# "
echo "验证数据"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "
select count(1) from ods_shihaihong.u_sale_pay_full;
"
echo "抽取完成"
任务调度：
编辑 GLUE IDE, 执行一次：
增量抽取： 
#!/bin/bash
day=$(date -d "yesterday" +%Y-%m-%d)
if [ $1 != "" ]; then
day=$(date -d "$1 -1 day" +%Y-%m-%d);
fi;
echo "抽取的日期为 $day"
echo "生成增量配置文件"
mkdir -p /zhiyun/shihaihong/jobs
echo '
{
"job": {
"content": [
{
"reader": {
"name": "mysqlreader",
"parameter": {
"column": ["*"],
"connection": [
{
"jdbcUrl": [
"jdbc:mysql://zhiyun.pub:233
06/erp?useSSL=false"
],
"table": [
"select u_sale_pay.*,stamp
from u_sale_pay left join u_sale_m on
u_sale_pay.saleno=u_sale_m.saleno where stamp between '\'"$day"
00:00:00\'' and '\'"$day" 23:59:59\'' and id>0 "
]
}
],
"password": "zhiyun",
"username": "zhiyun"
}},
"writer": {
"name": "hdfswriter",
"parameter": {
"column": [
{"name":
"id","type":"int"},
{"name":"saleno","type":"string"},
{"name":"cardno","type":"string"},
{"name":"netsum","type":"string"},
{"name":"paytype","type":"string"},
{"name":"bak1","type":"string"},
{"name":"stamp","type":"string"}
],
"defaultFS": "hdfs://cdh02:8020",
"fieldDelimiter": "\t",
"fileName": "erp_u_sale_pay_inc.data",
"fileType": "orc",
"path":
"/zhiyun/shihaihong/tmp/erp_u_sale_pay_inc",
"writeMode": "truncate"
}
}
}
],
"setting": {
"speed": {
"channel": "3"
}
}
}
}
' > /zhiyun/shihaihong/jobs/erp_u_sale_play_inc.json
echo "开始抽取"
hadoop fs -mkdir -p /zhiyun/shihaihong/tmp/erp_u_sale_play_inc
python /opt/datax/bin/datax.py
/zhiyun/shihaihong/jobs/erp_u_sale_play_inc.json
echo "hive 建表"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e 'create database if not exists ods_shihaihong location
"/zhiyun/shihaihong/ods";
create external table if not exists
ods_shihaihong.erp_u_sale_play_inc(
id int,
saleno string,
cardno string,
netsum string,
paytype string,
bak1 string
) partitioned by (stamp string)
row format delimited fields terminated by "\t"
lines terminated by "\n"
stored as orc
location "/zhiyun/shihaihong/ods/erp_u_sale_play_inc";
'
echo "加载数据"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "
load data inpath '/zhiyun/shihaihong/tmp/erp_u_sale_play_inc/*'
overwrite into table ods_shihaihong.erp_u_sale_play_inc
partition(stamp='"$day"');
"
echo "验证数据"
beeline -u jdbc:hive2://localhost:10000 -n root -p 123 -e "
show partitions ods_shihaihong.erp_u_sale_play_inc;
select count(1) from ods_shihaihong.erp_u_sale_play_inc where
stamp = '"$day"';
select * from ods_shihaihong.erp_u_sale_play_inc where stamp =
'"$day"' limit 5;
"
echo "抽取完成"
任务调度：
执行一次：

本文来自互联网用户投稿，该文观点仅代表作者本人，不代表本站立场。本站仅提供信息存储空间服务，不拥有所有权，不承担相关法律责任。如若转载，请注明出处：http://www.rhkb.cn/news/428904.html

如若内容造成侵权/违法违规/事实不符，请联系长河编程网进行投诉反馈email:809451989@qq.com，一经查实，立即删除！

相关文章

4.提升客户服务体验：ChatGPT在客服中的应用（4/10）

4.提升客户服务体验：ChatGPT在客服中的应用（4/10）

本文大纲旨在指导撰写一篇全面探讨ChatGPT如何通过优化客户服务流程、提供实际应用案例和用户反馈&#xff0c;以提升客户服务体验的深入博客文章。 引言 在当今竞争激烈的商业环境中&#xff0c;客户服务已成为企业成功的关键因素。优质的客户服务不仅能够增强客户满意度和忠…
阅读更多...
天池Fashion AI 比赛失败经历分享

天池Fashion AI 比赛失败经历分享

关联比赛: FashionAI全球挑战赛—服饰关键点定位 昨天是天池Fashion AI初赛Deadline&#xff0c; 成绩出来复赛都没能进&#xff0c;虽然结果很遗憾&#xff0c;但在比赛的过程中也接触到了不少的新东西&#xff0c;希望能在这里把我尝试过的方法都分享出来。作为对自己的总结…
阅读更多...
Why Is Prompt Tuning for Vision-Language Models Robust to Noisy Labels?

Why Is Prompt Tuning for Vision-Language Models Robust to Noisy Labels?

文章汇总 本文的作者针对了提示学习的结构设计进行了分析&#xff0c;发现了一些规律&#xff1a; 1)固定的类名令牌为模型的优化提供了强正则化&#xff0c;减少了由噪声样本引起的梯度。 2)从多样化和通用的web数据中学习到的强大的预训练图像文本嵌入为图像分类提供了强大…
阅读更多...
基于kubernetes-nmstate配置节点网络

基于kubernetes-nmstate配置节点网络

kubernetes-nmstate 简介 kubernetes-nmstate 通过 Kubernetes API 驱动的声明式节点网络配置。 随着混合云的出现&#xff0c;节点网络设置变得更加具有挑战性。不同的环境有不同的网络要求。 容器网络接口&#xff08;CNI&#xff09;标准实现了不同的解决方案&#xff0c;…
阅读更多...
简单多状态dp第二弹 leetcode -删除并获得点数 -粉刷房子

简单多状态dp第二弹 leetcode -删除并获得点数 -粉刷房子

740. 删除并获得点数 删除并获得点数 分析: 使用动态规划解决 这道题依旧是 打家劫舍I 问题的变型。 我们注意到题目描述&#xff0c;选择 x 数字的时候&#xff0c; x - 1 与 x 1 是不能被选择的。像不像 打家劫舍 问题中&#xff0c;选择 i 位置的金额之后&#xff0c;就不…
阅读更多...
更高效的搜索工具，国内免费好用的AI智能搜索引擎工具

更高效的搜索工具，国内免费好用的AI智能搜索引擎工具

搜索引擎是我们获取信息的重要渠道&#xff0c;然而由于搜索引擎搜索结果存在较多的广告以及一些无关内容&#xff0c;这使我们的搜索效率变得更低效。小编就和大家分享几款国内免费好用的AI智能搜索工具&#xff0c;提高搜索效率。 1.开搜AI搜索 开搜AI搜索是一款基于深度学…
阅读更多...
低版本SqlSugar的where条件中使用可空类型报语法错误

低版本SqlSugar的where条件中使用可空类型报语法错误

SQLServer数据表中有两列可空列&#xff0c;均为数值类型&#xff0c;同时在数据库中录入测试数据&#xff0c;Age和Height列均部分有值。   使用SqlSugar的DbFirst功能生成数据库表类&#xff0c;其中Age、Height属性均为可空类型。   开始使用的SqlSugar版本较低&…
阅读更多...
Zabbix 6.4添加中文语言

Zabbix 6.4添加中文语言

/usr/share/zabbix/include/locales .inc .phplocale -agrep “zh_CN" yum install langpacks-zh_CN.noarch y y y
阅读更多...
Linux | 探索 Linux 信号机制：信号的产生和自定义捕捉

Linux | 探索 Linux 信号机制：信号的产生和自定义捕捉

信号是 Linux 操作系统中非常重要的进程控制机制&#xff0c;用来异步通知进程发生某种事件。理解信号的产生、阻塞、递达、捕捉等概念&#xff0c;可以帮助开发者更好地编写健壮的应用程序&#xff0c;避免由于未处理的信号导致程序异常退出。本文将带你从基础概念开始&#x…
阅读更多...
利士策分享，自我和解：通往赚钱与内心富足的和谐之道

利士策分享，自我和解：通往赚钱与内心富足的和谐之道

利士策分享&#xff0c;自我和解&#xff1a;通往赚钱与内心富足的和谐之道 在这个快节奏、高压力的时代&#xff0c;我们往往在追求物质财富的同时&#xff0c;忽略了内心世界的和谐与平衡。 赚钱&#xff0c;作为现代生活中不可或缺的一部分&#xff0c;它不仅仅是生存的手段…
阅读更多...
【Godot4.3】胶囊形的偏移获取法

【Godot4.3】胶囊形的偏移获取法

概述 之前用半圆弧拼接的方式求过胶囊形&#xff0c;在逐渐熟练使用Geometry2D的过程中&#xff0c;发现通过线段求端点是圆角类型的偏移多边形&#xff0c;获得的就是胶囊形。 所以我们有了第二种胶囊形求法。 测试代码 tool extends Node2D## 横向宽度 export var width:…
阅读更多...
Java-数据结构-排序-(一) (｡･ω･｡)

Java-数据结构-排序-(一) (｡･ω･｡)

文本目录&#xff1a; ❄️一、排序的概念及引用&#xff1a; ➷ 排序的概念&#xff1a; ➷ 常见的排序算法&#xff1a; ❄️二、插入排序的实现&#xff1a; ➷ 1、直接插入排序&#xff1a; ☞ 直接插入排序的基本思想&#xff1a; ☞ 直接插入排序的实现&#xff1a; ▶…
阅读更多...
安全热点问题

安全热点问题

安全热点问题 1.DDOS2.补丁管理3.堡垒机管理4.加密机管理 1.DDOS 分布式拒绝服务攻击&#xff0c;是指黑客通过控制由多个肉鸡或服务器组成的僵尸网络&#xff0c;向目标发送大量看似合法的请求&#xff0c;从而占用大量网络资源使网络瘫痪&#xff0c;阻止用户对网络资源的正…
阅读更多...
HarmonyOS Next开发----使用XComponent自定义绘制

HarmonyOS Next开发----使用XComponent自定义绘制

XComponent组件作为一种绘制组件&#xff0c;通常用于满足用户复杂的自定义绘制需求&#xff0c;其主要有两种类型"surface和component。对于surface类型可以将相关数据传入XComponent单独拥有的NativeWindow来渲染画面。 由于上层UI是采用arkTS开发&#xff0c;那么想要…
阅读更多...
【医疗大数据】基于 B2B 的医疗保健系统中大数据信息管理的安全和隐私问题分析

【医疗大数据】基于 B2B 的医疗保健系统中大数据信息管理的安全和隐私问题分析

基于 B2B 的医疗保健系统中大数据信息管理的安全和隐私问题分析 1、引言 1-1 医疗大数据的特点 10 V模型&#xff1a;在医疗领域&#xff0c;大数据的特点被描述为10 V&#xff0c;包括价值&#xff08;Value&#xff09;、体量&#xff08;Volume&#xff09;、速度&#xf…
阅读更多...
Leetcode Hot 100刷题记录 -Day16（旋转图像）

Leetcode Hot 100刷题记录 -Day16（旋转图像）

旋转图像 问题描述&#xff1a; 给定一个 n n 的二维矩阵 matrix 表示一个图像。请你将图像顺时针旋转 90 度。 你必须在原地旋转图像&#xff0c;这意味着你需要直接修改输入的二维矩阵。请不要 使用另一个矩阵来旋转图像。 示例 1 输入&#xff1a;matrix [[1,2,3],[4,5,6]…
阅读更多...
Python学习——【4.2】数据容器：tuple元组

Python学习——【4.2】数据容器：tuple元组

文章目录 【4.2】数据容器&#xff1a;tuple元组一、元组的定义格式二、元组的特点三、元组的操作&#xff08;一&#xff09;常见操作&#xff08;二&#xff09;循环遍历 【4.2】数据容器&#xff1a;tuple元组 一、元组的定义格式 为什么需要元组 列表是可以修改的。如果想…
阅读更多...
【网络安全】分享4个高危业务逻辑漏洞

【网络安全】分享4个高危业务逻辑漏洞

未经许可，不得转载。 文章目录 正文逻辑漏洞1逻辑漏洞2逻辑漏洞3逻辑漏洞4其它正文 该目标程序是一家提供浏览器服务的公司，其核心功能是网页抓取和多账户登录操作，类似于浏览器中的隐身模式，但更加强大和高效。通过该平台，用户可以轻松管理并同时运行数百个隐身浏览器实…
阅读更多...
Navicate 链接Oracle 提示 Oracle Library is not loaded ,账号密码都正确地址端口也对

Navicate 链接Oracle 提示 Oracle Library is not loaded ,账号密码都正确地址端口也对

Navicate 链接Oracle 提示 Oracle Library is not loaded ,账号密码都正确地址端口也对的问题 解决办法 出现 Oracle Library is not loaded 错误提示&#xff0c;通常是因为 Navicat 无法找到或加载 Oracle 客户端库&#xff08;OCI.dll&#xff09;。要解决这个问题&#x…
阅读更多...
【自动驾驶】决策规划算法 | 数学基础（三）直角坐标与自然坐标转换Ⅱ

【自动驾驶】决策规划算法 | 数学基础（三）直角坐标与自然坐标转换Ⅱ

写在前面&#xff1a; &#x1f31f; 欢迎光临 清流君 的博客小天地&#xff0c;这里是我分享技术与心得的温馨角落。&#x1f4dd; 个人主页&#xff1a;清流君_CSDN博客&#xff0c;期待与您一同探索 移动机器人 领域的无限可能。 &#x1f50d; 本文系 清流君 原创之作&…
阅读更多...
最新文章
推荐文章

Copyright @ 2022~2023