作者：杨涛涛

背景

本文介绍 MySQL 8.0 shell 子模块 Util 的两个导入特性 importTable/import_table(JS和Python 版本的命名差异)、importJson/import_json的使用方法。其中 import_table 是通过传统 MySQL 协议来通信，Import_json 是通过 X 插件协议来通信。

MySQL 一直以来提供导入文件 SQL 命令 load data infile（单线程）以及对应的可执行文件 mysqlimport（多线程）。

比如我导入 100W 行示例数据到表 ytt.tl1，花了 24 秒。这个已经是 MySQL 默认导入来的最快的。

[root@mysql-dev ytt]# time mysqlimport --login-path=ytt_master --fields-terminated-by=, --use-threads=4 ytt /var/lib/mysql-files/tl1.csv
ytt.tl1: Records: 1048576 Deleted: 0 Skipped: 0 Warnings: 0
real 0m24.815s
user 0m0.013s
sys 0m0.031s

分析

那我们现在看下 mysqlimport 工具的升级版，mysqlshell 的 util 工具集。使用这两个工具之前，必须得临时开启 local_infile 选项。

1. import_table

[root@mysql-dev ytt]# mysqlsh
MySQL Shell 8.0.17
Copyright (c) 2016, 2019, Oracle and/or its affiliates. All rights reserved.
Oracle is a registered trademark of Oracle Corporation and/or its affiliates.
Other names may be trademarks of their respective owners.
Type 'help' or '?' for help; 'quit' to exit.

建立 3306 端口的新连接

MySQL JS > c admin@127.0.0.1:3306
Creating a session to 'admin@127.0.0.1:3306'
Fetching schema names for autocompletion... Press ^C to stop.
Your MySQL connection id is 56
Server version: 8.0.17 MySQL Community Server - GPL
No default schema selected; type \use 
<schema>
to set one.

我这里切换为 python 模式

MySQL 127.0.0.1:3306 ssl JS > py
Switching to Python mode...
MySQL 127.0.0.1:3306 ssl Py > \use ytt
Default schema set to `ytt`.

清空掉示例表 Ytt.tl1

MySQL 127.0.0.1:3306 ssl ytt Py > sql truncate tl1;
Fetching table and column names from `ytt` for auto-completion... Press ^C to stop.
Query OK, 0 rows affected (0.2354 sec)

示例表表结构

MySQL 127.0.0.1:3306 ssl ytt Py > sql show create table tl1G
*************************** 1. row ***************************
 Table: tl1
Create Table: CREATE TABLE `tl1` (
 `id` bigint(20) unsigned NOT NULL AUTO_INCREMENT,
 `r1` int(11) DEFAULT NULL,
 `r2` int(11) DEFAULT NULL,
 `r3` varchar(30) COLLATE utf8mb4_general_ci DEFAULT NULL,
 `r4` datetime DEFAULT NULL,
 PRIMARY KEY (`id`),
 UNIQUE KEY `id` (`id`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_general_ci
1 row in set (0.0011 sec)

import_table 有两个参数，第一个参数定义导入文件的路径，第二个定义相关选项，比如导入的格式，并发的数量等。

定义文件路径（参数1）

MySQL 127.0.0.1:3306 ssl ytt Py > y_file1='/var/lib/mysql-files/tl1.csv'

定义选项（参数2）

MySQL 127.0.0.1:3306 ssl ytt Py > y_options1={"schema":"ytt","table":"tl1","fieldsTerminatedBy":",","showProgress":True,"threads":4}

执行导入：

MySQL 127.0.0.1:3306 ssl ytt Py > util.import_table(y_file1,y_options1);
Importing from file '/var/lib/mysql-files/tl1.csv' to table `ytt`.`tl1` in MySQL Server at 127.0.0.1:3306 using 1 thread
[Worker000] ytt.tl1: Records: 1048576 Deleted: 0 Skipped: 0 Warnings: 0
100% (40.87 MB / 40.87 MB), 2.14 MB/s
File '/var/lib/mysql-files/tl1.csv' (40.87 MB) was imported in 16.7394 sec at 2.44 MB/s
Total rows affected in ytt.tl1: Records: 1048576 Deleted: 0 Skipped: 0 Warnings: 0

只花了不到 17 秒，比传统 mysqlimport 快了不少。

我们上面指定了显式指定了字段分隔符，那有没有已经定义好的组合格式呢？ 答案是有的，选项 dialect 可以指定以下格式：csv，tsv，json，csv-unix

那么上面的导入，我们可以更简单，

改下变量 y_options1 的定义

MySQL 127.0.0.1:3306 ssl ytt Py > sql truncate tl1;
Query OK, 0 rows affected (0.0776 sec)
MySQL 127.0.0.1:3306 ssl ytt Py > y_options1={"schema":"ytt","table":"tl1","dialect":"csv-unix","showProgress":True,"threads":4}
MySQL 127.0.0.1:3306 ssl ytt Py > util.import_table(y_file1,y_options1);
Importing from file '/var/lib/mysql-files/tl1.csv' to table `ytt`.`tl1` in MySQL Server at 127.0.0.1:3306 using 1 thread
[Worker000] ytt.tl1: Records: 1048576 Deleted: 0 Skipped: 0 Warnings: 0
100% (40.87 MB / 40.87 MB), 2.67 MB/s
File '/var/lib/mysql-files/tl1.csv' (40.87 MB) was imported in 14.1000 sec at 2.90 MB/s
Total rows affected in ytt.tl1: Records: 1048576 Deleted: 0 Skipped: 0 Warnings: 0

导入时间差不多。这里要说明下，dialect 选项的优先级比较低，比如添加了'linesTerminatedBy':'rn'， 则覆盖他自己的'n'。

选项 diaelect 还有一个可选值为 json，可以直接把 json 结果导入到文档表里。比如我新建一张表 tl1_json

MySQL 127.0.0.1:3306 ssl ytt Py > sql create table tl1_json(
 id bigint primary key,
 content json);
Query OK, 0 rows affected (0.3093 sec)

重新定义文件以及导入选项。

MySQL 127.0.0.1:3306 ssl ytt Py > y_file2='/var/lib/mysql-files/tl1.json'
MySQL 127.0.0.1:3306 ssl ytt Py > rows=['content']
MySQL 127.0.0.1:3306 ssl ytt Py > y_options2={"schema":"ytt","table":"tl1_json","dialect":"json","showProgress":True,"threads":4,'columns':rows}

导入 JSON 数据

MySQL 127.0.0.1:3306 ssl ytt Py > util.import_table(y_file2,y_options2)
Importing from file '/var/lib/mysql-files/tl1.json' to table `ytt`.`tl1_json` in MySQL Server at 127.0.0.1:3306 using 2 threads
[Worker000] ytt.tl1_json: Records: 464633 Deleted: 0 Skipped: 0 Warnings: 0
[Worker001] ytt.tl1_json: Records: 583943 Deleted: 0 Skipped: 0 Warnings: 0
100% (90.15 MB / 90.15 MB), 2.71 MB/s
File '/var/lib/mysql-files/tl1.json' (90.15 MB) was imported in 23.3530 sec at 3.86 MB/s
Total rows affected in ytt.tl1_json: Records: 1048576 Deleted: 0 Skipped: 0 Warnings: 0

速度也还可以，不到 24 秒。

那导入 json 数据，就必须得提到以 X 插件协议通信的工具 import_json了。

2. imort_json

我们切换到 mysqlx 端口

MySQL 127.0.0.1:3306 ssl ytt Py > c admin@127.0.0.1:33060
Creating a session to 'admin@127.0.0.1:33060'
Fetching schema names for autocompletion... Press ^C to stop.
Closing old connection...
Your MySQL connection id is 16 (X protocol)
Server version: 8.0.17 MySQL Community Server - GPL
No default schema selected; type \use 
<schema> to set one
.
 MySQL 127.0.0.1:33060+ ssl Py > \use ytt
Default schema `ytt` accessible through db.
-- 清空表tl1_json
 MySQL 127.0.0.1:33060+ ssl ytt Py > sql truncate tl1_json;
Query OK, 0 rows affected (0.1098 sec)

import_json 参数和 Import_table 参数类似，

这里我改下选项

MySQL 127.0.0.1:33060+ ssl ytt Py > y_file3=y_file2
MySQL 127.0.0.1:33060+ ssl ytt Py > y_options3={"schema":"ytt","table":"tl1_json",'tableColumn':'content'}
MySQL 127.0.0.1:33060+ ssl ytt Py > util.import_json(y_file3,y_options3)
Importing from file "/var/lib/mysql-files/tl1.json" to table `ytt`.`tl1_json` in MySQL Server at 127.0.0.1:33060
.. 517776.. 1032724.. 1048576.. 1048576
Processed 90.15 MB in 1048576 documents in 35.2400 sec (29.76K documents/s)
Total successfully imported documents 1048576 (29.76K documents/s)

我在手册上没有看到多线程的选项，所以单线程跑 35 秒慢了些。

查看刚刚导入的数据

MySQL 127.0.0.1:33060+ ssl ytt Py > sql select id,json_pretty(content) from tl1_json limit 1G
*************************** 1. row ***************************
 id: 1
json_pretty(content): {
 "id": 1,
 "r1": 10,
 "r2": 10,
 "r3": "mysql",
 "r4": "2019-09-16 16:49:50.000000"
}
1 row in set (0.0007 sec)

import_json 不仅仅可以导入 Json 数据，更重要的是可以在 BSON 和 JSON 之间平滑的转换，有兴趣的同学可以去 TRY 下。