Golang: Store Query Result in a Map

目录

  • [1. Golang: Store Query Result in a Map](#1. Golang: Store Query Result in a Map)
    • [1.1. Using Structs](#1.1. Using Structs)
    • [1.2. Using Maps](#1.2. Using Maps)

1. Golang: Store Query Result in a Map

注意: 使用这个可能会造成列名和列值乱串的现象,解决这个可以使用 AS 语法:

sql 复制代码
SELECT 
  TENANT_ID AS TENANT_ID,
  SVR_IP AS SVR_IP,
  SVR_PORT AS SVR_PORT,
  PLAN_ID AS PLAN_ID,
  SQL_ID AS SQL_ID,
  TYPE AS TYPE,
  DB_ID AS DB_ID,
  STATEMENT AS STATEMENT,
  PLAN_HASH AS PLAN_HASH,
  LAST_ACTIVE_TIME AS LAST_ACTIVE_TIME,
  ELAPSED_TIME AS ELAPSED_TIME
FROM GV$PLAN_CACHE_PLAN_STAT
WHERE LAST_ACTIVE_TIME > '%s' AND LAST_ACTIVE_TIME <= '%s' AND ELAPSED_TIME > %d

Converting the results of a SQL query to a struct in Go is trivial, but sometimes you don't know ahead of time exactly what columns and types you're going to be retrieving, and a map may be better suited for storing the results.

1.1. Using Structs

First, here's the standard way we can convert results to a struct:

go 复制代码
rows, _ := db.Query("SELECT ...") // Note: Ignoring errors for brevity
for rows.Next() {
    s := myStruct{}
    if err := rows.Scan(&s); err != nil {
        return err
    }
    
    // Do something with 's'
}

Easy enough, but storing your query result in a map is a bit trickier.

1.2. Using Maps

Let's say for example you're working with a user's database where you don't know the schema ahead of time. You can't write a struct to store the results, because you don't know what columns and data types you're going to be retrieving. What we want in this case is a map[string]interface{} where the key is the column name and the value could be any data type.

You might assume you can do the following:

go 复制代码
rows, _ := db.Query("SELECT ...") // Note: Ignoring errors for brevity
for rows.Next() {
    m := make(map[string]interface{})
    
    // This WON'T WORK
    if err := rows.Scan(&m); err != nil {
        // ERROR: sql: expected X destination arguments in Scan, not 1
    }
}

Basically the SQL package is thinking that you expect a single column to be returned and for it to be a map[string]interface{} compatible type, which isn't what we we're trying to do.

Instead what we have to do in order to make this work is the following:

go 复制代码
rows, _ := db.Query("SELECT ...") // Note: Ignoring errors for brevity
cols, _ := rows.Columns()

for rows.Next() {
    // Create a slice of interface{}'s to represent each column,
    // and a second slice to contain pointers to each item in the columns slice.
    columns := make([]interface{}, len(cols))
    columnPointers := make([]interface{}, len(cols))
    for i, _ := range columns {
        columnPointers[i] = &columns[i]
    }
    
    // Scan the result into the column pointers...
    if err := rows.Scan(columnPointers...); err != nil {
        return err
    }

    // Create our map, and retrieve the value for each column from the pointers slice,
    // storing it in the map with the name of the column as the key.
    m := make(map[string]interface{})
    for i, colName := range cols {
        val := columnPointers[i].(*interface{})
        m[colName] = *val
    }
    
    // Outputs: map[columnName:value columnName2:value2 columnName3:value3 ...] 
    fmt.Print(m)
}

So how does this work? Well first we query and get our rows as usual, but this time we use rows.Columns() to get a reference to all column names in the result.

Then for each row, we create a slice of interface{}'s called columns who's length matches the number of columns. Next we create a second slice with the same length called columnPointers, but this time we iterate over each element in columns and store a pointer to the interface{} element in our columnPointers slice. This is necessary because the sql package requires pointers when scanning. So now we have two slices, one of interface{}s and one of pointers to the interface{}s.

Now we can scan the row into the slice of interface{} pointers (ie. columnPointers).

Finally, we create our map[string]interface{}, and iterate over the column names. For each column name (colName), we deference the interface{} pointer at the current loop index from the columnPointers slice, which references the value in the columns slice. We take this dereferenced value and store it in the map as the value, with the key being the column name.

Now we can use the map however we need, essentially allowing us to perform queries dynamically, without requiring knowledge of the schema we're going to be querying when we write our code.

This may seem a little confusing at first, especially if you have no prior experience with pointers. If so, I'd recommend reading up on Pointers and trying the code out, inserting some debug logging to fully understand what's happening during each step of the process.

相关推荐
程序员岳焱32 分钟前
Java 与 MySQL 性能优化:Java 实现百万数据分批次插入的最佳实践
后端·mysql·性能优化
计算机毕设定制辅导-无忧学长1 小时前
西门子 PLC 与 Modbus 集成:S7-1500 RTU/TCP 配置指南(一)
服务器·数据库·tcp/ip
程序员柳2 小时前
基于微信小程序的校园二手交易平台、微信小程序校园二手商城源代码+数据库+使用说明,layui+微信小程序+Spring Boot
数据库·微信小程序·layui
梦在深巷、2 小时前
MySQL/MariaDB数据库主从复制之基于二进制日志的方式
linux·数据库·mysql·mariadb
IT乌鸦坐飞机2 小时前
ansible部署数据库服务随机启动并创建用户和设置用户有完全权限
数据库·ansible·centos7
IT_10242 小时前
Spring Boot项目开发实战销售管理系统——数据库设计!
java·开发语言·数据库·spring boot·后端·oracle
Johny_Zhao3 小时前
Ubuntu系统安装部署Pandawiki智能知识库
linux·mysql·网络安全·信息安全·云计算·shell·yum源·系统运维·itsm·pandawiki
祁思妙想3 小时前
八股学习(三)---MySQL
数据库·学习·mysql
惊骇世俗王某人4 小时前
1.MySQL之如何定位慢查询
数据库·mysql
秦歌6664 小时前
向量数据库-Milvus快速入门
数据库·milvus