dbplyr：合并两个表并将结果添加到数据库而不将它们加载到内存中

Question

请看一下post末尾的简单脚本。我有一个包含两个表的数据库，我使用 union_all 组合它们。有没有办法将结果添加到数据库而不收集数据，即将它们加载到内存中？非常感谢！




library(tidyverse)
library(DBI) # main DB interface
library(dbplyr) # dplyr back-end for DBs
#> 
#> Attaching package: 'dbplyr'
#> The following objects are masked from 'package:dplyr':
#> 
#>     ident, sql
library(RSQLite)


##create the databases

df1 <- tibble(x=1:20,y=rep(c("a", "b"), 10))


df2 <- tibble(x=101:120,y=rep(c("d", "e"), 10))




con <- dbConnect(drv=RSQLite::SQLite(), dbname="db.sqlite")

dbWriteTable(con,"mydata1",df1, overwrite=T)
dbWriteTable(con,"mydata2",df2, overwrite=T)

dbDisconnect(con) # closes our DB connection


con <- dbConnect(drv=RSQLite::SQLite(), dbname="db.sqlite")

mydb1 <- tbl(con, "mydata1")
mydb2 <- tbl(con, "mydata2")


mydb12 <- union_all(mydb1,mydb2)

#is there a way to add the union of mydb1 and mydb2 to the database without explicitly collecting the data?

^{由 reprex package (v0.3.0)}

于 2020-12-24 创建

Answer 1

由于您要处理 SQL，只需使用 SQL。

collect(mydb1) %>%
  nrow()
# [1] 20
DBI::dbExecute(con, "insert into mydata1 select * from mydata2")
# [1] 20
collect(mydb1) %>%
  nrow()
# [1] 40
collect(mydb1) %>%
  tail()
# # A tibble: 6 x 2
#       x y    
#   <int> <chr>
# 1   115 d    
# 2   116 e    
# 3   117 d    
# 4   118 e    
# 5   119 d    
# 6   120 e

如果您想要在新的 table 中合并数据，那么这里有一个替代方案。

DBI::dbExecute(con, "
  create table mydata12 as
    select * from mydata2 union all select * from mydata1")

dbplyr：合并两个表并将结果添加到数据库而不将它们加载到内存中

Dbplyr: combine two tables and add the to the result to the database without loading them in memory

r

dplyr

dbplyr