Menu
I want to add a column to my table using ALTER TABLE and UPDATE statements not to recreate the full table.
When using a subquery in my UPDATE statement I don't get the output I expect.
build reproducible data
library(dplyr)
library(dbplyr)
library(DBI)
con <- DBI::dbConnect(RSQLite::SQLite(), path = ":memory:")
copy_to(con, iris[c(1,2,51),],"iris")
tbl(con,"iris")
# # Source: table<iris> [?? x 5]
# # Database: sqlite 3.19.3 []
# Sepal.Length Sepal.Width Petal.Length Petal.Width Species
# <dbl> <dbl> <dbl> <dbl> <chr>
# 1 5.1 3.5 1.4 0.2 setosa
# 2 4.9 3.0 1.4 0.2 setosa
# 3 7.0 3.2 4.7 1.4 versicolor
create the new column in a separate table
DBI::dbSendQuery(con, "CREATE TABLE new_table AS SELECT t2.new_col from
iris t1 inner join
(SELECT Species, sum(`Sepal.Width`) as new_col FROM iris GROUP BY Species) t2
on t1.Species = t2.Species")
tbl(con,"new_table")
# # Source: table<new_table> [?? x 1]
# # Database: sqlite 3.19.3 []
# new_col
# <dbl>
# 1 6.5
# 2 6.5
# 3 3.2
create the new column in the old table
DBI::dbSendQuery(con, "ALTER TABLE iris ADD COLUMN new_col DOUBLE")
try to plug the new column from new_table there
DBI::dbSendQuery(con, "UPDATE iris SET new_col = (SELECT new_col FROM new_table)")
tbl(con,"iris")
# # Source: table<iris> [?? x 6]
# # Database: sqlite 3.19.3 []
# Sepal.Length Sepal.Width Petal.Length Petal.Width Species new_col
# <dbl> <dbl> <dbl> <dbl> <chr> <dbl>
# 1 5.1 3.5 1.4 0.2 setosa 6.5
# 2 4.9 3.0 1.4 0.2 setosa 6.5
# 3 7.0 3.2 4.7 1.4 versicolor 6.5
As you can see my new_col contains only the value 6.5 where I expected to have 3.2 on the last row. How can I fix this ?
1000
Like SELECT , the UPDATE statement can have a subquery in several places or clauses. In an UPDATE , the two clauses in which subqueries are used most commonly are SET and WHERE . The SET clause is where we define the new value for the column being modified by the UPDATE .
UPDATE operations with subqueries that reference the same table object are supported only if all of the following conditions are true: The subquery either returns a single row, or else has no correlated column references. The subquery is in the UPDATE statement WHERE clause, using Condition with Subquery syntax.
In SQLite, a subquery is a query within a query. You can create subqueries within your SQL statements. These subqueries can reside in the WHERE clause, the FROM clause, or the SELECT clause.
The rows in a table in a SQL database have no inherent order. So you cannot assign a "vector" of values like you would do it in R. However, You can modify your query slightly:
library(dplyr)
library(DBI)
con <- DBI::dbConnect(RSQLite::SQLite(), path = ":memory:")
copy_to(con, iris[c(1,2,51),],"iris")
DBI::dbSendQuery(con, "CREATE TABLE new_table AS
SELECT Species, sum(`Sepal.Width`) as new_col FROM iris GROUP BY Species")
tbl(con,"new_table")
#> # Source: table<new_table> [?? x 2]
#> # Database: sqlite 3.22.0 []
#> Species new_col
#> <chr> <dbl>
#> 1 setosa 6.5
#> 2 versicolor 3.2
DBI::dbSendQuery(con, "ALTER TABLE iris ADD COLUMN new_col DOUBLE")
DBI::dbSendQuery(con, "UPDATE iris SET new_col = (SELECT new_col FROM new_table t2
WHERE iris.Species = t2.Species)")
tbl(con,"iris")
#> # Source: table<iris> [?? x 6]
#> # Database: sqlite 3.22.0 []
#> Sepal.Length Sepal.Width Petal.Length Petal.Width Species new_col
#> <dbl> <dbl> <dbl> <dbl> <chr> <dbl>
#> 1 5.1 3.5 1.4 0.2 setosa 6.5
#> 2 4.9 3 1.4 0.2 setosa 6.5
#> 3 7 3.2 4.7 1.4 versicolor 3.2
If you have multiple computed columns, you can use UPDATE ... SET (c1, c2, ...) = (...) like this:
library(dplyr)
library(dbplyr)
library(DBI)
con <- DBI::dbConnect(RSQLite::SQLite(), path = ":memory:")
copy_to(con, iris[c(1,2,51),],"iris")
DBI::dbSendQuery(con, "CREATE TABLE aggs AS
SELECT Species,
SUM(`Sepal.Width`) AS sw_sum,
AVG(`Sepal.Width`) AS sw_avg
FROM iris GROUP BY Species")
tbl(con,"aggs")
#> # Source: table<aggs> [?? x 3]
#> # Database: sqlite 3.22.0 []
#> Species sw_sum sw_avg
#> <chr> <dbl> <dbl>
#> 1 setosa 6.5 3.25
#> 2 versicolor 3.2 3.2
DBI::dbSendQuery(con, "ALTER TABLE iris ADD COLUMN sw_sum DOUBLE")
DBI::dbSendQuery(con, "ALTER TABLE iris ADD COLUMN sw_avg DOUBLE")
DBI::dbSendQuery(con, "UPDATE iris
SET (sw_sum, sw_avg) = (SELECT sw_sum, sw_avg
FROM aggs WHERE iris.Species = aggs.Species)")
tbl(con,"iris")
#> # Source: table<iris> [?? x 7]
#> # Database: sqlite 3.22.0 []
#> Sepal.Length Sepal.Width Petal.Length Petal.Width Species sw_sum sw_avg
#> <dbl> <dbl> <dbl> <dbl> <chr> <dbl> <dbl>
#> 1 5.1 3.5 1.4 0.2 setosa 6.5 3.25
#> 2 4.9 3 1.4 0.2 setosa 6.5 3.25
#> 3 7 3.2 4.7 1.4 versico… 3.2 3.2
This should also work on Postgres, but probably not with SQL Server.
Actually, one does not need the intermediate table in this case:
library(dplyr)
library(dbplyr)
library(DBI)
con <- DBI::dbConnect(RSQLite::SQLite(), path = ":memory:")
copy_to(con, iris[c(1,2,51),],"iris")
DBI::dbSendQuery(con, "ALTER TABLE iris ADD COLUMN sw_sum DOUBLE")
DBI::dbSendQuery(con, "ALTER TABLE iris ADD COLUMN sw_avg DOUBLE")
DBI::dbSendQuery(con, "UPDATE iris
SET (sw_sum, sw_avg) =
(SELECT sw_sum, sw_avg FROM
(SELECT Species,
SUM(`Sepal.Width`) AS sw_sum,
AVG(`Sepal.Width`) AS sw_avg
FROM iris GROUP BY Species) aggs
WHERE iris.Species = aggs.Species)")
tbl(con,"iris")
#> # Source: table<iris> [?? x 7]
#> # Database: sqlite 3.22.0 []
#> Sepal.Length Sepal.Width Petal.Length Petal.Width Species sw_sum sw_avg
#> <dbl> <dbl> <dbl> <dbl> <chr> <dbl> <dbl>
#> 1 5.1 3.5 1.4 0.2 setosa 6.5 3.25
#> 2 4.9 3 1.4 0.2 setosa 6.5 3.25
#> 3 7 3.2 4.7 1.4 versico… 3.2 3.2
The intermediate table might be helpful in other cases, though. For example, when it is created using R as in the linked question.
152
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With
