这里有几个函数，create and update.tbl_lazy.
他们分别实现CREATE TABLE，这很简单，并且ALTER TABLE/UPDATE一对则不然：
CREATE
create <- function(data,name){
DBI::dbSendQuery(data$src$con,
paste("CREATE TABLE", name,"AS", dbplyr::sql_render(data)))
dplyr::tbl(data$src$con,name)
}
example:
library(dbplyr)
library(DBI)
con <- DBI::dbConnect(RSQLite::SQLite(), path = ":memory:")
copy_to(con, head(iris,3),"iris")
tbl(con,"iris") %>% mutate(Sepal.Area= Sepal.Length * Sepal.Width) %>% create("iris_2")
# # Source: table<iris_2> [?? x 6]
# # Database: sqlite 3.22.0 []
# Sepal.Length Sepal.Width Petal.Length Petal.Width Species Sepal.Area
# <dbl> <dbl> <dbl> <dbl> <chr> <dbl>
# 1 5.1 3.5 1.4 0.2 setosa 17.8
# 2 4.9 3 1.4 0.2 setosa 14.7
# 3 4.7 3.2 1.3 0.2 setosa 15.0
UPDATE
update.tbl_lazy <- function(.data,...,new_type="DOUBLE PRECISION"){
quos <- rlang::quos(...)
dots <- rlang::exprs_auto_name(quos, printer = tidy_text)
# extract key parameters from query
sql <- dbplyr::sql_render(.data)
con <- .data$src$con
table_name <-gsub(".*?(FROM (`|\")(.+?)(`|\")).*","\\3",sql)
if(grepl("\nWHERE ",sql)) where <- regmatches(sql, regexpr("WHERE .*",sql))
else where <- ""
new_cols <- setdiff(names(dots),colnames(.data))
# Add empty columns to base table
if(length(new_cols)){
alter_queries <- paste("ALTER TABLE",table_name,"ADD COLUMN",new_cols,new_type)
purrr::walk(alter_queries, ~{
rs <- DBI::dbSendStatement(con, .)
DBI::dbClearResult(rs)})}
# translate unevaluated dot arguments to SQL instructions as character
translations <- purrr::map_chr(dots, ~ translate_sql(!!! .))
# messy hack to make translations work
translations <- gsub("OVER \\(\\)","",translations)
# 2 possibilities: called group_by or (called filter or called nothing)
if(identical(.data$ops$name,"group_by")){
# ERROR if `filter` and `group_by` both used
if(where != "") stop("Using both `filter` and `group by` is not supported")
# Build aggregated table
gb_cols <- paste0('"',.data$ops$dots,'"',collapse=", ")
gb_query0 <- paste(translations,"AS", names(dots),collapse=", ")
gb_query <- paste("CREATE TABLE TEMP_GB_TABLE AS SELECT",
gb_cols,", ",gb_query0,
"FROM", table_name,"GROUP BY", gb_cols)
rs <- DBI::dbSendStatement(con, gb_query)
DBI::dbClearResult(rs)
# Delete temp table on exit
on.exit({
rs <- DBI::dbSendStatement(con,"DROP TABLE TEMP_GB_TABLE")
DBI::dbClearResult(rs)
})
# Build update query
gb_on <- paste0(table_name,'."',.data$ops$dots,'" = TEMP_GB_TABLE."', .data$ops$dots,'"',collapse=" AND ")
update_query0 <- paste0(names(dots)," = (SELECT ", names(dots), " FROM TEMP_GB_TABLE WHERE ",gb_on,")",
collapse=", ")
update_query <- paste("UPDATE", table_name, "SET", update_query0)
rs <- DBI::dbSendStatement(con, update_query)
DBI::dbClearResult(rs)
} else {
# Build update query in case of no group_by and optional where
update_query0 <- paste(names(dots),'=',translations,collapse=", ")
update_query <- paste("UPDATE", table_name,"SET", update_query0,where)
rs <- DBI::dbSendStatement(con, update_query)
DBI::dbClearResult(rs)
}
tbl(con,table_name)
}
示例1，定义 2 个新的数字列：
tbl(con,"iris") %>% update(x=pmax(Sepal.Length,Sepal.Width),
y=pmin(Sepal.Length,Sepal.Width))
# # Source: table<iris> [?? x 7]
# # Database: sqlite 3.22.0 []
# Sepal.Length Sepal.Width Petal.Length Petal.Width Species x y
# <dbl> <dbl> <dbl> <dbl> <chr> <dbl> <dbl>
# 1 5.1 3.5 1.4 0.2 setosa 5.1 3.5
# 2 4.9 3 1.4 0.2 setosa 4.9 3
# 3 4.7 3.2 1.3 0.2 setosa 4.7 3.2
示例2，修改现有列，创建 2 个不同类型的新列：
tbl(con,"iris") %>%
update(x= Sepal.Length*Sepal.Width,
z= 2*y,
a= Species %||% Species,
new_type = c("DOUBLE","VARCHAR(255)"))
# # Source: table<iris> [?? x 9]
# # Database: sqlite 3.22.0 []
# Sepal.Length Sepal.Width Petal.Length Petal.Width Species x y z a
# <dbl> <dbl> <dbl> <dbl> <chr> <dbl> <dbl> <dbl> <chr>
# 1 5.1 3.5 1.4 0.2 setosa 17.8 3.5 7 setosasetosa
# 2 4.9 3 1.4 0.2 setosa 14.7 3 6 setosasetosa
# 3 4.7 3.2 1.3 0.2 setosa 15.0 3.2 6.4 setosasetosa
实施例3，更新其中：
tbl(con,"iris") %>% filter(Sepal.Width > 3) %>% update(a="foo")
# # Source: table<iris> [?? x 9]
# # Database: sqlite 3.22.0 []
# Sepal.Length Sepal.Width Petal.Length Petal.Width Species x y z a
# <dbl> <dbl> <dbl> <dbl> <chr> <dbl> <dbl> <dbl> <chr>
# 1 5.1 3.5 1.4 0.2 setosa 17.8 3.5 7 foo
# 2 4.9 3 1.4 0.2 setosa 14.7 3 6 setosasetosa
# 3 4.7 3.2 1.3 0.2 setosa 15.0 3.2 6.4 foo
实施例4: 按组更新
tbl(con,"iris") %>%
group_by(Species, Petal.Width) %>%
update(new_col1 = sum(Sepal.Width,na.rm=TRUE), # using a R function
new_col2 = MAX(Sepal.Length)) # using native SQL
# # Source: SQL [?? x 11]
# # Database: sqlite 3.22.0 []
# Sepal.Length Sepal.Width Petal.Length Petal.Width Species x y z a new_col1 new_col2
# <dbl> <dbl> <dbl> <dbl> <chr> <dbl> <dbl> <dbl> <chr> <dbl> <dbl>
# 1 5.1 3.5 1.4 0.2 setosa 1 2 7 foo 6.5 5.1
# 2 4.9 3 1.4 0.2 setosa 1 2 6 setosasetosa 6.5 5.1
# 3 7 3.2 4.7 1.4 versicolor 1 2 6.4 foo 3.2 7
一般注意事项
代码使用用途dbplyr::translate_sql所以我们可以像以前一样使用 R 函数或本机函数mutate calls.
update只能在1次之后使用filter致电或一group_by调用或每个零，其他任何东西，你都会得到错误或意外的结果。
The group_by实现非常hacky，因此没有空间动态定义列或按操作分组，请坚持基础知识。
update and create都返回tbl(con, table_name)，这意味着您可以链接尽可能多的create or update根据您的意愿拨打电话，并支付适当的金额group_by and filter之间。事实上，我的 4 个例子都可以链接起来。
为了钉钉子，create不受同样的限制，您可以拥有尽可能多的dbplyr在调用之前按需要有趣。
我没有实现类型检测，所以我需要new_type参数，它被回收在paste的呼唤alter_queries我的代码中的定义，因此它可以是单个值或向量。
解决后者的一种方法是从translations变量，找到它们的类型dbGetQuery(con,"PRAGMA table_info(iris)")。然后我们需要所有现有类型之间的强制规则，我们已经设置好了。但由于不同的 DBMS 有不同的类型，我想不出通用的方法来做到这一点，我也不知道MonetDBLite.
