What I try to do is that bind rows in my PostgreSQL databes in matched columns like rbindlist's (from data.table) fill argument.

In short, the table I'd like to see in my database is like this;

a <- data.frame(no=c(234,235,346),year=2012:2014,col1=c(1,1,1))
b <- data.frame(no=c(333,353,324),year=2014:2016,col2=c(2,2,2))
union_data_frame <- data.table::rbindlist(list(a,b),fill=T)

union_data_frame

     no  year  col1  col2
1   234  2012     1    NA
2   235  2013     1    NA
3   346  2014     1    NA
4   333  2014    NA     2
5   353  2015    NA     2
6   324  2016    NA     2

I tried it in RPostgres in this way;

library(RPostgres)

a <- data.frame(no=c(234,235,346),year=2012:2014,col1=c(1,1,1))
b <- data.frame(no=c(333,353,324),year=2014:2016,col2=c(2,2,2))

drv <- dbDriver('Postgres')
con <- dbConnect(drv,user='postgres',dbname='dummy_db')


dbWriteTable(con,'dummy_table',a,append = T,row.names = F)
dbWriteTable(con,'dummy_table',b,append = T,row.names = F)

But it doesn't work and fields an error because the second table (b) doesn't have a column called col2.

How to append tables by only common columns ?

Thanks in advance.

CodePudding user response：

I think you need to:

1. identify which columns are missing,
2. alter table those new columns into existence, and then
3. upload the data, assuming all data in the first that are missing in the second are null-able.

### pg <- dbConnect(...)
dbWriteTable(pg, "some_table", a)
newcolumns <- setdiff(colnames(b), dbListFields(pg, "a"))
newcolumns
# [1] "col2"
addqry <- paste("alter table some_table",
                paste("add", newcolumns, dbDataType(pg, b[,newcolumns]),
                      collapse = ", "))
addqry
# [1] "alter table some_table add col2 DOUBLE PRECISION"
dbExecute(pg, addqry)
dbWriteTable(pg, "some_table", b, append = TRUE)
dbGetQuery(pg, "select * from some_table")
#    no year col1 col2
# 1 234 2012    1   NA
# 2 235 2013    1   NA
# 3 346 2014    1   NA
# 4 333 2014   NA    2
# 5 353 2015   NA    2
# 6 324 2016   NA    2