根据行 ID 将 R 中的值写入 PostgreSQL 表

Question

I have a PostgreSQL table Scores on a local server that looks like this:

ID   Score_X   Score_Y
1       NA        NA
2       NA        NA
3       NA        NA
4       NA        NA

I do a series of calculations in R that produces a dataframe Calc_Scores that looks like this:

ID   Score_X   Score_Y
1      0.53      0.81
4      0.75      0.95

I would like to write the scores that correspond with each ID from R to the PostgreSQL table such that the final PostgreSQL table should look like this:

ID   Score_X   Score_Y
1      0.53      0.81
2       NA        NA
3       NA        NA
4      0.75      0.95

I have a connection to the PostgreSQL table called connection which I setup using the function dbConnect() . The actual tables are quite big. What line/code in R could I use to write these scores to the PostgreSQL table? I have been looking for a similar question but couldn't find anything. I have tried

dbWriteTable(connection, "Scores", value = Calc_Scores, overwrite=T, append = F, row.names = F)

However, the entire table gets overwritten. I want only the scores to be updated.

Thank you.

Answer 1

Creating a temporary table could be an option:

# Create temporary table
dbWriteTable(connection, "ScoresTmp", value = Calc_Scores, overwrite=T, append = F, row.names = F)

# Update main table
dbExecute(connection,"
UPDATE Scores
SET Score_X = ScoresTmp.Score_X,
    Score_Y = ScoresTmp.Score_Y
FROM ScoresTmp 
WHERE Scores.ID = ScoresTmp.ID
")

# Clean up
dbExecute(connection,"DROP TABLE ScoresTmp")

Note that you should be able to create a real temporary table using the temporary=TRUE option: according to @Sirius comment below, it should work on a PostGreSQL database.
For users of an SQLServer database, this option doesn't work, but they can use the # prefix to create a temporary table.
In the example above, this would be:

 dbWriteTable(connection, "#ScoresTmp", value = Calc_Scores, overwrite=T, append = F, row.names = F)

Answer 2

One way of doing this relies on the SQL 'update' and in essence you do

- open a connection to your database
- loop over your changeset and for each row
    - form the update statement, i.e. for example via
      cmd <- paste('update table set x=', Score_x, ', y=', 
                   Score_y, ' where id=', id)
    - submit the cmd via eg `dbSendQuery`
- close the connection

There are examples in RPostgreSQL .