This example takes the same input and produces the same output as in Example 3: Custom R Function.
Instead of defining an R function, users create an R script that reads data in from a file and writes the output to a table. Users then use the runner function ta.source() to execute the script.
-
Create the R script "fruit_1104b.R" as shown here.
# Script file is "fruit_1104b.R" # File descriptor pointing to standard input in_table = file(description="stdin",open="r") # Read column values from stdin into a vector, nrows = -1 means read all rows while(1){ fields <- try(read.table(in_table,header=FALSE,sep="\t",quote="",nrows=-1) silent=TRUE ) if(inherits(fields,"try-error")) break #Fields are from table Fruit_sales s_date <- as.character(fields[,1]) orderid <- as.integer(fields[,2]) itemid <- as.character(fields[,3]) cost <- as.numeric(fields[,4]) salesprice <- as.numeric(fields[,5]) # do work profit <- salesprice - cost i <- data.frame(s_date, orderid, itemid, cost, salesprice, profit) DailySales <- aggregate(salesprice~s_date, i, sum) DailyProfit <- aggregate(profit~s_date, i, sum) outrec <- data.fame(DailySales[,1], DailySales[,2], DailyProfit[,2]) #write to standard output stream write.table(outrec, stdout(), col.names=FALSE, row.names=FALSE, quote=FALSE, sep="\t") } # END -
Create a table in the database based on the input data, and convert it to a virtual data frame.
ta.create(Fruit_sales, table="fruit_tmp", schemaName="public", tableType="dimension", row.names=FALSE, colTypes=NULL ) tadf.fruit <- ta.data.frame("fruit_tmp")
-
Install the script on the cluster.
ta.install.scripts("fruit_1104b.R") -
Create the vector defining the output column names and types.
outputs2 <- c("Date Date", "DailySales numeric", "DailyProfit numeric") -
Use the Aster R runner function ta.source() to run the script.
ta.source(data = tadf.fruit, script = "fruit_1104b.R", outputs = outputs2) Date Total Sales Total Profit 1 2010-05-01 29.80 5.65 2 2010-05-02 31.78 6.80 3 2010-05-03 19.25 4.72 4 2010-05-04 20.57 5.72