From f4767bcc7a9d1bdd301f054776aa45e7c9f344a7 Mon Sep 17 00:00:00 2001 From: Felix Cheung Date: Thu, 7 Jul 2016 15:21:57 -0700 Subject: [SPARK-16310][SPARKR] R na.string-like default for csv source ## What changes were proposed in this pull request? Apply default "NA" as null string for R, like R read.csv na.string parameter. https://stat.ethz.ch/R-manual/R-devel/library/utils/html/read.table.html na.strings = "NA" An user passing a csv file with NA value should get the same behavior with SparkR read.df(... source = "csv") (couldn't open JIRA, will do that later) ## How was this patch tested? unit tests shivaram Author: Felix Cheung Closes #13984 from felixcheung/rcsvnastring. --- R/pkg/inst/tests/testthat/test_sparkSQL.R | 32 +++++++++++++++++++++++++------ 1 file changed, 26 insertions(+), 6 deletions(-) (limited to 'R/pkg/inst') diff --git a/R/pkg/inst/tests/testthat/test_sparkSQL.R b/R/pkg/inst/tests/testthat/test_sparkSQL.R index a3aa26d9e7..a0ab719202 100644 --- a/R/pkg/inst/tests/testthat/test_sparkSQL.R +++ b/R/pkg/inst/tests/testthat/test_sparkSQL.R @@ -213,15 +213,35 @@ test_that("read csv as DataFrame", { mockLinesCsv <- c("year,make,model,comment,blank", "\"2012\",\"Tesla\",\"S\",\"No comment\",", "1997,Ford,E350,\"Go get one now they are going fast\",", - "2015,Chevy,Volt") + "2015,Chevy,Volt", + "NA,Dummy,Placeholder") writeLines(mockLinesCsv, csvPath) - # default "header" is false - df <- read.df(csvPath, "csv", header = "true") - expect_equal(count(df), 3) + # default "header" is false, inferSchema to handle "year" as "int" + df <- read.df(csvPath, "csv", header = "true", inferSchema = "true") + expect_equal(count(df), 4) expect_equal(columns(df), c("year", "make", "model", "comment", "blank")) - expect_equal(sort(unlist(collect(where(df, df$year == "2015")))), - sort(unlist(list(year = "2015", make = "Chevy", model = "Volt")))) + expect_equal(sort(unlist(collect(where(df, df$year == 2015)))), + sort(unlist(list(year = 2015, make = "Chevy", model = "Volt")))) + + # since "year" is "int", let's skip the NA values + withoutna <- na.omit(df, how = "any", cols = "year") + expect_equal(count(withoutna), 3) + + unlink(csvPath) + csvPath <- tempfile(pattern = "sparkr-test", fileext = ".csv") + mockLinesCsv <- c("year,make,model,comment,blank", + "\"2012\",\"Tesla\",\"S\",\"No comment\",", + "1997,Ford,E350,\"Go get one now they are going fast\",", + "2015,Chevy,Volt", + "Empty,Dummy,Placeholder") + writeLines(mockLinesCsv, csvPath) + + df2 <- read.df(csvPath, "csv", header = "true", inferSchema = "true", na.string = "Empty") + expect_equal(count(df2), 4) + withoutna2 <- na.omit(df2, how = "any", cols = "year") + expect_equal(count(withoutna2), 3) + expect_equal(count(where(withoutna2, withoutna2$make == "Dummy")), 0) unlink(csvPath) }) -- cgit v1.2.3