Merge pull request h2oai#17 from earlh/elh

R: pub-215 passes

R/tests/testdir_jira/runit_pub_180_ddply.R

@@ -35,11 +35,20 @@ ddplytest <- function(conn){
   h2o.addFunction(conn, fn4)
   df.h.4 <- ddply(df.h, .(colgroup, colgroup2), fn4)
 
+  Log.info('testing all column address modes')
+  df.h.4b <- ddply(df.h, c('colgroup', 'colgroup2'), fn4)
+  df.h.4c <- ddply(df.h, 1:2, fn4)
+  df.h.4d <- ddply(df.h, h2o..('colgroup', 'colgroup2'), fn4)
+
+
   Log.info('pulling data locally')
   df.1 <- as.data.frame( df.h.1 )
   df.2 <- as.data.frame( df.h.2 )
   df.3 <- as.data.frame( df.h.3 )
   df.4 <- as.data.frame( df.h.4 )
+  df.4b <- as.data.frame( df.h.4b )
+  df.4c <- as.data.frame( df.h.4c )
+  df.4d <- as.data.frame( df.h.4d )
 
   Log.info('avoid factor issues by making grouping columns into character')
   df.1$colgroup <- as.character(df.1$colgroup)
@@ -49,13 +58,22 @@ ddplytest <- function(conn){
   df.2$colgroup2 <- as.character(df.2$colgroup2)
   df.4$colgroup2 <- as.character(df.4$colgroup2)
 
+  df.4b$colgroup <- as.character(df.4b$colgroup)
+  df.4b$colgroup2 <- as.character(df.4b$colgroup2)
+  df.4c$colgroup <- as.character(df.4c$colgroup)
+  df.4c$colgroup2 <- as.character(df.4c$colgroup2)
+  df.4d$colgroup <- as.character(df.4d$colgroup)
+  df.4d$colgroup2 <- as.character(df.4d$colgroup2)
 
 
   # h2o doesnt sort
   df.1 <- df.1[order(df.1$colgroup), ]
   df.2 <- df.2[order(df.2$colgroup, df.2$colgroup2), ]
   df.3 <- df.3[order(df.3$colgroup), ]
   df.4 <- df.4[order(df.4$colgroup, df.4$colgroup2), ]
+  df.4b <- df.4b[order(df.4b$colgroup, df.4b$colgroup2), ]
+  df.4c <- df.4c[order(df.4c$colgroup, df.4c$colgroup2), ]
+  df.4d <- df.4d[order(df.4d$colgroup, df.4d$colgroup2), ]
 
   Log.info('testing')
   expect_that( dim(df.1), equals( c(3,2) ) )
@@ -79,6 +97,11 @@ ddplytest <- function(conn){
   expect_that(df.4[,2], equals(paste('group', c(1,1,3,1,2), sep='')) )
   expect_that(df.4[,3], equals(c(3,7,18,11,11)) )
 
+  # column addressing options
+  expect_that( all(df.4b == df.4), equals(T))
+  expect_that( all(df.4c == df.4), equals(T))
+  expect_that( all(df.4d == df.4), equals(T))
+
 
   testEnd()
 }

R/tests/testdir_jira/runit_NOPASS_pub_215_filenamesdash.R → R/tests/testdir_jira/runit_pub_215_filenamesdash.R

@@ -14,7 +14,7 @@ dash_filename_test <- function(conn){
   Log.info('printing from h2o')
   Log.info( head(df.h) )
 
-  res <- as.data.frame(table(df.h$l>0))
+  res <- as.data.frame(h2o.table(df.h$l>0))
 
   testEnd()
 }

R/tests/testdir_jira/runit_v_11_datemanipulation.R

@@ -0,0 +1,76 @@
+#
+# date parsing and field extraction tests
+#
+
+
+setwd(normalizePath(dirname(R.utils::commandArgs(asValues=TRUE)$"f")))
+source('../findNSourceUtils.R')
+
+
+
+datetest <- function(conn){
+
+  Log.info('uploading date testing dataset')
+  hdf <- h2o.importFile(conn, locate('smalldata/jira/v-11.csv'))
+  # df should be 5 columns: ds1:5
+
+  Log.info('data as loaded into h2o:')
+  Log.info(head(hdf))
+
+  # NB: columns 1,5 are currently unsupported as date types
+  # that is, h2o cannot understand:
+  # 1 integer days since epoch (or since any other date);
+  # 2 dates formatted as %d/%m/%y (in strptime format strings)
+
+  Log.info('adding date columns')
+  # NB: h2o automagically recognizes and if it doesn't recognize, you're out of luck
+
+  Log.info('extracting year and month from posix date objects')
+  hdf$year2 <- year(hdf$ds2)
+  hdf$year3 <- year(hdf$ds3)
+  hdf$year4 <- year(hdf$ds4)
+  hdf$mon2 <- month(hdf$ds2)
+  hdf$mon3 <- month(hdf$ds3)
+  hdf$mon4 <- month(hdf$ds4)
+  hdf$idx2 <- year(hdf$ds2) * 12 + month(hdf$ds2)
+  hdf$idx3 <- year(hdf$ds3) * 12 + month(hdf$ds3)
+  hdf$idx4 <- year(hdf$ds4) * 12 + month(hdf$ds4)
+
+  cc <- colnames(hdf)
+  nn <- c( paste('year', 2:4, sep=''), paste('month', 2:4, sep=''), paste('idx', 2:4, sep='') )
+  cc[ (length(cc) - length(nn) + 1):length(cc) ] <- nn
+  colnames(hdf) <- cc
+
+  Log.info('pulling year/month indices local')
+  ldf <- as.data.frame( hdf )
+
+  # build the truth using R internal date fns
+  rdf <- read.csv(locate('smalldata/jira/v-11.csv'))
+  rdf$days1 <- as.Date(rdf$ds1, origin='1970-01-01')
+  rdf$days2 <- as.Date(rdf$ds2, format='%Y-%m-%d')
+  rdf$days3 <- as.Date(rdf$ds3, format='%d-%b-%y')
+  rdf$days4 <- as.Date(rdf$ds4, format='%d-%B-%Y')
+  rdf$days5 <- as.Date(rdf$ds5, format='%d/%m/%y')
+
+  months <- data.frame(lapply(rdf[,6:10], function(x) as.POSIXlt(x)$mon))
+  years <- data.frame(lapply(rdf[,6:10], function(x) as.POSIXlt(x)$year))
+  idx <- 12*years + months
+
+  Log.info('testing correctness')
+  expect_that( ldf$year2, equals(years[,2]) )
+  expect_that( ldf$year3, equals(years[,3]) )
+  expect_that( ldf$year4, equals(years[,4]) )
+
+  expect_that( ldf$month2, equals(months[,2]) )
+  expect_that( ldf$month3, equals(months[,3]) )
+  expect_that( ldf$month4, equals(months[,4]) )
+
+  expect_that( ldf$idx2, equals(idx[,2]) )
+  expect_that( ldf$idx3, equals(idx[,3]) )
+  expect_that( ldf$idx4, equals(idx[,4]) )
+
+  testEnd()
+}
+
+
+doTest('date testing', datetest)

R/tests/testdir_munging/slice/runit_[[.R

@@ -0,0 +1,35 @@
+##
+##
+
+setwd(normalizePath(dirname(R.utils::commandArgs(asValues=TRUE)$"f")))
+source('../../findNSourceUtils.R')
+
+test.columndereference <- function(conn) {
+  Log.info('test column dereference')
+
+  hdf <- h2o.uploadFile.FV(conn, locate('smalldata/jira/pub-180.csv'))
+  otherhdf <- h2o.uploadFile.FV(conn, locate('smalldata/jira/v-11.csv'))
+
+  column <- 'colgroup2'
+
+  # get a single column out
+  expect_that( dim(hdf[['colgroup2']]), equals(c(12,1)) )
+  expect_that( dim(hdf[[column]]), equals(c(12,1)) )
+
+  # NULL if column name doesn't exist
+  expect_that( hdf[['col2group2']], equals(NULL))
+
+  # we can overwrite a column
+  hdf[['colgroup2']] <- hdf[['col2']]
+  ldf <- as.data.frame( hdf[[ column ]] )[,1]
+  expect_that(ldf, equals(c(2,4,6,11,3,4,6,11,2,4,6,11)) )
+
+  # don't overwrite with the wrong number of rows
+  expect_that( hdf[[column]] <- otherhdf[['ds1']], throws_error())
+  expect_that( hdf[[column]] <- hdf[,2:3], throws_error() )
+
+  testEnd()
+}
+
+doTest("test column dereference and assignment", test.columndereference)
+

R/tests/testdir_munging/slice/runit_cbind.R

@@ -0,0 +1,35 @@
+##
+##
+
+setwd(normalizePath(dirname(R.utils::commandArgs(asValues=TRUE)$"f")))
+source('../../findNSourceUtils.R')
+
+test.cbind <- function(conn) {
+  Log.info('test cbind')
+
+  hdf <- h2o.uploadFile(conn, locate('../../../smalldata/jira/pub-180.csv'))
+  otherhdf <- h2o.uploadFile(conn, locate('../../../smalldata/jira/v-11.csv'))
+
+  ##### WORKS #####
+  # cbind self to self
+  hdf2 <- cbind(hdf, hdf)
+  expect_that( dim(hdf2), equals(c(12, 8)) )
+
+  # cbind a sliced column to a sliced column
+  xx <- hdf[,1]
+  yy <- hdf[,2]
+  expect_that( dim(cbind(xx,yy)), equals(c(12,2)) )
+
+  # cbind unequal rows fails
+  expect_that(cbind(hdf, otherhdf), throws_error())
+
+
+  ##### FAILS #####
+  # cbind a df to a slice
+  #expect_that( dim(cbind(hdf, hdf[,1])), equals(12,5) )
+
+  testEnd()
+}
+
+doTest("test cbind", test.cbind)
+