How do I handle empty data frames when i'm using write.xlsx within a loop?
Below is what the loop looks like, where source("./Scripts/Analysis_details.R") is referencing the r file where the data frames are created.
library(xlsx)
for (A in unique(df_base$A)) {
df<- df_base[df_base$A==A,]
source("./Scripts/Analysis_details.R")
output_file = paste("./Output/report_", A, '_', Sys.Date(), ".xlsx", sep='')
write.xlsx(df1, file=output_file, sheetName="df1", append=TRUE, row.names=FALSE, showNA = FALSE)
write.xlsx(df2, file=output_file, sheetName="df2", append=TRUE, row.names=FALSE, showNA = FALSE)
write.xlsx(df3, file=output_file, sheetName="df3", append=TRUE, row.names=FALSE, showNA = FALSE)
write.xlsx(df4, file=output_file, sheetName="df4", append=TRUE, row.names=FALSE, showNA = FALSE)}
The error that I'm getting is...
Error in mapply(setCellValue, cells[seq_len(nrow(cells)), colIndex[ic]], : zero-length inputs cannot be mixed with those of non-zero length
I was able to work around this by putting a modified write.xlsx()
function into my script. If the data frame has zero rows, .write_block()
is skipped and and the file is saved with just the column names. Note that you'll also have to copy the original .write_block()
function into your script as well.
write.xlsx.custom <- function(x, file, sheetName="Sheet1",
col.names=TRUE, row.names=TRUE, append=FALSE, showNA=TRUE)
{
if (!is.data.frame(x))
x <- data.frame(x) # just because the error message is too ugly
iOffset <- jOffset <- 0
if (col.names)
iOffset <- 1
if (row.names)
jOffset <- 1
if (append && file.exists(file)){
wb <- loadWorkbook(file)
} else {
ext <- gsub(".*\\.(.*)$", "\\1", basename(file))
wb <- createWorkbook(type=ext)
}
sheet <- createSheet(wb, sheetName)
noRows <- nrow(x) + iOffset
noCols <- ncol(x) + jOffset
if (col.names){
rows <- createRow(sheet, 1) # create top row
cells <- createCell(rows, colIndex=1:noCols) # create cells
mapply(setCellValue, cells[1,(1+jOffset):noCols], colnames(x))
}
if (row.names) # add rownames to data x
x <- cbind(rownames=rownames(x), x)
if(nrow(x) > 0) {
colIndex <- seq_len(ncol(x))
rowIndex <- seq_len(nrow(x)) + iOffset
.write_block(wb, sheet, x, rowIndex, colIndex, showNA)
}
saveWorkbook(wb, file)
invisible()
}
I found the original functions at https://github.com/cran/xlsx/blob/master/R/write.xlsx.R