fwrite to excel csv

[hope-data-science]

fwrite is fast, but as an Excel user, sometimes the encoding is causing me much trouble. There is such a case, when I write files using fwrite, it is fast can could be read correctly in R, but not in Excel (the Chinese characters are in messy codes). But when I use readr::write_excel_csv, I get the correct format. Is there an option in fwrite to write Excel readable csv?

Thanks.

[shrektan]

fwrite only writes native encoding at present...

[MichaelChirico]

I believe this is a duplicate of #1770 -- please express your support there!

Feel free to re-open if you think the duplication was applied in error 👍

[shrektan]

Recently, I find this particular useful for me and is missing in data.table. This issue is actually a little bit different from #1770. #1770 is about UTF8 and this is about adding a BOM to UTF8 csv file.

I'm reopening this and going to implement this in the near future.

(a simple proto function about adding the BOM)

rewrite_csv_with_bom <- function(path, open = TRUE) {
  # here the open argument must be empty otherwise the content will
  # be converted to native encoding
  # In addition, the encoding must be UTF-8-BOM so that the BOM will
  # be removed (so that it's always safe to add a new BOM later)
  con <- file(path, encoding = 'UTF-8-BOM')
  content <- tryCatch(
    readLines(con), finally = close(con)
  )
  con <- file(path, open = 'wb')
  on.exit(close(con), add = TRUE)
  # we must write the binary to avoid the complexity with R's own internal
  # character handlings
  writeBin(as.raw(c(0xef, 0xbb, 0xbf)), con, useBytes = TRUE)
  writeLines(enc2utf8(content), con, useBytes = TRUE)
  invisible(path)
}

[shrektan]

Sorry, just realized that PR #3580 has already resolved the BOM issues. What we miss here is the ability to write UTF-8 csv files on Windows. On other OS, using fwrite(..., bom=TRUE) is enough.

It's a duplicate of #3488.