Example for user of how a combination of blank values may result in error reading a CSV

blanks.R

``` r
library(arrow)
library(dplyr)
library(stringr)

tf <- tempfile()

# values to save - note the space after the final new line
dodgy_vals <- "x,y\n0,1\n ,4"
cat(dodgy_vals)
#> x,y
#> 0,1
#>  ,4
writeLines(dodgy_vals, tf)

open_dataset(tf, format = "csv", schema = schema(x = int64(), y = int64()), skip = 1) %>%
  collect()
#> Error in `compute.Dataset()`:
#> ! Invalid: Could not open CSV input source '/tmp/RtmpF2Lxpf/file8b1d439d4116b': Invalid: In CSV column #0: Row #3: CSV conversion error to int64: invalid value ''
#> ℹ If you have supplied a schema and your data contains a header row, you should supply the argument `skip = 1` to prevent the header being read in as data.


open_dataset(tf, format = "csv", schema = schema(x = int64(), y = int64()), skip = 1, null_values = c(NA, " ", "")) %>%
  collect()
#> # A tibble: 2 × 2
#>       x     y
#>   <int> <int>
#> 1     0     1
#> 2    NA     4
```