Too many delimiters in a row causes final column to be "overstuffed" with values

[TMRHarrison]

Vroom doesn't fail, stop, or raise any errors when a file has a row with more columns than expected. Instead, any remaining values (separator and all) are forced into the final column of the output. A warning is given, but it's cryptic.

Take this tsv file:

num     chr     num2    num3
1       charab  123     434
2       charact         345     2345
3               chaaa   3123    1231

The function used and following result:

> vroom::vroom("test.tsv")
Rows: 3 Columns: 4
── Column specification ──────────────────────────────────────────────────────────────────────
Delimiter: "\t"
chr (3): chr, num2, num3
dbl (1): num

ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
# A tibble: 3 × 4
    num chr     num2  num3
  <dbl> <chr>   <chr> <chr>
1     1 charab  123   "434"
2     2 charact NA    "345\t2345"
3     3 NA      chaaa "3123\t1231"
Warning message:
One or more parsing issues, see `problems()` for details
> problems()
Error in vroom_materialize(x, replace = FALSE) :
  argument "x" is missing, with no default

The output I would expect is a more descriptive error, like data.table::fread() gives:

   num    chr num2 num3
1:   1 charab  123  434
Warning message:
In data.table::fread("test.tsv") :
  Stopped early on line 3. Expected 4 fields but found 5. Consider fill=TRUE and comment.char=. First discarded non-empty line: <<2	charact		345	2345>>

And either raising an error, discarding the offending rows, or stopping the read after the first offending row.

[TMRHarrison]

Upon reading into problems(), and passing the returned array as an argument, the error descriptions are sufficient, though a big obfuscated for my tastes. It would still be nice to have the offending rows dealt with in some other way then forcing all values into the last column.

[sbearrows]

Your data would actually be read correctly with readr::read_table() which handles whitespace delimited files with any number of whitespace characters between columns. Unfortunately, we are not currently pursuing replicating this feature in vroom (see https://github.com/tidyverse/vroom/issues/186).

text <- glue::glue(
'x\ty\tz\n
1\t2\t\t3\n
4\t\t5\t6\n')

tf <- withr::local_tempfile(lines = text)

# read_table() handles this messy data
readr::read_table(tf, show_col_types = FALSE)
#> # A tibble: 2 × 3
#>       x     y     z
#>   <dbl> <dbl> <dbl>
#> 1     1     2     3
#> 2     4     5     6

^{Created on 2022-08-26 by the reprex package (v2.0.1.9000)}