"One or more parsing issues" warning but problems() prints nothing

[zackw]

Consider the following CSV file. (This is cut down from an actual grade sheet exported from Canvas, with all the real data replaced by fake values, preserving structure.)

Name,ID,Assignment 1,Assignment 2,Assignment 3
,,Manual Posting,Manual Posting,
Points Possible,,25,25,40
Able,able@example.com,12,24,39
Baker,baker@example.com,25,25,37
Charlie,charlie@example.com,22,10,37

The line with "Manual Posting" in the first two assignment columns is junk -- I have no idea what it means or why it's even there -- but naturally enough it makes read_csv use character vectors for those columns. That's not what I want so I give a column type spec. No problem there.

The bug is that when I give a column type spec, I get a "One or more parsing issues" warning but problems() prints nothing:

> read_csv("grades.csv", col_types="ccddd")
# A tibble: 5 x 5
  Name            ID                `Assignment 1` `Assignment 2` `Assignment 3`
  <chr>           <chr>                      <dbl>          <dbl>          <dbl>
1 NA              NA                            NA             NA             NA
2 Points Possible NA                            25             25             40
3 Able            able@example.com              12             24             39
4 Baker           baker@example.com             25             25             37
5 Charlie         charlie@example.~             22             10             37
Warning message:
One or more parsing issues, see `problems()` for details 
> problems()
>

[zackw]

p.s. I imagine that the complaint that's not getting passed through is something like this:

> parse_double(c("", "Manual Posting", "12", "24"))
Warning: 1 parsing failure.
row col expected         actual
  2  -- a double Manual Posting

[1] NA NA 12 24
attr(,"problems")
# A tibble: 1 x 4
    row   col expected actual        
  <int> <int> <chr>    <chr>         
1     2    NA a double Manual Posting

It would be nice to be able to squelch this specifically. parse_double takes an argument with a list of strings to treat as NA with no complaint, but I don't see any way to pass that down from a cols() specification. I can supply that list directly to read_csv but then it applies to all columns which may be too broad. Maybe col_* could pass na=/locale=/trim_ws= arguments down to the corresponding parse_* function?

[sbearrows]

It looks like you are using problems() incorrectly. The documentation for problems says it requires a data frame to check for problems. But we agree that the output from read_csv() is a little confusing so we have an open issue here #1322 to update this warning message.

Having column specific NA options is also something we have considered but is not currently on the top of our todo list, but maybe in the future!

[zackw]

The documentation says that if you don't supply an argument to problems() it defaults to using .Last.value, so the demo I provided in the original bug report should have worked for you. Moreover, if I do supply an argument to problems() that doesn't change anything.

> d <- read_csv("grades.csv", col_types="ccddd")

Warning message:                   
One or more parsing issues, see `problems()` for details 
> problems(d)
>

Please reopen.

[jennybc]

I cannot reproduce this:

> library(readr)
> read_csv("grades.csv", col_types="ccddd")
# A tibble: 5 × 5                                                                                                             
  Name            ID                  `Assignment 1` `Assignment 2` `Assignment 3`
  <chr>           <chr>                        <dbl>          <dbl>          <dbl>
1 NA              NA                              NA             NA             NA
2 Points Possible NA                              25             25             40
3 Able            able@example.com                12             24             39
4 Baker           baker@example.com               25             25             37
5 Charlie         charlie@example.com             22             10             37
Warning message:
One or more parsing issues, see `problems()` for details 
> problems()
# A tibble: 2 × 5
    row   col expected actual         file                             
  <int> <int> <chr>    <chr>          <chr>                            
1     2     3 a double Manual Posting /Users/jenny/rrr/readr/grades.csv
2     2     4 a double Manual Posting /Users/jenny/rrr/readr/grades.csv

> d <- read_csv("grades.csv", col_types="ccddd")
Warning message:                                                                                                              
One or more parsing issues, see `problems()` for details 
> problems(d)
# A tibble: 2 × 5
    row   col expected actual         file                             
  <int> <int> <chr>    <chr>          <chr>                            
1     2     3 a double Manual Posting /Users/jenny/rrr/readr/grades.csv
2     2     4 a double Manual Posting /Users/jenny/rrr/readr/grades.csv