Check a rowwise assertion

[maia-sh]

Thanks for this great package! I'm trying to check a rowwise assertion and have gotten stuck. I previously posted this question on SO but am moving here after several days: https://stackoverflow.com/questions/70114227/check-rowwise-assertion-in-r-pointblank

I would like to check a rowwise assertion using {pointblank}. A row should "fail" if it is ALL NA; it should pass if ANY non-NA. In the reprex below, only the second row should fail. How would I write this assertion in {pointblank}? I am able to check each individual columns and whether ANY NA, but am stuck for ALL NA. Thank you for your help!

library(dplyr)
library(pointblank)

df <- tribble(
  ~a, ~b, ~c,
  1, 2, 3,
  NA, NA, NA,
  2, NA, 6
)

agent <-
  df %>%

  create_agent() %>%

  # Is there an NA in each column separately?
  col_vals_not_null(vars(a, b, c)) %>%

  # Is there an NA in any column?
  rows_complete(vars(a, b, c)) %>%

  interrogate()

Created on 2021-11-25 by the reprex package (v2.0.0)

[rich-iannone]

@maia-sh

For this, I'd recommend using the col_vals_expr() function to create an expectation that uses an expression:

library(dplyr)
library(pointblank)

df <- tribble(
  ~a, ~b, ~c,
  1, 2, 3,    # should pass (no NAs)
  NA, NA, NA, # should fail (completely NAs)
  2, NA, 6    # should pass (partial NAs)
)

# Create an agent and use `col_vals_expr()` to evaluate
# a more complex expression on a per-row basis
agent <-
  df %>%
    create_agent() %>%
    col_vals_expr(expr = ~ !(is.na(a) & is.na(b) & is.na(c))) %>%
    interrogate()

# This shows that only the second row fails
get_data_extracts(agent, i = 1)
#> # A tibble: 1 × 3
#>       a     b     c
#>   <dbl> <dbl> <dbl>
#> 1    NA    NA    NA

Unfortunately, rows_complete() doesn't do the job because it expects there aren't any missing values per row for a pass (this is something I could improve in the future with an any/all option).