catch all NULL in entry data

[erblast]

library(tsoa)
#> Loading required package: dplyr
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union

data("tsoa_data", package = "tsoa")

feats <- c(
  "autocorr",
  "average",
  "own_site_simil_score",
  "sd",
  "unique_value_count_relative",
  "lof",
  "range"
)

feats_collapse = paste(feats, collapse = ";")

data("tsoa_data", package = "tsoa")

tsoa_data$data <- tsoa_data$data %>%
  mutate(
    result = rep(as.numeric(NA), nrow(.))
  )

ls_tsoa <- process_a_study(
  data = tsoa_data$data,
  subjects = tsoa_data$subjects,
  parameters = tsoa_data$parameters,
  custom_timeseries = tsoa_data$custom_timeseries,
  custom_reference_groups = tsoa_data$custom_reference_groups,
  default_timeseries_features_to_calculate = feats_collapse,
  default_minimum_timepoints_per_series = 1,
  default_minimum_subjects_per_series = 3,
  default_max_share_missing_timepoints_per_series = 1,
  default_generate_change_from_baseline = FALSE,
  autogenerate_timeseries = TRUE
)
#> Error in `mutate()`:
#> ℹ In argument: `output = list(...)`.
#> Caused by error in `if (this_baseline == "cfb") ...`:
#> ! argument is of length zero
#> Backtrace:
#>      ▆
#>   1. ├─tsoa::process_a_study(...)
#>   2. │ └─... %>% ... at tsoa/R/tsoa.R:65:4
#>   3. ├─dplyr::select(...)
#>   4. ├─dplyr::mutate(...)
#>   5. ├─dplyr::mutate(...)
#>   6. ├─dplyr::ungroup(.)
#>   7. ├─tidyr::unnest(., "output")
#>   8. ├─dplyr::mutate(...)
#>   9. ├─dplyr:::mutate.data.frame(...)
#>  10. │ └─dplyr:::mutate_cols(.data, dplyr_quosures(...), by)
#>  11. │   ├─base::withCallingHandlers(...)
#>  12. │   └─dplyr:::mutate_col(dots[[i]], data, mask, new_columns)
#>  13. │     └─mask$eval_all_mutate(quo)
#>  14. │       └─dplyr (local) eval()
#>  15. ├─tsoa:::pick_timepoint_combos(...)
#>  16. └─base::.handleSimpleError(...) at tsoa/R/tsoa.R:774:2
#>  17.   └─dplyr (local) h(simpleError(msg, call))
#>  18.     └─rlang::abort(message, class = error_class, parent = parent, call = error_call)

^{Created on 2023-12-04 with reprex v2.0.2}

Since we allow NA in general and the standard behaviour is to return a named list with all NULL when no data matches the default criteria I would suggest the following unit test:

test_that("process_a_study - NA only", {

  data("tsoa_data", package = "tsoa")

  tsoa_data$data <- tsoa_data$data %>%
    mutate(
      result = rep(as.numeric(NA), nrow(.))
    )

  ls_tsoa <- process_a_study(
    data = tsoa_data$data,
    subjects = tsoa_data$subjects,
    parameters = tsoa_data$parameters,
    custom_timeseries = tsoa_data$custom_timeseries,
    custom_reference_groups = tsoa_data$custom_reference_groups,
    default_timeseries_features_to_calculate = feats_collapse,
    default_minimum_timepoints_per_series = 1,
    default_minimum_subjects_per_series = 3,
    default_max_share_missing_timepoints_per_series = 1,
    default_generate_change_from_baseline = FALSE,
    autogenerate_timeseries = TRUE
  )

  expect_true(all(purrr::map_lgl(ls_tsoa, is.null)))

})

[erblast]

possible solutions:

• 1. input check for all null and raise informative error
• 2. filter all data and make sure code runs through and named list with all NULL is returned (which is standard behaviour when all data gets filtered due to other timeseries requirements)

[pekkatii]

If test added into the main function which aborts execution if there are no result values in the data.