Review a study's definition

Lists, for the whole study, what is wrong with its definition (error: it cannot be used – every problem the constructors tfl_table_spec(), tfl_ard_spec(), tfl_listing_spec() and tfl_check_fig_design() stop or report on), what is valid but probably wrong (check), and what is missing and has to be set by hand (hand). Nothing stops: a definition that cannot be built is reviewed as the sheets it is. With facts, the definition is also checked against what the data hold (tfl_data_facts(): a column an analysis reads that its data do not have, a condition that keeps nothing, a code list against the values) and against what each report's ARD holds (facts$ard[[output_id]], tfl_ard_facts()). The rules are tfl_review_rules().

Usage

tfl_review_spec(
  spec = NULL,
  ard = NULL,
  listings = NULL,
  figures = NULL,
  output_id = NULL,
  facts = NULL,
  rules = NULL
)

Arguments

spec

The table definition: a tfl_table_spec() or a list of its sheets (tables, variables, cells ...).

ard

The ARD definition: a tfl_ard_spec() or a list of its five sheets.

listings

The listing definition: a tfl_listing_spec() or a list of its two sheets.

figures

The figures' designs: a list of tfl_fig_design()s (or their lists) named by their reports.

output_id

Only these reports; NULL reviews every report the sheets name, and the study-wide rows (output_id NA: the defaults, the data).

facts

The facts of the data (tfl_data_facts()), with the reports' ARD facts in facts$ard; NULL runs the rules that need the definition only.

rules

Only these rules: their ids ("T03") or areas ("table").

Value

A tfl_review: a data frame, one row per item – output_id, level (error, check, hand), area (report, codelists, ard, ard_run, table, listing, figure, data, spec), sheet, row (the row's key, as the sheet is keyed: the variable, the analysis_id, variable / context / row on cells), field (the column), message, template (the sentence the values fill: the rule's message or one of tfl_review_templates()), hint, rule, draft (FALSE: set by an import for the rows it brings), args (the values that fill template, for a translation) and fix (a one-step fix, where there is one: tfl_fig_apply_fix()). Errors first, then checks, then what to set by hand; within a level by report and sheet.

See also

tfl_review_rules(), tfl_data_facts(), tfl_ard_facts()

Examples

ard <- list(
  datasets = data.frame(dataset = "ADSL", path = "adsl.rds"),
  populations = data.frame(population_id = "SAF", dataset = "ADSL",
                           where = 'SAFFL == "Y"'),
  analyses = data.frame(output_id = "T-1", analysis_id = "AGE",
                        method = "continuous", population_id = "SAF",
                        dataset = "ADSL", by = "TRT01A",
                        variables = "AGE | TRT01A"))
r <- tfl_review_spec(ard = ard)
r[, c("output_id", "level", "rule", "message")]
#> <tfl_review> 0 errors, 1 to check, 0 to set by hand
#>   
summary(r)
#>   output_id error check hand
#> 1       T-1     0     1    0