Check a report's own code against the rules

A report's data step or content step may be the user's own R code (the report sheet's data_code / content_code: a step whose code is not blank is its code). The code follows rules the program around it relies on: the datasets it reads are the report's (registered in its data step: the program reads them first, by their names in lower case), this report's ARD rows are ard when the report reads its ARD, and the step leaves the object the next one reads – the data step data (a table or a listing; a table's may leave ard alone, which the program normalizes) or df (a figure), the content step content. The check reads the code; it does not run it.

Usage

tfl_check_code(
  code,
  step = c("data", "content"),
  type = c("table", "listing", "figure"),
  datasets = character(),
  catalog = character(),
  ard = FALSE,
  output_id = NA_character_
)

Arguments

code

The code: a string (lines joined by newlines) or a character vector of lines.

step

"data" or "content".

type

The report's type: "table", "listing" or "figure".

datasets

The datasets the report registers (any case).

catalog

The datasets of the study's catalog (any case): a name of these the code uses and does not make itself must be registered.

ard

Whether the report reads its ARD (ard is in reach).

output_id

The report, for the messages.

Value

A tfl_review (tfl_review_spec()'s rows: sheet report, field data_code / content_code), with the attribute datasets: the catalog's datasets the code uses (upper case), registered or not.

Details

The rules (tfl_review_rules(), area program):

• U01 (error): the code does not read as R.

• U02 (error): it names a dataset of the catalog the report does not register.

• U03 (error): it leaves nothing for the next step (no assignment to data / df / content).

• U04 (check): it reads files itself, attaches packages or changes the session (readRDS(), haven::read_xpt(), library(), setwd(), options() ...).

• U05 (error): it uses ard, but the report does not read its ARD.

Examples

code <- "df <- advs |> dplyr::filter(PARAMCD == 'SYSBP')"
r <- tfl_check_code(code, "data", "figure", datasets = "ADSL",
                    catalog = c("ADSL", "ADVS"), output_id = "F-1")
r[, c("rule", "level", "message")]
#> <tfl_review> 1 error, 0 to check, 0 to set by hand
#>   
attr(r, "datasets")
#> [1] "ADVS"