Run a single step in a workflow.

Runs a single step of an assessment workflow. This function is called by RunWorkflow for each step in the workflow. It prepares the parameters for the function call and then calls the function specified in lStep$name with the prepared parameters.

The primary utility of this function is to provide a prioritized parser for function parameterization. Parameters should be specified as a named list in lStep$params, where each element is a key-value pair that will be parsed and then passed to the specified function as a set of parameter names/values. Parameter values should be specified as scalar strings. Those values are then pulled from lMeta or lData when possible. When no matching lData or lMeta objects are found, parameter values are passed through as strings. Note that parsing vectorized parameters is not supported at this time; they are passed directly as character vectors. To pass a vector or list, we recommend saving it as an object in lData.

Full prioritization for parsing parameters is below:

If a single parameter value is equal to "lMeta", the the full lMeta object is passed to the function (for the given paramName).
If a single parameter value is equal to "lData", the full lData object is passed to the function.
If a single parameter value is equal to "lSpec", the full lSpec object is passed to the function.
If a single parameter value is found in names(lMeta), that property is pulled from lMeta (e.g. lMeta${paramVal}) and passed to the function.
If a single parameter value is found in names(lData), that property is pulled from lData (e.g. lData${paramVal}) and passed to the function.
Otherwise single parameter value is passed to the function as a string.
If the parameter value is a vector, the vector is passed to the function as a vector or strings.

Usage

RunStep(lStep, lData, lMeta, lSpec = NULL)

Arguments

lStep: list single workflow step (typically pulled from lWorkflow$steps). Should include the name of the function to run (lStep$name), name of the object where the function result should be saved (lStep$output) and configurable parameters (lStep$params) (if any)
lData: list a named list of domain level data frames.
lMeta: list a named list of meta data.
lSpec: list a data specification containing required columns. See vignette("gsm_extensions").

Value

list containing the results of the lStep$name function call should contain .$checks parameter with results from is_mapping_vald for each domain in lStep$inputs.

Examples

wf_mapping <- MakeWorkflowList(strPath = "workflow/1_mappings")
lStep <- MakeWorkflowList("kri0001")[["kri0001"]][["steps"]][[1]]
lMeta <- MakeWorkflowList("kri0001")[["kri0001"]][["meta"]]
lData <- list(
  dfSUBJ = clindata::rawplus_dm,
  dfAE = clindata::rawplus_ae,
  dfPD = clindata::ctms_protdev,
  dfCONSENT = clindata::rawplus_consent,
  dfIE = clindata::rawplus_ie
)
lMapped <- RunWorkflow(wf_mapping, lData)$lData
#> 
#> ── Initializing `_` Workflow ───────────────────────────────────────────────────
#> → Workflow `_` has no `steps` property.
#> → Workflow `_` has no `meta` property.
#> 
#> ── No spec found in workflow. Proceeding without checking data. 
#> 
#> ── Returning results from final step: NULL of length 0`. ──
#> 
#> ── Completed `_` Workflow ──────────────────────────────────────────────────────

ae_step <- RunStep(lStep = lStep, lData = lMapped, lMeta = lMeta)
#> 
#> ── Evaluating 1 parameter(s) for `ParseThreshold()` 
#> ✔ strThreshold = Threshold: Passing lMeta$Threshold.
#> 
#> ── Calling `ParseThreshold()` 
#> ✔ Parsed -2,-1,2,3 to numeric vector: -2, -1, 2, and 3