Shiny module development for DiagnosticsExplorer

[azimov]

The development of the DiagnosticsExplorer is likely daunting to newer developers due to the complexity of the codebase.
Consequently and new side branch - shiny-modules - is being developed with the intention of simplifying the codebase and making it easier for others to use and contribute.

The simplest example of how a module will work is in the Cohort Counts Module.

Sourcing this file there are two steps to utilize a module.

1. Defining block where the module will live within the shiny UI:

shiny::page(
  ... Insert Inputs for cohort and database selection ... 
   cohortCountsView(id = "counts")
)

3. Inside the calling server function

shiny::shinyServer(function(input, output, session) {
      ... bootstrap code ...

      selectedCohorts <- shiny::reactive({ cohortTable %>% dplyr::filter(.data$cohortId %in% input$cohorts ) })
      selectedDatabaseIds <- shiny::reactive({ databaseTable %>% dplyr::filter(.data$cohortId %in% input$databaseIds ) })

      countModule <- cohortCountsModule(id = "counts", #Should always match UI above
                                                                         dataSource = dataSource,
                                                                         cohortTable = cohortTable,
                                                                         databaseTable = databaseTable,
                                                                         selectedCohorts = selectedCohorts(),
                                                                         selectedDatabaseIds = selectedDatabaseIds(),
                                                                         cohortIds = cohortIds()) 
})

A big problem with the usability with these modules outside of CohortDiagnotics is the .. bootstrap code ... section (which intentionally left vague here). This loads required base tables (such as the cohortTable that contains all cohorts and the list of available databases).
Currently, these are not well typed objects. A dbplyr lazy loaded table solution but, currently, dbplyr is not yet supported by DatabaseConnector and any implemented solution may not work well with pooled database connections.

The steps this process takes are to load the required data model assets that are used within the diagnostics explorer. These are currently a large set of internally used functions that were not intended to be used outside of the shiny application.