rmdscaffold: Infrastructure to build user-defined Rmarkdown pipelines

[Bisaloo]

{rmdscaffold}: Infrastructure to build user-defined Rmarkdown pipelines

Description

Provides the necessary infrastructure to build user-defined Rmarkdown pipelines from isolated Rmarkdown chunks provided by a dependent package.

Target audience

• typical end-users: R package developers

• potential contributors: ?

• key collaborators: ?

Interoperability

• inputs: none

• outputs: a custom pipeline in the form of a Rmarkdown report

• imports:

  • {tinkr}, to build a single Rmd document from many chunks
  • {ymlthis}, to build the yaml header of this final Rmd document
  • {globals}, to identify global variables, i.e., dependencies to earlier chunks
  • {xmlparsedata}, as a possible alternative to {globals} and as a way to identify outputs as well
  • {xml2}, to manipulate objects returned by {tinkr} & {xmlparsedata}

• used by:

  • XXXX: the successor of episoap to allow to build pipelines for epidemiology in a more flexible way
  • XXXX: a package by/for @sbfnk to create flexible evaluation report templates for modelling hubs

• related projects

  • ?

Usage

Dependent package structure:

testpkg
├── DESCRIPTION
├── LICENSE.md
├── NAMESPACE
├── R
│   └── hello.R
├── README.Rmd
├── README.md
├── inst
│   └── rmdchunks
│       ├── epinow2_chunk.Rmd
│       ├── incidence2_chunk.Rmd
│       └── socialmixr_chunk.Rmd
├── man
│   └── hello.Rd
└── testpkg.Rproj

{rmdscaffold} usage:

rmdscaffold::build_scaffold(pkg = "testpkg")`

which opens a shiny app to create a custom pipeline from the Rmarkdown chunks provided in {testpkg}. From this shiny app, the user can generate a Rmd report for their custom pipeline.

Statement of need

{rmdscaffold} takes care of identifying the dependency relationships between the different packages / chunks and provides the interactive interface to drag&drop / select the elements to successively insert in the custom pipeline. It also summarises the list of dependencies in the form of library() calls at the top of the resulting Rmarkdown document.

Anticipated challenges

This forces developers of downstream packages to be super careful / rigorous / consistent in the variable names they use across chunks. The network visualisation can help identifying mistakes but it's still a little bit tedious.