fairenough

Convert tidy data into R packages with documentation websites and intelligent AI descriptions.

Installation

# install.packages("devtools")
devtools::install_github("openwashdata/fairenough")

Usage

One-Click Pipeline

For the simplest experience, move your datasets in an empty directory, then run the main fairenough() function:

library(fairenough)

# Set up your LLM chat object
Sys.setenv(OPENAI_API_KEY = "YOUR_API_TOKEN_HERE")
chat <- elmer::chat_openai(model = "gpt-4o-mini", api_args = list(temperature = 0.3))

# Run the complete pipeline with one command
fairenough(chat)

This automatically: - Sets up your R package structure - Processes all CSV/Excel files in the directory - Collects package metadata through interactive prompts - Builds documentation, citation files, README, and website - Generates intelligent data documentation using LLMs - Prepares your package for publication

Granular Control

For step-by-step control, use individual wrapper functions:

library(fairenough)

# Step 1: Initialize project structure
setup()

# Step 2: Process your data files
process()

# Step 3: Collect package metadata
collect()

# Step 4: Generate LLM-powered documentation
chat <- elmer::chat_openai(model = "gpt-4o-mini", api_args = list(temperature = 0.3))
generate(chat)

# Step 5: Build all package components
build()

Features

fairenough provides a complete pipeline for R data package creation, following these logical steps:

fairenough() - One-Click Pipeline

• Complete R data package creation with a single fairenough(chat) call

• Automated workflow from tidy data to finished package

• Args: Supports all flags that can be passed to the individual wrapper functions

Granular Control Options

• Individual wrapper functions: setup(), process(), collect(), generate(), build()

• Global support for args: overwrite = TRUE and verbose = TRUE

1. setup() - R Project Setup

• R package structure initialization with usethis

• Directory and files organization (data_raw, .gitignore, etc.)

2. process() - Automated Data Processing

• Processes all your tidy data with readr, readxl

• Validates data structure and formats

• Arg: auto_clean = TRUE for automated minimal data cleaning and tidying

3. collect() - Interactive Metadata Collection

• Guided prompts for package metadata (title, description, authors, etc.) using cli

• Saves directly to DESCRIPTION file with desc

• Arg: extended = TRUE for comprehensive documentation

4. generate() - LLM-Powered Documentation

• Data dictionary generation using elmer chat/LLM integration

• Variable descriptions uses the package’s description as context + actual data samples

• Arg: contex = "YOUR DATASETS CONTEXT"

5. build() - Complete Package Infrastructure

• Roxygen documentation generation with roxygen2

• Citation file creation with validation using cffr

• README generation with rmarkdown

• Package website building ready for deployment

License

MIT

References

Horst AM, Hill AP, Gorman KB (2020). palmerpenguins: Palmer Archipelago (Antarctica) penguin data. R package version 0.1.0. https://allisonhorst.github.io/palmerpenguins/. doi: 10.5281/zenodo.3960218.

Contributions

Welcome!

Philosophy

• Conventional commits! https://www.conventionalcommits.org/.

• Feature functions must maintain architectural consistency. This means that aspects like supported formats and path handling should be uniform across all functions. For more details, refer to R/utils.R.

• To test new functions:

roxygen2::roxygenise()
devtools::load_all()

• Due to their utility outside of {fairenough}, gendict.R, build_license.R and promptit.R have been kept general and not tied to the package’s architecture.