Download data from GitHub repository

batch_download_github.R

batch_download_github <- function(url, pattern, path, ...) {
  if(!require("rvest")) stop("Please install the 'rvest' package.")
  if(!require("RCurl")) stop("Please install the 'RCurl' package.")
  
  # Fetch file names
  github_page <- read_html(url)
  file_nodes <- html_nodes(github_page, ".content .css-truncate-target .js-navigation-open")
  file_names <- html_text(file_nodes)
  file_url <- html_attr(file_nodes, "href")[grep(pattern, file_names)]
  file_names <- file_names[grep(pattern, file_names)]
  
  file_url <- paste0("https://raw.githubusercontent.com", file_url)
  file_url <- gsub("blob/", "", file_url)
  
  data <- data.frame(
    file_name = file_names
    , file_content = sapply(file_url, getURL)
  )
  
  apply(data, 1, function(x) {
    writeLines(
      text = x["file_content"]
      , con = paste0(path, x["file_name"])
    )
  })
}