Obtaining licensing metadata for task-specific R packages on CRAN (…for the purpose of, say, avoiding GPL-licensed packages)

licenses.R

# devtools::install_github("RcppCore/Rcpp")
# devtools::install_github("hadley/xml2")
# devtools::install_github("hadley/rvest")
# devtools::install_github("metacran/crandb")
# devtools::install_github('ironholds/practice')
# install.packages('dplyr')

library(rvest); library(crandb)

get_licensing_for_task <- function(task_url) {
	elapsed <- system.time({
		task <- read_html(task_url)
		# Fetch packages listed in task
		task_packages <- task %>%
			html_nodes("body ul li a") %>%
			html_text %>%
			grep(pattern="\\n",x=.,invert=TRUE,value=TRUE) %>%
			unique
		# Fetch license metadata for each package in task
		licensing_metadata <- sapply(task_packages, . %>%
						practice:::get_package_metadata() %>%
						{ tail(.$versions,1)[[1]] } %>%
						{ c(Package=.$Package,License=.$License) }) %>%
						t %>% as.data.frame %>%
						{ rownames(.) <- 1:nrow(.); . }
	})['elapsed']
	cat("Finished processing",task_url,"(took",elapsed,"seconds)","\n")
	return(licensing_metadata)
}

# Fetch tasks:
tasks <- read_html("http://cran.r-project.org/web/views/")
tasks %<>% html_nodes("tr") %>% sapply(. %>% {
	c(html_nodes(.,"td") %>% { .[2] } %>% html_text,
	  html_nodes(., "a") %>% html_attr("href"))
}) %>% t
colnames(tasks) <- c("Task","Webpage")
#> head(tasks,1)
#      Task                 Webpage        
# [1,] "Bayesian Inference" "Bayesian.html"

# Obtain licensing information for the packages in each task:
licenses <- sapply(tasks[,'Webpage'],. %>%
			paste0('http://cran.r-project.org/web/views/',.) %>%
			get_licensing_for_task)

Check out get_package_names() for added simplification, in practise. I forsee this being a really fun exploratory tool \o/