Download CRAN database as needed.

Cache the results in a sensible manner.

DESCRIPTION

@@ -22,7 +22,8 @@ Imports:
     stackr,
     tidytext,
     tibble,
-    formattable
+    formattable,
+    memoise
 Remotes: lshep/stackr
 License: MIT + file LICENSE
 Encoding: UTF-8

NAMESPACE

@@ -3,6 +3,7 @@
 export(combine_metrics)
 export(cran_metrics)
 export(getGitHub)
+export(get_cran)
 export(get_pkgs)
 export(metrics_table)
 export(package_list_metrics)

R/combine_metrics.R

@@ -5,15 +5,14 @@
 #'
 #' @param package_name name of CRAN package - case-sensitive.
 #' @param includeSO Logical. Include metrics from StackOverflow?
-#'
+#' @inheritParams get_cran
 #' @return a tibble of package information.
 #' @export
 #' @examples
 #'
 #' combine_metrics("dplyr")
-#'
-combine_metrics <- function(package_name, includeSO = FALSE) {
-  c_pkg <- cran_metrics(package_name)
+combine_metrics <- function(package_name, includeSO = FALSE, forget = FALSE) {
+  c_pkg <- cran_metrics(package_name, forget)
   git_pkg <- scrape_github_package_page(package_name)
   res <- dplyr::full_join(c_pkg, git_pkg, by = "package")
   if (includeSO){

R/cran_metrics.R

@@ -11,10 +11,11 @@
 #'
 #' cran_metrics("dplyr")
 
-cran_metrics <- function(package_name) {
+cran_metrics <- function(package_name, forget = FALSE) {
   tv_packages <- c("broom",     "dplyr",     "forcats",   "ggplot2",   "haven",     "httr",      "hms",
                    "jsonlite",  "lubridate", "magrittr",  "modelr",    "purrr",     "readr",     "readxl",
                    "stringr",   "tibble",    "rvest",     "tidyr",     "xml2",      "tidyverse")
+  cran = get_cran(forget)
   cran %>%
   dplyr::filter(package %in% package_name) %>%
   dplyr::select(package,

R/data.R

@@ -1,8 +1,3 @@
-#' Cran data
-#'
-#' @format A dataframe of CRAN metadata
-"cran"
-
 #' Table of Packages
 #'
 #' @format A character vector of packages

R/getGitHub.R

@@ -1,52 +1,54 @@
-#' @title Get GitHub URLs for a set of package names (if they exist).
-#'
-#' @description
-#' Looks at both Package URL and BugReports URL, returns GitHub URL if present in either.
-#'
-#' @param packages character vector of CRAN package names - case-sensitive.
-#'
-#' @return a vector of URLs.
-#' @export
-#' @examples
-#'
-#' getGitHub("dplyr")
-
-getGitHub <- function(packages){
-
-  inCran(packages)
-
-  cran_urls <- cran %>%
-    dplyr::filter(package %in% packages) %>%
-    dplyr::select(package, url, bugreports)
-
-  url <- cran_urls$url[2] #temp
-  find_github <- function(url){
-    url <- gsub("\n", ",", url)
-    url <- gsub("^ | $", "", url)
-    url <- gsub(" ", ",", url)
-    url <- gsub("https", "http", url)
-    url <- gsub("http", "https", url)
-    url_list <- stringr::str_split(url, ",")[[1]]
-    github_url <- url_list[stringr::str_detect(url_list, "//github.com")]
-    github_url <- ifelse(length(github_url) == 0, NA, github_url)
-    github_url <- ifelse(stringr::str_sub(github_url, -1) == "/", stringr::str_sub(github_url, 1, -2), github_url)
-    return(github_url)
-  }
-
-
-  cran_urls %>%
-    dplyr::mutate(url_git = purrr::map_chr(url, find_github),
-           bug_git = purrr::map_chr(bugreports, find_github) %>% stringr::str_replace("/issues",""),
-           github_url = ifelse(is.na(url_git), bug_git, url_git)
-    ) %>%
-    dplyr::select(package, github_url) %>%
-    dplyr::mutate(ongithub = !is.na(github_url))
-
-}
-
-inCran <- function(package){
-  #stopifnot(package %in% cran$package)
-  if(! (package %in% cran$package )) stop("Package ",  package, " not in CRAN.  Only supply packages that exists in CRAN")
-}
-
-
+#' @title Get GitHub URLs for a set of package names (if they exist).
+#'
+#' @description
+#' Looks at both Package URL and BugReports URL, returns GitHub URL if present in either.
+#'
+#' @param packages character vector of CRAN package names - case-sensitive.
+#'
+#' @return a vector of URLs.
+#' @export
+#' @examples
+#'
+#' getGitHub("dplyr")
+
+getGitHub <- function(packages, forget = FALSE){
+
+  inCran(packages, forget)
+  # Don't need forget parameter, since forget =TRUE would have
+  # been triggered in inCran
+  cran_urls <- get_cran() %>%
+    dplyr::filter(package %in% packages) %>%
+    dplyr::select(package, url, bugreports)
+
+  url <- cran_urls$url[2] #temp
+  find_github <- function(url){
+    url <- gsub("\n", ",", url)
+    url <- gsub("^ | $", "", url)
+    url <- gsub(" ", ",", url)
+    url <- gsub("https", "http", url)
+    url <- gsub("http", "https", url)
+    url_list <- stringr::str_split(url, ",")[[1]]
+    github_url <- url_list[stringr::str_detect(url_list, "//github.com")]
+    github_url <- ifelse(length(github_url) == 0, NA, github_url)
+    github_url <- ifelse(stringr::str_sub(github_url, -1) == "/", stringr::str_sub(github_url, 1, -2), github_url)
+    return(github_url)
+  }
+
+
+  cran_urls %>%
+    dplyr::mutate(url_git = purrr::map_chr(url, find_github),
+           bug_git = purrr::map_chr(bugreports, find_github) %>% stringr::str_replace("/issues",""),
+           github_url = ifelse(is.na(url_git), bug_git, url_git)
+    ) %>%
+    dplyr::select(package, github_url) %>%
+    dplyr::mutate(ongithub = !is.na(github_url))
+
+}
+
+inCran <- function(package, forget = FALSE){
+  cran = get_cran(forget)
+  #stopifnot(package %in% cran$package)
+  if(! (package %in% cran$package )) stop("Package ",  package, " not in CRAN.  Only supply packages that exists in CRAN")
+}
+
+

R/get_cran.R

@@ -0,0 +1,33 @@
+get_memoise_cran <- memoise::memoise(
+  function() {
+    cran <- tools::CRAN_package_db()
+    # remove first instance of column name MD5Sum
+    cran <- cran[, -dplyr::first(which(names(cran) == "MD5sum"))]
+
+    # make it a tibble
+    cran <- dplyr::tbl_df(cran)
+    cran <- cran %>%
+      janitor::clean_names() %>%
+      janitor::remove_empty_cols()
+    cran
+  }
+)
+
+
+
+#' Get CRAN Packages
+#'
+#' Returns a tibble containing the current CRAN pacakges.
+#' For efficiency reasons, the tibble is cached, so future calls will just return
+#' the original tibble.
+#' @param forget logical, default \code{FALSE}. Reset the CRAN package cache.
+#' @export
+#' @examples
+#' @importFrom memoise memoise forget
+#' cran <- get_cran()
+get_cran <- function(forget = FALSE) {
+  if(forget)
+    memoise::forget(get_memoise_cran)
+  get_memoise_cran()
+}
+

R/package_list_metrics.R

@@ -4,15 +4,16 @@
 #' wrapper for combine_metrics for list of packages
 #'
 #' @param package_names_list list of CRAN packages - case-sensitive.
+#' @inheritParams get_cran
 #'
 #' @return a tibble of package information
 #' @export
 #' @examples
 #'
 #' package_list_metrics(list("dplyr", "tidyr", "ggplot2"))
 
-package_list_metrics <- function(package_names_list) {
-  purrr::map_df(package_names_list, combine_metrics)
+package_list_metrics <- function(package_names_list, forget = FALSE) {
+  purrr::map_df(package_names_list, combine_metrics, forget)
 }