From 62d3f014469e924154c167ce8aa854cb28168a3e Mon Sep 17 00:00:00 2001
From: Gabor Csardi <csardi.gabor@gmail.com>
Date: Mon, 27 Jul 2015 01:25:45 -0400
Subject: [PATCH] Support for R downloads

---
 R/cranlogs.R          | 40 ++++++++++++++++++++++++++++++++--------
 man/cran_downloads.Rd | 12 ++++++++++--
 2 files changed, 42 insertions(+), 10 deletions(-)

diff --git a/R/cranlogs.R b/R/cranlogs.R
index cb596be..04a0069 100644
--- a/R/cranlogs.R
+++ b/R/cranlogs.R
@@ -16,18 +16,23 @@ top_url   <- paste0(base_url, "top/")
 #'
 #' @param packages A character vector, the packages to query,
 #'   or \code{NULL} for a sum of downloads for all packages.
+#'   Alternatively, it can also be \code{"R"}, to query downloads
+#'   of R itself. \code{"R"} cannot be mixed with packages.
 #' @param when \code{last_day}, \code{last_week} or \code{last_month}.
 #'   If this is given, then \code{from} and \code{to} are ignored.
 #' @param from Start date, in \code{yyyy-mm-dd} format, or
 #'   \code{last-day}. It is ignored if \code{when} is given.
 #' @param to End date, in \code{yyyy-mm-dd} format, or
 #'   \code{last-day}. It is ignored if \code{when} is given.
-#' @return A data frame with columns:
+#' @return For packages a data frame with columns:
 #'   \item{\code{package}}{The package. This column is missing if
 #'     all packages were queried.}
 #'   \item{\code{date}}{Day of the downloads, it is a Date object.}
 #'   \item{\code{count}}{Download count.}
 #'
+#'   For downloads of R, there are also columns for the operating
+#'   system (\code{os}) and the R version (\code{version}).
+#'
 #' @family CRAN downloads
 #' @export
 #' @examples
@@ -46,6 +51,9 @@ top_url   <- paste0(base_url, "top/")
 #'
 #' ## Multiple packages
 #' cran_downloads(packages = c("ggplot2", "plyr", "dplyr"))
+#'
+#' ## R downloads
+#' cran_downloads("R")
 
 cran_downloads <- function(packages = NULL,
                            when = c("last-day", "last-week", "last-month"),
@@ -62,25 +70,30 @@ cran_downloads <- function(packages = NULL,
   }
 
   if (is.null(packages)) {
-    packages <- ""
+    ppackages <- ""
   } else {
-    packages <- paste(packages, collapse = ",")
-    packages <- paste0("/", packages)
+    if ("R" %in% packages && any(packages != "R")) {
+      stop("R downloads cannot be mixed with package downloads")
+    }
+    ppackages <- paste(packages, collapse = ",")
+    ppackages <- paste0("/", ppackages)
   }
 
-  r <- fromJSON(content(GET(paste0(daily_url, interval, packages)), as = "text"),
+  r <- fromJSON(content(GET(paste0(daily_url, interval, ppackages)), as = "text"),
                 simplifyVector = FALSE)
 
   if ("error" %in% names(r) && r$error == "Invalid query") {
     stop("Invalid query, probably invalid dates")
   }
 
-  to_df(r)
+  to_df(r, packages)
 
 }
 
-to_df <- function(res) {
-  if (length(res) == 1 && is.null(res[[1]]$package)) {
+to_df <- function(res, packages) {
+  if (length(res) == 1 && identical(toupper(packages), "R")) {
+    to_df_r(res[[1]])
+  } else if (length(res) == 1 && is.null(res[[1]]$package)) {
     to_df_1(res[[1]])
   } else {
     dfs <- lapply(res, to_df_1)
@@ -98,6 +111,17 @@ to_df_1 <- function(res1) {
   fill_in_dates(df, as.Date(res1$start), as.Date(res1$end))
 }
 
+to_df_r <- function(res1) {
+  df <- data.frame(
+    stringsAsFactors = FALSE,
+    date = as.Date(vapply(res1$downloads, "[[", "", "day")),
+    version = vapply(res1$downloads, "[[", "", "version"),
+    os = vapply(res1$downloads, "[[", "", "os"),
+    count = vapply(res1$downloads, "[[", 1, "downloads")
+  )
+  fill_in_dates(df, as.Date(res1$start), as.Date(res1$end))
+}
+
 fill_in_dates <- function(df, start, end) {
   if (start > end) stop("Empty time interval")
   if (end > Sys.Date()) warning("Time interval in the future")
diff --git a/man/cran_downloads.Rd b/man/cran_downloads.Rd
index b00edec..1f06f26 100644
--- a/man/cran_downloads.Rd
+++ b/man/cran_downloads.Rd
@@ -9,7 +9,9 @@ cran_downloads(packages = NULL, when = c("last-day", "last-week",
 }
 \arguments{
 \item{packages}{A character vector, the packages to query,
-or \code{NULL} for a sum of downloads for all packages.}
+or \code{NULL} for a sum of downloads for all packages.
+Alternatively, it can also be \code{"R"}, to query downloads
+of R itself. \code{"R"} cannot be mixed with packages.}
 
 \item{when}{\code{last_day}, \code{last_week} or \code{last_month}.
 If this is given, then \code{from} and \code{to} are ignored.}
@@ -21,11 +23,14 @@ If this is given, then \code{from} and \code{to} are ignored.}
 \code{last-day}. It is ignored if \code{when} is given.}
 }
 \value{
-A data frame with columns:
+For packages a data frame with columns:
   \item{\code{package}}{The package. This column is missing if
     all packages were queried.}
   \item{\code{date}}{Day of the downloads, it is a Date object.}
   \item{\code{count}}{Download count.}
+
+  For downloads of R, there are also columns for the operating
+  system (\code{os}) and the R version (\code{version}).
 }
 \description{
 Daily package downloads from the RStudio CRAN mirror
@@ -45,6 +50,9 @@ cran_downloads(from = "2014-06-30", to = "2014-08-08")
 
 ## Multiple packages
 cran_downloads(packages = c("ggplot2", "plyr", "dplyr"))
+
+## R downloads
+cran_downloads("R")
 }
 \seealso{
 Other CRAN.downloads: \code{\link{cran_top_downloads}}