Get and plot package dependencies using metacran services

package_deps.R

#' Search for packages that are dependent on a particular package
#'
#' @param pkg The name of the package
#' @param size Maximum number of dependencies to search for
#' @param ... Additional argumentst passed on to \link{seer}{see}
#'
#' @return A \code{tibble} with columns \code{package} and \code{relationship}
#' @export
#'
#' @examples
get_pkg_deps <- function(pkg, size = 10, ...) {
  if (!require("seer")) stop("You need to install the seer package from httsp://github.com/metacran/seer")
  if (!requireNamespace("purrr")) stop("You need to install the purrr package") 
  matches <- seer::see(pkg, size = size, ...)
  regex_string <- paste0("(^|\\s|,)", pkg, "(,|\\s|$)")
  purrr::map_df(matches$hits$hits, 
                ~ {
                  id <- .x[["_id"]]
                  src <- .x[["_source"]]
                  if (id == pkg) {
                    return(list(package = pkg, relationship = "Package"))
                  }
                  depends <- isTRUE(grepl(regex_string, src[["Depends"]]))
                  imports <- isTRUE(grepl(regex_string, src[["Imports"]]))
                  suggests <- isTRUE(grepl(regex_string, src[["Suggests"]]))
                  relationships <- c(Depends = depends, Imports = imports, Suggests = suggests)
                  if (!any(relationships)) return(NULL)
                  list(package = id, relationship = names(relationships[relationships]))
                })
}

#' Plot CRAN download logs of a package along with packages that depend on it.
#'
#' @param pkg The name of the package
#' @param when time period (one of "last-day", "last-week", or "last-month")
#' @param from Start date, in yyyy-mm-dd format, or last-day. It is ignored if when is given.
#' @param to End date, in yyyy-mm-dd format, or last-day. It is ignored if when is given.
#' @param by either "package" to show all packages on the plot, or "relationship" 
#' to group dependent packages by their relationship to the package ("Depends", "Imports", and "Suggests")
#' @param type Type of graph to dispaly - either "line" or "area"
#' @param size Maximum number of dependencies to search for
#' @param ... Additional argumentst passed on to \link{seer}{see}
#'
#' @return a ggplot2 object
#' @export
#'
#' @examples
plot_pkg_deps <- function(pkg, when = "last-month", from, to, 
                          by = c("package", "relationship"), 
                          type = c("area", "line"), size = 10, ...) {
  if (!requireNamespace("cranlogs")) stop("You need to install the cranlogs package from httsp://github.com/metacran/cranlogs")
  if (!requireNamespace("dplyr")) stop("You need to install the dplyr package") 
  if (!requireNamespace("ggplot2")) stop("You need to install the ggplot2 package") 
  deps <- get_pkg_deps(pkg, size = size, ...)
  by <- match.arg(by)
  type <- match.arg(type)
  
  downloads <- cranlogs::cran_downloads(unique(deps$package), when = when, from, to)
  downloads <- dplyr::left_join(downloads, deps, by = "package")
  
  if (by == "relationship") {
    downloads <- dplyr::group_by_(downloads, ~ relationship, ~ date)
    downloads <- dplyr::summarize_(downloads, count = ~ sum(count))
  }
  
  if (type == "line") {
    colour_breaks <- switch(by, 
                            "package" = c(pkg, setdiff(deps$package, pkg)), 
                            "relationship" = c("Package", "Depends", "Imports", "Suggests"))
    
    p <- ggplot2::ggplot(downloads, ggplot2::aes_string(x = "date", y = "count", colour = by, linetype = "relationship")) + 
      ggplot2::geom_line() + 
      ggplot2::scale_linetype_manual(values = c(Package = "solid", Depends = "longdash", Imports = "dashed", Suggests = "dotted"), 
                                 breaks = c("Package", "Depends", "Imports", "Suggests")) + 
      ggplot2::scale_colour_discrete(breaks = colour_breaks) +
      ggplot2::labs(title = paste("CRAN downloads for package:", pkg))
    
  } else {
    pkg_subset <- ifelse(by == "package", pkg, "Package")
    
    p <- ggplot2::ggplot(downloads[downloads[[by]] != pkg_subset, ], 
                         ggplot2::aes_string(x = "date", y = "count", fill = by)) + 
      ggplot2::geom_area() + 
      ggplot2::geom_line(data = downloads[downloads[[by]] == pkg_subset, ], 
                         ggplot2::aes_string(x = "date", y = "count", linetype = by), 
                         inherit.aes = FALSE) + 
      ggplot2::scale_linetype_manual(values = "solid", labels = pkg, name = "Package") +
      ggplot2::labs(title = paste("CRAN downloads for package:", pkg), 
                    fill = "Reverse Dependencies") + 
      ggplot2::guides(linetype = ggplot2::guide_legend(order = 1), fill = ggplot2::guide_legend(order = 2))  
  }
  
  p
}

usage_example.md

# install_github("metacran/seer")
library(devtools)
source_gist("https://gist.github.com/ateucher/8c1a16378cf9224d1b270512b7147080", filename = "package_deps.R")
#> Using GitHub PAT from envvar GITHUB_PAT
#> Sourcing https://gist.githubusercontent.com/ateucher/8c1a16378cf9224d1b270512b7147080/raw/6db9b5e7225aa92b7f2dac390471b2f5a7996c7c/package_deps.R
#> SHA-1 hash of file is 89d58320b3c892a06534c830c20d36bf1d984606

get_pkg_deps("rmapshaper")
#> Loading required package: seer
#> Loading required namespace: purrr
#> # A tibble: 7 x 2
#>              package relationship
#>                <chr>        <chr>
#> 1         rmapshaper      Package
#> 2           mregions     Suggests
#> 3          tmaptools      Imports
#> 4       leaflet.esri     Suggests
#> 5     leaflet.extras     Suggests
#> 6 MazamaSpatialUtils      Imports
#> 7           eechidna      Imports

plot_pkg_deps("rmapshaper")
#> Loading required namespace: cranlogs
#> Loading required namespace: ggplot2

plot_pkg_deps("rmapshaper", type = "line")

plot_pkg_deps("ggplot2", when = "last-week", by = "relationship")