276 create summary stats

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: GitStats
 Title: Get Statistics from Git Hosting Services
-Version: 0.1.0.9005
+Version: 0.1.0.901
 Authors@R: c(
     person(given = "Maciej", family = "Banaś", email = "[email protected]", role = c("aut", "cre")),
     person(given = "Kamil", family = "Koziej", email = "[email protected]", role = "aut"),

NAMESPACE

@@ -1,13 +1,16 @@
 # Generated by roxygen2: do not edit by hand
 
+S3method(gitstats_plot,commits_stats)
+S3method(gitstats_plot,repos_stats)
 export("%>%")
 export(create_gitstats)
 export(get_commits)
+export(get_commits_stats)
 export(get_orgs)
 export(get_repos)
+export(get_repos_stats)
 export(get_users)
-export(plot_commits)
-export(plot_repos)
+export(gitstats_plot)
 export(pull_commits)
 export(pull_repos)
 export(pull_repos_contributors)

NEWS.md

@@ -1,5 +1,7 @@
 GitStats (development version)
 
+- added `get_*_stats` functions to prepare summary stats from pulled data: repositories and commits (I: #276),
+- rename and refactor plot functions to one generic `gitstats_plot` which takes as an input `repos_stats` or `commits_stats` (I: #276),
 - changed names of `get_*` to `pull_*` functions; `get_*` functions are now to retrieve already pulled data from GitStats object (I: #294),
 - changed name of `setup` to `set_params` function (I: #294),
 - set new name for `set_connection` function: `set_host` as it is more informative (and shorter) (I: #271),

R/EngineGraphQLGitHub.R

@@ -20,7 +20,7 @@ EngineGraphQLGitHub <- R6::R6Class("EngineGraphQLGitHub",
       self$gql_query <- GQLQueryGitHub$new()
     },
 
-    #' @description Get all groups from GitLab.
+    #' @description Get all orgs from GitHub.
     pull_orgs = function() {
       end_cursor <- NULL
       has_next_page <- TRUE
@@ -134,7 +134,6 @@ EngineGraphQLGitHub <- R6::R6Class("EngineGraphQLGitHub",
       commits_table <- repos_list_with_commits %>%
         purrr::discard(~ length(.) == 0) %>%
         private$prepare_commits_table(org)
-
       return(commits_table)
     },
 
@@ -251,9 +250,7 @@ EngineGraphQLGitHub <- R6::R6Class("EngineGraphQLGitHub",
         repo$created_at <- gts_to_posixt(repo$created_at)
         repo$issues_open <- repo$issues_open$totalCount
         repo$issues_closed <- repo$issues_closed$totalCount
-        repo$last_activity_at <- difftime(Sys.time(), as.POSIXct(repo$last_activity_at),
-          units = "days"
-        ) %>% round(2)
+        repo$last_activity_at <- as.POSIXct(repo$last_activity_at)
         repo$organization <- repo$organization$login
         data.frame(repo)
       })
@@ -392,7 +389,7 @@ EngineGraphQLGitHub <- R6::R6Class("EngineGraphQLGitHub",
         commits_row
       }) %>%
         purrr::discard(~ length(.) == 1) %>%
-        rbindlist()
+        rbindlist(use.names = TRUE)
 
       if (nrow(commits_table) > 0) {
         commits_table <- commits_table %>%

R/EngineGraphQLGitLab.R

@@ -162,9 +162,7 @@ EngineGraphQLGitLab <- R6::R6Class("EngineGraphQLGitLab",
          repo$issues_open <- repo$issues$opened
          repo$issues_closed <- repo$issues$closed
          repo$issues <- NULL
-         repo$last_activity_at <- difftime(Sys.time(), as.POSIXct(repo$last_activity_at),
-                                           units = "days"
-         ) %>% round(2)
+         repo$last_activity_at <- as.POSIXct(repo$last_activity_at)
          repo$organization <- repo$group$name
          repo$group <- NULL
          data.frame(repo)

R/EngineRest.R

@@ -108,9 +108,7 @@ EngineRest <- R6::R6Class("EngineRest",
         repos_dt <- dplyr::mutate(repos_dt,
           id = as.character(id),
           created_at = as.POSIXct(created_at),
-          last_activity_at = difftime(Sys.time(), as.POSIXct(last_activity_at),
-            units = "days"
-          ) %>% round(2),
+          last_activity_at = as.POSIXct(last_activity_at),
           forks = as.integer(forks),
           issues_open = as.integer(issues_open),
           issues_closed = as.integer(issues_closed)

R/EngineRestGitHub.R

@@ -113,7 +113,7 @@ EngineRestGitHub <- R6::R6Class("EngineRestGitHub",
     pull_repos_contributors = function(repos_table) {
       if (nrow(repos_table) > 0) {
         if (!private$scan_all) {
-          cli::cli_alert_info("[GitHub][Engine:{cli::col_green('REST')}] Pulling contributors...")
+          cli::cli_alert_info("[GitHub][Engine:{cli::col_green('REST')}][org:{unique(repos_table$organization)}] Pulling contributors...")
         }
         repo_iterator <- paste0(repos_table$organization, "/", repos_table$name)
         user_name <- rlang::expr(.$login)

R/EngineRestGitLab.R

@@ -70,7 +70,7 @@ EngineRestGitLab <- R6::R6Class("EngineRestGitLab",
     pull_repos_contributors = function(repos_table) {
       if (nrow(repos_table) > 0) {
         if (!private$scan_all) {
-          cli::cli_alert_info("[GitLab][Engine:{cli::col_green('REST')}] Pulling contributors...")
+          cli::cli_alert_info("[GitLab][Engine:{cli::col_green('REST')}][org:{unique(repos_table$organization)}] Pulling contributors...")
         }
         repo_iterator <- repos_table$id
         user_name <- rlang::expr(.$name)

R/GitHost.R

@@ -114,6 +114,7 @@ GitHost <- R6::R6Class("GitHost",
         cli::cli_alert_info("[Host:{private$host}] {cli::col_yellow('Pulling commits from all organizations...')}")
       }
       commits_table <- purrr::map(private$orgs, function(org) {
+        commits_table_org <- NULL
         tryCatch({
           commits_table_org <- purrr::map(private$engines, ~ .$pull_commits(
             org = org,
@@ -138,12 +139,12 @@ GitHost <- R6::R6Class("GitHost",
                 date_until = date_until,
                 settings = settings
               )
-            })
+            }) %>%
+              purrr::list_rbind()
           } else {
             e
           }
         })
-
         return(commits_table_org)
       }, .progress = private$scan_all) %>%
         purrr::list_rbind()

R/get_stats.R

@@ -0,0 +1,86 @@
+#' @title Get Statistics On Repositories
+#' @name get_repos_stats
+#' @description Prepare statistics from the pulled repositories data.
+#' @param gitstats_obj A GitStats class object.
+#' @return A table of `repos_stats` class.
+#' @export
+get_repos_stats <- function(gitstats_obj){
+  repos_data <- data.table::copy(get_repos(gitstats_obj))
+  repos_stats <- repos_data %>%
+    dplyr::mutate(
+      fullname = paste0(organization, "/", name)
+    ) %>%
+    dplyr::mutate(
+      last_activity = difftime(
+        Sys.time(),
+        last_activity_at,
+        units = "days"
+      ) %>% round(2),
+      repository = fullname,
+      platform = retrieve_platform(api_url)
+    )
+  if ("contributors" %in% colnames(repos_data)) {
+    repos_stats <- dplyr::mutate(
+      repos_stats,
+      contributors_n = purrr::map_vec(contributors, function(contributors_string) {
+        length(stringr::str_split(contributors_string[1], ", ")[[1]])
+      })
+    )
+  } else {
+    repos_stats <- dplyr::mutate(
+      repos_stats,
+      contributors_n = NA
+    )
+  }
+  repos_stats <- dplyr::select(
+    repos_stats,
+    repository, platform, created_at, last_activity, stars, forks,
+    languages, issues_open, issues_closed, contributors_n
+  )
+  class(repos_stats) <- append(class(repos_stats), "repos_stats")
+  return(repos_stats)
+}
+
+#' @title Get Statistics On Commits
+#' @name get_commits_stats
+#' @description Prepare statistics from the pulled commits data.
+#' @param gitstats_obj A GitStats class object.
+#' @param time_interval A character, specifying time interval to show statistics.
+#' @return A table of `commits_stats` class.
+#' @export
+get_commits_stats <- function(gitstats_obj,
+                              time_interval = c("month", "day", "week")){
+  commits_data <- data.table::copy(get_commits(gitstats_obj))
+  time_interval <- match.arg(time_interval)
+
+  commits_stats <- commits_data %>%
+    dplyr::mutate(
+      stats_date = lubridate::floor_date(
+        committed_date,
+        unit = time_interval
+      ),
+      platform = retrieve_platform(api_url)
+    ) %>%
+    dplyr::group_by(stats_date, platform, organization) %>%
+    dplyr::summarise(
+      commits_n = dplyr::n()
+    ) %>%
+    dplyr::arrange(
+      stats_date
+    )
+  commits_stats <- commits_stats(
+    object = commits_stats,
+    time_interval = time_interval
+  )
+  return(commits_stats)
+}
+
+#' @noRd
+#' @description A constructor for `commits_stats` class
+commits_stats <- function(object, time_interval) {
+  stopifnot(inherits(object, "grouped_df"))
+  object <- dplyr::ungroup(object)
+  class(object) = append(class(object), "commits_stats")
+  attr(object, "time_interval") <- time_interval
+  object
+}

R/global.R

@@ -1,6 +1,8 @@
 # This is a solution for warnings in r-cmd checks of global variables
 globalVariables(c(
   ".", "fullname", "platform", "organization", "repo_url",
-  "name", "last_activity_at", "stats_date", "committed_date",
-  "api_url", "row_no", ".N"
+  "name", "created_at", "last_activity_at", "last_activity", "stats_date",
+  "committed_date", "commits_n", "api_url", "row_no", ".N", ".data",
+  "repository", "stars", "forks", "languages", "issues_open", "issues_closed",
+  "contributors_n"
 ))