Add sanitize() function, which wipes personal data and adds user URLs (…

…#29)


---------

Co-authored-by: Colin Gillespie <colin@jumpingrivers.com>

DESCRIPTION

@@ -1,14 +1,14 @@
 Type: Package
 Package: audit.connect
 Title: Posit Connect Health Check
-Version: 0.6.3
+Version: 0.7.0
 Authors@R:
     person("Jumping", "Rivers", , "info@jumpingrivers.com", role = c("aut", "cre"))
 Description: Posit Connect Health Check. Deploys various content types to
     assess whether Connect is functioning correctly.
 License: file LICENSE
 Imports: 
-    audit.base (>= 0.6.9),
+    audit.base (>= 0.6.10),
     cli,
     connectapi (>= 0.1.3.1),
     dplyr,
@@ -22,7 +22,6 @@ Imports:
     quarto (>= 1.3),
     rlang,
     rsconnect (>= 1.1.0),
-    serverHeaders,
     stringr,
     tibble
 Suggests:

NAMESPACE

@@ -21,6 +21,7 @@ export(create_config)
 export(get_quarto_locked_user_apps)
 export(get_quarto_old_users)
 export(get_quarto_user_roles)
+export(sanitise)
 import(audit.base)
 importFrom(dplyr,"%>%")
 importFrom(rlang,.data)

NEWS.md

@@ -1,3 +1,7 @@
+# audit.connect 0.7.0 _2023-10-18_
+ - feat: `audit_object |> audit.connect::sanitize()` removes personal user data
+ - feat: Add in user URL when summarising users
+
 # audit.connect 0.6.4 _2023-10-01_
  - feat: Check for Posit name leakage
 

R/quarto-helpers.R

@@ -9,7 +9,7 @@ get_quarto_old_users = function(out) {
     dplyr::filter(!.data$locked)
   old_users$email = tolower(old_users$email)
   old_users$domain = stringr::str_match(old_users$email, "@(.*)")[, 2]
-  old_users = dplyr::arrange(old_users, domain, email)
+  old_users = dplyr::arrange(old_users, .data$domain, .data$email)
 
   old_users = old_users %>%
     dplyr::mutate(last_log_on_diff = lubridate::interval(.data$active_time, lubridate::now()) / months(1), #nolint
@@ -24,7 +24,7 @@ get_quarto_old_users = function(out) {
     dplyr::reframe(email = paste(.data$email, collapse = ", ")) %>%
     dplyr::mutate(last_log_in = factor(.data$last_log_in,
                                        c("12 months+", "6 months+", "3 months+"),
-                                       ordered = T)) %>%
+                                       ordered = TRUE)) %>%
     dplyr::arrange(.data$last_log_in)
   old_users$n = stringr::str_count(old_users$email, ",") + 1
   old_users

R/sanitise.R

@@ -0,0 +1,12 @@
+#' Sanitise {audit.connect} object
+#'
+#' This function removes user-identifying data from an {audit.connect} object
+#' @param audit_connect_check An object from audit.connect::check()
+#' @export
+sanitise = function(audit_connect_check) {
+  # Wipe any user-identifiable data
+  for (value in c("email", "first_name", "last_name", "username")) {
+    audit_connect_check$users_details$user_list$users[, value] = NA_character_
+  }
+  audit_connect_check
+}

R/summarise_users.R

@@ -7,6 +7,7 @@ summarise_users = function(server, token, debug_level) {
   user_list = list()
   user_list$user_account_limit = settings$license$users
   user_list$users = suppress(connectapi::get_users(client, limit = Inf))
+  user_list$users$url = paste0(server, "connect/#/people/users/", user_list$users$guid)
 
   if (is_evaluation(settings)) {
     apps = NA

tests/testthat/test-check.R

@@ -3,15 +3,14 @@ test_that("High level test", {
   create_config(default = TRUE)
   expect_error(check(server = "aaa.bbb"))
   # Run standard check
-  rtn = check()
+  rtn = suppressMessages(check())
 
   expect_true(is.list(rtn))
   check_names = c("setup", "posit_version", "server_headers", "feature_usage",
                   "audit_details", "users_details", "versions", "sys_deps", "results") %in%
     names(rtn)
   expect_true(all(check_names))
-
-  expect_equal(ncol(rtn$results), 5)
+  expect_equal(ncol(rtn$results), 6)
 
   # Check Quarto report
   # Copy over necessary files

tests/testthat/test-sanitise.R

@@ -0,0 +1,17 @@
+describe("Checking sanitize function", {
+  testthat::skip_on_ci()
+  # Skip report checks for speed
+  create_config(default = FALSE, type = "force")
+  rtn = suppressMessages(check())
+  sanitize_rtn = sanitise(rtn)
+
+  it("Check users have been changed",
+     expect_false(identical(rtn[["users_details"]],
+                            sanitize_rtn[["users_details"]]))
+  )
+
+  it("Check everything else has stayed the same", {
+    rtn[["users_details"]] = sanitize_rtn[["users_details"]] = NULL
+    expect_identical(rtn, sanitize_rtn)
+  })
+})