who is visiting this content (table view) (#38)

* first commit

* fix summary sentence

* clarify comment

* add date formatting and column renaming

Author: toph-allen

extensions/who-is-visiting-this-content/.Rprofile

@@ -0,0 +1 @@
+source("renv/activate.R")

extensions/who-is-visiting-this-content/.gitignore

@@ -0,0 +1,2 @@
+.posit/
+app_cache/

extensions/who-is-visiting-this-content/app.R

@@ -0,0 +1,156 @@
+library(shiny)
+library(bslib)
+library(shinyjs)
+library(connectapi)
+library(dplyr)
+library(glue)
+library(lubridate)
+library(tidyr)
+
+shinyOptions(
+  cache = cachem::cache_disk("./app_cache/cache/", max_age = 60 * 60 * 8)
+)
+
+source("get_usage.R")
+
+ui <- page_fillable(
+  useShinyjs(),
+  theme = bs_theme(version = 5),
+
+  card(
+    card_header("Who is Visiting This Content?"),
+    layout_sidebar(
+      sidebar = sidebar(
+        title = "No Filters Yet",
+        open = FALSE,
+
+        actionButton("clear_cache", "Clear Cache", icon = icon("refresh"))
+      ),
+
+      textInput(
+        "content_guid",
+        "Content GUID"
+      ),
+
+      h4(
+        id = "guid_input_msg",
+        "Please enter a content GUID"
+      ),
+
+      textOutput("summary_message"),
+
+      tabsetPanel(
+        id = "content_visit_tables",
+        tabPanel(
+          "List of Visits",
+          tableOutput("all_visits")
+        ),
+        tabPanel(
+          "Aggregated Visits",
+          tableOutput("aggregated_visits")
+        )
+      )
+    )
+  )
+)
+
+server <- function(input, output, session) {
+  # Cache invalidation button ----
+  cache <- cachem::cache_disk("./app_cache/cache/")
+  observeEvent(input$clear_cache, {
+    print("Cache cleared!")
+    cache$reset()  # Clears all cached data
+    session$reload()  # Reload the app to ensure fresh data
+  })
+
+  observe({
+    if (nchar(input$content_guid) == 0) {
+      show("guid_input_msg")
+      hide("content_visit_tables")
+    } else {
+      hide("guid_input_msg")
+      show("content_visit_tables")
+    }
+  })
+
+  # Loading and processing data ----
+  client <- connect()
+
+  # Default dates. "This week" is best "common sense" best represented by six
+  # days ago thru the end of today. Without these, content takes too long to
+  # display on some servers.
+  date_range <- reactive({
+    list(
+      from_date = today() - ddays(6),
+      to_date = today()
+    )
+  })
+
+  content <- reactive({
+    # Grab the entire content data frame here and filter it using the pasted-in
+    # GUID to obtain content title and other metadata, rather than making a
+    # request to `v1/content/{GUID}`. If this were a prod, standalone dashboard,
+    # might be better to call that endpoint.
+    get_content(client)
+  }) |> bindCache("static_key")
+
+  user_names <- reactive({
+    get_users(client) |>
+      mutate(full_name = paste(first_name, last_name)) |>
+      select(user_guid = guid, full_name, username)
+  }) |> bindCache("static_key")
+
+  usage_data <- reactive({
+    get_usage(
+      client,
+      from = date_range()$from_date,
+      to = date_range()$to_date + hours(23) + minutes(59) + seconds(59)
+    )
+  }) |> bindCache(date_range()$from_date, date_range()$to_date)
+
+  # Compute data
+  all_visits_data <- reactive({
+    usage_data() |>
+      filter(content_guid == input$content_guid) |>
+      left_join(user_names(), by = "user_guid") |>
+      replace_na(list(full_name = "[Anonymous]")) |>
+      arrange(desc(timestamp)) |>
+      select(timestamp, full_name, username)
+  }) |> bindCache(date_range()$from_date, date_range()$to_date, input$content_guid)
+
+  aggregated_visits_data <- reactive({
+    usage_data() |>
+      filter(content_guid == input$content_guid) |>
+      group_by(user_guid) |>
+      summarize(n_visits = n()) |>
+      left_join(user_names(), by = "user_guid") |>
+      replace_na(list(full_name = "[Anonymous]")) |>
+      arrange(desc(n_visits)) |>
+      select(n_visits, full_name, username)
+  }) |> bindCache(date_range()$from_date, date_range()$to_date, input$content_guid)
+
+  summary_message <- reactive({
+    content_title <- content() |>
+      filter(guid == input$content_guid) |>
+      pull(title)
+    hits <- all_visits_data()
+    glue(
+      "Content '{content_title}' had {nrow(hits)} between ",
+      "{min(hits$timestamp)} and {max(hits$timestamp)}."
+    )
+  })
+
+
+  output$summary_message <- renderText(summary_message())
+  output$all_visits <- renderTable(
+    all_visits_data() |>
+      transmute(timestamp = format(timestamp, "%Y-%m-%d %H:%M:%S"), full_name, username) |>
+      rename("Time" = timestamp, "Full Name" = full_name, "Username" = username)
+  )
+  output$aggregated_visits <- renderTable(
+    aggregated_visits_data() |>
+      rename("Total Visits" = n_visits, "Full Name" = full_name, "Username" = username)
+  )
+}
+
+shinyApp(ui, server)

extensions/who-is-visiting-this-content/get_usage.R

@@ -0,0 +1,60 @@
+library(connectapi)
+
+# This file contains functions that ultimately will more likely be part of
+# connectapi. As such, I'm not using dplyr or pipes here.
+
+NA_datetime_ <- vctrs::new_datetime(NA_real_, tzone = "UTC")
+NA_list_ <- list(list())
+
+usage_dtype <- tibble::tibble(
+  "id" = NA_integer_,
+  "user_guid" = NA_character_,
+  "content_guid" = NA_character_,
+  "timestamp" = NA_datetime_,
+  "data" = NA_list_
+)
+
+# A rough implementation of how a new firehose usage function would work in
+# `connectapi`.
+get_usage_firehose <- function(client, from = NULL, to = NULL) {
+  usage_raw <- client$GET(
+    connectapi:::unversioned_url("instrumentation", "content", "hits"),
+    query = list(
+      from = from,
+      to = to
+    )
+  )
+
+  # FIXME for connectapi: This is slow, it's where most of the slowness is with
+  # the new endpoint.
+  usage_parsed <- connectapi:::parse_connectapi_typed(usage_raw, usage_dtype)
+
+  usage_parsed[c("user_guid", "content_guid", "timestamp")]
+}
+
+get_usage_legacy <- function(client, from = NULL, to = NULL) {
+  shiny_usage <- get_usage_shiny(client, limit = Inf, from = from, to = to)
+  shiny_usage_cols <- shiny_usage[c("user_guid", "content_guid")]
+  shiny_usage_cols$timestamp <- shiny_usage$started
+
+  static_usage <- get_usage_static(client, limit = Inf, from = from, to = to)
+  static_usage_cols <- static_usage[c("user_guid", "content_guid")]
+  static_usage_cols$timestamp <- static_usage$time
+
+  bind_rows(shiny_usage_cols, static_usage_cols)
+}
+
+get_usage <- function(client, from = NULL, to = NULL) {
+  from <- format(from, "%Y-%m-%dT%H:%M:%SZ")
+  to <- format(to, "%Y-%m-%dT%H:%M:%SZ")
+  tryCatch(
+    {
+      print("Trying firehose usage endpoint.")
+      get_usage_firehose(client, from, to)
+    },
+    error = function(e) {
+      print("Could not use firehose endpoint; trying legacy usage endpoints.")
+      get_usage_legacy(client, from, to)
+    }
+  )
+}