Skip to content

Instantly share code, notes, and snippets.

@trestletech
Last active February 2, 2022 09:47
Show Gist options
  • Save trestletech/9793754 to your computer and use it in GitHub Desktop.
Save trestletech/9793754 to your computer and use it in GitHub Desktop.
A Shiny app combining the use of dplyr and SQLite. The goal is to demonstrate a full-fledged, database-backed user authorization framework in Shiny.
library(shiny)
library(dplyr)
library(lubridate)
# Load libraries and functions needed to create SQLite databases.
library(RSQLite)
library(RSQLite.extfuns)
saveSQLite <- function(data, name){
path <- dplyr:::db_location(filename=paste0(name, ".sqlite"))
if (!file.exists(path)) {
message("Caching db at ", path)
src <- src_sqlite(path, create = TRUE)
copy_to(src, data, name, temporary = FALSE)
} else {
src <- src_sqlite(path)
}
return (src)
}
# Load/create some data and put it in SQLite. In practice, the data you want
# likely already exists in the databse, so you would just be reading the data
# in from the database, not uploading it from R.
# Load and upload flights data
library(hflights)
hflights_db <- tbl(hflights_sqlite(), "hflights")
# Create a user membership data.frame that maps user names to an airline
# company.
membership <- data.frame(
user = c("kim", "sam", "john", "kelly", "ben", "joe"),
company = c("", "DL", "AA", "UA", "US", "DL"),
role = c("manager", rep("user", 5)))
membership_db <- tbl(saveSQLite(membership, "membership"), "membership")
airlines <- data.frame(
abbrev = c("AA", "DL", "UA", "US"),
name = c("American Airlines", "Delta Air Lines",
"United Airlines", "US Airways")
)
airline_db <- tbl(saveSQLite(airlines, "airline"), "airline")
#' Get the full name of an airline given its abbreviation.
airlineName <- function(abbr){
as.data.frame(select(filter(airline_db, abbrev == abbr), name))[1,1]
}
shinyServer(function(input, output, session) {
#' Get the current user's username
user <- reactive({
curUser <- session$user
# Not logged in. Shiny Server Pro should be configured to prevent this.
if (is.null(curUser)){
return(NULL)
}
# Look up the user in the database to load all the associated data.
user <- as.data.frame(
filter(membership_db, user==curUser)
)
# No user in the database
if (nrow(user) < 1){
return(NULL)
}
user[1,]
})
#' Determine whether or not the current user is a manager.
isManager <- reactive({
if (is.null(user())){
return(FALSE)
}
role <- user()$role
return(role == "manager")
})
#' Get the company of which the current user is a member
userCompany <- reactive({
if (is.null(user())){
return(NULL)
}
if (isManager()){
# If the user is a manager, then they're allowed to select any company
# they want and view its data.
if (is.null(input$company)){
return(as.data.frame(airline_db)$abbrev[1])
}
return(input$company)
}
# Otherwise this is just a regular, logged-in user. Look up what company
# they're associated with and return that.
user()$company
})
#' Get the data the current user has permissions to see
#' @return a dplyr tbl
companyData <- reactive({
# Trim down to only relevant variables
delays <- select(hflights_db, Month, DayofMonth, DepDelay, UniqueCarrier)
# Trim down to only values that we have permissions to see
comp <- userCompany()
delays <- filter(delays, UniqueCarrier == comp)
delays
})
#' Of the data a user is allowed to see, further refine it to only include the
#' date range selected by the user.
filteredData <- reactive({
# Get current month and day
curMonth <- month(now())
curDay <- day(now())
# Get the previous month and day based on the slider input
prevMonth <- month(now()-days(input$days))
prevDay <- day(now()-days(input$days))
# Filter to only include the flights in between the selected dates.
data <- filter(companyData(),
(Month > prevMonth | (Month == prevMonth & DayofMonth >= prevDay)) &
(Month < curMonth | (Month == curMonth & DayofMonth <= curDay)))
as.data.frame(data)
})
output$title <- renderText({
if(is.null(user())){
return("ERROR: This application is designed to be run in Shiny Server Pro and to require authentication.")
}
paste0("Airline Delays for ", airlineName(userCompany()))
})
output$userPanel <- renderUI({
if (isManager()){
# The management UI should have a drop-down that allows you to select a
# company.
tagList(
HTML(paste0("Logged in as <code>", user()$user,
"</code> who is a <code>", user()$role ,"</code>.")),
hr(),
p("As a manager, you may select any company's data you wish to view."),
selectInput("company", "", as.data.frame(airline_db)$abbrev)
)
} else{
# It's just a regular user. Just tell them who they are.
HTML(paste0("Logged in as <code>", user()$user, "</code> with <code>",
airlineName(userCompany()),"</code>."))
}
})
#' Print a boxplot of the selected data.
output$box <- renderPlot({
boxplot(
lapply(
split(filteredData(), as.factor(
paste0(filteredData()$Month, "/", filteredData()$DayofMonth))),
function(dayData){
dayData$DepDelay
}
), ylab = "Delay (minutes)"
)
})
})
library(shiny)
shinyUI(
fluidPage(
# Setup the page title
tagList(tags$head(tags$title("Airline Delays")), h1(textOutput("title"))),
sidebarLayout(
sidebarPanel(
uiOutput("userPanel"),
hr(),
sliderInput("days", "Prior days to include:", 1, 30, 7, 1),
hr(),
helpText("The graph on the right shows a boxplot of the departure " ,
"delays for the airline(s) your username is allowed to view.")
),
mainPanel(
plotOutput("box")
)
)
)
)
@KasperSkytte
Copy link

You can install removed packages from the CRAN archives: https://cran.r-project.org/src/contrib/Archive/RSQLite.extfuns/

@vzhomeexperiments
Copy link

usually I am encrypting passwords that are used in Shiny to connect with Databases. It can be done using package openssl or in case more people are working on that using package secret

@gadepallivs
Copy link

Is this a working example? I run into errors every time? Is there any updated version. Thank you

@mishaborys
Copy link

@trestletech Could you, please, say where can get working code?

Copy link

ghost commented Nov 15, 2018

i've seen a recent working example at https://github.com/paulc91/shinyauthr maybe it could suffice the same requirement

@wikithink
Copy link

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment