app.R

# This is a Shiny web application. You can run the application by clicking
# the 'Run App' button above.
# version to get data from S3 bucket files

library(shiny)
library(tidyverse)
library(rtweet)
library(ggthemes)
library(aws.s3)
library(lubridate)
library(DT)
# setting DB connection and getting users and tweets
setwd("~/aupolitics")
# Use local copy of data

# load tweets from local file and compare by status_id with Db
load("tweets_app.RData")


# Define UI for application that draws a histogram
ui <- fluidPage(
   # Google Analytics code
   tags$head(includeHTML("ga.html")),
   # Application title
   titlePanel("Australian Members of Parliament Twitter statistics. Updated daily"),
   
   # Sidebar with a selector for party
   sidebarLayout(
      sidebarPanel(
         selectInput("party", "Select party(s), use 'Backspace' key to delete",
                      c("Australian Labor Party", "Liberal Party of Australia", "The Nationals"),
                     selected=c("Australian Labor Party", "Liberal Party of Australia", "The Nationals"), 
                     selectize = TRUE, multiple = TRUE),
         dateInput("start_date", "Start date:", value=today()-30, min=today()-90, max=today()-1)
      ),
      
      # Show a plot of the generated distribution
      mainPanel(
        tabsetPanel(
          tabPanel("Devices",plotOutput("devPlot")),
          tabPanel("Timeline", plotOutput("timePlot")),
          tabPanel("Top tweets",
                   # Scatterplot
                   plotOutput(outputId = "tweetsPlot", brush = "plot_brush"),
                   tags$br(),
                   # Show data table
                   dataTableOutput(outputId = "tweetsTable")
                   ),
          tabPanel("Top tweets table", DT::dataTableOutput("toptweetsTable"))
        )
      )
   ),
   # Footer notes
   hr(),
   div(class = "footer",
       includeHTML("footer.html")
   )
)

# Define server logic required to draw a histogram
server <- function(input, output) {
   
   
   # Cross-session reactive file reader. 
   fileData <- reactiveFileReader(1000, NULL, 'tweets_app.RData', load)
   
   
   ## End(Not run)
  
  # reactive function to update data frame based on partie(s) and date
  t1 <- reactive({
    req(input$party, input$start_date) # ensure input$ vars available
    filter(t, as.Date(created_at)>as.Date(input$start_date)) %>% filter(party %in% input$party)
    })
  
  #tab 1 plot  
   output$devPlot <- renderPlot({
      # generate col chart for devices used per party based on input$party from ui.R
           
           clients <- t1() %>% group_by(source=as.factor(source)) %>% 
                   summarise(count = n()) %>% 
                   top_n(5, count) %>% arrange(desc(count)) 
           # draw the col chart
           clients %>% ggplot(aes(reorder(source,count),count, fill=source, label = count))+
                   geom_col()+
                   labs(x="", y="Number of tweets", fill="Source", title = "Twitter sources/devices")+
                   geom_text(size = 3, position = position_stack(vjust = 0.5))+
                   theme_economist()
           
      
   })
   output$timePlot <- renderPlot({
     # status updates by party since date selected
     
     ts_plot(t1(), by = "1 day", trim=1, tz="Australia/Sydney") +
       ggplot2::theme_light() +
       ggplot2::theme(plot.title = ggplot2::element_text(face = "bold")) +
       ggplot2::labs(
         x = NULL, y = NULL,
         title = paste0("Frequency of selected party MPs tweets, since ", as.character(input$start_date)),
         subtitle = "Tweets counts aggregated using 1-day intervals, trimmed",
         caption = "\nSource: Data collected from Twitter's REST API via rtweet"
       )+
       theme_economist()
     
   })
   
   output$tweetsPlot <- renderPlot({
     # get likes vs retweets, original only 
     t2 <- t1() %>% ungroup() %>%
       filter (is.na(retweet_status_id)) %>% plain_tweets() %>%
       mutate (link=paste0('<a href=', status_url, ' target="_blank" >On Twitter</a>')) %>%
       select (text, screen_name, retweets=retweet_count, likes=favorite_count, link, party)   
     
     # draw scatterplot
     ggplot(t2, aes_string(x = t2$likes, y = t2$retweets)) +
       geom_point(aes(color=party), size=3)+
       ggplot2::theme_light() +
       ggplot2::theme(plot.title = ggplot2::element_text(face = "bold")) +
       scale_x_log10()+
       scale_y_log10()+
       ggplot2::labs(
         x = 'Likes', y = 'Retweets',
         title = "Likes vs retweets",
         caption = "\nSource: Data collected from Twitter's REST API via rtweet"
       )+
       theme_economist()
   })
      # Create data table
      output$tweetsTable <- renderDataTable({
        
        t2 <- t1() %>% ungroup() %>%
        filter (is.na(retweet_status_id)) %>% plain_tweets() %>%
        mutate (link=paste0('<a href=', status_url, ' target="_blank" >On Twitter</a>')) %>%
        select (text, screen_name, retweets=retweet_count, likes=favorite_count, link, party)    
        b <- brushedPoints(t2, brush = input$plot_brush, xvar="likes", yvar="retweets") %>%
        select(text, screen_name, party, likes, retweets, link)  
        print (b)
     
   })
   
   output$toptweetsTable <- DT::renderDataTable({
     # find top 200 tweets for the party over given period of time
      
       t2 <- t1() %>% ungroup() %>%
       filter (is.na(retweet_status_id)) %>% plain_tweets() %>%
       mutate (link=paste0('<a href=', status_url, ' target="_blank" >On Twitter</a>')) %>%
       select (text, screen_name, retweets=retweet_count, likes=favorite_count, link, party) %>%
       top_n (200, likes+retweets) %>% arrange(likes) 
       DT::datatable(t2, escape = FALSE, 
                     options = list(order = list(list(4, 'desc'), list(3, 'desc')))
                     )
   })
}

# Run the application 
shinyApp(ui = ui, server = server)