# WITH TYPE AS RADIOBUTTON

library(shiny)
library(ggplot2)
library(dplyr)
library(lubridate)
library(tidyr)
library(dygraphs)
library(xts)
library(zoo)

# Read in the csv into a dataframe
kai_data<-read.csv("shiny_kaibosh.csv", stringsAsFactors = FALSE, header = TRUE)
kai_data$Month=factor(kai_data$Month)

print(kai_data)

##User Interface
ui <- fluidPage(

  br(),
  tags$img(src = "/kaibosh_logo_colour2_64770.png", width="300px"),
  br(), br(),

  sidebarLayout(
    sidebarPanel(
      uiOutput("unitOutput"),
      uiOutput("regionOutput"),
      uiOutput("typeOutput")
    ),
    mainPanel(
      dygraphOutput("dygraph"),
      # plotOutput("kaiboshplot"),
      br(), br(),
      tableOutput("results")
    )
  )
)

server <- function(input, output) {
  output$regionOutput <- renderUI({
    selectInput("regionInput", "Choose Region",
                unique(sort(kai_data$Region)),
                selected = "City")
  })

  output$unitOutput <- renderUI({
    radioButtons("unitInput", "Choose Unit",
                   choices = c("KG",
                               "Emissions",
                               "Meals",
                               "%"),
                    selected = "KG")
  })

  output$typeOutput <- renderUI({
    checkboxGroupInput("typeInput", "Choose Food Type",
                unique(sort(kai_data$Type)))
  })

  filtered <- reactive({
    if (is.null(input$regionInput)) {
      return(NULL)
    }
    if (is.null(input$typeInput)) {
      return(NULL)
    }

    # Filter by selected region
    kai_data <- filter(kai_data, Region == input$regionInput)
    # Update the dataset based on unit input
    if (input$unitInput == "KG") {
      kai_data
    } else if (input$unitInput == "Emissions") {
      kai_data["Weight"] = kai_data["Weight"] / 0.35
    } else if (input$unitInput == "Meals") {
      kai_data["Weight"] = kai_data["Weight"] * 0.361
    } else if (input$unitInput == "%") {
      kai_data
    }

    #Find all the selected checkboxes
    target <- c(input$typeInput)
    # now we have a list of selected types so we filter the dataset to just these
    filtered_kai <- filter(kai_data, Type %in% target)
    # Now we split into separate dataframes to convert to time series -> now we have a list of dataframes
    split_kai <- split(filtered_kai, filtered_kai$Type, drop = FALSE)
    # Let's loop through convert to time series!
    kai_as_xts <- lapply(split_kai, function(kai) xts(kai, order.by = as.Date(as.yearmon(kai$Month, "%Y-%m"))))
    # use cbind on every element in the list
    final_data <- do.call(cbind, kai_as_xts)
    return (final_data)

  })

  output$dygraph <- renderDygraph({
    dygraph(filtered())
  })

  # output$results <- renderTable({
  #   filtered()
  # })
}

##final line of code
shinyApp(ui = ui, server = server)