library(shiny)
library(ggplot2)
library(dplyr)
library(tidyr)

ui <- fluidPage(
  titlePanel("クライアント型電子計算機の分析可視化（連動型）"),
  sidebarLayout(
    sidebarPanel(
      # 残存関数パラメータ
      sliderInput("alpha", "残存関数のパラメータ alpha:", 
                  min = 1, max = 4, value = 2.58, step = 0.01),
      sliderInput("lambda", "残存関数のパラメータ lambda:", 
                  min = 5, max = 15, value = 8.0, step = 0.1),
      
      # ストック推計関連パラメータ
      sliderInput("base_shipment", "基準年出荷台数（万台）:", 
                  min = 800, max = 1200, value = 1000, step = 10),
      sliderInput("decline_rate", "出荷台数減少率（%/年）:", 
                  min = 0, max = 5, value = 2, step = 0.1),
      
      # TEC関連パラメータ
      sliderInput("base_tec", "基準TEC値（kWh）:", 
                  min = 15, max = 35, value = 25, step = 0.5),
      sliderInput("tec_improvement", "TEC改善率（%/年）:", 
                  min = 0, max = 10, value = 3, step = 0.1),
      
      # TEC分布共通インタラクティブ項目
      sliderInput("screen_min", "画面サイズ（最小）", 
                  min = 8, max = 14.9, value = 8, step = 0.1),
      sliderInput("screen_max", "画面サイズ（最大）", 
                  min = 15.1, max = 18, value = 18, step = 0.1),
      numericInput("mean_tec", "TEC平均値", value = 20, min = 0),
      numericInput("sd_tec", "TEC標準偏差", value = 5, min = 0.1),
      checkboxInput("show_medians", "中央値ラベルを表示", value = TRUE),
      checkboxInput("show_vline", "15インチ区切り線を表示", value = TRUE)
    ),
    mainPanel(
      tabsetPanel(
        tabPanel("残存関数", plotOutput("residualPlot")),
        tabPanel("ストック推計", plotOutput("stockPlot")),
        tabPanel("エネルギー消費量", plotOutput("energyPlot")),
        tabPanel("TEC分布（ノートPC）", plotOutput("tecNotePlot")),
        tabPanel("TEC分布（Pスコア）", plotOutput("tecPscorePlot")),
        tabPanel("TEC分布（デスクトップ）", plotOutput("tecDesktopPlot")),
        tabPanel("TEC分布（分離型PC）", plotOutput("tecSplitPlot")),
        tabPanel("パラメータ感度分析", plotOutput("sensitivityPlot"))
      )
    )
  )
)


server <- function(input, output) {
  
  # Survival function calculation
  residual_function <- reactive({
    n <- 0:30
    residual_rate <- exp(- (n / input$lambda)^input$alpha)
    data.frame(Year = n, ResidualRate = residual_rate)
  })
  
  # Stock estimation calculation (linked with survival function)
  stock_estimation <- reactive({
    years <- 2000:2030
    n_years <- length(years)
    
    # Actual data (2000-2022)
    actual_stock <- c(55000000, 61000000, 67000000, 73000000, 80000000, 86000000, 
                      91000000, 96000000, 101000000, 106000000, 111000000, 115000000, 
                      118000000, 120000000, 121000000, 120000000, 116000000, 113000000,
                      112000000, 111000000, 108000000, 107000000, 106000000)
    
    # Estimated data (2023-2030) - Based on survival function parameters
    estimated_stock <- numeric(8)
    base_year_stock <- 106000000  # Based on 2022 value
    
    for (i in 1:8) {
      # Stock estimation using survival function
      age_distribution <- residual_function()
      survival_rate <- mean(age_distribution$ResidualRate[1:min(i+5, nrow(age_distribution))])
      
      # Consider shipment decline
      shipment_factor <- (1 - input$decline_rate/100)^i
      
      estimated_stock[i] <- base_year_stock * survival_rate * shipment_factor
    }
    
    # Create data frame
    data.frame(
      Year = years,
      Actual = c(actual_stock, rep(NA, 8)),
      Estimated = c(rep(NA, 23), estimated_stock)
    )
  })
  
  # Energy consumption calculation (linked with stock estimation and TEC values)
  energy_consumption <- reactive({
    stock_data <- stock_estimation()
    years <- 2006:2030
    
    # Actual data (2006-2022)
    actual_consumption <- c(10000, 10300, 10400, 10200, 10100, 10000, 9800, 9400, 9200, 8900,
                            8500, 8100, 7700, 7200, 6800, 6300, 6000)
    
    # Estimated data (2023-2030) - Based on stock and TEC improvement
    estimated_consumption <- numeric(8)
    
    for (i in 1:8) {
      year_idx <- 23 + i  # Index for 2023 onwards in stock_data
      stock_value <- stock_data$Estimated[year_idx]
      
      if (!is.na(stock_value)) {
        # Consider TEC improvement
        tec_factor <- input$base_tec * (1 - input$tec_improvement/100)^i
        
        # Energy consumption = Stock × TEC × Operating hours factor
        # Unit conversion: units × kWh → GWh
        operating_hours_factor <- 2000  # Annual operating hours
        estimated_consumption[i] <- (stock_value * tec_factor * operating_hours_factor) / 1e9
      } else {
        estimated_consumption[i] <- NA
      }
    }
    
    data.frame(
      Year = years,
      Actual = c(actual_consumption, rep(NA, 8)),
      Estimated = c(rep(NA, 17), estimated_consumption)
    )
  })
  
  # Plot outputs
  output$residualPlot <- renderPlot({
    df <- residual_function()
    
    ggplot(df, aes(x = Year, y = ResidualRate)) +
      geom_line(color = "darkorange", size = 1.2) +
      geom_point(color = "darkorange", size = 2) +
      labs(title = "Survival Function", 
           subtitle = paste("alpha =", input$alpha, ", lambda =", input$lambda),
           x = "Years (n)", y = "Survival Rate") +
      theme_minimal() + 
      ylim(0, 1.2) +
      theme(text = element_text(family = "sans"))
  })
  
  output$stockPlot <- renderPlot({
    df <- stock_estimation()
    df_long <- df %>% 
      mutate(Year = as.character(Year)) %>%
      pivot_longer(cols = c(Actual, Estimated), names_to = "Category", values_to = "Stock") %>%
      filter(!is.na(Stock))
    
    ggplot(df_long, aes(x = Year, y = Stock, fill = Category)) +
      geom_col(position = "dodge") +
      scale_fill_manual(values = c("Actual" = "steelblue", "Estimated" = "darkorange")) +
      scale_y_continuous(labels = scales::comma_format()) +
      theme_minimal() + 
      theme(axis.text.x = element_text(angle = 90, hjust = 1),
            text = element_text(family = "sans")) +
      labs(title = "Stock Estimation (Linked with Survival Function)", 
           subtitle = paste("Decline Rate:", input$decline_rate, "%/year"),
           x = "Year", y = "Stock (units)")
  })
  
  output$energyPlot <- renderPlot({
    data <- energy_consumption()
    data_long <- data %>% 
      pivot_longer(cols = c("Actual", "Estimated"), names_to = "Category", values_to = "Consumption") %>%
      filter(!is.na(Consumption))
    
    ggplot(data_long, aes(x = Year, y = Consumption, fill = Category)) +
      geom_col(position = "dodge") +
      scale_fill_manual(values = c("Actual" = "steelblue", "Estimated" = "darkorange")) +
      scale_y_continuous(labels = scales::comma_format()) +
      labs(title = "Energy Consumption Trends (Linked with Stock & TEC)", 
           subtitle = paste("TEC Improvement Rate:", input$tec_improvement, "%/year, Base TEC:", input$base_tec, "kWh"),
           x = "Year", y = "GWh/year") +
      theme_minimal() + 
      theme(axis.text.x = element_text(angle = 45, hjust = 1),
            text = element_text(family = "sans"))
  })
  
  # Sensitivity analysis plot
  output$sensitivityPlot <- renderPlot({
    # Current parameters for 2030 estimation
    current_stock <- tail(stock_estimation()$Estimated[!is.na(stock_estimation()$Estimated)], 1)
    current_energy <- tail(energy_consumption()$Estimated[!is.na(energy_consumption()$Estimated)], 1)
    
    # Parameter ranges for sensitivity analysis
    alpha_range <- seq(1.5, 3.5, 0.1)
    lambda_range <- seq(6, 12, 0.5)
    
    sensitivity_data <- expand.grid(alpha = alpha_range, lambda = lambda_range)
    sensitivity_data$stock_2030 <- NA
    sensitivity_data$energy_2030 <- NA
    
    for (i in 1:nrow(sensitivity_data)) {
      # Simplified calculation (streamlined calculation logic)
      alpha_temp <- sensitivity_data$alpha[i]
      lambda_temp <- sensitivity_data$lambda[i]
      
      survival_rate <- exp(-(8/lambda_temp)^alpha_temp)  # 8-year survival rate
      stock_factor <- (1 - input$decline_rate/100)^8
      
      sensitivity_data$stock_2030[i] <- 106000000 * survival_rate * stock_factor
      sensitivity_data$energy_2030[i] <- (sensitivity_data$stock_2030[i] * 
                                            input$base_tec * (1 - input$tec_improvement/100)^8 * 2000) / 1e9
    }
    
    ggplot(sensitivity_data, aes(x = alpha, y = lambda, fill = energy_2030)) +
      geom_tile() +
      scale_fill_gradient2(low = "blue", mid = "white", high = "red", 
                           midpoint = median(sensitivity_data$energy_2030, na.rm = TRUE)) +
      geom_point(aes(x = input$alpha, y = input$lambda), 
                 color = "black", size = 4, shape = 21, fill = "yellow") +
      labs(title = "Parameter Sensitivity Analysis (2030 Energy Consumption)",
           subtitle = "Yellow dot: Current setting",
           x = "Alpha", y = "Lambda", fill = "GWh/year") +
      theme_minimal()
  })
  
  # Existing TEC plots
  output$tecNotePlot <- renderPlot({
    set.seed(123)
    screen_size <- c(runif(25, input$screen_min, min(14.9, input$screen_max)),
                     runif(15, max(15.1, input$screen_min), input$screen_max))
    tec_value <- c(rnorm(25, input$mean_tec - 5, input$sd_tec),
                   rnorm(15, input$mean_tec + 5, input$sd_tec))
    df1 <- data.frame(ScreenSize = screen_size, TEC = tec_value)
    medians1 <- data.frame(
      ScreenSize = c(mean(c(input$screen_min, 14.9)), mean(c(15.1, input$screen_max))),
      TEC = c(input$mean_tec - 5, input$mean_tec + 5),
      Label = sprintf("%.2f", c(input$mean_tec - 5, input$mean_tec + 5))
    )
    
    p <- ggplot(df1, aes(x = ScreenSize, y = TEC)) +
      geom_point(color = "blue") +
      labs(title = "TEC Distribution: Notebook PC (Category J・K)", x = "Screen Size", y = "TEC[kWh]") +
      theme_minimal()
    
    if (input$show_vline) {
      p <- p + geom_vline(xintercept = 15, linetype = "dashed")
    }
    if (input$show_medians) {
      p <- p + geom_text(data = medians1, aes(label = Label), vjust = -1, color = "red", size = 4)
    }
    p
  })
  
  output$tecPscorePlot <- renderPlot({
    set.seed(124)
    df2 <- data.frame(P_score = runif(30, 8, 13), TEC = rnorm(30, 20, 5))
    ggplot(df2, aes(x = P_score, y = TEC)) + 
      geom_point(color = "darkgreen") +
      geom_segment(aes(x = 10.5, xend = 12.5, y = 11.34, yend = 11.34), color = "red") +
      annotate("text", x = 12.7, y = 11.34, label = "11.34") +
      labs(title = "TEC Distribution: P-score 8 or Higher (Category L)", x = "P-score", y = "TEC[kWh]") +
      theme_minimal()
  })
  
  output$tecDesktopPlot <- renderPlot({
    set.seed(125)
    df3 <- data.frame(P_score = runif(40, 2, 15), TEC = runif(40, 30, 180))
    ggplot(df3, aes(x = P_score, y = TEC)) + 
      geom_point() +
      geom_vline(xintercept = 8, linetype = "dashed") +
      annotate("text", x = 5, y = 180, label = "Category M") +
      annotate("text", x = 12, y = 180, label = "Category N") +
      geom_segment(aes(x = 3, xend = 7, y = 39.87, yend = 39.87), color = "red") +
      geom_segment(aes(x = 10, xend = 14, y = 53.32, yend = 53.32), color = "red") +
      annotate("text", x = 7.1, y = 39.87, label = "39.87") +
      annotate("text", x = 14.1, y = 53.32, label = "53.32") +
      labs(title = "TEC Distribution: Integrated Desktop (M/N)", x = "P-score", y = "TEC[kWh]") +
      theme_minimal()
  })
  
  output$tecSplitPlot <- renderPlot({
    set.seed(126)
    df4 <- data.frame(Volume = runif(40, 0, 20), TEC = runif(40, 20, 140))
    ggplot(df4, aes(x = Volume, y = TEC)) + 
      geom_point() +
      geom_vline(xintercept = 10, linetype = "dashed") +
      annotate("text", x = 4, y = 140, label = "Category O") +
      annotate("text", x = 14, y = 140, label = "Category P") +
      geom_segment(aes(x = 3, xend = 8, y = 29.55, yend = 29.55), color = "red") +
      geom_segment(aes(x = 11, xend = 15, y = 31.33, yend = 31.33), color = "red") +
      annotate("text", x = 8.1, y = 29.55, label = "29.55") +
      annotate("text", x = 15.1, y = 31.33, label = "31.33") +
      labs(title = "TEC Distribution: Separate PC (O/P)", x = "Volume[L]", y = "TEC[kWh]") +
      theme_minimal()
  })
}

shinyApp(ui = ui, server = server)