## server app for shiny to create plots based on inputs from ui
## data are loaded from .Rdata file
## Load libraries

library(shiny)
library(ggplot2)
# library(mapproj)
library(plyr)
library(scales)

## Load data, specify path when needed
load(file = "data/MWTL_Schelde_bewerkt.Rdata")
# load(file = "data/map.Rdata")
alllocs <- as.character(unique(rws_dat$locatie))

## Define a server for the Shiny app
shinyServer(function(input, output) {
  
  output$substanceControls <- renderUI({
    periodSelection <- subset(rws_dat, 
                              rws_dat$year >= input$interval[1] & rws_dat$year <= input$interval[2])
    substances <- unique(periodSelection$variable)
    selectInput("substance", "Substance:",
                choices = substances, selected = 'ammonium'
                #                    helpText("select substance"),
    )
  })
  
  output$locationControls <- renderUI({
    locationSelection <- subset(rws_dat, 
                                rws_dat$year >= input$interval[1] & rws_dat$year <= input$interval[2] &
                                  rws_dat$variable == input$substance)
    locations <- unique(locationSelection$locatie)
    selectInput("location", "Location:",
                choices = locations, selected = "Schaar van Ouden Doel")
    #                    helpText("select location"),
  })
  
  
  output$timePlot <- renderPlot({
    ## Select data that meet the criteria defined in the UI
    subdf = subset(rws_dat, rws_dat$variable == input$substance &  #"phosphate" &
                     rws_dat$locatie == input$location &   #  "Groote Gat noord"  &
                     (rws_dat$year >= input$interval[1] & rws_dat$year <= input$interval[2])
    )
    unit = unique(subdf$eenheid)
    subdf = subdf[!is.na(subdf$waarde),]
    
    ## Create statistics information needed in the plot annotation
    if(input$analysis == "trend") {
      x = as.numeric(subdf$datetime); y = subdf$waarde    
      regression <- summary(glm(y ~ x + I(cos(2 * pi * x / 31556995)) + I(sin(2 * pi * x / 31556995))))
      slope <- regression$coefficients[2,1]  #slope of regression including season effect
      pvalue <- format(regression$coefficients[2,4], digits = 2)
      intercept <- regression$coefficients[[1]]
    }
    
    ## make parameters for positioning of plot annotation
    minscale <- min(subdf$datetime)
    maxscale <- max(subdf$datetime)
    yposition <- quantile(subdf$waarde, 0.99, na.rm = T)
    #     q90 <- function(x) {quantile(x,probs=0.9)}
    
    subdf_year <- ddply(subdf, ~ year + season, summarize, perc = quantile(waarde, probs = 0.9), na.rm = T)
    subdf_year$year <- as.POSIXct(paste(as.character(subdf_year$year), "-07-01 00:00:00", sep = ""))
    
    summer90 <- quantile(subset(subdf, subdf$season == "summer")$waarde,  probs = 0.9, na.rm = T)
    winter90 <- quantile(subset(subdf, subdf$season == "winter")$waarde,  probs = 0.9, na.rm = T)
    
    ## create plot 
    
    pl <- ggplot(subdf, aes(datetime, waarde))
    pl <- pl + facet_grid(variable ~ locatie, scales = "free")
    #            pl <- pl + geom_smooth(aes(color = season) , method="lm", size = 2)
    if(input$analysis == "trend") {   # in case trend is choice of analysis
      pl <- pl + geom_point(aes(), color = "seagreen4", alpha = 0.5)
      pl <- pl + geom_smooth(aes(), method='lm', 
                             formula = y ~ x+I(cos(2*pi*x/31622400))+I(sin(2*pi*x/31622400)), 
                             size = 0.7, alpha = 0.3, n = 1000)
      pl <- pl + geom_abline(intercept = intercept, slope = slope , color = "darkblue", size = 1)
      pl <- pl + annotate("text", label = paste("slope =", format(slope*31622400, digits = 2), unit, "per year | ", "p=", pvalue),
                          x = maxscale - 0.5 * (maxscale - minscale), y = yposition)
    }
    if(input$analysis == "per90") {  # in case 90 percentile is choice of analysis
      pl <- pl + geom_point(aes(color = season), size = 2, alpha = 0.4)
      pl <- pl + geom_crossbar(data = subdf_year, aes(year, perc, color = season), size = 0.75, ymin =F, ymax = F)
      #       pl <- pl + geom_hline(yintercept = summer90, color = "orange", size = 1, alpha = 0.5)
      #       pl <- pl + geom_hline(yintercept = winter90, color = "turquoise4", size = 1, alpha = 0.5)
      if(input$substance == "chlorofyl-a"){
        if(input$location %in% c("Bocht van Watum", "Groote Gat noord")){
          pl <- pl + geom_hline(yintercept = 18, size = 1,  linetype = 21, alpha = 0.5) +
            annotate("text", label = "maatlat", 
                     x = maxscale - 0.5 * (maxscale - minscale), y = 17)
        }
        if(input$location %in% c("Huibertgat oost")){
          pl <- pl + geom_hline(yintercept = 21, size = 1,  linetype = 21, alpha = 0.5) +
            annotate("text", label = "maatlat", 
                     x = maxscale - 0.5 * (maxscale - minscale), y = 20)
        }
      }
    }
    if(input$analysis == "loess") {   # in case trend is choice of analysis
      pl <- pl + geom_point(aes(), color = "seagreen4", alpha = 0.5)
      pl <- pl + geom_smooth(aes(), method='loess', span = input$lspan, size = 1)
    }
    
    pl <- pl + theme(text = element_text(size = 16), legend.position = "bottom")
    pl <- pl + xlab("date") + ylab(unit)
    print(pl)
    
  })
  output$boxPlot <- renderPlot({
    
    ## Select data that meet the criteria defined in the UI
    subdf = subset(rws_dat, rws_dat$variable == input$substance &
                     rws_dat$locatie == input$location &
                     rws_dat$waarde < 500 &
                     rws_dat$year >= input$interval[1] &
                     rws_dat$year <= input$interval[2])
    unit = unique(subdf$eenheid)
    
    minscale <- min(subdf$datetime)
    maxscale <- max(subdf$datetime)
    
    summer90 <- quantile(subset(subdf, subdf$season == "summer")$waarde,  probs = 0.9)
    winter90 <- quantile(subset(subdf, subdf$season == "winter")$waarde,  probs = 0.9)
    subdf_season <- ddply(subdf, ~ season, summarize, perc = quantile(waarde, probs = 0.9))
    
    ## Create plot
    q <- ggplot(subdf,aes(month,waarde))
    if(input$analysis == "trend") {   # in case trend is choice of analysis
      q <- q + geom_jitter(position = position_jitter(width = .5), colour = "seagreen4", alpha = 0.3)
      q <- q + geom_boxplot(outlier.colour = "orange", alpha = 0.5, color = "orange", notch = F, notchwidth = 0.5, outlier.size = F)
      q <- q + geom_smooth(aes(group = 1), method='lm', formula = y ~ I(cos(2*pi*x/12))+I(sin(2*pi*x/12)), size = 0.7, alpha = 0.2)
    }
    if(input$analysis == "per90") {  # in case 90 percentile is choice of analysis
      #       q <- q + geom_violin(outlier.colour = "orange", alpha = 0.5, color = "orange", notch = F, notchwidth = 0.5)
      q <- q + geom_jitter(aes(color = season), position = position_jitter(width = .5), alpha = 0.6)
      q <- q + geom_hline(data = subdf_season, aes(yintercept = perc, color = season), size = 0.75)
      #       q <- q + geom_hline(yintercept = summer90, color = "orange", size = 1, alpha = 0.5)
      #       q <- q + geom_hline(yintercept = winter90, color = "turquoise4", size = 1, alpha = 0.5)
      if(input$substance == "chlorofyl-a"){
        if(input$location %in% c("Bocht van Watum", "Groote Gat noord")){
          q <- q + geom_hline(yintercept = 18, size = 1,  linetype = 21, alpha = 0.5) #+
          annotate("text", label = "maatlat", 
                   x = maxscale - 0.5 * (maxscale - minscale), y = 17)
        }
        if(input$location %in% c("Huibertgat oost")){
          q <- q + geom_hline(yintercept = 21, size = 1,  linetype = 21, alpha = 0.5) #+
          #             annotate("text", label = "maatlat", 
          #                      x = maxscale - 0.5 * (maxscale - minscale), y = 20)
        }
      }
    }
    if(input$analysis == "loess") {   # in case trend is choice of analysis
      q <- q + geom_jitter(position = position_jitter(width = .5), colour = "seagreen4", alpha = 0.3)
      q <- q + geom_violin(alpha = 0.5, color = "orange")
    }    
    q <- q + facet_grid(variable ~ locatie, scales = "free") +
      theme(text = element_text(size = 16), legend.position = "bottom") +
      xlab("month") + ylab(unit)
    print(q)
    
  })
  
  #   output$map <- renderPlot({
  #     ## Select data that meet the criteria defined in the UI
  #     maploc   <-   input$location
  #     locmapsub  <- subset(locmap, locmap$locatie == maploc)
  #     
  #     alllocmap <- subset(locmap, locmap$locatie %in% alllocs)
  #     
  #     ## Create map
  #     map <- ggplot() +
  #       geom_polygon(data = NL_df, aes(x=long, y=lat, group=group),
  #                    color = "lightgrey", fill = "darkgrey") +
  #       geom_polygon(data = GER_df, aes(x=long, y=lat, group=group),
  #                    color = "lightgrey", fill = "darkgrey")
  #     
  #     xxlim <- c(5,7.35)   ## selected range suitable for Ems estuary
  #     yylim <- c(53,54)
  #     
  #     map +
  #       coord_map("ortho", xlim = xxlim, ylim = yylim, orientation=c(55, 10, 0)) +
  #       geom_point(aes(X_WGS, Y_WGS), data = alllocmap, alpha = 0.4, size = 3, color = "blue") +
  #       geom_point(aes(X_WGS, Y_WGS), data = locmapsub, size = 15, shape = "*", color = "red") +
  #       geom_text(aes(X_WGS, Y_WGS, label = locatie),hjust = 1, vjust = 0, data = locmapsub, size = 4) +
  #       theme(axis.text = element_blank(),
  #             axis.ticks = element_blank(),
  #             axis.title = element_blank())
  #     
  #   })
  
})