##
# script for interpolating tracks using crawl prior to SSF resampling

####packages####
rm(list=ls())
library(crawl)
library(dplyr)
library(sf)
library(ggplot2)
library(rgdal)
library(raster)
library(mapview)
library(tidyr)

###get data####
#for finding ladder #one large file
data_frame <- read.csv('data/find ladder/all_approaches_60mins.csv')
#check data
names(data_frame)
head(data_frame)

#rename some columns
data_frame$fish_id <- data_frame$id

#for some reason discharge columns were weird, use whatever lines are needed

#data_frame$discharge <- data_frame$discharge_x
data_frame$discharge_nearest_10 <- data_frame$discharge_nearest_10_x


#drop extra discharge colums
data_frame <- subset(data_frame,select=-c(discharge_x,discharge_y,
                                          discharge_nearest_10_x,discharge_nearest_10_y))

summary(data_frame)
str(data_frame)

#get time as posixct
data_frame$time2 <- as.POSIXct(data_frame$Time)

#get raster files for checking if in river
raster_files <- list.files("rasters")  #identify file names

#get substrate data for checking if in river #actually dont need too much but eh keep it
substrate_poly  <-readOGR('substrate/Materials_hydraulicModel.shp')


####LOOP####
#since runs fast

####prepare for crawl####
#
#
#
data <- data_frame
fish_id_list <-unique(data$fish_id)

#with river shape

river_shape <-readOGR('rivershp/new_river_shapefile.shp')
river_shape2 <- fortify(river_shape)

for(q in 1:length(fish_id_list)){
    
    data_crawled_fish <- data.frame(matrix(ncol=30,nrow=0))
    colnames(data_crawled_fish) <- c("TimeNum","locType","Time","id",  "tag",
                                     "lat","lon", "error_x","error_y" ,"error_xy",
                                     "nobs",   "signal_strength" ,
                                     "noise_rec" ,"quality_score","species", 
                                     "dist.from.fishway" ,"near_entrance" , "approach" ,
                                     "if.within.15m.of.ladder",
                                     "most_track_near_ladder","fish_id" ,
                                     "discharge" ,"discharge_nearest_10", "time2", 
                                     "x",   "y",   "mu.x","nu.x",   "mu.y",   "nu.y"   )
    
    errors_small <- data.frame(matrix(ncol=4,nrow=0))
    colnames(errors_small) <-c('fish_id','approach','track','reason')
    errors_small_blank <- errors_small
    
    
    
    ####crawl for new positions prior to ssf####
    #do this fish by fish actually <3 no loop
    #set id number
    
    #set fish id
    j <-fish_id_list[q]
    data_j <-filter(data,fish_id==j)
    
    #make blank df to append to
    
    #remove approaches outside migration period for Altusried
    
    if(data_j$species[1] == "Barbel"){
      data_j <- filter(data_j,time2<"2018-06-30"&time2>"2018-04-16")
      
    }
    
    if(data_j$species[1] == "Grayling"){
      data_j <- filter(data_j,time2<"2018-05-15"&time2>"2018-03-01")
      
    }
    
    approaches <- unique(data_j$approach)
    
    
    ##ok first: code to split approaches where >30s between points so crawl
    #doesnt interpolate there
    
    ####loop####
    starttime <- Sys.time()
    print(paste0('started fish id ',j,' at time: ',starttime))
    for (i in approaches) {
      #select approach
      start_approachtime <- Sys.time()
      data2 <- filter(data_j, approach==i)
      data2$time_diff <- as.numeric(data2$time2 - lag(data2$time2),units='secs')
      #remove points with less than 1.1s burst
      
      data2 <-filter(data2,time_diff>=1.1)
      
      print(paste0('Started approach ',i,' for fish ',j,' at time ',start_approachtime))
      #remove if no data
      #remove if most in ladder
      if(length(data2$Time)>1){
        if (data2$most_track_near_ladder[1]=='No'){
          data2$track_part <- 1
          for (p in 2:length(data2$id)){ 
            
            data2$track_part[p] <- ifelse(data2$time_diff[p]>60,
                                          data2$track_part[p-1]+1,
                                          data2$track_part[p-1])
            
          }
          
          #add in if for total time of track as well?
          start_time <- min(data2$time2) #get start time
          end_time <- max(data2$time2)
          track_time <- as.numeric(end_time - start_time,units='secs')
          #if track lasts longer than 10mins? 15mins?
          if(track_time>900){
            #so if track is >30 mins 
            #then split into track parts
            num_tracks <-unique(data2$track_part)
            for(z in num_tracks){
              #get track
              data3 <-filter(data2,track_part==z)
              if (length(data3$Time)>10){ #if >10 detections in a part
                
                #THEN filter data for if in river or not <3
                #get coords
                data.points <-data3[,c(14,15)]
                data.points.fin <- data.points
                
                
                for(x in 1:length(raster_files)) { 
                  #line below is to get the heading name for pasting to dataframe
                  c <-gsub('.tif','', raster_files[x]) 
                  data.points.fin[[paste0(c)]] <- raster::extract(raster(paste0("rasters/",
                                                                                raster_files[x])),data.points)
                }
                
                for(y in 1:length(data3$fish_id)) {
                  #get discharge at i
                  discharge_start_10s <- data3$discharge_nearest_10[y]
                  if (discharge_start_10s>80){
                    discharge_start_10s = 80
                  }
                  #getheadings for the discharge
                  headings <- names(data.points.fin)[grepl(discharge_start_10s,names(data.points.fin))]
                  
                  #individual headings for each variable
                  #this means can direct r where to TAKE VALUES FROM
                  depth_head  <- headings[grepl('depth',headings )]
                  vel_head  <- headings[grepl('vel_',headings )]
                  velgrad_head  <- headings[grepl('svg',headings )]
                  
                  #then add to df
                  #IT WORKS!!
                  data3[y, 'depth'] <- data.points.fin[y, depth_head]
                  data3[y, 'water_velocity'] <- data.points.fin[y, vel_head]
                  data3[y, 'svg'] <- data.points.fin[y, velgrad_head]
                  
                }
                
                #add substrate
                
                
                substrate_extract <- raster::extract(substrate_poly,data.points)
                substrate_type <- substrate_extract$MATNAME
                
                
                #duplicated data to be safe and check its ok
                data_with_substrate<- cbind(data3, substrate_type)
                
                #filter to remove nas
                
                data_frame_filtered <- drop_na(data_with_substrate,c(svg,depth,
                                                                     water_velocity,substrate_type))
                
                
                #crawl?
                #make into sf data type
                sf_data <- st_as_sf(data_frame_filtered,coords=c('x','y')) %>% st_set_crs(32632)
                #get crawl errors
                crawl_model <- crwMLE(data=sf_data,Time.name='time2')
                
                #make time sequence
                start_time <- min(sf_data$time2) #get start time
                end_time <- max(sf_data$time2) #get end time #possibly do as +5 to ensure get the end?
                
                times <- seq.POSIXt(from=start_time, to=end_time, by='20 sec') #by 5s as easier to filter to 10s
                
                #this uses the model make above with crwMLE to predict positions at the specified time interval
                crawled_data <- crwPredict(crawl_model,times)
                data_crawled_fish <- rbind(data_crawled_fish,crawled_data)
              }
              else{
                print(paste0('ERROR, fish id ',j,' approach ',i,' track number ',z, ' has less than 10 data points'))
                errors_small2 <-errors_small_blank
                errors_small2[1,] <- c(j,i,z,'<10 points')
                errors_small <- rbind(errors_small,errors_small2)
              }
            }
          }
          else{
            print(paste0('ERROR, fish id ',j,' approach ',i,' is less than 15mins'))
            errors_small2 <-errors_small_blank
            errors_small2[1,] <- c(j,i,'NA','<15mins')
            errors_small <- rbind(errors_small,errors_small2)
          } 
        }
        else{
          print(paste0('ERROR, fish id ',j,' approach ',i,' is mostly in the area near ladder'))
          errors_small2 <-errors_small_blank
          errors_small2[1,] <- c(j,i,'NA','most points within 15m of ladder')
          errors_small <- rbind(errors_small,errors_small2)
        }
      }
      else{
        print(paste0('ERROR, fish id ',j,' approach ',i,' has 0 data points after filtering on burst interval'))
        errors_small2 <-errors_small_blank
        errors_small2[1,] <- c(j,i,'NA','0 detections after filter on burst interval')
        errors_small <- rbind(errors_small,errors_small2)
      }  
    }
    
    #then save to a csv.
    #but 1 csv per fish
    
    endtime <- Sys.time()
    print(paste0('Finished fish id ',j,' at time: ',endtime))
    print(paste0('time taken: ',as.numeric(endtime-starttime,units='mins'), ' minutes'))
    

    ####save csv####
    write.csv(data_crawled_fish,paste0('data/find ladder/data near ladder CRAWLED/',data_crawled_fish$fish_id[1],'_60min_track_crawl_20s.csv'),row.names = F)

    write.csv(errors_small,paste0('data/find ladder/ERRORS when crawl - if too small/',data_crawled_fish$fish_id[1],'_60min_small_approaches.csv'),row.names=F)
    
    ####actual plot####
    
    #final approaches
    fin_approaches <- unique(data_crawled_fish$approach)
    #fin_approaches <-fin_approaches[fin_approaches<20]
    for (i in fin_approaches){
      data_i <- subset(data_crawled_fish,approach==i)
      data_i <- as_tibble(data_i)
      data_i <- mutate(data_i,num=row_number())
      plot <- ggplot(data_i,aes(x=mu.x,y=mu.y))+
        geom_point(aes(col=num))+
        geom_path(aes(col=num))+
        geom_polygon(data=river_shape2,aes(long,lat),col='black', fill=NA)+
        #coord_cartesian(xlim=c(591660,591690),ylim=c(5296870,5296895))+
        #point to represent the ladder
        geom_point(aes(x=591683.2,y=5296867.5), colour='red')+
        labs(x='x',y='y',title=paste0('approach number ',i,' for fish id',data_i$fish_id[1]))
      print(plot)
      
    }
    
}