#Load necessary libraries
library(igraph)
library(tidyverse)

#Load in the boater matrix we're using.
boats = read.csv("Kao's boater networks/boats1.csv")

#Lake info spreadsheet--contains info on infestation statuses of all lakes, their county affiliation, their dow numbers and names, etc.
AIS = read.csv("lake_info.csv")
AIS$DOW = as.numeric(AIS$DOW) #Warning can be ignored here.

#Filter out and rename columns
AIS = dplyr::select(AIS, DOW, zm=zm2019, ss=ss2019, ew=ew2019, sf=sf2019)

#Join the two files together so we have infestation statuses by edge.
boats2 = left_join(boats, AIS, by = c("dow_origin" = "DOW"), relationship = "many-to-many")
boats3 = left_join(boats2, AIS, by = c("dow_destination" = "DOW"), suffix = c(".donor", ".receiver"), relationship = "many-to-many")

#Remove self edges and duplicated edges from the joins, then calculate b.ij indicator variable with and without SWF included for each edge. Bij is > 0 for every 1 species that could move along that edge that the originating lake has and the receiving lake does not.
boats4 = boats3 %>% 
  filter(dow_origin != dow_destination) %>% 
  distinct(dow_origin, dow_destination, .keep_all = TRUE) %>% 
  rowwise() %>% 
  mutate(
                bij.swf = sum(
                  c(zm.donor, ss.donor, ew.donor, sf.donor) * 
                    (1-c(zm.receiver, ss.receiver, ew.receiver, sf.receiver))),
                bij.noswf = sum(
                  c(zm.donor, ss.donor, ew.donor) * 
                    (1-c(zm.receiver, ss.receiver, ew.receiver)))
        )

#Select down to just the essential columns and filter out all edges not carrying risky boats.
boats_reduce = boats4 %>% 
  dplyr::select(from = dow_origin, to = dow_destination, weight, bij = bij.swf) %>% 
  filter(bij > 0)

boats_reduce_noswf = boats4 %>% 
  dplyr::select(from = dow_origin, to = dow_destination, weight, bij = bij.noswf) %>% 
  filter(bij > 0)

#Save the resulting files.
saveRDS(boats_reduce, "boats_reduce")
saveRDS(boats_reduce_noswf, "boats_reduce_noswf")