#install.packages('countrycode')

library(countrycode)

# The Correlates of War (COW) uses a particular set of country codes that is different
# from the ISO3 codes supported by various graphing and charting libraries.
# This function uses the R countrycode package for the necessary conversion from COW to ISO3. 
# However, the package does not have any data for Kosovo which is present in the COW Contiguity 
# dataset and so this is handled separately by converting it to Serbia
# Refer https://en.wikipedia.org/wiki/Kosovo%E2%80%93Serbia_relations
lookupISO3byCOWC <- function(cowcVal) {
  if (cowcVal == c('KOS')) {
    iso3cVal <- c('SRB')
    #iso3cVal <- c('KSV')
  } else {
    iso3cVal <- countrycode(cowcVal, 'cowc', 'iso3c')
  }
  return(iso3cVal)
}

# Uses the countrycode_data dataframe from the countrycode package to dump the list
# of ISO3 codes for all countries. Again Kosovo is added separately. Refer to the 
# description of the above function lookupISO3byCOWC
generateListOfAllCountriesISO3 <- function(outFilename) {
  
  df_countries <- countrycode_data[, c('cowc','iso3c', 'country.name.en')]

  #kos_add <-data.frame("KOS","KSV", "Kosovo")
  #names(kos_add)<-c('cowc','iso3c', 'country.name.en')

  #df_countries <- rbind(df_countries, kos_add)
  
  all_countries_iso3 <- subset(df_countries, select=c('iso3c'), !is.na(iso3c))
  
  N <- nrow(all_countries_iso3)

  cat("Number of countries/nodes (N) = ", N, '\n')
  
  write.table(all_countries_iso3,file=outFilename, quote=F, row.names=F, col.names=FALSE)
  cat("Generated output data file containing list of all countries (ISO3) :", outFilename, '\n')
}

# Processes the Correlates of War Contiguity dataset and prints of pairs of countries
# (ISO3 codes) that share a land or river border as on the latest timeframe 2001612. Refer to the data 
# dictionary for details of contiguity types
generateNeighborPairs <- function(inDataFilename, outFilename) {
  df <- read.csv(inDataFilename, stringsAsFactors=FALSE)
  df_edg <- subset(df, conttype==1 & end==201612, select=c('statelab','statehab'))
  states <- unique(c(df_edg$statelab,df_edg$statehab))

  Nbar <-length(states)
  M <- nrow(df_edg)

  cat("Number of countries with land or river border(s) (Nbar) =", Nbar, "; Number of borders/edges M =", M, '\n')

  df_edg$statelab <- sapply(df_edg$statelab, lookupISO3byCOWC )
  df_edg$statehab <- sapply(df_edg$statehab, lookupISO3byCOWC)

  write.table(df_edg,file=outFilename, quote=F,sep=" ",row.names=FALSE, col.names=FALSE)
  cat("Generated output data file containing neighbor country pairs (ISO3) :", outFilename,'\n')
  
}

##### Main #####
generateListOfAllCountriesISO3("data/all-countries-iso3.dat")
generateNeighborPairs("data/contdir.csv", "data/country-borders-landorriver-iso3.dat")