#! /usr/bin/env Rscript
# Original 2017 geocoding script

library(dplyr)
library(ggmap)

allmed <- read.csv("../Data/facilities20170901.csv", stringsAsFactors=FALSE)

gppractice <- filter(allmed, Facility.Type.Name == "GP Practice")
gppractice[gppractice$DHB.Name == "NULL", "DHB.Name"] <- "Missing"
DHBs <- unique(gppractice$DHB.Name)
DHBsToGet <- 13

# Notes:
# * Clean up missing values
# * Seems to fail for Taranaki

for (dhb in DHBs[DHBsToGet]) {
	print(dhb)
	gpdhb <- filter(gppractice, DHB.Name == dhb)
	print(nrow(gpdhb))
	geodhb <- geocode(gpdhb$Address, output <- "latlona") %>% rename("Calculated.Address" = address)
	gpdhb <- cbind(gpdhb, geodhb)
	write.csv(gpdhb, file = paste0("../Data/", dhb, ".csv"), row.names=FALSE)
}

# gphv <- filter(gppractice, DHB.Name == "Hutt Valley District Health Board")
# 
# geocodes <- geocode(gphv$Address, output = "latlona")
# geocodes <- rename(geocodes, "Calculated.Address" = address)
# gphv <- cbind(gphv, geocodes)
# 
# write.csv(gphv, file="../Data/HuttDHB.csv")

alldhbs <- NULL

for (dhb in DHBs) {
	alldhbs <- rbind(alldhbs, read.csv(paste0("../Data/", dhb, ".csv"), stringsAsFactors=FALSE))
}

write.csv(alldhbs, file="../Data/AllDHBsGeocoded.csv", row.names=FALSE)

# Read in allDHB data
alldhbs <- read.csv("../Data/AllDHBsGeocoded.csv", stringsAsFactors=FALSE)

# Calc found and notfound
found <- filter(alldhbs, !is.na(lon))
notfound <- filter(alldhbs, is.na(lon))

# Write found and not found
write.csv(found, file="../Data/FoundGPs.csv", row.names=FALSE)
notfound$lat <- NULL
notfound$lon <- NULL
notfound$Calculated.Address <- NULL
write.csv(notfound, file="../Data/NotFoundGPs.csv", row.names=FALSE)

# Read found and not found
found <- read.csv("../Data/FoundGPs.csv", stringsAsFactors=FALSE)
notfound <- read.csv("../Data/NotFoundGPs.csv", stringsAsFactors=FALSE)

#notfound$Address <- paste0(notfound$Name, ", ", notfound$Address)
nfgeo <- geocode(notfound$Address, output = "latlona")
nfgeo <- rename(nfgeo, "Calculated.Address" = address)
notfound <- cbind(notfound, nfgeo)


# Join found and not found
#notfound <- cbind(notfound, lon=NA, lat=NA, Calculated.Address=NA)
alldhbs <- rbind(found, notfound)

bad <- !grepl("new zealand", alldhbs$Calculated.Address)

badcode <- subset(alldhbs, bad )
goodcode <- subset(alldhbs, (!bad))
notfound <- badcode
found <- goodcode