Geocoding in R using Google's API

gistfile1.r

# Geocoding in R using Google's API
require(RCurl)
require(XML)
require(data.table)
require(dplyr)

ridbLocs <- data.table(read.delim(file = "Data/ridbLocations.tsv"))
setnames(ridbLocs,c("original","locationType","simplified1","gCode","simplified","lat","lon"))

#' Model formula from my Google Sheet:
#' =IMPORTXML("https://maps.googleapis.com/maps/api/geocode/xml?address="&E2&"&key={insert api key here}","//result[1]/geometry[1]/location[1]/lng[1]")

# Dulani's Google Geocoding API Key 
apiKey <- "{insert api key here}"
baseURL <- "https://maps.googleapis.com/maps/api/geocode/"

# queryString <- "ECHTERDINGEN ARMY AIR FIELD, GERMANY"

geoQuery <- function(baseURL,apiKey,queryString,queryType=c("XML","JSON")) {
  #Model URL:
  #   https://maps.googleapis.com/maps/api/geocode/xml?address=ECHTERDINGEN ARMY AIR FIELD, GERMANY&key=AIzaSyB_QN2SLYO1TCKd_MUPVXEXVwe3jgo6myQ
  
  queryType <- switch(queryType[1],
                      XML = "xml",
                      JSON = "json")
  
  #Build URL
  url <- paste(baseURL,queryType,"?address=",queryString,"&key=",apiKey,sep="")
  
  #Replace spaces with pluses (Google's API doesn't seem to care. But that's what their own app does so I'm mimicking). 
  url <- gsub(pattern = " ",replacement = "+",x = url)
  
  result <- getURL(url)
  
  #Should do some parsing here of XML or JSON and return something more useful in R.  
  return(result)
}

setnames(ridbLocs,c("original","locationType","simplified1","gCode","simplified","lat","lon"))

for(curRow in 1:nrow(ridbLocs)) {
  # curRow <- 1
  cat(paste("Searching for:",ridbLocs[curRow,simplified],"\t\t"))
  fullXML <- geoQuery(baseURL,apiKey,queryString=ridbLocs[curRow,simplified])
  
  parsedXML <- xmlParse(fullXML)
  status <- xpathSApply(parsedXML, "//GeocodeResponse[1]/status[1]",xmlValue)
  
  if(status=="OK") {
    loc <- xpathSApply(parsedXML, "//result[1]/formatted_address[1]",xmlValue)
    cat(paste("Found: ",loc, "\n"))
  }
  if(status=="ZERO_RESULTS") {
    loc <- status
    cat(paste(status,"\n"))
  }
  
  ridbLocs <- ridbLocs[curRow,XML:=fullXML]  
  ridbLocs <- ridbLocs[curRow,locationFound:=loc]  
  Sys.sleep(time = 0.25) #Google's "free" limit is 5 per second and 2,500 per day. 
}

#For some reason, the lat/lon don't appear to be updated: 
getGeoCoord <- function(xml,coord=c("lat","lon")) {
#   browser()
#   xml <- ridbLocs[3,XML]
  #Extract the latitude or longitude from the XML output of Google's Geocoding API.
  xml <- xmlParse(xml)
  
  coord <- switch(coord[1],
                  lat = "lat",
                  lon = "lng" #Google uses the abbreviation 'lng' for longitude rather than the more common(?) 'lon'
  )
   
  geoCoord <- xpathSApply(xml, sprintf("//result[1]/geometry[1]/location[1]/%s[1]",coord),xmlValue)
  if(is.null(geoCoord)) {
    geoCoord <- NA
  } else {
    geoCoord <- as.numeric(geoCoord)    
  }
  return(geoCoord)
}

ridbLocs <- ridbLocs %>% 
  mutate(lon=sapply(ridbLocs[,XML],getGeoCoord,"lon",USE.NAMES = F,simplify = T)) %>%
  mutate(lat=sapply(ridbLocs[,XML],getGeoCoord,"lat",USE.NAMES = F,simplify = T)) %>%
  mutate(id=1:nrow(ridbLocs)) %>%
  mutate(goodMatch = ifelse(locationFound=="ZERO_RESULTS",F,T)) %>% #Add a flag for the rows that need a manual review
  select(-simplified1) #Drop a column.

View(select(ridbLocs,-XML))
save(ridbLocs,file="Data/Geocoded RIDB Locations.RData")