get air quality data from EEA

get_eea.R

library(readr)
library(glue)
library(RCurl)
library(lubridate)
library(dplyr)
library(futile.logger)
library(httr)
library(data.table)

# https://discomap.eea.europa.eu/map/fme/AirQualityExport.htm
poll_codes <- read_csv("https://dd.eionet.europa.eu/vocabulary/aq/pollutant/csv")

get_eea_pollutantcodes <- function(descr) {
  i <- match(descr, poll_codes$Notation)
  basename(poll_codes$URI)[i]
}

list_eea_pollutants <- function() {
  sort(unique(poll_codes$Notation))
}

get_eea_data <- function(CountryCode="IT",
                         CityName="",
                         Pollutant="",
                         Year_from="",
                         Year_to=Year_from,
                         Station="",
                         Samplingpoint="",
                         Source="E1a",
                         Output="TEXT",
                         UpdateDate="",
                         TimeCoverage="Year",
                         TimeZone="Etc/GMT-1") {
  if(!is.numeric(Pollutant)) {
    Pollutant <- get_eea_pollutantcodes(Pollutant)
  }
  address <- glue(
    "https://fme.discomap.eea.europa.eu/fmedatastreaming/AirQualityDownload/AQData_Extract.fmw?",
    "CountryCode={CountryCode}&CityName={CityName}&Pollutant={Pollutant}&",
    "Year_from={Year_from}&Year_to={Year_to}&Station={Station}&",
    "Samplingpoint={Samplingpoint}&Source={Source}&Output={Output}&",
    "UpdateDate={UpdateDate}&TimeCoverage={TimeCoverage}")
  uu <- gsub('[^\x20-\x7E]', '', strsplit(getURL(address), split="\r\n")[[1]])
  dd <- lapply(uu, function(x) read_csv(url(x)))
  out <- data.table::rbindlist(dd)
  out %>%
    mutate(DatetimeBegin=as_datetime(gsub("\\:","",DatetimeBegin), 
                                     format = "%Y-%m-%d %H%M%S %z", tz=TimeZone),
           DatetimeEnd=as_datetime(gsub("\\:","",DatetimeEnd), 
                                   format = "%Y-%m-%d %H%M%S %z", tz=TimeZone))
}

get_eea_metadata <- function(
    filein = "http://discomap.eea.europa.eu/map/fme/metadata/PanEuropean_metadata.csv") {
  ana <- NULL
  st <- HEAD(filein)$status
  if(st==200) {
    ana <- fread(
      filein, sep="\t", header = T, stringsAsFactors = F, 
      na.strings = c("-9999","-999","-9900","-100","-99","nan"), dec=".", 
      encoding = "Latin-1")
  } else {
    flog.warn(glue("Problems with {filein} - code: {st}"))
  }
  if(nrow(ana)>0) {
    ana %>% 
      filter(!is.na(Longitude), !is.na(Latitude)) %>%
      distinct(Namespace,AirQualityStationEoICode,AirPollutantCode,
               Sample, .keep_all=TRUE) -> ana
  } else {
    flog.warn(glue("File {filein} is empty"))
  }
  return(ana)
}