#!/usr/bin/env sh

# target filename
tn="dhl-ds-versendung.csv"

# todo: download file (its UTF-8 encoded)

fn="dhl-ds-versendung-3.csv"
# backup
cp -f "$fn" "dhl.csv"

# replace special spaces and hyphens
sed -i -e 's/[\u00A0\u202F[:space:]]+/ /g' -e 's/[\u2010\u2011\u2012\u2013\u002D]+/-/g' -e 's/[\u00AD\uFEFF]+//g' "$fn"

# todo: convert encoding from UTF-8 to CP1252 "Windows"

# a possibility for conversion? seems not to be reliant
#~ iconv -o "$tn" -f UTF-8 -t CP1252 "$fn" 

# todo: validate length for entries

# todo: convert country to countrycode as in ISO3
iso3countrycode () {
	if test "$#" -eq 0 ; then
		echo "argument missing. ($0)"
	else
		case "$1" in
			Belgien) echo "BEL";;
			Deutschland) echo "DEU";;
			England) echo "GBR";;
			Luxemburg|Luxembourg) echo "LUX";;
			Niederlande) echo "NLD";;
			Österreich) echo "AUT";;
			Schweiz) echo "CHE";;
			Ungarn) echo "HUN";;
			*) echo "unknown country: $1";exit 1;; 
		esac
	fi
}

testiso3 () {
	# covering data in https://doku.ccc.de/index.php?title=Attribut:Chaostreff-Country&limit=500&offset=0
	iso3countrycode "Belgien"
	iso3countrycode "Deutschland"
	iso3countrycode "England"
	iso3countrycode "Luxembourg"
	iso3countrycode "Luxemburg"
	iso3countrycode "Niederlande"
	iso3countrycode "Österreich"
	iso3countrycode "Ungarn"
	iso3countrycode "Schweiz"
	iso3countrycode "Simbabwe" # should fail
}
testiso3