107 lines
2.8 KiB
Bash
Executable File
107 lines
2.8 KiB
Bash
Executable File
#!/usr/bin/env sh
|
|
dependencies="sed test cp libreoffice"
|
|
assert_tools () {
|
|
while test $# -gt 0; do
|
|
which $1 >/dev/null 2>/dev/null || {
|
|
echo "tool missing: "$1
|
|
exit 2
|
|
}
|
|
shift
|
|
done
|
|
}
|
|
assert_tools ${dependencies}
|
|
|
|
# target filename
|
|
tn="dhl-cp1252.csv"
|
|
|
|
# todo: download file (its UTF-8 encoded)
|
|
|
|
fn="dhl.csv"
|
|
# backup
|
|
cp -f "${fn}" "${fn}.bak"
|
|
|
|
# replace special spaces and hyphens
|
|
sed -i -e 's/[\u00A0\u202F[:space:]]+/ /g' -e 's/[\u2010\u2011\u2012\u2013\u002D]+/-/g' -e 's/[\u00AD\uFEFF]+//g' "$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: $1";exit 1;;
|
|
esac
|
|
fi
|
|
}
|
|
|
|
testcmp () {
|
|
# usage: testcmp "cmpresult" "cmdfunctionname" [ "argument1", "argument2", […] ]
|
|
if test "$#" -eq 0 ; then
|
|
echo "arguments missing."; exit 1
|
|
fi
|
|
case $2 in
|
|
#whitelisted fn
|
|
iso3countrycode)
|
|
expected="$1"
|
|
shift
|
|
fn="$1"
|
|
shift
|
|
for a in "$@"; do
|
|
result=$("$fn" "$a")
|
|
if test "$result" = "$expected" ; then
|
|
echo "$fn passed for: $a."
|
|
else
|
|
echo "$fn failed for: $a."
|
|
fi
|
|
done
|
|
;;
|
|
*)
|
|
echo "no test performed, function or command not allowd."
|
|
esac
|
|
}
|
|
|
|
testiso3 () {
|
|
# covering data in https://doku.ccc.de/index.php?title=Attribut:Chaostreff-Country&limit=500&offset=0
|
|
testcmp "BEL" "iso3countrycode" "Belgien"
|
|
testcmp "DEU" "iso3countrycode" "Deutschland"
|
|
testcmp "GBR" "iso3countrycode" "England"
|
|
testcmp "LUX" "iso3countrycode" "Luxembourg" "Luxemburg"
|
|
testcmp "NLD" "iso3countrycode" "Niederlande"
|
|
testcmp "AUT" "iso3countrycode" "Österreich"
|
|
testcmp "HUN" "iso3countrycode" "Ungarn"
|
|
testcmp "CHE" "iso3countrycode" "Schweiz"
|
|
testcmp "" "iso3countrycode" "Simbabwe" # should fail
|
|
}
|
|
#~ testiso3
|
|
|
|
convcsv () {
|
|
if test $# -eq 0 ; then echo "argument missing."; exit 1; fi
|
|
if test -e $1 ; then echo "file already exists."; exit 2; fi
|
|
# charset windows codepage 1252
|
|
# doc for headless conversion options https://wiki.openoffice.org/wiki/Documentation/DevGuide/Spreadsheets/Filter_Options
|
|
if="${1}"
|
|
of=$(echo "${if}"|rev|cut -d. -f2-|rev)".ods"
|
|
libreoffice --headless --convert-to ods --infilter=CSV:59,34,76,1 "${if}" && {
|
|
mv "${if}" "${if}.bak"
|
|
libreoffice --headless --infilter=CSV:59,34,76,1 --convert-to csv:"Text - txt - csv (StarCalc)":"44,34,1,0,1,,0" "${tf}" && {
|
|
rm "${tf}" "${if}.bak"
|
|
}
|
|
}
|
|
}
|
|
convcsv "$fn"
|