load-terminology

#!/usr/bin/bash
set -euo pipefail
function error() { echo $@ >&2; exit 1; }

[[ $# -ge 2 ]] || error "Usage: $0 BARTOC-URI RDF-URL [format]"
uri=$1
id="${uri##*/}"
url=$2
ext="${url##*.}"
format=${3:-$ext}

[[ "$uri" =~ ^http://bartoc.org/en/node/[0-9]+$ ]] || error "Invalid BARTOC URI!"
[[ "$url" =~ ^https?:// ]] || error "Invalid URL"

case "$format" in
    ttl)
        format=turtle
        ext=ttl
        ;;
    nt)
        format=ntriples
        ext=nt
        ;;
    xml | rdf)
        format=rdfxml
        ext=rdf
        ;;
    ndjson | jskos)
        format=jskos
        ext=ndjson
        ;;
    *)
        error "RDF serialization format unknown or not supported"
esac

# update list of terminologies
make terminologies

# check whether terminology is known
found=$(jq -c --arg uri "$uri" '.[]|select(.uri==$uri).uri' terminologies.json)
[[ -z "$found" ]] && error "Terminology $uri not found in N4O Terminologies (http://bartoc.org/en/node/18961)" 

# download terminology
dir=import-voc/$id
mkdir -p $dir
input=$dir/original.$ext
unique=$dir/unique.nt

curl -sL "$url" > $input

# convert JSKOS to N-Triples
if [[ "$format" = "jskos"  ]]; then
    npm run --silent -- jsonld2rdf -c jskos-context.json "$input" > "$dir/original.nt"
    input="$dir/original.nt"
    ext=nt
    format=ntriples
fi

# TODO: this is duplicated in receive
tmp=$(mktemp)
rapper -q -i turtle "$input" | sort | uniq > "$tmp"

if [[ ! -s "$tmp" ]]; then
    rm "$tmp"
    echo "$input ist syntaktisch nicht korrekt oder leer!"
    exit 1
fi

mv "$tmp" "$unique"
echo "$input ist syntaktisch korrektes RDF. "
echo
echo "Anzahl unterschiedlicher Tripel: **$(<$unique wc -l)**"
echo

# TODO: filtern/cleanup to only have JSKOS compatible fields (SKOS etc.)

./load-rdf-graph $uri $unique