rls-discogs/search-discogs-elastic.fish

#!/usr/bin/env -S fish
#set fish_trace on

function usage
    begin
        echo "Reads tsv release from sdtin, outputs JSON"
        echo
        echo "Usage: $0 [-v][-eURL]"
        echo
        echo " -eURL/--endpoint=URL     Set ElasticSearch /_search endpoint URL"
        echo " -v/--verbose             Print query and warnings"
    	exit 1
	end >&2
end

argparse -n (status current-filename) 'v/verbose' 'h/help' 'e/endpoint=?' -- $argv
if test $status -ne 0
    usage
end

if set -q _flag_help
    usage
end

if not set -q _flag_endpoint
    set _flag_endpoint 'http://localhost:9200/_all/_search'
end

curl -s -o /dev/null -G "$_flag_endpoint"
if test $status -ne 0
    echo "ERR: Search API Endpoint $_flag_endpoint failed"
    exit 1
end

set OUT (mktemp -t rls.ES.XXXXXXXX.json)

function discord_format
	switch $argv[1]
        case (seq 100)VINYL VINYL VLS Vinyl vinyl
            printf '%s' "Vinyl"
        case (seq 100)CD CD CDA CDS CDM CDEP EP
            printf '%s' "CD"
        case CDR CDREP
            printf '%s' "CDr"
        case TAPE
            printf '%s' "Cassette"
		case WEB SINGLE-WEB EP-WEB
		    printf '%s' "File"
		case '*'
		    printf '%s' "*"
	end
end

while read -l -d (printf '\t') ORG A T C B F E Y
    set _in_A "$A"; set _in_T "$T"; set _in_C "$C";
    set _in_B "$B"; set _in_F "$F"; set _in_E "$E"; set _in_Y "$Y"
    if set -q A; and test -n "$A"
        set A (string trim "$A")
        switch "$A"
            case VA Va va "V.A." OST "O.S.T" "Various Artists"
                set Q "q=artists.name:\"Various\""
            case '*'
                set A (string replace -a ')' ' ' "$A")
                set A (string replace -a '(' ' ' "$A")
                set A (string lower "$A")
                set Q "q=artists.name:\"$A\""
        end
    end
    if set -q T; and test -n "$T"
        set T (string replace -a '(' ' ' "$T")
        set T (string replace -a ')' ' ' "$T")
        set T (string trim "$T")
        set T (string lower "$T")
        set -a Q " AND title:\"$T\"~"
    end
    if set -q C; and test -n "$C"
        set -a Q " AND (identifiers.value:$C~ OR labels.catno:$C)"
    end
    if set -q B; and test -n "$B"
        set B (string trim "$B")
        set -a Q " AND (identifiers.value:$B~ OR labels.catno:$B)"
    end
    if set -q F; and test -n "$F"
        switch "$F"
            case CDS
                set -a Q " AND formats.name:CD AND formats.descriptions:Single"
            case CDM
                set -a Q " AND formats.name:CD AND formats.descriptions:Maxi-Single"
            case CDA
                set -a Q " AND formats.name:CD AND formats.descriptions:Album"
            case '*'
                set -a Q " AND formats.name:$(discord_format $F)"
        end
    else
        # No format given, at least exclude Vinyl which would usually be named in rls
        set -a Q " AND NOT formats.name:Vinyl"
    end
    if set -q Y
        set -a Q " AND released:[$Y-01-01 TO $Y-12-31]"
    end

    # TODO Editions

    if set -q Q; and test -n "$Q"
        if set -q _flag_verbose
            echo "$Q" >&2
        end
        set CODE (curl -s -G -w '%{http_code}' -o "$OUT" --data-urlencode "$Q" --data-urlencode size=100 "$_flag_endpoint")
        if test $status -ne 0
            echo "ERR: curl request failed" >&2
            rm "$OUT"
            continue
        end

        if test $CODE -ne 200
            begin
                echo "ERR: ES status code $CODE"
                echo $ORG
                echo $Q
            end >&2
            rm "$OUT"
            continue
        end

        set ES_COUNT (jq -r .hits.total.value "$OUT")
        if test $status -ne 0
            begin
                echo "ERR: Failed parsing result"
                echo $ORG
                echo $Q
            end >&2
            rm "$OUT"
            continue
        end

        if test $ES_COUNT -eq 0
            echo '{}' | jq -SMc --arg rls "$ORG" --arg query "$Q" --arg cnt "$ES_COUNT" \
                --arg a "$_in_A" \
                --arg t "$_in_T" \
                --arg c "$_in_C" \
                --arg b "$_in_B" \
                --arg f "$_in_F" \
                --arg e "$_in_E" \
                --arg y "$_in_Y" \
            '
            {
                "_scene_release_name": $rls,
                "_query": $query,
                "_search_result_count": $cnt | tonumber,
                "_artist": $a,
                "_title": $t,
                "_catno": $c,
                "_barcode": $b,
                "_format": $f,
                "_edition": $e,
                "_year": $y
            }'
        else if test $ES_COUNT -eq 1
            jq -SMc --arg rls "$ORG" --arg query "$Q" --arg count "$ES_COUNT" '.hits.hits[0] |
                ._source + {
                "_scene_release_name": $rls,
                "_query": $query,
                "_search_result_count": $count | tonumber,
            }' "$OUT"
        else if test $ES_COUNT -gt 1
            # skip results with mixed mastere ids
            set UNIQUE_MASTERS (jq -r '[.hits.hits[]._source.master_id] | unique | length' $OUT)
            if test $UNIQUE_MASTERS -gt 1
                if set -q _flag_verbose
                    echo "WARN: Mixed results $ORG" >&2
                end
                rm "$OUT"
                continue
            else
                jq -SMc --arg rls "$ORG" --arg query "$Q" --arg count "$ES_COUNT" '.hits.hits[0] |
                    ._source + {
                    "_scene_release_name": $rls,
                    "_query": $query,
                    "_search_result_count": $count | tonumber,
                }' "$OUT"
            end
        end
        rm "$OUT"
    end
end