IT:Italia/Progetto del Mese/2026/Risultati gennaio

USE_DOCKER=true
DAILY_CSV_FILE="count.csv"
MONTH_BEGINNING="italy-251231.osm.pbf"
last_total=0

if $USE_DOCKER; then
    OSMIUM="docker run --rm -v $(pwd):/data:delegated iboates/osmium:1.18.0"
    DATA_DIR="/data"
else
    OSMIUM="osmium"
    DATA_DIR="."
fi

echo 'Day,Date,wikidata=* count,wikidata=* diff,Secondary count,Secondary diff,Total distinct tags,Tags diff' > $DAILY_CSV_FILE
for file in italy-*.pbf ; do
    date="20${file:6:2}-${file:8:2}-${file:10:2}"

    direct_file="wikidata_$file"
    if [ ! -f "$direct_file" ]; then
        echo "Filtering $file to $direct_file"
        # https://docs.osmcode.org/osmium/latest/osmium-tags-filter.html
        $OSMIUM tags-filter --omit-referenced -o "$DATA_DIR/$direct_file" "$DATA_DIR/$file" "wikidata=*"
    fi

    secondary_file="secondary_$file"
    if [ ! -f "$secondary_file" ]; then
        echo "Filtering $file to $secondary_file"
        # https://docs.osmcode.org/osmium/latest/osmium-tags-filter.html
        $OSMIUM tags-filter --omit-referenced -o "$DATA_DIR/$secondary_file" "$DATA_DIR/$file" "*:wikidata=*"
    fi

    count_file="count-$file.tsv"
    if [ ! -f "$count_file" ]; then
        echo "Counting in $count_file"
        # https://docs.osmcode.org/osmium/latest/osmium-tags-count.html
        $OSMIUM tags-count --sort=count-desc --output="$DATA_DIR/$count_file" "$DATA_DIR/$file" 'wikidata=*' '*:wikidata=*'
    fi
    total_tags=`cat $count_file | wc -l`
    
    echo "Counting $direct_file"
    # https://docs.osmcode.org/osmium/latest/osmium-fileinfo.html
    nodes=`$OSMIUM fileinfo -e -g "data.count.nodes" "$DATA_DIR/$direct_file"`
    ways=`$OSMIUM fileinfo -e -g "data.count.ways" "$DATA_DIR/$direct_file"`
    rels=`$OSMIUM fileinfo -e -g "data.count.relations" "$DATA_DIR/$direct_file"`
    total=$((nodes + ways + rels))
    
    echo "Counting $secondary_file"
    # https://docs.osmcode.org/osmium/latest/osmium-fileinfo.html
    secondary_nodes=`$OSMIUM fileinfo -e -g "data.count.nodes" "$DATA_DIR/$secondary_file"`
    secondary_ways=`$OSMIUM fileinfo -e -g "data.count.ways" "$DATA_DIR/$secondary_file"`
    secondary_rels=`$OSMIUM fileinfo -e -g "data.count.relations" "$DATA_DIR/$secondary_file"`
    total_secondary=$((secondary_nodes + secondary_ways + secondary_rels))

    if [[ $MONTH_BEGINNING == $file ]]; then
        diff='('$((total - last_total))')'
        month_beginning_total=$total
        secondary_diff='('$((total_secondary - last_total_secondary))')'
        month_beginning_total_secondary=$total_secondary
        tags_diff='('$((total_tags - last_total_tags))')'
        month_beginning_total_tags=$total_tags
        day="00"
    elif [[ $file == italy-25* ]]; then 
        diff=''
        secondary_diff=''
        tags_diff=''
        day=''
    else
        diff=$((total - month_beginning_total))
        secondary_diff=$((total_secondary - month_beginning_total_secondary))
        tags_diff=$((total_tags - month_beginning_total_tags))
        day="${file:10:2}"
    fi
    last_file=$file
    last_total=$total
    last_total_secondary=$total_secondary
    last_total_tags=$total_tags

    echo "$day,$date,$total,$diff,$total_secondary,$secondary_diff,$total_tags,$tags_diff" >> $DAILY_CSV_FILE
done