Statistics json file generator from Presto hive stats

gen_stats.sh

#!/bin/bash

set -euo pipefail

function generate() {
    local stats_file=$1
    local json_file=`echo $1 | sed 's/.stats//'`.json
    while read line; do
        line=$(echo $line | sed 's/"//g')
        local rows=$(echo $line | cut -d, -f5)
        if [[ ! -z $rows ]]; then
            break
        fi
    done < $stats_file

    cat << EOF > $json_file
 {
    "rowCount" : $rows,
     "columns" : {
EOF

    while read line; do
        line=$(echo $line | sed 's/"//g')
        local columnName=$(echo $line | cut -d, -f1)
        local dataSize=$(echo $line | cut -d, -f2)
        local ndv=$(echo $line | cut -d, -f3)
        local nulls=$(echo $line | cut -d, -f4)
        if [[ ! -z $nulls ]]; then
            nulls=$(python -c "print($nulls*$rows)")
        fi
        local min=$(echo $line | cut -d, -f6 | sed 's/^\./0./' | sed 's/\.//' | sed 's/^0*//')
        local max=$(echo $line | cut -d, -f7 | sed 's/^\./0./' | sed 's/\.//' | sed 's/^0*//')
        if [[ -z $columnName ]]; then
            cat << EOF >> $json_file
        "dummyColumn" : {}
    }
}
EOF
        else 
            cat << EOF >> $json_file
        "$columnName" : {
EOF
            if [[ ! -z $dataSize ]]; then
                cat << EOF >> $json_file
             "dataSize" : $dataSize,
EOF
            fi

            if [[ -z $min ]]; then
                cat << EOF >> $json_file
             "nullsCount" : $nulls,
             "distinctValuesCount" : $ndv
        },
EOF
            elif (echo $min | grep -q '.-..-') ; then
                cat << EOF >> $json_file
             "distinctValuesCount" : $ndv,
             "nullsCount" : $nulls,
             "min" : "$min",
             "max" : "$max"
        },
EOF
            else
                cat << EOF >> $json_file
             "distinctValuesCount" : $ndv,
             "nullsCount" : $nulls,
             "min" : $min,
             "max" : $max
        },
EOF
            fi
        fi
    done < $stats_file
}

function presto_cli() {
    ~/presto/presto-cli/target/presto-cli-*-executable.jar "${@}"
}

#server="${1}"
#schema="${2}"
#
#set -x
#presto_cli --server "${server}" --execute "show tables from hive.${schema}" --output-format TSV > tables.txt
#
#while read table; do
#    presto_cli --server "${server}" --execute "show stats for hive.${schema}.${table}" > "${table}".stats
#done < tables.txt
#set +x

for f in *stats; do
    echo $f
    generate $f
done