#!/usr/bin/env bash

##########################################################
# Constants
##########################################################
GREEN=$(tput setaf 2)
RED=$(tput setaf 1)
CYAN=$(tput setaf 6)
CLRESET=$(tput sgr0)
prog_name="nowayout"
dbBuild="03182024"
dbPath="/hpc/db/${prog_name}/$dbBuild"
taxonomyPath="$dbPath/taxonomy"

usage()
{
    echo
    echo usage: "$0" [-h] 
    echo
    echo "Check for species presence in ${prog_name} database(s)."
    echo
    echo 'Example usage:'
    echo
    echo 'dbcheck -l'
    echo 'dbcheck -g Cathartus'
    echo 'dbcheck -d mitomine -g Cathartus'
    echo 'dbcheck -d mitomine -s "Cathartus quadriculus"'
    echo
    echo 'Options:'
    echo " -l        : List ${prog_name} databases"
    echo ' -d        : Search this database. Default: mitomine.'
    echo ' -g        : Genus to search for.'
    echo ' -s        : "Genus Species" to search for.'
    echo ' -h        : Show this help message and exit'
    echo
    echo "$1"
}

while getopts ":d:g:s:l" OPT; do
    case "${OPT}" in
        l)
            listdb="list"
            ;;
        d)
            dbname=${OPTARG}
            ;;
        g)
            genus=${OPTARG}
            ;;
        s)
            species=${OPTARG}
            ;;
        ?)
            usage
            exit 0
            ;;
    esac
done



if [ -n "$listdb" ]; then
    num_dbs=$(find "$taxonomyPath" -type d | tail -n+2 | wc -l)
    echo "=============================================="
    
    db_num="1"
    find $taxonomyPath -type d | tail -n+2 | while read -r db; do
        dbName=$(basename "$db")
        echo "${db_num}. $dbName"
        db_num=$(( db_num + 1 ))
    done
    echo "=============================================="
    echo "Number of ${prog_name} databases: $num_dbs"
    echo "=============================================="

    exit 0
fi



if [ -z "$dbname" ]; then
    dbname="mitomine"
fi

if [[ -n "$genus" && -n "$species" ]]; then
    usage "ERROR: Only one of -g or -s needs to be defined!"
    exit 1
elif [ -n "$genus" ]; then
    check="$genus"
elif [ -n "$species" ]; then
    check="$species"
else 
    check=""
fi

if [ -z "$check" ]; then
    usage "ERROR: -g or -s is required! check:$check"
    exit 1
fi

lineages="$taxonomyPath/$dbname/lineages.csv"

echo
echo -e "Checking ${dbname} for ${CYAN}${check}${CLRESET}...\nPlease wait..."
echo

num=$(grep -F ",$check," "$lineages" | cut -f1 -d, | sort -u | wc -l)
num_species=$(tail -n+2 "$lineages" | cut -f8 -d, | sort -u | wc -l)
num_entries=$(tail -n+2 "$lineages" | wc -l)

echo "$dbname brief stats"
echo "=============================================="
echo "DB Build: $dbBuild"
echo "Number of unique species: $num_species"
echo "Number of accessions in database: $num_entries"
echo "=============================================="


if [ "$num" -gt 0 ]; then
    echo
    echo "${GREEN}$check is present in ${dbname}${CLRESET}."
    echo "Number of accessions representing $check: $num"
    echo "=============================================="
else
    echo "${RED}$check is absent in ${dbname}${CLRESET}."
    echo -e "No worries. Please request the developer of\n${prog_name} to augment the database!"
    echo "=============================================="
fi