#!/usr/bin/env bash
#WAZNE: sprawdzic czy ten skrypt dobrze liczy gaps. mozliwe ze liczy razem z query i subject
query_folder="${1}"
out_folder="${2}"
id="${3}"
All_hspidentities=0
hspidentities=0
All_hspgaps=0
hspgaps=0
All_hsplen=0
hsplen=0
seq=""
mkdir $out_folder
printf "Name, Length, Mismatches, Deletions, Identities" > stats.csv
for query_file in $query_folder/*; do
	hspidentities="$(xmlstarlet sel -t -v 'sum(//Hsp_identity)' $query_file)"
	((All_hspidentities+=hspidentities))
	hspgaps="$(xmlstarlet sel -t -v 'sum(//Hsp_gaps)' $query_file)"
	((All_hspgaps+=hspgaps))
	hsplen="$(xmlstarlet sel -t -v 'sum(//Hsp_align-len)' $query_file)"
	((All_hsplen+=hsplen))
	((hspmismatches=hsplen-hspidentities-hspgaps))
	printf "\n$query_file, $hsplen, $hspmismatches, $hspgaps, $hspidentities" >> stats.csv
	seq="$(xmlstarlet sel -t -v //Hsp_hseq $query_file)"
	printf ">$(basename $query_file .xml)\n$seq" > $out_folder/"218_$(basename $query_file .xml).fasta"
done
((All_hspmismatches=All_hsplen-All_hspidentity-All_hspgaps))
printf "\n,Total length $All_hsplen, Mismatches $All_hspmismatches, Deletions $All_hspgaps, Identities $All_hspidentities" >> stats.csv