#!/bin/bash

T=$(mktemp)
unzip -p "$1" tellico.xml | tr -d '\n' | sed -e $'s/<entry id="/\\\n<entry id="/g' -e $'s/<\/entry>/<\/entry>\\\n/g' > "$T"


hf=$(cat "$T" | grep -Pe '^<entry id="' | grep '<hasflatworms>yep</hasflatworms>')
# nejvyssi dosazene ID
mid=$(cat "$T" | grep -Pe '^<entry id="' | grep -oe ' <localityid>.*</localityid> ' | grep -oPe '[0-9]+' | sort | tail -n 1)
mid=$(($mid+1))


function ll()
{
	while IFS='' read -r r || [[ -n "$r" ]]; do
		t=$(echo "$r" | grep -oe ' <taken-date>.*</taken-date> ' | sed -E 's/( )*<(\/)?taken-date>( )*//g' | sed -E 's/([0-9]+)\.([0-9]+)\.([0-9]{4})/\3-\2-\1/g' | sed -E 's/-([0-9])-/-0\1-/g' | sed -E 's/-([0-9])$/-0\1/g')
		l=$(echo "$r" | grep -oe '<locality>.*</locality>' | grep -oPe '[0-9]+\.[0-9]+[, ]*[0-9]+\.[0-9]+')
		 echo "$t@$l"
	done <<< $(echo "$1")
}

srt=$(ll "$hf" | sort -u)

uid=$mid
while IFS='' read -r r || [[ -n "$r" ]]; do
	l=$(echo "$r" | grep -oPe '@.*$' | tr -d '@')
	lid=$(cat "$T" | grep  " <locality>$l</locality> " | grep -oe ' <localityid>.*</localityid> ' | grep -oPe '[0-9]+')
	if [[ "$lid" = "" ]]; then

		sedexp="s/( <locality>$l<\/locality> .*<hasflatworms>yep<\/hasflatworms> )/ \\1 <localityid>$uid<\/localityid> /g"
		#echo "$sedexp"
		sed -E "$sedexp" -i "$T"
		uid=$(($uid+1))
	fi
done <<< $(echo "$srt")

D=$(mktemp -d)
cp -f "$T" "$D/tellico.xml"
zip -j "./micro_new.tc" "$D/tellico.xml"

rm -rfv "$T"
rm -rfv "$D"