#!/bin/bash
kanji_ref="Kanji.txt"
composite_ref="composites"

countpipes(){
	var="$1"
	res="${var//[^|]}"
	echo "${#res}"
}

for i in {1..2200}; do
	kanji="$(grep -P "^$i\t" "$kanji_ref" | cut -f2)"
	regexp_composites="^L-\d{1,4}\t$kanji"
	composites_line="$(grep -P "$regexp_composites" "$composite_ref" )"
	#echo "$composites_line"
	ON_readings="$( echo "$composites_line" | cut -f3 |  tr '\n' '|' | uniq | sed 's/|$//')"
	if [ "$ON_readings" = "" ]; then
		#echo "No composites found for the kanji $kanji"
		continue
	fi
	composites="$( echo "$composites_line" | cut -f4 | awk '!seen[$0]++' | tr '\n' '|' | sed 's/|$//')"
	composites_kana="$( echo "$composites_line" | cut -f5 | awk '!seen[$0]++' | tr '\n' '|' | sed 's/|$//')"
	composites_meaning="$( echo "$composites_line" | cut -f6 | awk '!seen[$0]++' | tr '\n' '|' | sed 's/|$//')"

	#CHECK FOR CORRECT FORMATTING
	if [ "$ON_readings" = "" ];then
		echo missing on reading for no. $i
	fi
	if [ "$composites" = "" ];then
		echo missing composites for no. $i
	fi
	if [ "$composites_kana" = "" ];then
		echo missing composites kana no. $i
	fi
	if [ "$( echo "$composites_kana"  | grep -E '(\[|\])')" != "" ];then
		echo composite likely in composites_kana for no.$i
		echo "composite_kana: $composites_kana"
	fi
	if [ "$composites_meaning" = "" ];then
		echo missing composite meaning for $i
	fi
	flag=0
	if [ "$(countpipes "$ON_readings")" != "$(countpipes "$composites")" ]; then
		flag=1
	elif [ "$(countpipes "$composites")" != "$(countpipes "$composites_kana")" ]; then
		flag=1
	elif [ "$(countpipes "$composites_kana")" != "$(countpipes "$composites_meaning")" ]; then
		flag=1
	fi
	if [ $flag -eq 1 ]; then
		echo line-mismatch for kanji no.$i
	fi

	#WANTED OUTPUT
		printf '%s\t%s\t%s\t%s\t%s\n' "$i" "$ON_readings" "$composites" "$composites_kana" "$composites_meaning"
	#sleep 0.2
done