#!/bin/bash
# Seriously, if you know BaSH well, DO NOT read this code.
# You gonna kill yourself, or maybe kill me, if you do.
# --- STOP READING THE CODE ---

TABS=0

temp=$(mktemp -d)
cd $temp

GLOBAL_IGNORE='wordpress|facebook|baidu|google'

[[ -z $MAX_DEPTH ]] && MAX_DEPTH=50

function getLinksByURL {
	curl --max-time 10 -sL "$1" | awk -F'href=' '{print $2}' | awk -F'>| ' '{print $1}' | sort | uniq | grep 'http' | awk -F"/" '{print $3}' | tr -d "\"'" | sort | uniq
}

function showTabs {
	for i in $(seq 1 $1)
	do
		echo -n "  "
	done
}

function showSiteLinkOut {
	let TABS++
	mkdir "$1" 2> /dev/null; cd "$1"
	getLinksByURL "$1" | grep -vE "$1|$(cat $temp/DONE)" | while read -r out
	do
		echo -n "|$1" >> $temp/DONE
		[[ $TABS -lt $MAX_DEPTH && -z $(grep -E "$(cat $temp/DONE)" <<< "$out") && ! "$1" == "$out" && -z $(grep -E "$GLOBAL_IGNORE" <<< "$out") && ! "$1" == "$out" ]] && echo "[$TABS]$(showTabs $TABS)>>> $out" && showSiteLinkOut "$out" && echo && let TABS-- && echo "[$TABS]$(showTabs $TABS)<<< $out"  && cd ..
		echo 0 > /dev/null
	done
}

echo "working at $temp"
echo ">>> TRACE to $1 START"
echo -n "$1" > $temp/DONE
showSiteLinkOut $1
cd $temp
tree -d $1 > "$(dirname "$0")"/tree.txt