#!/bin/bash

VERSION=0.99.1

usage() {
	cat <<EOF
noobfarm2fortune [-options]

noobfarm2fortune v$VERSION by B. Watson (urchlay@slackware.uk), released
under the WTFPL. Do WTF you like with this. For full license, see
http://sam.zoy.org/wtfpl/

Scrapes quotes from noobfarm.org, formats them as a fortune database.
Output is a text file called "noobfarm" and an index called
"noobfarm.dat", created in the current directory (or wherever the -o
option says).

Options:
   -h --help        This usage message.
   -f --first       First quote. Default is 1.
   -l --last        Last quote. Default is to use the highest numbered
                    on the site.
   -n --no-cache    Wipe the cache directory before starting.
   -s --sleep N     Sleep N seconds between fetches. Minimum is 2.
   -o --output-dir  Write the noobfarm and noobfarm.dat files here. This
                    could be /usr/share/games/fortunes to write directly
                    to the system-wide fortune database. Default is the
                    current directory.

This script is NOT supported by the owners of noobfarm.org. Do NOT
contact them if you have problems.
EOF
}

sanity_check() {
	local exe
	local bad=0

	for exe in strfile links sed wget touch grep id getopt; do
		# use 'type' instead of 'which', it's a bash builtin.
		# also if getopt became a bash builtin one day, this
		# code wouldn't break.
		if ! type $exe &>/dev/null; then
			echo "Can't find $exe in path"
			bad=1
		fi
	done

	if [ "$bad" = "1" ]; then
		echo "Install the required external programs or fix PATH so they're visible."
		echo "PATH is set to: $PATH"
		exit 1
	fi
}

die() {
	echo "$@" 1>&2
	exit 1
}

extract_quote() {
	local file="$1"
	links -html-margin 0 -dump $file > $file.tmp
	local added=$( sed -n 's/.*\(Added:\)/\1/p' $file.tmp )
	sed -e '1,/^$/d' -e '/Home || Add Quote/,$d' < $file.tmp >> $QUOTEFILE
	( echo ; echo "$INDENT-- noobfarm.org, quote #$quote" ; echo "$INDENT   $added " ) >> $QUOTEFILE
	if [ "$quote" -ne "$LASTQUOTE" ]; then
		echo '%' >> $QUOTEFILE
	fi
	rm -f $file.tmp
}

# This is completely stupid, but amusing.
spinchars="|/-\\"
spinner() {
	local step
	local pos
	local char

	let step=$1/10
	let pos=$step%4
	char=${spinchars:$pos:1}
	echo -ne "[ $char ]\r"
}

# main() would start here in a real programming language

sanity_check

OPTS=$(getopt -n noobfarm2fortune -o hno:s:f:l: -l help,no-cache,output-dir:,sleep:,first:,last: -- "$@")

if [ "$?" != "0" ]; then
	usage
	exit 1
fi

while true; do
	case "$1" in
		-h|--help) usage; exit 0 ;;
		-n|--no-cache) WIPECACHE=1 ; shift ;;
		-o|--output-dir) OUTDIR="$2" ; shift 2 ;;
		-s|--sleep) SLEEPTIME="$2" ; shift 2 ;;
		-f|--first) FIRSTQUOTE="$2" ; shift 2 ;;
		-l|--last) LASTQUOTE="$2" ; shift 2 ;;
		""|--) shift ; break ;;
		*) echo "Unknown argument: $1" ; usage ; exit 1 ;;
	esac
done
if [[ "$@" != "" ]]; then
	usage
	exit 1
fi

if [ "$( id -u )" = "0" ]; then
	CACHEDIR=/var/cache/noobfarm2fortune
else
	CACHEDIR=/tmp/noobfarm2fortune."$( id -nu )"
fi

echo "Using cache dir: $CACHEDIR"

OUTDIR=${OUTDIR:-$( pwd )}
QUOTEFILE=$CACHEDIR/noobfarm
SLEEPTIME=${SLEEPTIME:-2}
INDENT="                                     " # default is 37 spaces

OUTDIR="$( readlink -m "$OUTDIR" )"
echo "Using output dir: $OUTDIR"

case "$SLEEPTIME" in
	1|1.*|.*|0*|"") echo "Ignoring sleep time $SLEEPTIME, using 2 sec"
	             SLEEPTIME=2 ;;
	[0-9]*) ;;
	*) echo "Invalid sleep time $SLEEPTIME" ; usage ; exit 1 ;;
esac

if [ "$WIPECACHE" != "" ]; then
	rm -rf $CACHEDIR
	echo "Wiped cache"
fi

mkdir -p $CACHEDIR
cd $CACHEDIR || die "Couldn't create $CACHEDIR"
rm -f $QUOTEFILE

if [ "$LASTQUOTE" = "" ]; then
	LASTQUOTE="$( wget -q -O- http://noobfarm.org/ | sed -n 's/.*div *id *= *"quote_\([0-9]*\)">.*/\1/p' | sort -n | tail -1 )"

	if [ -z "$LASTQUOTE" ]; then
		die "can't get last quote number from site (page layout changed?)"
	fi

	if echo "$LASTQUOTE" | grep -q '[^0-9]'; then
		die "got weirdness \"$LASTQUOTE\" instead of last quote number (page layout changed?)"
	fi
fi

echo "Last quote is $LASTQUOTE"

FIRSTQUOTE=${FIRSTQUOTE:-1}
quote=$FIRSTQUOTE

# The .inprogress marker is there in case the user presses ^C
# while wget's waiting for data (resulting in a 0-byte HTML file).
while [ "$quote" -le "$LASTQUOTE" ]; do
	spinner $quote
	if [ -e "$quote".inprogress ]; then
		rm -f "$quote".inprogress "$quote".html
	fi

	if [ ! -e "$quote".html ]; then
		echo -ne "Getting quote $quote...        \b\b\b\b\b\b\b\b"
		touch "$quote".inprogress
		wget -q -O "$quote".html "http://noobfarm.org/index.php?id=$quote"
		if [ "$?" = "0" ]; then
			echo -ne "OK\r"
		else
			echo "FAIL"
		fi
		rm -f "$quote".inprogress
		sleep $SLEEPTIME
	fi
	quote=$(( quote + 1 ))
done

quote=$FIRSTQUOTE

while [ "$quote" -le "$LASTQUOTE" ]; do
	spinner $quote
	if grep -q '"quote_output"' $quote.html; then
		extract_quote $quote.html
	elif grep -q 'That quote does not exist' $quote.html; then
		echo -ne "      Quote #$quote does not exist\r"
	else
		echo "Quote #$quote seems malformed, check $CACHEDIR/$quote.html"
	fi
	quote=$(( quote + 1 ))
done

set -e
echo
mkdir -p "$OUTDIR"
cd "$OUTDIR"
cp $QUOTEFILE .
strfile $( basename $QUOTEFILE )