gibi

#!/bin/sh -e

# Recai Oktaş <roktas@debian.org>
# This file is in public domain.

PROGRAM=${0##*/}
BASEURI='labs.google.com'
THESMALL="Small+Set"
THELARGE="Large+Set"

# Portable which(1).
pathfind () {
    ifs_save="$IFS"
    IFS=:
    for _p in $PATH; do
        if [ -x "$_p/$*" ] && [ -f "$_p/$*" ]; then
            IFS="$OLDIFS"
            return 0
        fi
    done
    IFS="$ifs_save"
    return 1
}

dump () {
	for p in wget lynx w3m curl links w3c; do
		if pathfind $p; then
			dumper=$p
			break
		fi
	done
	# Setup proper options.
	case "$dumper" in
	wget)  opt="-O-" ;;
	lynx)  opt="-source" ;;
	w3m)   opt="-dump_source" ;;
	curl)  opt="" ;;
	links) opt="-source" ;;
	w3c)   opt="-n -get" ;;
	"")    printf "Couldn't locate a program to fetch files from net " >&2
	       printf "(e.g. wget, w3m, lynx, w3c, or curl).\n" >&2
	       exit 1 ;;
	esac
	$dumper $opt "$@"
}

for term; do
	i=$(($i + 1))
	term=$(echo $term | tr [[:blank:]] '+')
	query_string="${query_string}q${i}=${term}&"
done

size="$THESMALL"
case "$PROGRAM" in
gibigibi) size="$THELARGE" ;;
esac

QUERY="http://${BASEURI}/sets?${query_string}btn=${size}"

echo "Querying for the terms: $@..." >&2
echo >&2

dump "$QUERY" 2>/dev/null |
awk '
	/<tr>.*<a href=.+\.google\.com\/search.+<\/a>/ {
		term = $0
		sub(/<\/a>.+$/, "", term)
		sub(/^.+>/, "", term)
		re = "/q=" term "/"

		if (re)	print term
	}
'