getallimages() { _getallimages_inner "href" "$@" } getallembed() { _getallimages_inner "src" "$@" } _getallimages_inner() { TAG="$1" shift while test "$#" -gt 0; do URL="$1" DIR=`echo "$URL" | sed 's/[^a-zA-Z0-9_.-]/-/g'` PREFIX=`echo "$1" | sed 's/?.*//' | sed 's@^\(.*/\).[^/]*$@\1@'` ROOT=`echo "$1" | sed 's@\(://[^/]*\)/.*$@\1@'` mkdir -p "$DIR" ( cd "$DIR" wget --cache=no -O - -U 'Mozilla/5.0' "$URL" | tr '>' '\n' | grep -i "$TAG" | sed 's/.*'"$TAG"' *=[ "]*//i ; s/[" ].*//g' | grep -i '\(jpe\|jpg\|jpeg\|gif\|png\|mpg\|mpeg\|avi\|wmv\)$' | grep -v '/src[.]cgi/' | sort | uniq | while read IMAGE do if echo "$IMAGE" | grep -qi "^/"; then IMAGE="$ROOT/$IMAGE"; fi if echo "$IMAGE" | grep -vqi "^http"; then IMAGE="$PREFIX/$IMAGE"; fi echo "$IMAGE" wget -c -U 'Mozilla/5.0' --header "Referer: $URL" "$IMAGE" 2>&1 | grep '\(\[\|fully retrieved\)' done ) # cd .. shift done }