soundbash

#!/usr/bin/env bash

# Checks if a page is a single track page or not.
#
# Input:
# $1 - URL of the page
# Return: 0 if the page is a single track page, 1 otherwise
is_single_track_page() {
	wget -O - -- "$1" 2>/dev/null | grep '<meta content="soundcloud:sound"' >/dev/null 2>&1
}

# Extracts all stream URLs from a given page.
#
# Input:
# $1 - URL of the page
# Output:
# stdout - streaming URLs separated by newlines
get_stream_URLs() {
	wget -O - -- "$1" 2>/dev/null | sed -ne '/window\.SC\.bufferTracks\.push/{s/.*uri":"\([^"]*\)".*streamUrl":"\([^"?]*\).*/\2?\1/;p;}'
}

# Extracts the page count for a page.
#
# Input:
# $1 - URL of the page
# Output:
# stdout - number of pages or an empty string if there aren’t multiple pages
get_page_count() {
	local count="$(wget -O - -- "$1" 2>/dev/null | grep -o 'page=[0-9]*' | cut -d= -f 2 | sort -g | tail -n 1)"
	if ! (( count >= 1 )); then
		count=1
	fi
	echo "$count"
}

# Shows a description of the script and a list of parameters
#
# Output:
# stdout - short description and a list of parameters
show_help() {
	cat <<EOF
Usage: soundbash soundcloud_link
This utility extracts all SoundCloud streaming URLs from a given page. It should work with pretty much everything (track URLs, user favorites, searches).
The streaming URL contains the track path as a parameter. Prepend the part after the quotation mark with http://soundcloud.com to get the track URL.

Parameters:
    -h -?       print this help and exit.
    -m number   limits the output to a maximum of number pages. Use 0 or a negative number to fetch all results. Default: 10
    -l URL      extract the streaming URLs from the given link.
    -u user     extract the streaming URLs from the track page of the given user
    -f user     extract the streaming URLs from the favorites page of the given user
    -g group    extract the streaming URLs from a group
EOF
}

# Extracts the streaming URL from a single track page.
#
# Input:
# $1 - URL
# Output:
# stdout - the streaming URL
get_single_track_URL() {
	get_stream_URLs "$1" | head -n 1
}

# Extracts streaming URLs from a multi track page. A maximum number of pages can be specified.
#
# Input:
# $1 - URL
# $2 - maximum number of pages. Pass nothing, a number <= 0 or a string to get all pages.
# Output:
# stdout - the streaming URLs from the page
get_multi_track_URLs() {
	count="$(get_page_count "$1")"
	if ! (( count >= 1 )); then
		count=1
	fi
	if (( "$2" >= 1 && "$2" < count )); then
		count="$2"
	fi
	for (( i=1; i <= count; i++ )); do
		get_stream_URLs "${1}?page=${i}"
	done
}

# Extracts streaming URLs from a page. If a single track page is given, only the URL for the track is returned, otherwise as many track pages as
# specified are returned.
#
# Input:
# $1 - URL
# $2 - maximum number of pages to extract streaming URLs from
# Output:
# stdout - the streaming URL(s) for the track(s)
handle_url() {
	if is_single_track_page "$1"; then
		get_single_track_URL "$1"
	else
		get_multi_track_URLs "$1" "$2"
	fi
}

# Extracts streaming URLs from the tracks page of a given user.
#
# Input:
# $1 - user name
# $2 - maximum number of pages to extract
# Output:
# stdout - the streaming URLs
handle_user() {
	get_multi_track_URLs "http://soundcloud.com/${1}/tracks" "$2"
}

# Extracts streaming URLs from the favorites page of a given user.
#
# Input:
# $1 - user name
# $2 - maximum number of pages to extract
# Output:
# stdout - the streaming URLs
handle_favorites() {
	get_multi_track_URLs "http://soundcloud.com/${1}/favorites" "$2"
}

# Extracts streaming URLs from a given group.
#
# Input:
# $1 - the group name
# $2 - maximum number of pages to extract
# Output:
# stdout - the streaming URLs
handle_group() {
	get_multi_track_URLs "http://soundcloud.com/groups/${1}/tracks" "$2"
}


max_pages=10

while (( $# > 0 )); do
	if [[ "${1:0:1}" == "-" ]]; then
		params="${1:1}"
		shift
		while (( ${#params} > 0 )); do
			case "${params:0:1}" in
				h|\?)
					show_help
					exit 0
					;;

				m)
					if (( $# < 1 )); then
						echo "-m needs a parameter" >&2
						exit 1
					fi
					max_pages="$1"
					shift
					;;

				u)
					if (( $# < 1 )); then
						echo "-u needs a parameter" >&2
						exit 1
					fi
					handle_user "$1" "$max_pages"
					shift
					;;

				f)
					if (( $# < 1 )); then
						echo "-f needs a parameter" >&2
						exit 1
					fi
					handle_favorites "$1" "$max_pages"
					shift
					;;

				g)
					if (( $# < 1 )); then
						echo "-g needs a parameter" >&2
						exit 1
					fi
					handle_group "$1" "$max_pages"
					shift
					;;

				l)
					if (( $# < 1 )); then
						echo "-l needs a parameter" >&2
						exit 1
					fi
					handle_url "$1" "$max_pages"
					shift
					;;

				*)
					echo "Unknown parameter: ${params:0:1}" >&2
					exit 1
					;;
			esac
			params="${params:1}"
		done
	else
		handle_url "$1" "$max_pages"
		shift
	fi
done