#!/bin/sh
###############################################################################
##									     ##
##	File:     getyoutube						     ##
##	Author:   Wolfgang S. Rupprecht <wolfgang.rupprecht@gmail.com>       ##
##	Created:  Wed Jul  4 14:44:59 PDT 2007				     ##
##	Contents: quick shell script to grab and play youtube videos         ##
##									     ##
##	Copyright (c) 2007 Wolfgang S. Rupprecht.			     ##
##	Released under GPL version 3					     ##
##									     ##
##	$Id: getyoutube,v 1.24 2007/11/17 07:37:47 wolfgang Exp $
###############################################################################

# usage:
# getyoutube http://www.youtube.com/watch?v=DCQ7dmGuAU8 [...]
# getyoutube http://www.youtube.com/view_play_list?p=472DE700A3CC70A4 [...]
# getyoutube http://www.youtube.com/results?search_query=openmoko [...]
# getyoutube http://www.youtube.com/profile_videos?user=openmoko [...]
#
# The "&" in the youtube URLS and anything past them is optional.
# Freel free to not cut and paste it and save yourself the trouble of
# having to quote the lines (since the &'s have meaning to the shell.)
#
# We can deal with a few types of youtube html files.
# /watch?v=                (individual video page)
# /view_play_list?p=       (playlist of the above files)
# /results?search_query=   (search results)
# /profile_videos?user=    (a user's video list)
#
# Requires wget(1) and mplayer(1).  Fedora users may need to install
# the optional mplayer codecs from the livna repository.

set -eu

TMP=/tmp
LOCK=/tmp/mplayer-lock

GETVIDEO="http://youtube.com/get_video.php?"
WATCHURL="http://youtube.com/watch?v="

testflag=''

mkdir -p $TMP/YouTube
cd $TMP/YouTube

playone ()
{
    vcache="$(echo $1 | sed -n -e 's:.*/watch?v=\([^\&]*\).*:v-\1:p')"

    if [ -z "$vcache" ]
    then
	echo "That isn't a \"/watch?v=XXXX\" type url.  Quitting."
	exit 1
    fi

    if [ ! -f "$vcache.html" ]
    then
	# echo "wget $1 -O $vcache.html"
	wget $1 -O "$vcache.html~"
	mv $vcache.html~ $vcache.html
    fi

    gvext="$(grep -a fullscreen $vcache.html | cut -d? -f2 | cut -d\" -f1)"

    if [ -z "$gvext" ]
    then
	if [ -n "${COOKIES-}" ]
	then
	    rm $vcache.html
	    echo "Couldn't get filename. Retrying with cookies..."
	    wget --load-cookies $COOKIES $1 -O "$vcache.html~"
	    mv $vcache.html~ $vcache.html
	    gvext="$(grep -a player2.swf $vcache.html | cut -d? -f2 | cut -d\" -f1)"
	    if [ -z "$gvext" ]
	    then
		echo "Retried with cookies.  Still couldn't find video's filename in the html file."
		mv -f $vcache.html $vcache.html~
		exit 1
	    fi
	else
	    mv -f $vcache.html $vcache.html~
	    echo "Couldn't find video's filename in the html file.  Is this a flagged video?"
	    echo "Try setting COOKIES to a valid cookie file containing the login cookies and the is_adult cookie."
	    exit 1
	fi
    fi

    gvext="$GETVIDEO$gvext"

    # <meta name="title" content="Rumorsontheinternets">
    # grab the title and clean it up a bit.
    title=$(grep -a 'meta name="title"' $vcache.html | sed -n -e 's/.*content="\(.*\)".*/\1/p' | sed -e 's/[^a-z0-9A-Z]/_/g' -e 's/___*/_/g' -e 's/_$//' -e 's/^_//')

    if [ -z "$title" ]
    then
	echo "get_video: '$gvext'"
	echo "Couldn't find video's title in the html file."
	exit 1
    fi

    # echo "$1 -> $title.flv"

    if [ ! -f "$title.flv" ]
    then
	wget "$gvext" -O "$title.flv~"
	if [ -s "$title.flv~" ]
	then
	    mv "$title.flv~" "$title.flv"
	else
	    rm "$title.flv~"
	fi
    fi

    if [ -f $LOCK ]
    then
	echo "Another mplayer is running, waiting"

	while [ -f $LOCK ]
	do
	    sleep 5
	done
    fi

    if [ -z "$testflag" ]
    then
	touch $LOCK
	mplayer -fs "$title.flv"
	rm $LOCK
    fi

    echo "cd $PWD; mplayer -fs \"$title.flv\""

    # uncomment to not permanently cache the v= html file.
    # rm $vcache.html
}


# called with URL and TMPFILENAME
playmulti()
{
    if [ -z "$2" ]
    then
	echo "$1 filename parse failure"
	exit 1
    fi

    if [ ! -f "$2.html" ]
    then
	wget $1 -O "$2.html~"
	mv $2.html~ $2.html
    fi

    # grab these lines avoiding dups with "<img" directives.
    # <a href="/watch?v=DCQ7dmGuAU8&feature=PlayList&p=472DE700A3CC70A4&index=0">Openmoko 1</a>

    vurllist=$(grep '/watch?v=' $2.html | grep -v '<img '| sed -n -e 's:.*href=\"/watch?v=\([^\&\"]*\).*:\1:p')

    if [ -z "$vurllist" ]
    then
	echo "Couldn't find video's filename in the html file."
	exit 1
    fi

    for vurl in $vurllist
    do
	# echo "playone $WATCHURL$vurl";
	playone $WATCHURL$vurl
    done

    # uncomment to not permanently cache the html file.
    # rm $2.html
}

for origurl in "$@"
do

    case $origurl
    in
	http://www.youtube.com/*) ;;
	http://youtube.com/*) ;;
	--test) testflag=true ;;
	*) echo "$origurl is not a valid youtube.com url"
	continue
	;;
    esac

    case $origurl
    in
	*/watch?v=*)
	playone $origurl
	;;
	*/view_play_list?p=*)
	playmulti $origurl $(echo $origurl | sed -n -e 's/&page=/-p-/' -e 's:.*?p=\([^\&]*\).*:p-\1:p' )
	;;
	*/results?search_query=*)
	playmulti $origurl $(echo $origurl | sed -n -e 's/&page=/-p-/' -e 's:.*?search_query=\([^\&]*\).*:s-\1:p' )
	;;
	*/profile_videos?user=*)
	playmulti $origurl $(echo $origurl | sed -n -e 's/&page=/-p-/' -e 's:.*?user=\([^\&]*\).*:pv-\1:p' )
	;;
	--test)
	;;
	*) echo "$origurl is not a youtube watch, playlist or search url"
	continue
	;;
    esac
done

#
# end
#